EUROPE AND CENTRAL ASIA STUDIES CRITICAL CONNECTIONS Promoting Economic Growth and Resilience in Europe and Central Asia David Michael Gould Critical Connections Critical Connections Promoting Economic Growth and Resilience in Europe and Central Asia David Michael Gould © 2018 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 1 2 3 4 21 20 19 18 This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creative​ commons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: Gould, David Michael. 2018. Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia. Europe and Central Asia Studies. Washington, DC: World Bank. doi:10.1596/978-1-4648-1157-9. License: Creative Commons Attribution CC BY 3.0 IGO. Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. ISBN (paper): 978-1-4648-1157-9 ISBN (electronic): 978-1-4648-1158-6 DOI: 10.1596/978-1-4648-1157-9 Cover and interior design: Debra Naylor, Naylor Design. Cover images: inset: © sigdem/shutterstock; background image: © VTT Studio/shutterstock. All images used with permission. Further permission required for reuse. Library of Congress Cataloging-in-Publication Data has been requested. Contents About the Authors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii Foreword. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvii Regional Classifications Used in This Report. . . . . . . . . . . . . . . . . . . . . . . . . . xxxi Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Main Findings of Critical Connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Multidimensional Connectivity Is a Key to Europe and Central Asia’s Development and Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Europe and Central Asia Connectivity Is a Critical Source of Knowledge Transfers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Foreign-Owned and -Managed Firms Tend to Perform Better and Contribute to Local Firms’ Productivity. . . . . . . . . . . . . . . . . . . . . . . 15 Economic Migration Has Been Beneficial to Europe and Central Asia . . . . 22 Strong Infrastructure Transport Links Provide Important Support for Connectivity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 The Growth of Supply Chains Reflects Greater Connectivity and Has Facilitated Increased International Knowledge Flows . . . . . . . . 31 European and Central Asian Countries Have Moved toward More Open Policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Considerable Scope Remains for Improving Policies to Increase Connectivity in Europe and Central Asia. . . . . . . . . . . . . . . . . . . . . . . . . 38 Annex OA.  Selected Indicators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Multidimensional Connectivity: Pathways to Growth 1  and Shared Prosperity in Europe and Central Asia. . . . . . . . 47 Main Messages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Trends in Economic Connectivity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Connectivity and Income Growth. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 Trade-Offs and Resilience to Shocks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 v vi  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Annex 1A. Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Annex 1B. Network Graph Methodology. . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Annex 1C. Multiplex PageRank Centrality . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Annex 1D. Centrality Indicator. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Spotlight 1: Trends in Foreign Direct Investment in Europe and ­ Central Asia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 Reference. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 Knowledge Transfers from International Openness 2  in Trade and Investment: The European Case. . . . . . . . . . . . . 87 Main Messages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 Knowledge Creation in Europe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .88 Knowledge and Learning from Trade, Investment, and GVCs: Insights from the Economic Literature. . . . . . . . . . . . . . . . . . . . . . . . . . . 92 Knowledge Diffusion in Europe: The Two-Stage Process of Technology Transfer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Spotlight 2: Attracting Foreign Direct Investment: The Role of Deep Preferential Trade Agreements . . . . . . . . . . . . . . 111 Deep PTAs in ECA: A Snapshot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 Linking Deep Agreements with FDI: Empirical Strategy. . . . . . . . . . . . . . . 114 Linking Deep Agreements with FDI: Results. . . . . . . . . . . . . . . . . . . . . . . .114 Annex S2A. Definition of Country Groups and Methodology . . . . . . . . . . 117 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 3 Connectivity and Firms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 Main Messages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 Firm Connectivity in ECA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 Annex 3A. Coverage of Orbis Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 Annex 3B. Additional Tables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 Spotlight 3: Reaping Digital Dividends through Complementary ­Investments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 Contents ●  vii 4 Migration and Connectivity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 Main Messages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 Migration Patterns in Europe and Central Asia . . . . . . . . . . . . . . . . . . . . . 162 Migration Patterns in ECA Are Likely to Change. . . . . . . . . . . . . . . . . . . . 171 Policies Should Aim to Improve the Integration of Migrants. . . . . . . . . . . 176 Emigration Generates Net Benefits in ECA Origin Countries . . . . . . . . . . 180 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 Annex 4A. Gravity Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 Annex 4B. Additional Tables and Figures. . . . . . . . . . . . . . . . . . . . . . . . . . 189 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 5 Infrastructure Linkages: Cost, Time, and Networks. . . . . . 199 Main Messages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 Connecting Cities and Neighbors: A Vicinity View of Transport Services in ECA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 From First Neighbors to Transport Networks: Connectivity as a Policy Objective. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 Connectivity as a Collective Challenge: Centrality and Criticality . . . . . . . 216 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224 Annex 5A. Methodology and Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 Supply Chains in Europe and Central Asia: Connectivity 6  through Cross-Border Production Fragmentation. . . . . . . . 237 Main Messages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237 Factory Europe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238 Are There Only Benefits from Increased Interdependence of Countries?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249 Different Policies for GVC Upgrading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 Annex 6A. Elasticities of Value Added in Exports, Gross Exports, and Fragmentation Intensity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 Annex 6B. Interdependence of Countries . . . . . . . . . . . . . . . . . . . . . . . . . 263 Annex 6C. Regression of Backward- and Forward-Participation Indexes over a Set of Policy Variables . . . . . . . . . . . . . . . . . . . . . . . . . . 263 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 ECA Policies for Improving Connectivity. . . . . . . . . . . . . . . . .269 7  Main Messages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 MFN Tariffs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 Foreign Direct Investment Policies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 viii  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Preferential Trade Agreements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284 Bilateral Investment Agreements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288 Product Market Regulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292 EBRD Transition Indicators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299 Policy Comovements—Are Policies Consistent? . . . . . . . . . . . . . . . . . . . . 301 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 Boxes O.1 Global Value Chain Spillovers in Romania. . . . . . . . . . . . . . . . . . . . 12 O.2 Marius Stefan of Autonom Romania: Knowledge transfers through travel and studies abroad . . . . . . . . . . . . . . . . . . . . . . . . . 25 1.1 A Better Way of Measuring Network Connectivity . . . . . . . . . . . . . 66 1.2 Example of Using Connectivity Measures for Investment Decisions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 S2A.1 Methodology for the Estimation of the Impact of Deep Integration on FDI Flows. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 4.1 The Globalization of Education. . . . . . . . . . . . . . . . . . . . . . . . . . . 173 4.2 Nicolas Catena Zapata and the Malbec: Technology Transfer through Migration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 4.3 Emigration Can Improve Political Institutions in the Home Country. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 5.1 Measuring Market Access. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212 5.2 Linkages and Integration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 5.3 Centrality and Criticality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217 6.1 Global Value Chain Spillovers in Romania. . . . . . . . . . . . . . . . . . . . 250 Figures O.1 Exports of manufactured goods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 O.2 Framework and logical flow of chapters for this report. . . . . . . . . . . 6 O.3 Multidimensional connectivity combines many channels of connectivity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 O.4 Europe and Central Asia’s connectivity has grown, but there are wide variations across subregions . . . . . . . . . . . . . . . . . . . . . . . 9 O.5 Connectivity’s effects on overall and bottom-40 growth. . . . . . . . . 10 O.6 A shock originating in Germany has the largest impact on countries that gain their global connectivity through Germany . . . 11 O.7 Europe lags behind the frontier in services . . . . . . . . . . . . . . . . . . 13 O.8 How technology flows from European frontier firms (global value chain lead firms) to the remaining European firms . . 14 O.9 Productivity growth was lower in Central and Eastern Europe during the crisis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 O.10 Foreign-owned and foreign-managed firms in ECA, 2013 . . . . . . . 16 Contents ●  ix O.11 Large firms are more likely to be foreign owned in ECA. . . . . . . . . 17 O.12 There is no clear relationship between a firm’s age and the likelihood of its being foreign owned. . . . . . . . . . . . . . . . . . . . . . . 17 O.13 Foreign-owned and -managed firms perform better than local firms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 O.14 Foreign affiliates tend to have better management practices than local firms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 O.15 The positive spillovers of well-managed foreign firms seem weaker for small and young firms. . . . . . . . . . . . . . . . . . . . . . . . . . 21 O.16 Foreign firms’ employment decisions are less procyclical than those of their domestic peers. . . . . . . . . . . . . . . . . . . . . . . . . 22 O.17 ECA migration is driven by geography, language, historical ties, and past migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 O.18 Transport connectivity (cost and time) between and within ECA countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 O.19 Cost-driven criticality in container network for Europe and Central Asia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 O.20 Participation in global value chains is correlated with higher labor productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 O.21 Among the transition countries, greater production fragmentation is associated with a more rapid increase in the flows of value added in exports . . . . . . . . . . . . . . . . . . . . . . . . 34 O.22 ECA ranks among the top regions in regard to the number of trade agreements and investment treaties . . . . . . . . . . . . . . . . . 36 1.1 Trends in intraregional trade in ECA. . . . . . . . . . . . . . . . . . . . . . . . 52 1.2 Exports of manufactured goods . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 1.3 Foreign direct investment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 1.4 Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 1.5 Airline connectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 1.6 Internet and communication technologies . . . . . . . . . . . . . . . . . . . 58 1.7 Portfolio financial flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 B1.1.1 Examples of network connectivity and the modified PageRank . . . 66 1.8 Multidimensional connectivity network. . . . . . . . . . . . . . . . . . . . . . 67 1.9 Multidimensional network connectivity. . . . . . . . . . . . . . . . . . . . . . 68 1.10 Europe and Central Asia’s connectivity has grown, but there are wide variations across subregions. . . . . . . . . . . . . . . 69 B1.2.1 Kazakhstan’s connectivity ranking change. . . . . . . . . . . . . . . . . . . . 71 1.11 Simulated impact on individual countries’ connectivity measure (modified PageRank) of a 10 percent decline in trade, foreign direct investment, and migration in Germany, the Russian Federation, and the United States . . . . . . . . . . . . . . . . . . . . . . . . . 74 S1.1 The relevance of ECA as both a destination and an origin of FDI has fallen since 2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 S1.2 World FDI inflows into ECA are relatively more diversified by ECA destination than ECA FDI outflows to the world. . . . . . . . . . . 83 S1.3 ECA’s share of world FDI inflows is greater than its share of world GDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 x  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia S1.4 FDI attraction patterns increase with development levels but vary by country. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 S1.5 Services and manufacturing dominate FDI inflow patterns across ECA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 S1.6 Germany and the United States dominate EU investment; France and China lead elsewhere . . . . . . . . . . . . . . . . . . . . . . . . . 86 2.1 Differences between frontier and laggard firms vary across sectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 2.2 Europe lags behind the frontier in services . . . . . . . . . . . . . . . . . . 91 2.3 Technology transfer tends to follow a typical sequence . . . . . . . . . 98 2.4 How technology flows from European frontier firms (global value chain lead firms) to the remaining European firms . . . . . . . 100 2.5 Firms’ international connectivity and technology transfer follow three stages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 2.6 GVC participation is particularly high in Central and Eastern Europe. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 2.7 Import intensity varies over time for Central and Eastern European EU countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 2.8 Productivity growth was lower in Central and Eastern Europe during the crisis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 S2.1 The European Union shows the greatest depth of agreements among ECA country groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 S2.2 Sectoral and customs-related provisions are the most frequent WTO provisions in ECA PTAs . . . . . . . . . . . . . . . . . . . . . 113 S2.3 Among WTO+ provisions, Competition Policy, Movement of Capital, and Intellectual Property Rights are the most frequent WTO+ provisions included in ECA PTAs . . . . . . . . . . . . 114 S2.4 The impact of deep agreements on FDI. . . . . . . . . . . . . . . . . . . . 115 S2.5 Deep agreements are more helpful in facilitating FDI in culturally distant destinations for manufacturing, while the opposite is true for services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 3.1 The presence of foreign firms varies substantially across ECA countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 3.2 Large firms are more likely to be foreign owned in ECA. . . . . . . . 128 3.3 There is no clear relationship between a firm’s age and the likelihood of its being foreign owned. . . . . . . . . . . . . . . . . . . . . . 129 3.4 Foreign affiliates owned by tax haven countries are small and medium sized. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 3.5 The Average Management Index is strongly correlated with the WMS Management Index . . . . . . . . . . . . . . . . . . . . . . . . 136 3.6 Foreign affiliates tend to have better management practices than local firms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 3.7 More foreign affiliates are owned by countries with better management indexes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 3.8 Foreign firms’ employment decisions are less procyclical than those of their domestic peers. . . . . . . . . . . . . . . . . . . . . . . . 139 3.9 Foreign-owned and -managed firms perform better than local firms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 Contents ●  xi 3.10 The positive spillovers of well-managed foreign firms seem weaker for small and young firms. . . . . . . . . . . . . . . . . . . . . . . . . 145 S3.1 Firms’ online sales rise with more efficient logistics and payment systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 S3.2 Internet use by firms is associated with the intensity of local competition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 4.1 Top destinations of emigrants and share of total who have completed tertiary education, 2010 . . . . . . . . . . . . . . . . . . . . . . . 165 4.2 Age composition of native-born and immigrant populations, 2010. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 4.3 Percentage of women among emigrants, 2010. . . . . . . . . . . . . . . 169 4.4 The share of high-skilled immigrants to high-income ECA OECD member countries increased between 2000 and 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 4.5 The share of temporary employment increased in Europe and Central Asia between 2002 and 2016 . . . . . . . . . . . . . . . . . . 172 4.6 The share of temporary migration is positively related to the share of temporary employment. . . . . . . . . . . . . . . . . . . . . . . 173 B4.1.1 Most top ECA destinations attracted more international tertiary students in 2014 than in 2004. . . . . . . . . . . . . . . . . . . . . . 174 B4.1.2 Most source countries of international tertiary students are in ECA . . . 174 B4.1.3 ECA hosted half of the world’s tertiary students in 2014 . . . . . . . 175 B4.1.4 Top 10 corridors of international tertiary students with ECA hosts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 4.7 Unemployment rates are higher for foreign-born than for native-born workers in most countries in Europe and Central Asia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 4.8 Tertiary education rates in the European Union are about the same among native- and foreign-born working-age populations (ages 25–54) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 4.9 Migrant Integration Policy Index overall, labor market integration, and political participation scores in ECA and selected countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 4.10 Many countries in Europe and Central Asia depend on remittances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 4.11 High-income ECA countries are much more permissive toward international mobility than the United States . . . . . . . . . . . . . . . . 186 4B.1 Share of tertiary educated among emigrants from EU15+ countries, 2000 and 2010 . . . . . . . . . . . . . . . . . . . . . . . . . 193 4B.2 Share of tertiary educated among emigrants from Central and Eastern European countries, 2000 and 2010 . . . . . . . . . . . . . 193 4B.3 Share of tertiary educated among emigrants in Central Asia and the South Caucasus, 2000 and 2010 . . . . . . . . . . . . . . . . . . . 194 5.1 Domestic connectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 5.2 Neighbor connectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204 5.3 Nonlinear impact of connecting with neighbors of neighbors . . . 206 5.4 Cost-based connectivity of the ECA road transport network, by region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 xii  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia 5.5 Cost and time connectivity in the ECA network . . . . . . . . . . . . . . 209 5.6 Container cost connectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 5.7 Realized potential of connectivity to ECA markets relative to advanced Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 5.8 Realized potential of connectivity to markets by country, 2016. . . . . 214 5.9 Linkages and overall integration . . . . . . . . . . . . . . . . . . . . . . . . . 216 5.10 Centrality in the ECA network for container transport. . . . . . . . . . 218 5.11 Cost-driven criticality in the ECA network for container transport. . . . 219 5.12 Romania: Impact of a 30 percent decrease in container transport costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 5.13 Kazakhstan: Costs and potential indexes before and after decrease in costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 5A.1 Cost of and time required for freight transport services . . . . . . . . 231 6.1 Three clusters of countries emerge: “Factory Europe” around Germany, “Factory North America” around the United States, and “Factory Asia” around China . . . . . . . . . . . . . . . . . . . . . . . . . 239 6.2 Smaller European countries, like the Czech Republic, are dominated by trade with Germany, while Germany is the headquarters of “Factory Europe,” which trades more globally . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240 6.3 Higher production fragmentation due to supply chains is associated with more rapid growth in value added in exports over 2000–11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242 6.4 Among the transition EU13 countries, greater production fragmentation is associated with a more rapid increase in the flows of value added in exports . . . . . . . . . . . . . . . . . . . . . 243 6.5 EU countries buy more foreign value added (backward linkages) than they sell to third countries (forward linkages) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244 6.6 Participation in supply chains is heterogeneous among ECA countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245 6.7 Supply chain trade with close partners remains high . . . . . . . . . . 246 6.8 European FDI inflows into the EU13 countries increased dramatically after the 2003 entry of those countries into the European Union . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249 6.9 Participation in global value chains is correlated with higher labor productivity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251 6.10 Participation in global value chains is correlated with higher domestic value added at the sector level. . . . . . . . . . . . . . . . . . . 252 6.11 Sectors from advanced EU15, transition EU13, and non-EU countries are interdependent, but the most central sectors are from EU15 countries . . . . . . . . . . . . . . . . . . . . . . . . . . 253 6.12 Sectors that are more integrated in a production trade network move more together. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255 6.13 Imports of intermediate goods are more volatile than imports of final goods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 6.14 Many EU13 countries have high backward participation but low forward participation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257 Contents ●  xiii 6.15 ECA Non-EU countries mainly import final goods and export little . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258 6.16 Heterogeneous participation in trade of domestic and foreign firms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 6.17 Many ECA countries can still reduce trade barriers and improve logistics and the business environment to increase their supply chain participation. . . . . . . . . . . . . . . . . . . . . . . . . . . 260 6A.1 Elasticities of exports of value added and gross exports, and of the ratio of gross exports to value added, for the EU13, EU15, and all countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 7.1 Tariffs across global regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278 7.2 Tariffs in the main ECA subregions . . . . . . . . . . . . . . . . . . . . . . . . 278 7.3 Tariffs in ECA regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 7.4 Tariffs in ECA countries by income group. . . . . . . . . . . . . . . . . . . 279 7.5 FDIRRI in ECA and selected countries. . . . . . . . . . . . . . . . . . . . . . 280 7.6 FDIRRI in ECA subregion I. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281 7.7 FDIRRI in ECA subregion II. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281 7.8 FDIRRI in ECA countries by income group . . . . . . . . . . . . . . . . . . 282 7.9 FDIRRI for communications and transport in ECA and selected countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 7.10 FDIRRI for communications and transport in ECA subregions. . . . 283 7.11 FDIRRI for transport in ECA countries by income group. . . . . . . . 283 7.12 FDIRRI for communications in ECA countries by income group. . . . 284 7.13 Preferential trade agreements across global regions. . . . . . . . . . . 285 7.14 Preferential trade agreements across global regions: Intraregional integration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286 7.15 Preferential trade agreements across global regions: Extraregional integration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286 7.16 Preferential trade agreements in ECA subregions. . . . . . . . . . . . . 287 7.17 Preferential trade agreements in ECA countries by income group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287 7.18 Preferential trade agreements in ECA subregions: Intra- versus extraregional integration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288 7.19 Preferential trade agreements in ECA countries by income group: Intra- versus extraregional integration. . . . . . . . . . 288 7.20 Bilateral investment treaties across global regions. . . . . . . . . . . . 289 7.21 Bilateral investment treaties across global regions: Intraregional integration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290 7.22 Bilateral investment treaties across global regions: Extraregional integration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290 7.23 Bilateral investment treaties in ECA subregions . . . . . . . . . . . . . . 291 7.24 Bilateral investment treaties in ECA countries by income group. . . . 291 7.25 Bilateral investment treaties in ECA subregions: Intra- versus extraregional integration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292 7.26 Bilateral trade agreements in ECA countries by income group: Intra- versus extraregional integration. . . . . . . . . . . . . . . . 292 7.27 Horizontal product market regulation in ECA and selected countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294 xiv  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia 7.28 Horizontal product market regulation in ECA subregions. . . . . . . 294 7.29 Horizontal product market regulation in ECA countries by income group. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295 7.30 Product market regulation: Aggregate energy, transport, and communications regulations in ECA and selected countries. . . . . 295 7.31 Product market regulation: Aggregate energy, transport, and communications regulations in ECA subregions. . . . . . . . . . . . . . 296 7.32 Product market regulation: Regulations regarding airlines in ECA subregions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296 7.33 Product market regulation: Regulations involving railways in ECA subregions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 7.34 Product market regulation: Regulations regarding telecommunications in ECA subregions . . . . . . . . . . . . . . . . . . . . 297 7.35 Product market regulation: Aggregate energy, transport, and communications regulations in ECA countries by income group. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298 7.36 Technical barriers to trade and sanitary and phytosanitary notifications in ECA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299 7.37 Technical barriers to trade and sanitary and phytosanitary notifications in ECA across income regions. . . . . . . . . . . . . . . . . . 299 7.38 Aggregate EBRD Transition Indicators for infrastructure in ECA subregions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 7.39 Aggregate EBRD Transition Indicators for infrastructure in ECA countries by income group. . . . . . . . . . . . . . . . . . . . . . . . 301 7.40 Global evolution of selected policy comovements . . . . . . . . . . . . 303 7.41 Evolution of selected policy comovements across income groups. . . . 304 7.42 Tariff–Mobility Barriers Index comovements across countries imposing mobility restrictions. . . . . . . . . . . . . . . . . . . . . . . . . . . . 305 7.43 Tariff–bilateral investment treaty comovements across ECA subregions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305 7.44 Tariff–bilateral investment treaty comovements across global regions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306 7.45 Tariff–bilateral investment treaty comovements across ECA macrosubregions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306 Maps O.1 Emigration and immigration shares have seen the highest increase in Europe and Central Asia . . . . . . . . . . . . . . . . . . . . . . . . 24 S2.1 The European Union and North America show the deepest forms of integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 4.1 Emigration shares have seen the highest increase in ECA. . . . . . .163 4.2 Immigration shares are significant in many ECA countries . . . . . . 164 5A.1 Typologies used in the analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 228 6.1 Which countries are the most central in the ECA production network?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255 Contents ●  xv Tables O.1 Multidimensional Connectivity Varies by ECA Subregion, with the Highest Connectivity in the Western Part of the Region and the Lowest in the Eastern Part. . . . . . . . . . . . . . . . . . . . 8 O.2 EU and Non-EU Countries Most Affected by Brexit. . . . . . . . . . . . .11 O.3 Most Foreign Firms in ECA Are Owned by German and US Firms . . . 18 OA.1 Multidimensional Connectivity Indexes (on an Absolute Basis). . . . 42 OA.2 Multidimensional Connectivity Indexes (on a Per Capita Basis). . . . 44 1.1 Connectivity Effects on Overall Income Growth . . . . . . . . . . . . . . . 63 1.2 Connectivity Effects on Bottom-40 Income Growth . . . . . . . . . . . . 64 1.3 Correlation between Connectivity Layers Is High, Except in the Case of Portfolio Financial Flows . . . . . . . . . . . . . . . . . . . . . . . 65 1.4 Multidimensional Connectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 1.5 Multidimensional Connectivity Varies by ECA Subregion, with the Highest Connectivity in the Western Part of the Region and the Lowest in the Eastern Part. . . . . . . . . . . . . . . . . . . 69 B1.2.1 Potential Markets and Their Connectivity Indexes. . . . . . . . . . . . . . 70 B1.2.2 Kazakhstan’s New Multidimensional Connectivity Index after Investing $100 Million Each in Various Markets. . . . . . . . . . . . . . . . 70 1.6 ECA Countries Most and Least Affected by Brexit . . . . . . . . . . . . . 76 1.7 Transmission of Trade, Migration, and FDI Shocks to ECA Subregions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 1A.1 Long-Term Growth Determinants . . . . . . . . . . . . . . . . . . . . . . . . . . 77 1A.2 Network Country Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 S1.1 ECA Is the Main Investor in ECA. . . . . . . . . . . . . . . . . . . . . . . . . . . 85 2.1 Europe Specializes in Several Sectors with Below-Average R&D Intensity and Growth. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 S2A.1 Country Groups and Subgroups. . . . . . . . . . . . . . . . . . . . . . . . . . 117 S2A.2 Percentage of ECA PTAs Including WTO Provisions, by Subgroup. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 S2A.3 WTO and WTO+ Policy Areas in PTAs . . . . . . . . . . . . . . . . . . . . . 119 S2A.4 Percentage of ECA PTAs Including WTO+ Provisions, by Subgroup. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 S2A.5 Countries and Economies Included in the Estimations . . . . . . . . . 120 S2A.6 Regression Results: Deep Agreements and Foreign Direct Investment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 S2A.7 Foreign Direct Investment and Depth: Interactions with ECA. . . . 121 S2A.8 Foreign Direct Investment and Depth, by Group of Provisions . . . . . 122 S2A.9 Foreign Direct Investment and Depth, by Technology Level (OECD Rev. 3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 S2A.10 Foreign Direct Investment, Depth, and Distance. . . . . . . . . . . . . .122 S2A.11 Foreign Direct Investment and Depth: Triple Interactions. . . . . . . 123 3.1 Most Foreign-Owned Firms in ECA Are Owned by Germany and the United States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 3.2 Determinants of Foreign Ownership and Management . . . . . . . . 131 3.3 Better-Managed Foreign Affiliates Perform Better . . . . . . . . . . . . 141 xvi  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia 3.4 Foreign Firms Locate in Regions and Sectors with Larger Local Firms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 3.5 Domestic Firms in ECA Grow Faster in Regions with Well-Managed Foreign Firms . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 3B.1 Foreign-Owned and Foreign-Managed Firms Perform Better Than Local Firms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 3B.2 Firms with Better Foreign Managers Perform Better. . . . . . . . . . . 148 3B.3 Spillover Effects of Foreign-Owned Firms on Domestic Firms. . . . 149 3B.4 Firm Growth over the Business Cycle . . . . . . . . . . . . . . . . . . . . . . 150 4.1 The Availability of Programs Designed to Integrate Migrants Varied in ECA, 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 4.2 Nearly All EU13 Countries Have Developed Policies to Encourage the Return of Their Citizens. . . . . . . . . . . . . . . . . . . . . 182 B4.3.1 Impact of Emigration on Institutions of Origin Countries, by Type of Destination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 4B.1 Gravity Regression Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 4B.2 Emigration in ECA Countries (excluding EU15+), 2000. . . . . . . . . 190 4B.3 Emigration in ECA Countries (excluding EU15+), 2010 . . . . . . . . 191 4B.4 Immigration in ECA Countries (excluding EU15+), 2000. . . . . . . . 191 4B.5 Immigration in ECA Countries (excluding EU15+), 2010. . . . . . . . 192 4B.6 Economies Included in Figure B4.1.3 . . . . . . . . . . . . . . . . . . . . . . 193 5.1 Changes in Centrality Due to a Decrease in Container Transport Costs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 5.2 Segments Affected by 33 Percent Cost Reduction in Container Transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 6.1 The Importance of Imported Value Added in Exports (Backward Linkages) and Exported Value Added in Third Countries’ Exports (Forward Linkages) Differs by Region . . . . . . . 244 6.2 What Sectors or Countries Are Expected to Have the Largest Impact on the Rest of the ECA Economies When They Face Either a Positive or a Negative Shock? . . . . . . . . . . . . . . . . . . . . . 254 6B.1 Sectors That Are More Integrated in the Production Network Are More Correlated. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 6C.1 Variables for Global Value Chain Participation and Forward Linkages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 7.1 Policy Measures and Indicators. . . . . . . . . . . . . . . . . . . . . . . . . . . 276 7.2 Comovements of Connectivity Policies. . . . . . . . . . . . . . . . . . . . . 302 About the Authors David M. Gould (World Bank) is currently Lead Economist in the World Bank’s Europe and Central Asia Region and the lead author of the ECA Critical Connections flagship. He is the author of several books and peer-reviewed journal articles on international trade and finance, migration, and economic policy. Currently, he is leading Europe and Central Asia regional studies on the development impact of disruptive technologies. During his 15 years at the World Bank, he has led teams to deliver country development strategies and analytical and lending operations in Europe, Latin America, and South Asia. Prior to joining the World Bank, he served as the Director of Global Economic Analysis at the Institute of International Finance and as Senior Economist and Policy Advisor at the US Federal Reserve. He has held visiting research positions at the Central Banks of Mexico and Chile. He holds a PhD in International Economics (with honors) from the University of California at Los Angeles and is a Chartered Financial Analyst charter holder. Megersa Abate (World Bank) is a Transport Economist in the World Bank’s Transport and Digital Development Global Practice. He has extensive expertise and interest in various topics, including freight demand modeling, air transport regulation, and transport connectivity. His research has been published in leading transportation economics journals. Before joining the Bank in 2016, he worked as a researcher at VTI, the Swedish National Road and Transport Research Institute, and at the VU University of Amsterdam. Earlier in his career, he worked at the Ethiopian Civil Aviation Authority as an air transport expert. He received his PhD in Transport Economics from the Technical University of Denmark in 2013; during his PhD studies, he was also a visiting student in the Institute of Transport Studies at the University of Leeds. Erhan Artuc (World Bank) is a Senior Economist in the World Bank’s Development Research Group. Prior to joining the World Bank in 2011, he was a faculty member at Koç University in Istanbul, Turkey. His most recent research focuses on interna- tional trade and migration policies and their effects on labor markets and development. His work has been published in leading academic and policy journals such as the Journal of International Economics, Economic Journal, and American Economic Review. He received his undergraduate degree from Bilkent University and a PhD in Economics from the University of Virginia. Omar Bamieh (University of Vienna) is an Assistant Professor of Economics at the University of Vienna. His research focuses on labor market institutions and the xvii xviii  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia interaction between legal frameworks and labor markets. He holds an MS in Economics and Social Sciences from Bocconi University in Milan and a PhD in Economics from the European University Institute in Florence, where he also worked as a Research Fellow in Global Economics at the Robert Schuman Centre for Advanced Studies. Cecilia Briceno-Garmendia (World Bank) is a Lead Economist in the World Bank’s Transport and ICT Global Practice, where she leads the economic research agenda in logistics and transport in the Latin America region and globally advises teams and governments in strategic issues pertaining to prioritization and plan- ning of infrastructure investments, including aspects related to spending effi- ciency, green trucking and trucking sector performance, multimodal development corridors, and adaptation of transport networks to climate change. Previously, she led the economic team in the Office of the Director for Sustainable Development of the Latin America region, where she provided leadership for the analytical agenda on infrastructure, urban, and disaster risk management and climate change adaptation issues. She has worked extensively on issues pertaining to connectivity, logistics, and port performance and co-led the pathbreaking Africa Infrastructure Country Diagnostic. She has worked on projects and research in more than 70 countries. Before joining the World Bank, she worked in software engineering and the design of information and organizational systems for both private and public sector enterprises in República Bolivariana de Venezuela. She has an MBA from the Instituto de Estudios Superiores en Administración in Caracas, República Bolivariana de Venezuela, and a doctorate in Economics from Georgetown University. Matteo Fiorini (European University Institute) is a Research Fellow in Global Economics at the Robert Schuman Centre for Advanced Studies of the European University Institute in Florence, Italy. His research focuses on international trade, trade policy, migration, and development. Prior to joining the Schuman Centre, he worked as a researcher at the Institute’s Migration Policy Centre, the World Trade Organization, and Bocconi University. He holds an MS in Economics and Social Sciences from Bocconi University in Milan and a PhD in Economics from the European University Institute in Florence. Bernard Hoekman (European University Institute) is a Professor at the Robert Schuman Centre for Advanced Studies of the European University Institute. He is also a Research Fellow at the Centre for Economic Policy Research, a member of the World Economic Forum Council on Trade and Investment, and a Senior Fellow at the Centre for International Governance Innovation. His research focuses on trade and development, economic integration, and the multilateral trading system. Dror Y. Kenett (World Bank) is a multidisciplinary financial economist and an expert on financial networks, financial stability, and systemic risk. He is a consultant to the World Bank, Adjunct Professor at Johns Hopkins University, a research asso- ciate at the London School of Economics Systemic Risk Centre, and a visiting researcher at Boston University and at the Israel Securities Authority. He has also About the Authors ●  xix held a researcher position in the US Department of the Treasury’s Office of Financial Research. He applies his scientific background to financial stability questions, focusing on network-based models, market structure, financial contagion and spill- overs, and correlation-based models. He has extensive policy experience and has contributed to the Office of Financial Research Financial Stability Report and par- ticipated in the development of the Office’s monitoring tools. He has published more than 40 papers in financial, physics, and engineering journals, including the Journal of Banking and Finance, Journal of Risk and Financial Management, Quantitative Finance, Nature Physics, and Scientific Reports. He has a PhD in Physics from Tel Aviv University in Israel. Mathilde Lebrand (World Bank) is an Economist in the World Bank’s Transport Global Practice. Currently she is working on the Belt and Road Initiative, eco- nomic corridor development, and connectivity. Previously she worked for the Europe and Central Asia Chief Economist office and contributed to several upcoming regional studies. Her research focuses on economic geography, international trade and global value chains, networks, and political economy. She has taught at the University of Montreal and has worked at the World Trade Organization in Geneva. She is a Research Fellow at the Center for Economic Studies ifo Institute (CESifo). She holds a PhD in economics from the European University Institute. Paloma López-Garcia (European Central Bank) has been a Senior Economist in the Directorate-General—Economics at the European Central Bank since 2015. Before that she was Coordinator of the Competitiveness Research Network (CompNet) in the Directorate-General—Research, and she has also worked at the Instituto de Empresa Business School and in the Research and Economics Department of the Central Bank of Spain. She has published articles in the European Economic Review, Small Business Economics, and Economics of Innovation and New Technology, among other peer-reviewed journals. Her research topics are microanalysis of productivity and employment growth, innova- tion, and trade and competitiveness. She earned her PhD at the London School of Economics in 2003. Çag˘lar Özden (World Bank) is a Lead Economist in the World Bank’s Research Department. His research explores the nexus of globalization of product and labor markets, government policies, and economic development. His current research projects explore the determinants and patterns of global labor mobility; impacts of migrants on destination labor market outcomes; linkages between migration, trade, and foreign direct investment flows; medical brain drain; and linkages between aging and global economic integration. He has edited three books and published numerous papers in leading academic journals such as the American Economic Review and the Economic Journal. He is a Fellow of the Institute of Labor Economics (IZA) and of the Centre for Research and Analysis of Migration (CreAM) and serves on the advisory board of the Economic Research Forum. He received his undergraduate degrees in Economics and Industrial Engineering from Cornell University and his PhD in Economics from Stanford University. xx  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Georgi Panterov (World Bank) is a Research Analyst in the World Bank’s Office of the Chief Economist for Europe and Central Asia. His research interests are focused on machine learning, econometrics, blockchain, and cryptocurrencies. During his time at the World Bank, he has contributed to the Golden Aging flag- ship report, the Critical Connections flagship report, and the Europe and Central Asia economic update reports. Before joining the World Bank, he worked at Google, the US Department of Agriculture, and American University. He is cur- rently a PhD candidate in Economics at American University in Washington, DC. Nadia Rocha (World Bank) is a Senior Economist in the World Bank’s Macroeconomics, Trade and Investment Global Practice. Prior to joining the Bank in 2016, she worked for five years in the World Trade Organization’s Economic Research and Statistics Division. She was seconded to the Colombian Ministry of Trade to serve as a Senior Advisor on Trade during 2015. Her current work focuses on regionalism, trade costs, global value chains, and trade and gender. She holds a BA in economics from Bocconi University in Milan, an MA in Economics from Pompeu Fabra University of Barcelona, and a PhD in International Economics from the Graduate Institute, Geneva. Daria Taglioni (World Bank) is the Principal Economist for the Europe and Central Asia and East Asia and Pacific Regions in the Economics and Private Sector Development Vice Presidency of the World Bank Group’s International Finance Corporation. Prior to joining the World Bank Group, Daria worked at the European Central Bank and at the Organisation for Economic Co-operation and Development. Her research focuses on trade and competitiveness. She has published articles in the American Economic Review and Journal of International Economics, among other peer-reviewed journals. She holds a PhD in International Economics from the Graduate Institute of Geneva. Shawn Tan (World Bank) is an Economist in the World Bank’s Finance, Competitiveness and Innovation Global Practice and is currently working on pri- vate sector development and trade issues in the countries of Eastern Europe and the Western Balkans. He has worked on reports such as the World Development Report 2016: Digital Dividends, Reaping Digital Dividends: Leveraging the Internet for Development in Europe and Central Asia, and the high-growth entreneurship report and has written papers on international trade, firm productivity, and high- growth firms. Before joining the World Bank, he worked at the Singapore Economic Development Board, where he was a negotiator for Singapore’s free trade agree- ments and bilateral investment treaties and worked on trade facilitation issues for multinational companies in Singapore. His research interests are broadly in inter- national trade, economic geography, and firm productivity and performance. He holds a PhD in Economics from the University of Melbourne. Gonzalo Varela (World Bank) is a Senior Economist in the Global Trade and Regional Integration Unit of the World Bank’s Macroeconomics, Trade and Investment Global Practice. Prior to joining the World Bank, he was a Lecturer at the University of Sussex and at Uruguay’s Ministry of Industry, Energy, and Mining. About the Authors ●  xxi His work agenda focuses on global integration and economic performance and on the analysis of trade policy and competitiveness. He holds a BSc in Economics from the Universidad de la República in Uruguay and both an MA in International Economics and a PhD in Economics from the University of Sussex. Hernan Winkler (World Bank) is a Senior Economist in the World Bank’s Jobs Group. He specializes in applied microeconomics, with a particular focus on issues related to labor markets, technological change, and the sources and conse- quences of poverty and inequality. His research has been published in peer- reviewed economics journals, including the Review of Economics and Statistics and the Journal of Development Economics. He was a lead author of the World Bank regional report Reaping Digital Dividends: Leveraging the Internet for Development in Europe and Central Asia. He has been part of the core teams of several regional reports, including Diversified Development, Golden Aging, and Risk and Returns. He was previously a researcher at the Center for Distributive, Labor and Social Studies (CEDLAS) at the National University of La Plata in Argentina, where he conducted research on poverty and distributional issues affecting countries in Latin America and the Caribbean. He holds a master’s degree in Economics from the National University of La Plata and a PhD in Economics from the University of California at Los Angeles. Thea Yde-Jensen (World Bank) is a Researcher in the World Bank Group’s Poverty and Equity Global Practice, where she conducts research on issues related to liveli- hoods, labor market outcomes, and displacement. Her expertise and research interests particularly focus on examining the interlinkages of labor markets and inequality and poverty. Previously she worked as a Researcher in the Bank’s Office of the Chief Economist for Europe and Central Asia, focusing on issues related to employment and firms’ access to finance and international networks. Prior to join- ing the World Bank, she worked in the International Monetary Fund’s Statistics Department and in the Department of Economics at Copenhagen Business School. She has a BS and an MS in Economics from the University of Copenhagen. Foreword In mid-2014 when Critical Connections was first contemplated, the Europe and Central Asia (ECA) region was still emerging from the global financial crisis, growth was uncertain and tepid, and policy makers were largely focused on mitigating further financial and macroeconomic risks from ongoing weakness in the banking sector and large fiscal deficits. Appropriately, the policy discourse was largely targeted to shoring up near-term challenges, rather than to assembling the building blocks that would provide the foundation for restoring the promise of long-term resilient growth. Critical Connections was born out of the desire to help policy makers focus their attention on their long-term goals of regional and global integration to capture the benefits of connectivity, from which ECA countries had advanced so far during the early years of market expansion in the 1990s and early 2000s. What started simply as an exploration into policies to capture the gains of specialization and knowledge transfers has taken on much greater meaning in recent times. The trend toward regional and global integration is under serious threat as many voters, particularly in high-income countries, see nationalism and protection as a remedy to greater economic uncertainty. But as former UK Prime Minister Gordon Brown noted in a 2015 speech, “the problems that give rise to nationalism can’t be solved by nationalism and in an interdependent world the problems that give rise to isolationism and protectionism cannot be solved by isolationism and protectionism.” While Critical Connections does not provide answers to assuage all the concerns about our changing global economy, it does provide an invaluable insight into understanding—at the firm and country level—the interdependence of our world and how it has historically operated, and currently operates, to advance economic growth and shared prosperity. A key insight of this report is that ECA’s international connectivity through trade, foreign direct investment, migration, telecommunications, transportation, and other avenues facilitates the transfers of knowledge and technology that are critical to long-term growth and shared prosperity. These connections complement one another because of the tacit (learning by doing), rather than explicit (contained in books or blueprints), nature of knowledge transfers. Migration, for example, enhances knowledge spillovers through trade and foreign investment by migrants transferring information on foreign markets and supporting connections to them. Similarly, the internet and efficient transport links are both necessary for successful e-commerce. xxiii xxiv  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Moreover, the depth of ECA’s connections and the geographic composition of the connections both matter. Knowledge transfers are greater from countries that themselves have strong links to third countries. These transfers also emerge from linkages between firms in global value chains as well as foreign ownership and management practices that generate local spillovers. While these connections are important for prosperity, however, one should not be naive about their impact. Despite its overall benefits, increased connectivity exposes ECA countries to shocks, particularly those emanating from countries at the center of international economic transactions, which may have contributed to economic insecurity. However, by providing alternative sources of external demand and financing, a broad range of connections can reduce those risks and help countries cope with both domestic and external shocks. The European side of the ECA region is an ideal laboratory for observing the role of multidimensional connectivity in action. Regional supply chains are strong, and links between countries across the various forms of connectivity allow observations on how connectivity opens doors for the knowledge transfers that support resilient growth. Nonetheless, in many European countries, progress on deepening connectivity has stalled since the global financial crisis, and productivity growth attributed to connectivity has suffered. In Central Asia, despite recent moves toward building greater interconnectedness, the region remains among the least connected globally. Because of both its geographical position and its limited infrastructure, many Central Asian countries are only weakly connected to other ECA countries and the global economy. The vast distances between Central Asia, Europe, and East Asia will remain an obstacle to connectivity. However, infrastructure investments and policies to improve integration through freer trade, infrastructure, and investment policies are likely to provide large growth benefits in Central Asia. Many ECA countries can be proud of what they have achieved in building greater connectivity and advancing development during the past 25 years. But because the economic benefits of connectivity through knowledge and technology transfers are not obvious, while the challenge of economic uncertainty is, building the case for deepening connections requires solid and clear evidence. By recognizing the challenges as well as making explicit the potential opportunities of greater connectivity through various channels, Critical Connections can assist ECA’s policy makers in building the foundations for deepening important connections in the coming decades. Cyril Muller Vice President Europe and Central Asia Region World Bank Group Acknowledgments This report was written by a team led by David Michael Gould, Lead Economist in the World Bank’s Office of the Chief Economist for Europe and Central Asia. The core team members were Erhan Artuc, Cecilia Briceno-Garmendia, Bernard Hoekman (European University Institute), Mathilde Lebrand, Çag ˘lar Özden, Georgi Panterov, Nadia Rocha, William Shaw, Daria Taglioni, Shawn Tan, Ekaterina Ushakova, Gonzalo Varela, Hernan Winkler, and Thea Yde-Jensen. The work was carried out under the overall supervision and guidance of Hans Timmer, Chief Economist for the Europe and Central Asia Region. The Macroeconomics, Trade, and Investment team benefited from the guid- ance of Jose Guilherme Reis and comments and discussions with Jean Francois Arvis, Cordula Rastogi, and Daniel Saslavsky. The Transport team appreciates the general guidance of Juan Gaviria and comments and discussions with Baher El-Hifnawi, Carolina Monsalve, and the Global Practice Transport team at large. Many thanks go to Rashmi Shankar for her extremely helpful contributions during the early stages of the report and Moritz Meyer for his generous time and discus- sions on applied network analysis. Peer reviewers Luis-Felipe Lopez-Calva, Caroline Freund, Bill Maloney, Aaditya Mattoo, and Russell Hillberry provided very helpful advice and comments on the report. This report would not have been timely or relevant without the insights and inputs of European Central Bank staff members, who provided data, analysis, and technical support for the work on “Knowledge Transfers from International Openness in Trade and Investment: The European Case,” and staff members of the International Civil Aviation Organization, particularly Dr. Ananthanarayan Sainarayan and colleagues, who generously provided data on origin-to-­destination airline connectivity. Private and public sector organizations and experts in Romania and Moldova also provided significant inputs and insights; in Romania: Startnet, Softelligence, Fondul Proprietatea, the Foreign Investors Council, Oracle România, the Ministry of Transport, the Ministry of External Commerce, the Ministry of Communications and Information Society, UPC Romania, Dr. Adrian Curaj (UNESCO), AmCham, H. Essers, Robin Martens (International Project Management), Kuijken Logistics Group, and Autonom Rent-A-Car; in Moldova: Ionel, Andragrup SRL, the European Business Association, GIZ (German Cooperation for International Development), the Ministry of Transport and Roads Infrastructure, APIUS, Moldova Investment and Export Promotion Organization, the Ministry of Finance, Star Legal Consulting, AmCham, Danube Logistics, and the Bureau for Relations with the Diaspora. xxv xxvi  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Many people participated in the writing of the report. The main authors and contributors were • Overview: David Michael Gould • Chapter 1: David Michael Gould, Dror Kenett, and Georgi Panterov (with con- tributions from Angel Bogoev [American University], Michael Danziger [Center for Complex Network Research and Department of Physics, Northeastern University], Dobrina Gogova, Xin Yuan [Boston University], and Tlek Zeinullayev [Harvard University]) • Chapter 2: Paloma López-Garcia (European Central Bank) and Daria Taglioni (with contributions from Francesco Chiacchio [European Central Bank], Alvaro Espitia, Katerina Gradeva [European Central Bank], Laura Gomez-Mera, Asier Mariscal [Carlos III University of Madrid], Nadia Rocha, and Gonzalo Varela) • Chapter 3: Shawn Tan, Hernan Winkler, and Thea Yde-Jensen • Chapter 4: Erhan Artuc and Çag ˘ lar Özden (with contributions from Gnanaraj Chellaraj, Julio Elias, David Michael Gould, Bingjie Hu, Zovanga Kone, Tu Chi Nguyen, and Michael Packard) • Chapter 5: Cecilia Briceno-Garmendia, Mathilde Lebrand, and Megersa Abate (with contributions from Rodrigo Archondo, Gözde Isik, and Tetyana Kuchma) • Chapter 6: Mathilde Lebrand • Chapter 7: Omar Bamieh (University of Vienna), Matteo Fiorini (European University Institute), and Bernard Hoekman (European University Institute) • Spotlights 1 and 2: Gonzalo Varela and Nadia Rocha • Spotlight 3: Hernan Winkler • Content and Technical Editing: William Shaw • Overview Editing: Richard Alm Ekaterina Ushakova oversaw the production and support of the report. Many thanks go to all the commentators and reviewers in the initial stages of the report, particularly Europe and Central Asia Country Directors and Managers and Cyril Muller, Vice President of the World Bank’s Europe and Central Asia Region. Abbreviations AMI average management index BIT bilateral investment treaty BRI Belt and Road Initiative BvD Bureau van Dijk CIS Commonwealth of Independent States CMEA Council for Mutual Economic Assistance DCFTA Deep and Comprehensive Free Trade Agreements EBRD European Bank for Reconstruction and Development ECA Europe and Central Asia EEA European Economic Association EEC European Economic Community EU European Union twenty-eight member countries EU13  Bulgaria, Croatia, Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Romania, Slovakia, and Slovenia EU15  Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, Sweden, and the United Kingdom EU15+ EU15 plus Norway and Switzerland FDI foreign direct investment FDIRRI FDI Regulatory Restrictiveness Indicators GATS General Agreement on Trade in Services GDP gross domestic product GVC global value chain HIC high-income country ICT information and communication technologies LMIC lower-middle-income country MBI Mobility Barriers Index MDC multidimensional connectivity MFN most favored nation MIPEX Migrant Integration Policy Index MNE multinational enterprise NACE European Classification of Economic Activities NAFTA North American Free Trade Agreement NTM nontariff measures NUTS-3 nomenclature of territorial units for statistics (Nomenclature des Unités territoriales statistiques), level 3 xxvii xxviii  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia OECD Organisation for Economic Co-operation and Development OLS ordinary least squares PMR product market regulation PPML Poisson pseudo–maximum likelihood PTA preferential trade agreement RTA revealed technology advantage SPS sanitary and phytosanitary TBT Technical Barriers to Trade TFP total factor productivity TiVA Trade in Value Added TRIMS Trade Related Investment Measures UMIC upper-middle-income country WDI World Development Indicators WEF World Economic Forum WITS World Integrated Trade Solution WMS World Management Survey WTO World Trade Organization Countries and Economies International Organization for Standardization three-letter country codes; italics designate countries in the Europe and Central Asia region AFG Afghanistan ALB Albania ARE United Arab Emirates ARG Argentina ARM Armenia ATG Antigua and Barbuda AUS Australia AUT Austria AZE Azerbaijan BEL Belgium BEN Benin BFA Burkina Faso BGD Bangladesh BGR Bulgaria BHS Bahamas, The BIH Bosnia and Herzegovina BLR Belarus BLZ Belize BRA Brazil BRB Barbados BWA Botswana CAN Canada CHE Switzerland CHL Chile Abbreviations ●  xxix CHN China COL Colombia CMR Cameroon CRI Costa Rica CYP Cyprus CZE Czech Republic DEU Germany DNK Denmark DOM Dominican Republic DZA Algeria ECU Ecuador EGY Egypt, Arab Rep. ESP Spain EST Estonia ETH Ethiopia FIN Finland FRA France GAB Gabon GBR United Kingdom GEO Georgia GHA Ghana GRC Greece GUY Guyana HKG Hong Kong SAR, China HRV Croatia HUN Hungary IDN Indonesia IND India IRL Ireland ISL Iceland ISR Israel ITA Italy JAM Jamaica JOR Jordan JPN Japan KAZ Kazakhstan KEN Kenya KGZ Kyrgyz Republic KWT Kuwait LBN Lebanon LTU Lithuania LUX Luxembourg LVA Latvia MDA Moldova MEX Mexico MKD Macedonia, FYR MLT Malta xxx  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia MNE Montenegro MOZ Mozambique MUS Mauritius MYS Malaysia NAM Namibia NGA Nigeria NIC Nicaragua NLD Netherlands NOR Norway NZL New Zealand OMN Oman PAK Pakistan POL Poland PRT Portugal PRY Paraguay ROU Romania RUS Russian Federation SAU Saudi Arabia SGP Singapore SRB Serbia SWE Sweden SVK Slovak Republic SVN Slovenia SWZ Swaziland THA Thailand TJK Tajikistan TKM Turkmenistan TTO Trinidad and Tobago TUR Turkey TZA Tanzania UKR Ukraine UZB Uzbekistan XKX Kosovo (not listed as an ISO standard country; the unofficial two- and three-digit codes are used by the European Commission and others until an ISO code is assigned) YUG Serbia and Montenegro (former Yugoslavia) ZAF South Africa ZMB Zambia Regional Classifications Used in This Report Europe and Central Asia Northern Europe Southern Europe Central Europe Western Europe Western Balkans Denmark Greece Bulgaria Austria Albania Estonia Italy Croatia Belgium Bosnia and Herzegovina Finland Portugal Czech Republic France Kosovo Latvia Spain Hungary Germany Macedonia, FYR Lithuania Cyprus Poland Ireland Montenegro Sweden Malta Romania Luxembourg Serbia Slovak Republic Netherlands Slovenia United Kingdom South Caucasus Central Asia Russian Federation Turkey Other Eastern Europe Armenia Kazakhstan Belarus Azerbaijan Kyrgyz Republic Moldova Georgia Tajikistan Ukraine Turkmenistan Uzbekistan xxxi Overview The countries of the Europe and Central Asia (ECA) region, along with much of the rest of the world, find themselves engaged in a revival of one of the fundamental questions of economic policy: how much to open to the rest of the world. At the turn of the century, the issue seemed largely settled, and most nations viewed greater openness as a key component of the path to prosperity. In these heady days, the European Union (EU) deepened with a drive toward greater integration and expanded by incorporating nations transitioning to market-based economies. More recent events—most notably, the global financial crisis and the tough times that followed—sowed the seeds of doubts about the benefits of globalization, leading to a rise of protectionist and nationalist economic sentiments, exemplified by Britain’s referendum to withdraw from the EU. In 2018, how much to open to the rest of the world now dominates the political economy of the ECA region, not just within the advanced EU economies, but also among the emerging economies of the region. Deciding where to draw the line between openness and protection- ism has become a pivotal and divisive issue, often tinged with emotion. With this publication, the World Bank offers new research on the process of economic ­ integration, showing its potential benefits without ignoring the downsides. Main Findings of Critical Connections • The ECA region’s international connectivity through trade, foreign direct invest- ment (FDI), migration, telecommunications, transportation, and other avenues 1 2  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia facilitates the transfers of knowledge and technology that are critical to long- run growth and shared prosperity. These connections complement each other. For example, migration encourages trade and foreign investment by providing knowledge spillovers between host and home country markets and supporting connections to them. Similarly, the internet and efficient transport links are both necessary for successful e-commerce. Therefore, a balanced approach to increasing all dimensions of connectivity is desirable. • The depth of overall connections and the geographic composition of the connections both matter. Knowledge transfers are greater from countries that themselves have strong links to third countries. These transfers also emerge from firm linkages in global value chains as well as foreign ownership and management that generate local spillovers. • Deep integration of countries into the EU along many dimensions has gener- ated important benefits to growth through knowledge transfers. Central Asia, the South Caucasus, and the Western Balkans have benefited from regional connections as well, but the gains have been less pronounced. Much of the difference is due to the lack of direct and indirect connectivity to the wider global economy in the eastern part of ECA. • Despite its overall benefits, increased connectivity has encountered opposition— most notably, Britain’s June 2016 vote to exit the EU. National challenges often contribute to the backlash, but increased connectivity can expose ECA countries to external shocks, particularly those emanating from countries at the center of international economic transactions. By providing alternative sources of external demand and financing, however, a broad range of connections can reduce those risks and help countries cope with both domestic and external shocks. Introduction The ECA region has a rich history of regional integration and connectivity to the broader world economy, which has stimulated the growth of knowledge and tech- nological innovation. Indeed, through migration, trade, investments, and other interactions, ECA countries have depended on, and benefited from, connectivity with other countries for centuries. The Silk Road, formally established during China’s Han Dynasty in the second century BCE, facilitated more than the exchange of commercial goods. It was also a conduit for art, religion, philosophy, technology, language, science, and architecture (Starr 2015). Similarly, the Age of Discovery (1453–1660 CE) led to the deepening of a global community that was associated with profound advances in commerce and culture. As new navigation technology made sailing long distances possible, Europeans took to the seas to forge direct trading relationships with China, Indonesia, and Japan. Historians contend that it was the spice trade that fueled the development of faster boats, encouraged the discovery of new lands, and fostered new diplomatic relationships between East and West (Parthesius 2010; Bernstein 2013). In recent times, the most prominent feature of ECA connectivity has been regional integration through the gradual expansion of what is now the EU. The 1951 European Coal and Steel Community, a sectoral integration initiative among six European states, led to a much more ambitious agreement to form a European Overview ●  3 Economic Community in 1957. Over the next half century, the Community grew incrementally in geographic reach, issue coverage, and depth of policy cooperation. Within the EU, economic connections have progressively deepened from the initial lowering of trade barriers through the Single Market’s convergence of regulation and finally the adoption of the euro as a common currency by 19 member states. Today, the 28-country EU incorporates the free movement of goods, services, capi- tal, and people, with associated supranational common institutions—all the hard- won results of a multigenerational push toward greater connectivity. A major feature of European integration in the past 20 years has been the process of EU accession—most notably by 10 Baltic and Central European coun- tries (the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, the Slovak Republic, and Slovenia in 2004, followed by Bulgaria and Romania in 2007). Until the dissolution of the Soviet Union in 1991, the 10 nations that joined the EU had in one form or another been part of the ECA region’s second major regional bloc: the Council for Mutual Economic Assistance. The perceived advantages of con- nectivity led to a looser form of economic integration and cooperation between Russia and the former Soviet republics—the Commonwealth of Independent States (CIS). In the past decade, Russia has sought to deepen the CIS into a com- mon market and economic union and pursued a process of deepening economic integration with a subset of its neighbors through the creation of a Eurasian Economic Union. However, while progress has been made, the strength of global connectivity in the CIS remains much lower than in the EU. The ECA region’s growing participation in global and regional supply chains has greatly increased the importance and variety of international economic con- nections across the region. These forces have expanded ECA countries’ regional connections more rapidly than their connections outside the region. Nevertheless, as shown in the example of trade connectivity (figure O.1), many ECA countries have achieved substantial increases in global connectivity through their links to other ECA countries, such as Germany (DEU in the figure), France (FRA), or the United Kingdom (GBR), that have strong global connections. The ECA region’s persistent efforts to integrate reflect at the very least an intui- tive appreciation of the potential benefits from greater connectivity. More formally, economists have recognized the superiority of openness over autarky. In studying linkages between nations, they have focused on how knowledge transfers through international connectivity boost long-term growth, rather than one-time jumps in output due to gains from specialization (Romer 1990). Much of the knowledge gain from connectivity comes from “tacit” knowledge—the kind that comes through learning by doing and face-to-face interactions. Unlike “explicit” knowl- edge, it cannot be transferred in texts and blueprints. When looking at connectivity and knowledge transfers, analysts typically con- sider one channel at a time—such as trade, FDI, migration, telecommunications, or transport links. While many cross-country studies find, for example, that the level of trade or FDI relative to gross domestic product (GDP) is positively associ- ated with growth, they generally do not consider how many forms of connectivity work together. For example, it is hard to imagine trade taking place on the historic Silk Road without migration and transportation networks, or the recent develop- ment of e-commerce without high-speed internet and an efficient means of trans- ferring goods from one country to another. 4  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE O.1  Exports of manufactured goods a. 2000 b. 2014 Source: Calculations based on data from the United Nations Conference on Trade and Development. Note: The size of each country node reflects the total volume of trade. Each node has two outgoing links, which point to the country’s two top ­ export partners. Countries in the Europe and Central Asia region are shown in shades of blue. The methodology for plotting the countries attempts to show clearly the connections between countries in the global network of countries. The largest country nodes are pulled to the outer boundaries of the figure, but the pull is counterbalanced by the number and strength of connections with partner countries. Consequently, country nodes will tend to be grouped ­ ­ together if they share common connections. The importance of each connectivity channel for growth is likely to be affected by the strength of other channels—particularly when technology transfers depend on both tacit and explicit knowledge. For example, FDI by higher-income ECA economies in those with lower incomes can be an important source of knowledge transfers through exposure to sophisticated production techniques and manage- ment styles that are learned “on the job.” Migrants often learn important skills working abroad, and workers and managers from the investing country typically accompany the FDI. Thus, FDI and migration can work together to accelerate technology transfers within ECA. In Moldova, for example, connections developed through migration to Northern, Western, and Southern Europe in the 1990s sub- sequently generated Italian investment in the garment industry as well as German investment in factories for the assembly of electronic car components. Because of these initial connections and foreign investments, Moldova is now developing a service and manufacturing industry for the local market, creating its own brands, and exporting to other ECA countries. In addition to being mutually reinforcing, connectivity channels vary in depth and geographic composition. Being well connected to highly connected countries can provide benefits beyond being well connected to comparatively isolated countries. The advanced economies in Europe have provided a gateway for knowledge Overview ●  5 transfers from outside of ECA. Poland, for example, leveraged its growing ties to Germany to develop connections with that country’s trading partners and expand trade to broader markets within Europe and beyond. In the ECA region and other parts of the world, greater connectivity has delivered overall economic benefits for growth and development. Regional and global connectivity have been a tremen- dous “convergence machine,” raising living standards in lower-income countries to those of wealthier middle- to high-income countries (see World Bank 2012). The gains, however, are not evenly distributed or universally recognized. The 2007 global economic crisis and various commodity price shocks underscored the importance of understanding the potential risks of increased connectivity transmit- ting shocks from one country to another.1 Voters, both in Europe and elsewhere, are now questioning whether the benefits of greater connectivity are worth the costs. In addition to the United Kingdom’s 2016 vote to exit the EU (Brexit), recent elections in several European countries reflect an underlying skepticism regarding the benefits of deepening cooperation, with voters increasingly favoring parties seeking greater national autonomy instead of greater regional integration. Some analysts have attributed the lack of enthusiasm to concerns over the large migration flows and recent influx of refugees. Certainly, large sudden shifts in migrant flows, due to natu- ral disasters or wars, bring critical societal issues into play for domestic policy consid- eration. But larger questions have been raised about the downsides of regional integration and globalization in general and the role that deeper integration initia- tives have played as a driver in the rise of populism (see, for example, Rodrik 2018). Thus far, the skepticism has not led to a widespread retreat from integration among ECA countries. The institutions and policies that promote regional and global connectivity remain largely intact, with most countries continuing to benefit. However, ECA integration has been slowing, and the chal- lenges and questions call for a better understanding of ECA connectiv- While recognizing ity and its economic impacts. Analyzing the evolution of ECA’s regional the benefits of greater and global connectivity calls for paying particular attention to how connectivity, it is the various types of connections have interacted with one another important to acknowl- edge the potential and why connectivity in the region and in the larger global network downside risks through has played a key role in boosting growth and living standards. While the transmission of recognizing the benefits of greater connectivity, it is important to economic shocks. acknowledge the potential downside risks through the transmission of economic shocks as well as the choices countries face regarding which types of connections to strengthen with various partner countries. This Overview summarizes the main findings of the World Bank flagship study Critical Connections.2 The flagship’s primary purpose is to offer a deep analysis of ECA connectivity and how it has evolved over the past two decades. The frame- work and logical flow of chapters for this report is shown in figure O.2. In a key innovation of the study, a network analysis measure of multidimensional connectiv- ity captures the relationship between different forms of ­ connectivity and their joint impacts on growth and the transmission of shocks (chapter 1). The next step is to examine how knowledge flows through trade and investment channels from the ECA’s frontier firms to less technically advanced companies, improving productiv- ity at the firm level (chapter 2). In the ECA region, firm connectivity does not just exert its influence through foreign investment; it also works through enhancing 6  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE O.2  Framework and logical flow of chapters for Multidimensional Connectivity: Overall Pathways to Growth and Shared Prosperity in Europe and Central Asia this report connectivity (chapter 1) Knowledge Transfers from Informational Connectivity and Firms Migration and Connectivity International Openness in Trade channels (chapter 3) (chapter 4) and Investment (chapter 2) Conduits for Infrastructure Linkages connectivity (chapter 5) Organizational Supply Chains in Europe and Central Asia outcomes (chapter 6) ECA Policies for Improving Connectivity Policies (chapter 7) management practices. Ties between firms are associated with better outcomes in foreign-owned or -managed firms as well as with spillover effects that improve outcomes in locally owned and managed firms (chapter 3). Another complemen- tary channel of ECA connectivity is migration (chapter 4). A new methodology for filling in large gaps in our knowledge of ECA migration, particularly regarding skills and gender, provides insights into the trends and determinants of migration and migration’s economic impact on the region. Facilitating the movement of people and goods across the ECA region is the focus of the next layer of connectivity: infrastructure linkages (chapter 5). Another key innovation looks at the time and cost involved in moving goods and people across the region, rather than the kilometers and density of roads and rail links. This network analysis yields a richer perspective on the ECA’s transport links. The development of supply chains has been a key organizational outcome of the depth of ECA informational channels and conduits for connectivity (chapter 6). The development of Europe’s supply chains (“Factory Europe”), and the efficiency gains they provide, reflects the successes of narrowing policy barriers to trade, investment, migration, information and communication technology (ICT), and transport. Finally, ECA countries’ policy progress in supporting international con- nectivity over time and relative to other countries is evaluated to guide future policy actions (chapter 7). Multidimensional Connectivity Is a Key to Europe and Central Asia’s Development and Growth International connections include trade, FDI, migration, ICT, and transport links. Most studies measure the impact of each of these channels individually. This study takes a different approach, creating an indicator that combines all channels (networks) in a functional form that recognizes their complementarity—​ the multidimensional connectivity index (represented in figure O.3 as the MDC network). The measure reflects both the depth of each channel between each Overview ●  7 FIGURE O.3  Multidimensional connectivity combines many channels of connectivity Trade network FDI network N-network MDC network Note: This figure presents an indicative representation of the multidimensional connectivity (MDC) network that incorporates the relationship between all networks—trade, FDI, and other measured ­ global networks (N)—into a single collapsed network. A modified form of PageRank centrality for each country (node) is developed based on this collapsed network and used as an indicator of how overall connectivity influences growth overall and growth of the bottom 40 percent of the income distribution. FDI = foreign direct investment. pair of countries (e.g., the size of bilateral trade relative to each country’s GDP) and the benefits a country may reap from being connected to another well-­ connected country (e.g., Croatia’s trade with Germany is likely to boost ­knowledge spillovers more than Croatia’s trade with Albania owing to Germany’s wider global connections in addition to its higher level of technology). Compared with traditional approaches, this method more accurately mea- sures a country’s exposure to knowledge flows via direct and indirect interna- tional connections. The analysis presented in this study emphasizes the importance of complementary and balanced connectivity across the various channels. The impact on growth of any single connectivity channel is expected to decline as additional knowledge gains from the channel diminish—unless other channels of connectivity grow as well. In other words, policies to improve trade without complementary policies to improve investment and transport will have diminishing returns. Thus, promoting balanced connectivity across trade, transport, foreign investment, and other channels is likely to be more beneficial than focusing on enhancing only one channel. 8  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE O.1  Multidimensional Connectivity Varies by ECA Subregion, with the Highest Connectivity in the Western Part of the Region and the Lowest in the Eastern Part Multidimensional ECA subregions connectivity Trade FDI Migration ICT Airline Portfolio flows High connectivity Western Europe 6 6 6 9 9 15 19 Northern Europe 12 12 17 26 21 23 22   of which Baltics 30 28 36 38 50 28 21 Southern Europe 25 24 26 21 28 23 22 Central Europe 31 27 34 36 41 46 46 Medium connectivity Russian Federation 55 53 61 28 63 64 83 Turkey 57 51 67 33 73 79 40 Eastern Europe 62 59 60 81 54 57 76 Low connectivity Western Balkans 88 75 97 45 88 86 99 Central Asia 94 99 93 101 101 103 101 South Caucasus 104 104 102 64 104 104 93 Note: The table shows global rankings, from best to worst, in combined per capita connectivity, with lower values indicating better connectivity. Subregion indicators are median values of the subregion’s countries ECA = Europe and Central Asia; FDI = foreign direct investment; ICT = information and communication technology. The various channels exhibit some degree of substitutability, but complemen- tarity dominates. In some contexts, for example, trade may substitute for FDI because firms can either export a product to a foreign market or invest in the for- eign market to produce there. However, the information flows from trade tend to complement those from FDI. Firms may discover opportunities to export to a for- eign market because of their exposure through investing there. Thus, improving connectivity in one dimension improves connectivity through other channels. In terms of per capita levels in ECA subregional multidimensional connectivity (table O.1), Western Europe has the highest global ranking, followed by Northern, Central, and Southern Europe, while Russia, Turkey, and Eastern Europe are in the middle range, and the Western Balkans, Central Asia, and the South Caucasus have the lowest levels of overall connectivity. Not surprisingly, higher per capita levels of connectivity are associated with higher levels of development, reflect- ing both the number and depth of connections a country has. Tables OA.1 and OA.2 show individual country rankings of multidimensional connectivity on an absolute and a per capita basis, respectively. Central Asia and the South Caucasus rank low on overall connectivity, but because they started from a low base, they also saw the greatest improvement from 2000 to 2014 (figure O.4). The South Caucasus saw connectivity increase by nearly 75 percent, while Central Asia saw connectivity increase by more than 40 percent. Eastern Europe and the Western Balkans, although also starting from relatively low levels, have not seen increases as rapid, with connectivity increasing only 20 percent and 10 percent, respec- tively. The key challenge for these regions is to find ways to improve balanced Overview ●  9 FIGURE O.4  Europe and Central Asia’s connectivity has grown, but there are wide variations across subregions Growth in connectivity, percent, 2000–14 80 70 60 50 40 30 20 10 0 Central Central Eastern Northern Russian South Southern Turkey Western Western Global Asia Europe Europe Europe Federation Caucasus Europe Balkans Europe Note: Subregional and global indicators are median growth rates. connectivity, particularly through easing domestic constraints on doing business and facilitating trade, FDI, airline, and ICT connectivity. For the ECA region as a whole, while improvements in connectivity have slowed since 2008, it still has grown faster than global connectivity since 2000, reflecting the EU integration process as well as strides taken in transition economies. Using this study’s new MDC measure, we find a closer association with growth than when considering individual connectivity indexes separately. In the ECA region, the depth of a country’s international connections in 2000 contributed to growth over the subsequent 16 years, after accounting for other fundamental determinants of growth typically used in cross-country studies (such as initial GDP, education, size of government, inflation, investment rate, and quality of governance). This is because a deepening of each channel tends to increase the growth impact of other channels. The association between MDC and growth is shown in figure O.5, along with each individual component connectivity channel. The level of trade connectivity has the most significant individual impact on growth, followed by measures of connectivity through FDI, migration, and airline flights. Trade connectivity is statistically significant and associated with more rapid income growth of the bottom 40 percent of the income distribution, but the other connectivity indicators are not, perhaps because the bottom 40 percent benefit more directly from trade and less so from other forms of connectivity. The increase in international connectivity over the past decade has occurred in tandem with severe disruptions to the international economy, most notably the global financial crisis. Greater connectivity may have increased ECA countries’ exposure to such shocks, but it may also have increased countries’ ability to cope with them. Least vulnerable to shocks are countries with very high levels of connec- tivity and countries with very low levels of connectivity. The former can more easily find alternative export markets or sources of finance, and the latter are more insu- lated from the global economy. Countries in the middle of the range tend to be the most vulnerable to shocks for lack of easy alternatives to compensate for declines 10  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE O.5  Connectivity’s a. Single-dimension connectivity effects on overall and bottom-40 growth effects on overall and 1.6 *** bottom-40 growth 1.4 1.2 1.0 0.8 0.6 *** *** 0.4 * 0.2 * 0 –0.2 –0.4 Trade FDI connectivity Migration ICT connectivity Portfolio flows Airline connectivity connectivity connectivity Overall growth Bottom-40 growth b. Multidimensional connectivity effects on overall and bottom-40 growth 1.6 *** 1.4 1.2 1.0 0.8 *** 0.6 0.4 0.2 0 Multidimensional connectivity Overall growth Bottom-40 growth Note: All coefficients (except those on multidimensional connectivity) are estimated with ordinary least squares regression; multidimensional connectivity is estimated using a maximum likelihood procedure. The connectivity variables, including multidimensional connectivity, are normalized using the standard normal distribution; therefore, the size of the coefficient represents the annual growth impact of a one- standard-deviation change. FDI = foreign direct investment; ICT = information and communication technology. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. in connectivity. One example is countries highly dependent on a well-connected country—as shown in figure O.6 in the case of a shock originating in Germany. Using multidimensional connectivity to better understand the transmission of shocks also indicates that ripples radiate across countries indirectly con- nected to places experiencing hard times. The impacts are not always obvious. Take, for example, the potential impact of Brexit. In this example, a 20 percent drop in all connections between the United Kingdom and the 27 remaining EU countries (EU27) is simulated. As expected, the United Kingdom suffers the greatest harm, followed by small countries of the EU27, such as Malta and Ireland, that get most of their global connectivity through the UK. However, other countries outside the EU that are not directly affected by Brexit, such as Norway, Senegal, Libya, and Fiji, are nonetheless indirectly affected through Overview ●  11 FIGURE O.6  A shock originating in Germany has the largest impact on countries that gain their global connectivity through Germany Shock: 10 percent fall in all types of connectivity 0 SGP URY NZL HKG CAN PAN AUS BLZ PRY TTO ARG JAM SLV MYS IRL KGZ GUY KWT ECU GTM MEX OMN BLR THA –1 ATG BRN SWZ JOR QAT CRI SAU NOR SWE BWA FIN GBR BEN GAB ALB MUS ARE LUX JPN AZE GEO MOZ BHS DOM PER ISR IDN BRA CHN ARM CYP NGA MAR CHL FRA AFG CMR KEN COL USA TGO ETH GHA ESP BEL Change in level of connectivity (percent) BFA TJK NAM DZA EST EGY DNK LBN LVA ZAF BRB PRT IND CHE –2 TUN LTU ITA PAK BHR BGR NLD BIH RUS MKD KAZ GRC –3 SVK CZE Most affected by shock HUN due to high reliance –4 MLT on connectivity HRV AUT with Germany SYR TUR MDA BGD –5 LKA POL UKR –6 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Initial level of connectivity: Multidimensional index TABLE O.2  EU and Non-EU Countries Most Affected by Brexit   1. United Kingdom 15. Italy   2. Malta 16. Poland   3. Ireland 17. Germany   4. Cyprus 18. Latvia   5. Netherlands 19. Finland   6. Denmark 20. Hungary   7. Luxembourg 21. Czech Republic   8. Sweden 22. Senegal   9. France 23. Libya 10. Spain 24. Suriname 11. Norway 25. Slovenia 12. Greece 26. Fiji 13. Portugal 27. Iceland 14. Belgium 28. Austria Note: The table ranks countries according to the impact on the countries from Brexit, with a ranking of 1 indicating the greatest impact. EU = European Union. the connections of connections (table O.2). Interestingly, these countries are more affected by Brexit than some EU countries, such as Austria, Estonia, Lithuania, Romania, and Bulgaria. How connectivity transmits shocks is relevant to the past decade’s shift in public sentiment away from openness and toward a more inward-looking stance. 12  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia At the same time, the new mood increases the need to better understand how connectivity works to improve economic performance. The next few sections describe the ECA region’s recent experiences with the key channels of greater connectivity: knowledge transfers, foreign ownership and management, migra- tion, infrastructure, and supply chains. Europe and Central Asia Connectivity Is a Critical Source of Knowledge Transfers ECA connections with other countries through trade, investment, and production sharing are important because they increase access to technology and ideas critical to growth. Importing firms learn from exposure to more diverse and sophisticated inputs to their production, and exporters learn through opportunities to achieve economies of scale, upgrade workers’ skills, and improve products to compete in foreign markets. Local firms involved in FDI learn through technology transfers and exposure to high- skilled workers. Moreover, local firms not involved in trade or FDI also may learn through exposure to, or competition from, more internationally connected firms. All of this emphasizes the critical importance of openness to international transactions for gaining access to the knowledge essential for growth and productivity enhancements. Romania’s greater openness after it joined the EU led a Belgian logistics provider and an American software company to extend global value chains (GVCs) into the country, creating spillovers that benefited the local economy (box O.1). BOX O.1 Global Value Chain Spillovers in Romania H. Essers and Oracle are two examples of foreign Oracle is a major multinational company head- companies investing in Romania that illustrate quartered in the United States, specializing in benefits from foreign investments. developing and marketing database software and H. Essers is a leading European logistics firm with technology, cloud-engineered systems, and enter- headquarters in Belgium, focusing on chemicals, prise software products. In the mid-2000s, it pharmaceuticals, health care, and high-quality prod- opened a branch in Bucharest and began to hire ucts. After its integration with a Dutch company local engineers for routine software development. already doing business in Romania, the Belgian firm In addition to short-term spillovers, Oracle’s entry increased its presence in Romania, with an eye on has spurred a new generation of entrepreneurs Eastern Europe and Central Asia. Knowledge and who got their start at the company’s operations in know-how coming from traditional logistics hubs like Bucharest and went on to create their own busi- the Netherlands and Belgium subsequently nesses. One of them is Softelligence, a Romanian improved Romania’s logistics performance. Logistics software company that designs tailored mobile is the backbone of supply chains, making produc- applications for financial institutions. The low cost tion fragmentation and the smooth coordination of of entry for new entrepreneurs in this industry— its stages possible. Knowledge spillovers occur coupled with competitive wages, a qualified work- through clients’ learning good practices in quality force, and excellent internet connectivity—has norms, information technology, and cold chains. boosted this sector and diversified the economy. Overview ●  13 Europe compares well to the global frontier in manufacturing productivity, but it lags behind the global frontier in services and in some innovation-based growth industries. Technology creation in European manufacturing is very simi- lar to that in other advanced economies, as measured by the gap in labor productivity growth between frontier firms in Europe and the Organisation for Economic Co-operation and Development (OECD) countries (figure O.7). 3 However, labor productivity growth is lower in European services firms than in firms at the OECD frontier. The numbers suggest that the continent could be served well by pursuing better connectivity to the global frontier firms in the services sector. Similar sectoral differences between the most advanced European and global firms can be seen in the intensity of investment in research and development (R&D). FDI inflows, FDI outflows, and imports are all important to productivity growth across the world; as such, they are conduits for the transmission of both tacit and explicit knowledge. Empirical evidence suggests that when a pair of countries are linked through FDI or trade, an increase in R&D investment in one is associated with an increase in total factor productivity (TFP) in the other. In other words, FDI and trade help international R&D spillovers materialize.4 When the focus is trade as a conduit of R&D spillovers, evidence reveals that the quality of domestic insti- tutions is an important factor that facilitates spillovers. Better business environ- ments, the quality of tertiary education systems, and stronger patent protection are associated with stronger R&D spillovers. FIGURE O.7  Europe lags 3.0 behind the frontier in services 2.5 Average annual labor productivity growth, percent, 2010–13 2.0 1.5 1.0 0.5 0 Manufacturing Services European frontier OECD frontier Source: Calculations based on data from the Organisation for Economic Co-operation and Development (OECD) and Amadeus. Note: Sample is based on firms with more than 20 employees. The European frontier is among the EU15 (that is, the original core countries of the European Union). The technology gap is proxied by the differ- ence in productivity growth between frontier firms and other firms (laggards) in the same sector and year. 14  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Productivity increases in most firms are generated by the absorption of knowl- edge from other sources, rather than through their own investment in creating new knowledge. A firm’s potential to learn from existing knowledge can be measured by the difference between the firm’s TFP (or its TFP growth rate) and that of the most advanced firms in the sector. In Europe, advanced firms tend to be larger, have higher levels of capital relative to labor, invest more in intangibles (such as marketing practices), and have more-educated workers than other firms, although some of these differences vary by sector. The transfer of knowledge from international sources tends to follow a two- stage process (figure O.8). First, the most advanced domestic firms absorb knowl- edge from the most advanced firms globally, often through participation in GVCs that involve production sharing through trade, investment ties, and contractual agreements. Second, less advanced domestic firms absorb this knowledge through their exposure to the most advanced domestic firms. By contrast, the direct tech- nology transfer between the most advanced global firms and the less advanced domestic firms tends to be limited. Econometric evidence for the ECA region con- firms that a rise in TFP growth among advanced domestic firms (defined as the top 20 percent of domestic firms by TFP in each sector) leads to a similar increase in TFP among other domestic firms, but an increase in TFP among the most advanced global firms has little direct impact on the less advanced domestic firms. This analysis sheds light on the productivity slowdown in many ECA coun- tries after the global economic crisis. Productivity growth in Central and Eastern European EU members fell by 8.2 percentage points in 2008–14, compared to 2000–07 ­ (figure O.9). The crisis was transmitted through global supply chains and sharply reduced domestic firms’ engagement in these chains, which serve advanced markets in Europe and the United States. This decreased opportuni- ties for firms in Central and Eastern Europe to learn, led to a fall in their R&D spending as a share of GDP, and lowered their propensity to introduce new FIGURE O.8  How technology flows from European frontier firms (global value chain lead firms) to the remaining European firms STAGE 1: National Remaining GVC Integration of frontier: firms, • Predominantly through frontier: national firms in core STAGE 2: Top-producivity exposed to domestic networks Global lead production processes Outsourcing of firms technology firms in of GVC leads noncore • Some direct access to participating from non-CEE capabilities foreign multinationals Exposure to directly in national countries learning GVCs frontier Basic capabilities Irregular engagement of FDI with nonnational frontier firms on noncore GVC activities requiring noncore activities Note: CEE = Central and Eastern Europe; FDI = foreign direct investment; GVC = global value chain. Overview ●  15 FIGURE O.9  Productivity growth was lower in Central and Eastern Europe during the crisis Difference between labor productivity growth in Central European EU countries and that in Eastern European EU countries, 2000–07 versus 2008–14 2 0 –2 –4 –6 –8 –10 Latvia Estonia Lithuania Slovenia Romania Czech Slovak Bulgaria Croatia Hungary Non-CEE Poland Republic Republic EU Capital deepening Labor quality Total factor productivity Labor productivity Source: Calculations based on Conference Board data. Note: Overall macroeconomic data may reflect both sectoral changes and within-firm productivity growth. “Non-CEE EU” refers to the unweighted average for Austria, Belgium, Cyprus, Denmark, Finland, France, Germany, Ireland, Italy, Luxembourg, Malta, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom. CEE = Central and Eastern Europe; EU = European Union. products or processes—as shown in the World Bank’s Enterprise Surveys. In this instance, exposure to international volatility was a major driver of slower growth following the crisis. However, the “cure” of reducing firms’ exposure to international volatility through restrictions on trade or FDI would be worse than the disease, because such policies would diminish firms’ opportunities to learn through participation in global supply chains and other international transac- tions. This would particularly depress growth in less advanced countries, where firms are further from the productivity frontier and thus have greater opportuni- ties to learn through connections with foreign markets. International knowledge flows and their productivity impacts take place within companies—so their internal operating characteristics are likely to be important in determining whether connectivity gains are large or small. A critical factor is management. Looking at micro data, the next section focuses on how foreign management, regardless of ownership, can influence firm outcomes. Foreign-Owned and -Managed Firms Tend to Perform Better and Contribute to Local Firms’ Productivity The share of ECA firms owned by foreigners (excluding firms owned by parent companies located in tax havens) ranges from negligible, in countries such as Belgium, Bulgaria, Hungary, Russia, Ukraine, and most Southern European 16  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia ­ ountries, to 5 ­ c percent or more in most of Central Europe, Latvia, Lithuania, and the Western Balkans, to more than 32 percent in Ireland and the United Kingdom. More than half of foreign-owned firms in ECA also have predominantly foreign management (figure O.10). Across the ECA, foreign-owned firms tend to be larger than domestic firms, although the age of foreign-owned firms is not, on average, significantly different from that of local firms (figures O.11 and O.12). Many ECA firms are owned by people or firms in large, rich economies, such as Germany or the United States. However, geographic proximity, common lan- guage, cultural heritage, trade ties, and immigration from the source country are also important determinants of foreign ownership (table O.3). Firms that are foreign owned or foreign managed tend to achieve higher growth in operating revenues, employment, and average wages than other firms (figure O.13). Foreign-owned firms with foreign management have 28.3 percent higher growth in operating revenue, 19.6 percent higher job growth, and 16.8 ­percent higher wage growth than local firms. Foreign affiliates with local managers also perform better than local firms, although less so than foreign firms with foreign management. However, it is unclear whether the foreign firms’ better performance reflects the impact of foreign own- ership or management or foreign companies’ tendency to invest in the most productive regions, sectors, or firms. FIGURE O.10  Foreign-owned and foreign-managed firms in ECA, 2013 35 30 25 Share of all firms (percent) 20 15 10 5 0 SVK CZE ROU SVN POL HRV BGR HUN DNK LVA EST FIN SWE LTU ISL UKR RUS GRC ITA ESP PRT BIH SRB GBR IRL AUT DEU FRA NLD BEL Central Europe Northern Europe Southern Europe Western Western Europe Other Eastern Federation Russian Balkans Europe Foreign-owned firms Foreign-managed firms Foreign-owned and foreign-managed firms Source: Calculations using Orbis data. Note: Sample excludes firms with owners in tax haven countries. ECA = Europe and Central Asia. Overview ●  17 FIGURE O.11  Large firms are more likely to be foreign owned in ECA Share of foreign-owned firms by number of employees, 2013 20 18 18.0 17.7 17.2 16 14.8 14 12 11.7 11.2 Percent 10 8.6 8.5 8 6.7 6.9 6 4.6 3.8 4.1 3.8 4 2 2.0 1.6 1.8 2.0 2.3 0.6 0 0.0 0.1 0.2 0.3 0.0 0.1 0.1 0.5 Central Europe Northern Europe Ukraine Russian Federation Southern Europe Western Balkans Western Europe 1–9 employees 10–49 employees 50–249 employees 250+ employees Source: Calculations using Orbis data. Note: Sample excludes firms with owners in tax haven countries. ECA = Europe and Central Asia. FIGURE O.12  There is no clear relationship between a firm’s age and the likelihood of its being foreign owned Share of foreign-owned firms by age of firm, 2013 8 7.0 7 6 5.9 5.6 5.2 5 4.9 Share or firms (percent) 4.2 4 3.5 3.4 3.5 3.0 3 2.9 2.8 2.6 2.7 2.1 2.1 2 1.6 1.4 1.2 1.3 1 0.7 0.1 0.1 0.0 0.1 0.0 0.1 0.1 0 Central Europe Northern Europe Ukraine Russian Federation Southern Europe Western Balkans Western Europe 1–4 years 5–9 years 10–29 years 30+ years Source: Calculations using Orbis data. Note: Data for 2013. Sample excludes firms with owners in tax haven countries. 18  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE O.3  Most Foreign Firms in ECA Are Owned by German and US Firms Most common global ultimate owner Others (from left to right, top to bottom): Denmark, Norway, Russian Federation, United Belgium, Croatia, Slovenia, Region Germany United States Kingdom Netherlands Austria France Italy Finland Sweden and Japan Central Europe Northern Europe Ukraine Russian Federation Southern Europe Western Balkans Western Europe Note: Sample excludes firms owned by tax haven countries. Each row in the table shows the five (or more, if there is a tie) most common countries of ownership, among the top ten countries of ownership, for each country or region at left. For the Russian Federation and Ukraine, the rows show the five countries with the largest ownership shares. ECA = Europe and Central Asia. Overview ●  19 FIGURE O.13  Foreign-owned 30 and -managed firms perform 25 better than local firms 20 Percent 15 10 5 0 Growth in operating revenues Growth in jobs Growth in average wages Locally owned, foreign managed Foreign owned, locally managed Foreign owned and managed Note: Each bar in the figure represents the difference in growth (of the type labeled) between the type of firm depicted in that bar and that of firms that are both locally managed and locally owned. underlying coefficients are statistically significant. All ­ Transfer of management practices from the source country is likely an important reason for the better performance of foreign firms. For example, US-owned firms in Europe have management practices that place more emphasis on merit in determining career success, which is associated with greater use of ICT, than do domestic firms or firms owned by other countries (Bloom et al. 2018). On average in the ECA region, but not in the most advanced ECA economies, foreign-owned firms tend to have better management practices than local firms (figure O.14). The source country’s management quality is significantly related to the performance of its foreign affiliates: foreign affiliates from countries with better management practices perform better than other foreign affiliates, even after differences ­ between the source countries’ income levels, financial development, population, and stock of immigrants are taken into account. Local firms without foreign ownership or management can also benefit from the presence of foreign-owned firms (figure O.15). Local firms can learn from observ- ing management practices and technology in foreign affiliates, or through hiring workers trained in foreign affiliates. However, evidence of such effects across ­ industries is mixed. Better-performing foreign affiliates also may affect local firms through competition—by forcing them either to improve or to exit the market. Local firms tend to achieve significantly higher growth in operating revenues and wages in regions or sectors in ECA countries with higher shares of foreign firms than in sectors in which foreign affiliates are less prevalent. For small and young firms, there is no statistically significant relationship between the share of foreign firms in a sector and local firms’ employment growth. A possible interpretation of this result is that the presence of foreign firms forces some successful small and young local firms to become more efficient by increasing capital relative to labor, slowing job creation. In addition, ­ other firms that cannot compete shed labor (to more efficient firms) or close. Again, these relationships may in part reflect foreign owners’ decisions to invest in better-­performing sectors or regions. 20  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE O.14  Foreign 6 affiliates tend to have better management practices IRL FIN SWE than local firms DEU BEL DNK ITA LTU PRT GBR HUN AUT AMI of foreign-owned firms GRC POL FRA NLD CZE ESP HRV SVK 5 BGR RUS SVN ROU BIH 4 3 3 4 5 6 AMI of locally owned firms Western Europe Western Balkans Southern Europe Russian Federation Central Europe Other Eastern Europe Northern Europe 45-degree line Source: Calculations using data from the World Economic Forum (WEF) Global Competitiveness ­ Survey and Orbis. Note: A country’s Average Management Index (AMI) is based on the WEF Global Competitiveness survey that measures the quality of national business schools and the reliance of professional ­ ­ management. This index is also highly correlated with the World Management Survey, which is more comprehensive, but not as widely available. Because of competition from better-managed companies, larger and older firms are more likely to upgrade and adjust compared with younger and smaller firms. Increases in the quality of management in foreign affiliates are associated with faster growth in operating revenues, wages, and employment in local firms more than four years old and having more than 50 employees, but lower growth in these performance indicators in younger and smaller firms. Foreign ownership of firms tends to reduce the level of employment volatility in a country’s domestic economy. Interestingly, once a number of variables that influ- ence firm performance are controlled for, the performance of an average foreign firm in the ECA region is not statistically correlated with local economic growth. Foreign firms are less responsive to local economic conditions than local firms, regardless of whether foreign and local economies have the same business cycle. This could reflect better access to finance during an economic upswing in the parent country or a search for opportunities abroad when the parent ­ company’s profits at home are limited relative to the destination country. Regardless of the economic conditions in the parent country, foreign companies’ employment decisions seem to be less procyclical with respect to the domestic economy than domestic compa- nies: the former tend to create fewer jobs when the local (host) economy expands (figure O.16). Likewise, this also means that foreign companies tend to destroy fewer jobs than domestic firms when the local economy experiences a recession, possibly reflecting access to external factors that allow foreign companies to buffer Overview ●  21 FIGURE O.15  The positive spillovers of well-managed foreign firms seem weaker for small and young firms a. Regional spillover effects 60 2 Changes due to a one-unit increase in the Average Management Index All-firm impact (right axis) score of foreign firms in the region (percentage points) 1.5 40 1 20 0.5 0 0 –0.5 –20 –1 –40 –1.5 –60 –2 Mature and Small and Mature and Small and Mature and Small and larger firms young firms larger firms young firms larger firms young firms Change in operating revenue Change in number of employees Change in average wage b. Sectoral spillover effects 20 1.5 Changes due to a one-unit increase in the Average Management Index All-firm impact (right axis) 15 1.0 score of foreign firms in the sector (percentage points) 10 0.5 5 0 0.0 –5 –0.5 –10 –15 –1.0 –20 –1.5 –25 –2.0 –30 –35 –2.5 Mature and Small and Mature and Small and Mature and Small and larger firms young firms larger firms young firms larger firms young firms Change in operating revenue Change in number of employees Change in average wage Note: Small firms are those with 49 employees or fewer; young firms are those four years old or younger. Each bar represents the effect of increasing by one point the Average Management Index scores of foreign firms. The bars in each panel show the baseline effect (mature and larger firms), the baseline effect plus the interaction term associated with size, and the baseline effect plus the interaction terms associated with size and age simultaneously. 22  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE O.16  Foreign firms’ 10 employment decisions are less procyclical than those of 8 Foreign firms’ employment growth relative to local firms' GDP growth and difference between foreign-firm and their domestic peers 6 Local GDP growth local-firm employment growth (percent) 4 2 0 –2 –4 –6 –8 –10 t1 t4 t7 t10 t13 t16 t19 t22 t25 t28 t31 t34 t37 t40 t43 Time period the impact of the decline in economic activity. In other words, while foreign firms seem to contribute less to job creation than their local counterparts when the local economy is growing, they seem to bring more stability to the labor market during a downturn in economic activity because they lay off workers to a lesser extent than local companies do. It is just a short step from foreign owners and managers to the broader topic of migration, a hot-button issue in recent years. The next section focuses on how the ECA region’s increasing migration facilitates trade, knowledge transfers, and other benefits associated with greater connectivity. Economic Migration Has Been Beneficial to Europe and Central Asia In general, openness to migration, including that by foreign managers, helps many countries gain the skills, technology, and resources required to improve efficiency and compete in an increasingly complex, globalized world. In destina- tion countries, workers who are close substitutes for immigrants (e.g., they have similar skills) may lose as a result of lower wages or diminishing job opportuni- ties. At the same time, workers with skills complementary to those of immigrants may benefit. While the net economic effect for the country overall is positive, income distribution impacts may be positive or negative depending on the skill mix of the native and immigrant populations. Outside of economic consider- ations, large sudden shifts in migrant flows as a result of natural disasters or wars, such as the recent Syrian refugee crisis, bring humanitarian and local social impacts into play for the host country. These are critical societal issues for domes- tic policy consideration but are outside the purview of this analysis. Overview ●  23 Both emigration and immigration rates in many ECA countries are higher than the global average (map O.1), mostly driven by the removal of barriers to mobility within the EU and large migration flows following the opening up of Eastern-bloc countries. High levels of ECA migration have encouraged greater cross-border investment and trade (for example, by helping firms learn about foreign markets) and have facilitated the sharing of technology and knowledge between countries (for example, through schooling and language skills attained abroad). A large diaspora can generate substantial economic benefits for many origin countries in the ECA region. Remittances are an important source of income, have a positive impact on long-term economic growth and poverty reduction, and can improve access to capital markets. Diasporas are also a significant source of invest- ment, export demand, and knowledge transfers for ECA economies, particularly given the disproportionately high flows of skilled emigrants from regional coun- tries (see box O.2). Finally, the increasing share of migrants going to the United States and Northern, Western, and Southern Europe may have contributed to improving institutions in ECA transition economies by increasing their populations’ exposure to the norms of competitive democratic countries. What determines migration connectivity? To answer this question, this report develops a global bilateral migration matrix showing the number of migrants between all country pairs. Constructing the matrix requires estimating migration flows for the many countries missing such data and then estimating the global relationship between population flows and various drivers of migration (­figure O.17). Most migrants move to countries with similar or higher levels of per capita income. Migration tends to increase as the distance between countries decreases. A large share of low-skilled migrants move to neighboring countries, but high-skilled migrants are more likely to move to nonneighboring countries, reflecting the tendency for high-skilled emigrants from developing countries to move to high- income OECD countries (especially English-speaking countries). Sharing a similar language also has a positive effect on migration flows, particularly for skilled migrants, whose jobs often require strong language skills. Finally, the existence of a diaspora tends to increase the flow of migrants (particularly for unskilled workers) by reducing the costs of information, financing movements, and perhaps reducing the risks involved in migration. Some characteristics of ECA migration differ from these global patterns. Migrants from ECA countries (other than the advanced European economies), regardless of education level or gender, tend to move to other countries within the region. By contrast, migrants in other regions are no more likely to move within the region, once distance between countries and common borders are taken into account. High-skilled (but not low-skilled) migrants from former Soviet Union countries tend to go to other former Soviet Union countries, where they find similar institutions (a legacy of the Soviet Union) and close economic inte- gration. Differences in the size of the working-age population (those 25–65 years of age) are also important determinants of migration flows. Countries with larger and younger working-age populations tend to have larger emigration to coun- tries with smaller and older working-age populations. Because of aging popula- tions in Central and Eastern European countries, a smaller working-age population emigrates from those countries. Central Asia, however, remains a 24  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia MAP O.1  Emigration and immigration shares have seen the highest increase in Europe and Central Asia a. Emigration shares, 2010 IBRD 43819 | JULY 2018 Emigrants as a percentage of the population, 2010 20–100 10–20 5–10 1–5 0–1 No data b. Immigration shares, 2010 IBRD 43820 | JULY 2018 Immigrants as a percentage of the population, 2010 20–100 10–20 5–10 1–5 0–1 No data Source: World Bank 2018. Note: Reported data for Central Asia for 2010 are particularly spotty; therefore these maps rely heavily on an estimation methodology developed by the World Bank staff. Overview ●  25 BOX O.2 Marius Stefan of Autonom Romania: Knowledge transfers through travel and studies abroad Marius Stefan, a Romanian national, graduated with than three million Romanians emigrated—most an MBA from the University of Maryland in 2004.a moving from the economically depressed region His proud parents flew in from Romania for the cer- around Piatra Neamt to Spain and Italy. Many of the emony, and Marius decided to show them around migrants returned to Romania frequently, and they the US East Coast on a short road trip. Marius did it were glad to have the opportunity to rent a car. the way he learned from friends: he rented a car. When Marius realized the potential, he decided to His father was amazed that a private i­ndividual— act on a lesson from business school: after a pilot from a foreign country, no less—could rent a car project is successful, focus all your efforts on scaling so easily. Such businesses were unknown in up as rapidly as possible. Romania. Marius had a good understanding of the Marius’s brother had been educated in France car rental market from one of the case studies from and had been working in Paris as an international his MBA program, and he explained the business consultant. After a year of successful business devel- model to his father. opment, Marius added his brother’s experience and After seeing how it was done and listening to knowledge to the business, and they have been his son, Marius’s father made an unexpected pro- working together as the enterprise has expanded posal: that they would open a car rental business in beyond anyone’s dreams. Romania. Marius’s family was from Piatra Neamt, a Autonom now operates in 46 locations, employs small town in one of the poorest regions in north- 300 workers, and offers more than 5,000 cars for rent. east Romania. It had almost no tourism and little The company is going through a stage of accelerated economic activity at the time. While starting a busi- growth. It is developing a division for long-term rent- ness was exciting, Marius thought there was no als (operating leases) both organically and through potential and quickly forgot the conversation. acquisitions. It recently acquired the operational leas- Marius returned to Romania and worked in sev- ing division of Banca Transilvania, the largest bank in eral small businesses in Bucharest, the country’s capi- Romania, and the plan is to double the company’s tal. But his father kept reminding him about his turnover and assets in 2018. While developing the proposal to start a car rental business in his home business across Romania, the company started to town. Marius finally relented and agreed to ship three expand abroad. It started Autonom Hungary three vehicles to Piatra Neamt to gauge the potential for years ago and Autonom Serbia in May 2018—in car rentals. To his surprise, the rent-a-car business effect, transitioning from a national champion to a became an overnight success. Within a few weeks, regional player. employees at the business were calling Marius in None of this would have happened had the Bucharest, asking why the company had just three Stefan brothers not migrated abroad. An idea cars. There were always more people wanting to rent, sparked by a routine car rental in another country, but too few cars to satisfy the demand. combined with the knowledge gained by studying Marius traveled from Bucharest to Piatra Neamt abroad, helped Marius and his brother build a suc- to investigate and found a simple explanation. cessful company that testifies to the power of con- When the country joined the European Union, more nectivity and its knowledge transfers. a. This box was based on discussions with Marius Stefan in 2018. The author thanks Mr. Stefan for being so generous with his time. 26  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE O.17  ECA 0.8 migration is driven by geography, language, 0.6 historical ties, and past 0.4 migration 0.2 0 –0.2 –0.4 –0.6 –0.8 –1.0 Diaspora Distance Colony Contiguity Language similarity Male unskilled Male skilled Female unskilled Female skilled Note: Calculations are for 2010. “Skilled” migrants are those completing tertiary education. The size of the bars in the figure represents the coefficient of a regression equation and the percentage point impact on migration from a percentage point change in the migrant and home/host country attributes. Regressions also include regional dummies, such as migration within Europe and Central Asia (ECA) subregions and within the rest of the world. subregion with a relatively young working-age population and has relatively large emigration. In addition, while women make up slightly less than half of global emigrants, they are the majority of ECA emigrants, possibly because of their higher skill levels relative to the global pool of emigrants. Migration and the other connectivity components discussed in previous sec- tions require highways, railroads, air links, and cargo transport to reach their full potential. The next section examines the role infrastructure plays in enhancing the ECA region’s connectivity. While all forms of transportation are important in this regard, this research project focuses on the differences in cost and time needed to move goods and people among various countries. Strong Infrastructure Transport Links Provide Important Support for Connectivity Transport infrastructure forms the bedrock for international (and domestic) connectivity. This is obvious in the cases of trade in goods and migration, but ser- vices trade and investment flows also require supportive logistics, travel-related ser- vices, and infrastructure and may be linked to goods trade. Measuring the extent to which a country’s transport infrastructure facilitates connectivity through the chan- nels mentioned is complicated. Many traditional infrastructure indicators—such as kilometers of roads and rail, their condition, or the density of connections—provide only limited information on the economic value of transport connections. To get an alternative view, this study looks at new data on transport services and network analysis tools to measure the economic value of ECA countries’ connections through roads, railroads, and to a lesser extent maritime transport. Overview ●  27 The economic value of connections is measured here by the cost of trans- port, the time required, and the importance of the destination country in the overall transport network. Time is a particularly important consideration when the type of product (for example, perishable goods or parts and components traded within supply chains) or the nature of the passenger trip (for example, urgent business) requires rapid and reliable transport. Information on these dimensions is therefore essential for evaluating infrastructure investments’ real impact on growth and welfare. With some exceptions, domestic travel in the most advanced economies tends to take less time and be more affordable for residents (given the high incomes in these countries) than travel in other countries (figure O.18). By contrast, the time required to deliver a container from one main city to another within a country dif- fers little across ECA subregions, except in the case of Russia, for which distance between major cities plays a large role. However, time required for cargo ship- ments between countries is quite high in a few countries that serve as gateways to their neighbors (e.g., Belgium, Luxembourg, and the Netherlands), likely owing to the high level of congestion in their highway systems. The cost and time required for passengers to connect with neighboring coun- tries varies greatly within ECA, from the highest levels in Central Asia to the lowest in the Western Balkans. Advanced Europe’s high-level road and rail infra- structure also delivers fast travel times at relatively low cost for residents, given their high incomes. Transport takes up a larger share of income in poorer ECA FIGURE O.18  Transport connectivity (cost and time) between and within ECA countries Regional averages Passengers a. Within-country cost and travel time b. Between-neighbor-country cost and travel time South Caucasus Western Balkans Western Balkans Western Balkans Western Balkans Advanced Europe Central Europe Central Europe Central Asia South Caucasus Eastern Europe Advanced Europe Eastern Europe Central Europe South Caucasus Eastern Europe Central Europe Eastern Europe Advanced Europe Turkey Advanced Europe Turkey Turkey South Caucasus Russian Turkey Central Asia Central Asia Federation Russian Russian Russian Federation Federation Federation Central Asia 0 10 20 30 40 50 0 5 10 15 20 25 0 50 100 150 0 10 20 30 40 50 Average passenger Average travel Average passenger Average travel cost (euros) time (hours) cost (euros) time (hours) continued 28  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE O.18  continued Freight c. Within-country cost and delivery time d. Between-neighbor-country cost and delivery time Western Balkans Advanced Europe Western Balkans Central Europe South Caucasus Central Asia Central Europe Eastern Europe Central Europe Central Europe Eastern Europe Western Balkans Eastern Europe Eastern Europe South Caucasus Advanced Europe Advanced Europe South Caucasus Advanced Europe South Caucasus Turkey Turkey Central Asia Turkey Central Asia Western Balkans Turkey Central Asia Russian Russian Russian Russian Federation Federation Federation Federation 0 500 1,000 1,500 2,000 0 0.5 1.0 1.5 2.0 0 1,000 2,000 3,000 0 0.5 1.0 1.5 2.0 2.5 Average container Average container Average container Average container sending cost (euros) delivery time (days) sending cost (euros) delivery time (days) Note: Within-country transport connectivity as measured here is multimodal, averaging across road, rail, and bus modes the price that must be paid to travel to a representative main city in the country. Transport connectivity between neighboring countries as measured here is multimodal between a country’s capital city and main cities of neighboring countries. Only countries with complete data for time and cost for all modes (road, bus, and rail) are included. Within-country freight transport connectivity for a given city is measured as the average price to send a container from that city to the other main cities within a country. Between-neighbor-country freight transport connectivity is measured as the average price to send a container from a country’s capital city to the main cities of neighboring countries. “Advanced Europe” includes countries in Western, Southern, and Northern Europe that signed the Maastricht Treaty or joined the European Union before 1995. ECA = Europe and Central Asia. countries, with most Central European and Baltic countries in the middle of the pack. The average cost and time required to ship a container from a country’s capital city to the main city of neighboring countries varies little among most subregions, except for higher costs and time in Central Asia, Russia, and Turkey. Central Asian countries have much higher travel costs for both road and container transport and much longer travel times than other ECA regions. These countries might get help in integrating and improving their connectivity through recent or expected infrastructure projects gathered under the Belt and Road Initiative, a long-term project designed to reduce the cost of transport from China to Europe. Along with Portugal and Spain, the island countries Cyprus, Ireland, and Malta are also among the countries with the highest costs and longest time to reach the rest of the ECA network. South Caucasus per- forms better in terms of costs compared to time, whereas in Western Europe the opposite is the case. Central and Eastern European countries have rela- tively cheaper and faster connections to the rest of the network. The similarity in these rankings largely reflects road transport costs, which are determined in part by infrastructure quality and its impact on average speeds. The cost of Overview ●  29 moving containers is another matter. Unlike passenger road costs, this cost reflects other parameters, such as logistics costs, the presence of rent seekers, and the degree of competitiveness among service providers. As a result, coun- tries’ cost versus time performance is more diverse when looking at the cost of moving containers rather than people. Countries like Armenia, Greece, Kosovo, the former Yugoslav Republic of Macedonia, and Turkey have relatively better connectivity in terms of container costs than they do in terms of time. Montenegro, Norway, Slovenia, and others have relatively better time connec- tivity than cost connectivity. Understanding country specificities requires a deeper look into institutional factors, the quality of logistics, and the competi- tiveness of the transport sector. In addition to cost and time, determining the economic value of connections requires considering destination countries’ place in the network. The availability of close connections varies greatly within the ECA region: the total number of a country’s neighbors (or neighbors of neighbors) ranges from 2 to 22, and the level of aggregate GDP in the countries adjacent to a country varies significantly from country to country. Thus, some connections have a higher value than others in terms of access to a larger market. So in planning transport investments, a country should consider whether a particular investment improves access to a relatively isolated country or to a country with connections to a wider network. Transportation infrastructure channels the movement of goods and people along major cross-country networks and, within networks, corridors. The com- prehensive nature of the economic benefits that accrue to countries from being on a particular corridor, or at a specific crossroads of a network, remains an open question. A key question is whether roads or rail that pass though coun- tries provide economic benefits if ancillary businesses associated with the cor- ridor fail to materialize. If a country’s economic and business environment is sufficiently attractive for investment, however, transit flows may increase export and import opportunities for firms along these routes (or corridors), develop new sectors such as logistics services, and generate nonmaterial benefits (flows of ideas and knowledge) to boost productivity. Firms located in transport hub countries may benefit from lower production costs and an improved ability to deliver on time. Higher transport network connectivity might be desirable for increasing a country’s participation in regional and GVCs, attracting FDI, or increasing its participation in development corridors. It is important to identify the most critical countries in transport networks. Doing so reveals which countries have more control over transportation net- works’ operability and what shocks in these countries imply for other con- nected countries. Measuring critical transport networks can help countries target investments to reduce their vulnerability to specific country shocks that might impede access to markets or other areas of the network. More generally, critical countries in the transport network are those where disruption would have a major impact on subnetworks or countries that can be de facto discon- nected. For container costs, Russia is the most critical country in the network in Eurasia (figure O.19). A country’s cost-driven criticality reflects the increase in costs that a shipper (in the case of containers) would incur if it had to avoid that 30  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE O.19  Cost-driven CYP criticality in container network for Europe and ALB KSV Central Asia MNE MKD BIH GRC SRB HRV BGR TUR MLT ROM SVN HUN MDA ITA AUT PRT ARM SVK ESP GEO UKR CHE CZE FRA POL LUX GBR AZE DEU BLR BEL IRL RUS LTU LVA NLD EST DNK NORFIN KAZ SWE KGZ ISL TJK UZB TKM Note: The results shown in the figure capture only the Europe and Central Asia (ECA) transport net- work and do not include connectivity to countries outside ECA. Consequently, large ports (e.g., Rotterdam) that are connected to the United States or China will not appear as critical, although they are in the larger global context. Circle size indicates level of criticality (larger diameter = greater criti- cality). For illustrative purposes, the circles representing the top five countries in criticality are colored in green. Lines between nodes indicate the presence in the physical network of an optimal corridor connecting countries. Locations of ­ circles and countries are not linked to geography in any way. Results for time-driven criticality are not presented, as the results are very similar to those presented. country in shipping cargo. Germany, Ukraine, Hungary, and Poland are among the five most network-critical countries. As expected, islands or isolated coun- tries have a very low criticality. While France is not a top-five country in terms of network criticality, disruptions in the French transport network would affect connectivity to the rest of the ECA region for Spain, Portugal, the United Kingdom, and Ireland: Portugal’s connection to the European network is con- tingent upon Spain’s, and so forth. Different goals for transport networks may imply quite different investment priorities: • Countries may choose to strengthen partnerships—for example, to reach large markets, participate in supply chains, or connect to countries with high levels of technology (so the potential for learning is greater). As revealed by the cost and time of freight transport, ECA countries can be grouped into three categories in terms of partnerships: (a) the Western Balkans and Central Europe incur lower costs to reach the largest economies of Europe; (b) countries in Overview ●  31 Central Asia and South Caucasus together with Russia incur lower costs to reach countries with similar technology levels; and (c) countries in Advanced and Eastern Europe, as well as Turkey, incur lower costs to reach either the larg- est ECA economies or countries with more sophisticated technology. • Another possible goal among countries is to maximize the size of the market made accessible by their transport systems. Advanced Europe captures the largest amount of GDP per unit cost of transport (a container, private car, or railway ticket) among ECA subregions, followed by Eastern and Central Europe (85–90 percent of Advanced Europe’s market potential), South Caucasus and Turkey (50 percent), and Central Asia (40 percent). While the size of investments and the quality of services are important, many countries’ ability to increase their market connectivity by improving transport is limited by long distances from markets and difficult terrain. • Countries also may choose their investments to maximize their integration within the ECA transport network. Some connections contribute more to a country’s overall connectivity with the region than others. The Czech Republic, the Slovak Republic, and Austria are the three most integrated countries in the ECA network, while Central Asia remains poorly integrated. Factors over which a country has no control, such as the number of neighboring countries it has, are also key elements in determining its degree of integration, but its transportation network is more important. More cooperation among countries, especially along corridors, could increase the global benefits of transport investments. When it comes to the Belt and Road Initiative, for example, the benefit of investments for the network as a whole varies by individual segment. Reducing the cost of shipping a container in the Kazakhstan- China segment would have the largest impact on Kazakhstan’s ability to reach foreign markets, but Russia and Germany also would benefit significantly. Improving the Belarus-Russia segment would mostly benefit Belarus. A cost reduc- tion on the Poland-Belarus segment would have the smallest impact on the network as a whole. Although it would provide broad support for all aspects of ECA connectivity, bet- ter infrastructure has particular relevance for cross-border supply chains. Today’s busi- nesses, rather than being concentrated in a single country, find their production of goods is now divided among plants in different countries, with each assigned pro- duction of particular components or the assembly of components from other plants. The final consumer product may thus reflect inputs from a number of countries. The Growth of Supply Chains Reflects Greater Connectivity and Has Facilitated Increased International Knowledge Flows Regional supply chains are deepening around the world and are focused in three clusters—Asia, Europe, and North America. Europe’s supply chain is largely focused on Germany, particularly in motor vehicles, retail, and machinery and equipment. Despite enormous reductions in the prices of transport and communication, 32  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia geographic proximity continues to be important in coordinating production through- out supply chains. In fact, the importance of regional supply chains increased some- what faster than that of global supply chains from 1995 to 2011. Proximity remains important for several reasons. Suppliers often need com- plex information (tacit knowledge) that cannot easily be codified in blueprints or instruction manuals (explicit knowledge), requiring frequent, face-to-face communications with lead firms that are difficult and costly to achieve when plants are separated by thousands of kilometers. Proximity is associated with similarities in culture and l ­anguage as well as migration networks—both of which facilitate these detailed information transfers. On-time delivery and reli- able quality are critical in supply chains, where the lack of an intermediate input can slow production all along the chain. Therefore, lead firms may place greater emphasis on allocating production to close-by firms they know, rather than seeking cheaper locations at greater d ­ istance. Locating plants in proximity to one another can help improve the allocation of workers and machines across firms, facilitate the transfer of knowledge across plants, and enable more effi- cient use of infrastructure and other public services. Finally, regional integra- tion agreements such as the EU encourage supply chain production through regional partners by lowering costs at the border, establishing similar legal and regulatory frameworks, and providing confidence in the stability of integration frameworks over time. The growth of supply chains has generated substantial benefits. Participation in supply chains can expand the range of goods produced in developing countries. For example, a poor country that finds it difficult to compete with more sophisti- cated firms in the production of complex electronic products may be able to exploit its advantage of low-cost labor to assemble such products from compo- nents produced elsewhere. The transfer of knowledge is heightened in supply chains, which often involve intensive contacts with more sophisticated firms through trade, investment, and the movement of technicians and managers. Exposure to such knowledge can improve productivity. The growth of supply chains can also increase productivity through more intense competition, greater specialization (which can improve worker performance through learning by doing), and access to increased diversity of inputs. In OECD countries, growth in a coun- associated with growth in its real labor produc- try’s participation in supply chains is ­ tivity (figure O.20), although this association may partly reflect the fact that more productive countries are more likely to participate in supply chains. Increased participation in European (and Asian) supply chains has been asso- ciated with rising revenues from exports of both goods and services, even after subtracting the value of associated imports. To be more precise, participation in supply chains boosts the gross exports recorded in the balance of payments statistics. However, a substantial share of these export revenues must be devoted to paying for imported inputs used in the production of the goods, so the funds channeled to domestic profits and wages (i.e., domestic value added) may be a small fraction of gross export revenues. ECA countries that have increased their participation in supply chains tend to achieve a more rapid increase than other countries in the domestic value added generated by exports. From 2000 to 2011, for example, Turkey and Poland experienced some of the largest Overview ●  33 FIGURE O.20  Participation 0.10 in global value chains is correlated with higher labor productivity Growth in labor productivity (percent) 0.05 0 –0.05 –0.10 –0.20 –0.10 0 0.10 Growth in GVC participation (percent) Source: World Bank labor productivity data and country global value chain participation index for member countries of the Organisation for Economic Co-operation and Development over the period 2009–11. Note: Each dot in the figure represents one country for one year. GVC = global value chain. percentage increases in the share of exports through supply chains among tran- sition countries (figure O.21, panel a) as well as the highest growth rate of exports of value added (figure O.21, panel b). By contrast, Slovenia, Russia, and Hungary saw decreases in the share of exports through supply chains over the same period and only modest growth of value-added exports. Greater participation in supply chains also tends to increase a country’s depen- dence on other countries, potentially raising economic volatility in some segments of the country’s economy. For example, a natural disaster in Indonesia that inter- rupts the production of an intermediate good may idle workers in the Czech Republic and reduce the profits of German retailers. Finding the central sectors and the major cross-border links is important to understanding how positive or adverse shocks spread through production net- works in the ECA region. A country or a sector that is central might be able to spread ideas to the rest of the network, but it might also more frequently receive shocks from the rest of the network. The ECA production network is organized around several clusters that include sectors (for example, retail trade and motor vehicles) from different parts of the region. Having sectors from different regions in the same produc- tion cluster illustrates the interdependence of country-sectors across most ECA countries through input-output linkages. By evaluating which countries and sec- tor clusters are most critical for production in ECA, it becomes clear that motor vehicles in Germany are the most central sector in the ECA production network. This sector largely relies on wide-reaching regional value chains to organize its production. The retail sectors in Italy, France, Germany, and Russia are all 34  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE O.21  Among the a. Production fragmentation, 2000–11 transition countries, greater production fragmentation is HUN associated with a more rapid RUS increase in the flows of value SVN added in exports ROU SVK DEU BGR POL CZE TUR –20 –10 0 10 20 30 Change in the ratio of exports to value added (percentage points) b. Exports of value added, 2000–11 SVN HUN DEU RUS CZE SVK BGR ROU POL TUR 0 5 10 15 20 Average growth per year (percent) Source: Calculations based on Organisation for Economic Co-Operation and Development Trade in Value Added database. important as well, but Germany’s machinery and equipment sector is among the most critical value chain sectors. Outside of the mature EU countries, manufac- turing clusters in Poland, Russia, and Turkey play a secondary role. By far, France, Germany, and Italy are the most important centers for the ECA trade production network, followed by Russia and Turkey. The least central countries are the Baltic countries, the Eastern European countries, and Portugal. To this point, Critical Connections has focused on important aspects of con- nectivity in the ECA region. The phenomenon is multidimensional, with its vari- ous components working together to increase productivity. The gains, not shared equally among or within countries, are spurred by foreign ownership and management, migration, vital infrastructure, and supply chains. Overview ●  35 Connectivity, however, carries risks and ignites changes in national economies. The remaining task for the study centers on ECA policies: first, the ECA region’s recent record on promoting connectivity, and second, the challenges and opportunities that remain. European and Central Asian Countries Have Moved toward More Open Policies While most economic policies can affect international connectivity in some way, this study’s evaluation of ECA policy progress focuses on a set of areas that have important implications for openness to international connections. These include policies governing import tariffs, preferential trade agreements (PTAs), inward FDI, bilateral investment treaties (BITs) that protect investors from expropriation and adverse changes in investment policies, product market regulations, sectoral domestic regulations, visa regimes, and integration of migrants. ECA countries have made important progress on policies to boost international connectivity. In part, this has reflected individual countries’ efforts to open them- selves up to the global economy, particularly following the collapse of the Soviet Union. However, regional agreements have also played a critical role, including the increasing sectoral coverage and depth of agreements within the EU, the expan- sion of the EU membership, and the formation of the Eurasian Economic Union, composed of Armenia, Belarus, Kazakhstan, and Russia. The EU also has entered into more than 50 PTAs with countries in and around Europe. Over time, these agreements have shifted their primary focus away from reducing trade barriers, expanding to include liberalizing trade in services, pub- lic procurement markets, and cross-border investment. In addition, they have added provisions governing how the agreements are implemented through national regulatory regimes (figure O.22). This broader agenda requires more complex decisions than negotiations over tariff levels. For example, regulations designed to achieve such objectives as safeguarding the health and safety of consumers (which are consistent with a trade agreement) have a greater scope than regulatory measures that serve only to protect domestic producers (which are inconsistent with the market integration goals of a trade agreement). Studies have found that participation in deeper and more comprehensive trade agree- ments is related to a country’s ability to attract FDI. Each provision added to an agreement between a pair of countries (particularly in the areas of competition policies, investment, movement of capital, and intellectual property rights) is associated with an average 3 percent increase in FDI flows between the agree- ment partners. ECA countries have, on average, made considerable progress in reducing barriers to trade and investment: • Tariff levels applying to countries outside of preferential agreements have fallen steadily in ECA countries—from an average of 7 percent in 1988–96 to 5 percent in 2006–15. ECA’s average tariff in the latter period was lower than that in all other regions except North America (Canada and the United States). 36  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE O.22  ECA ranks 50 among the top regions in regard to the number of trade agreements and 40 investment treaties 30 20 10 0 Total number of Total number of Percentage of bilateral investment treaties preferential trade agreements most-favored-nation tariffs (country average) (country average) (country average) East Asia and Pacific Latin America and the Caribbean North America Sub-Saharan Africa Europe and Central Asia Middle East and North Africa South Asia Note: ECA = Europe and Central Asia. Although tariff rates were reduced in all ECA subregions over this period, they were particularly low in high-income countries. • ECA countries are among the least restrictive globally toward inward FDI, according to the OECD’s FDI Regulatory Restrictiveness Index (although data are not available for the Western Balkans and the South Caucasus subregions). EU countries are the least restrictive, but all ECA subregions achieved some reduction in restrictions from 1997 to 2015. • ECA’s average score on an index of the intensity of use of PTAs dwarfs those of all other regions, driven by regional integration among EU member states and—to a lesser extent—an expansion of PTAs by Turkey and Russia. • The average ECA country has entered into more BITs—almost 50 from 2000 to 2016—than the average country in other global regions. Almost 60 percent of all BITs signed by the average ECA country involve a partner in the ECA region. ECA countries have also made progress in reducing product market restrictions that hamper international connectivity, but the advances have slowed in recent years. A measurement of product market restrictions that includes barriers to entrepreneurship, trade, and investment, as well as the impact of the scope and nature of state control of the economy, shows that the restrictiveness of the aver- age European country has declined since 1998—but progress has stalled since 2008. Similarly, an index of the degree of policy restrictiveness in the ECA region implied by domestic regulatory regimes for energy, transport, and communica- tions has improved since the mid-1980s (though the country coverage of this index is limited)—but progress has been negligible over the past 10 years. According to both indicators, the ECA region has more restrictive policies, on average, than Canada and the United States. Overview ●  37 The transition countries (mostly not covered in the two regulation indicators discussed in the previous paragraph) have made significant progress in improving market-friendly regulations, as measured by the Transition Indicator Database developed and managed by the European Bank for Reconstruction and Development. The market openness indicators in formerly centrally planned econ- omies in the Baltic countries and Central Europe improved markedly from 1989 to 2000, although the pace of reform slowed in the subsequent decade. Other sub- regions also made significant progress, but scores on these indicators of open markets vary considerably. For example, the gap between scores for the average country in Central Asia and the OECD benchmark is four times larger than the gap between the average country in the Baltics and Central Europe and the OECD standard. Barriers to cross-border movement of people in high-income ECA countries—as measured by visa requirements, visa-issuing Barriers to cross-border practices, and consular services—declined moderately from movement of people in 2006 to 2012, and they remain substantially less than those high-income ECA imposed by the United States. However, obtaining a formal ­countries—as measured sector job is more difficult for immigrants in high-income ECA by visa requirements, countries than in the United States. The decline in visa restric- visa-issuing practices, tions in both the ECA and the United States has applied mostly and consular services— to nationals of high- and upper-middle-income countries, while declined moderately from 2006 to 2012. nationals from poorer countries have seen little decline in restric- tions on moving to high-income countries. In many of the region’s countries, limited integration of immigrants has impaired their contributions to host economies. For exam- ple, unemployment rates in the region tend to be higher among the foreign- born than the native-born population. Support for integration is even more important for refugees, who tend to take longer after entry to participate in the labor force. Although scores vary within the region, the ECA fares poorly on average when compared with other regions regarding policies supporting migrant integration (as measured by the Migrant Integration Policy Index). The average ECA country has had less success in integrating migrants than three of the four comparator countries selected for this exercise: Canada, the Republic of Korea, and New Zealand. (The ECA region has had more success than the fourth comparator country, Japan). The Central Europe subregion and Turkey have the worst performance, and while average scores in Western, Southern, and Northern Europe are closer to those of the best performers outside ECA, no ECA subregion has a score above the best-performing regions. In summary, ECA countries have been global leaders in cooperation through PTAs and BITs and are comparable to high-income countries in terms of poli- cies toward immigration. However, the average ECA country is more restrictive than countries in non-ECA regions regarding domestic regulations and migrant integration policies. The trend toward more open policies has slowed signifi- cantly, however, particularly since the first decade of this century. Little prog- ress has been made in tariff liberalization (as of the beginning of the 2000s), the use of BITs (as of the end of the 2000s), or reductions in FDI regulatory restrictions and product market liberalization (as of 2010). 38  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Considerable Scope Remains for Improving Policies to Increase Connectivity in Europe and Central Asia Over the past few decades, steps toward increasing connectivity have brought economic growth and greater productivity to most countries in the ECA region. It would be beneficial to develop these connections more broadly and deeply to support broad-based growth. This conclusion rests firmly on Critical Connections’ primary innovation: a multidimensional approach that examines the depth and breadth of ECA countries’ connections, both within the region and globally. It explicitly recognizes the complementarity of the individual channels of ­ connectivity: trade, FDI, migration, ICT, portfolio flows, and transport. The princi- pal message is that diversity in country connections and balance in all channels of international connectivity are critical for achieving the highest impact on growth and economic resilience. It is not enough to focus on a few countries for connectiv- ity or on one or two channels. The deepening of one channel can boost the impact of the other channels on growth. In addition to recognizing that channels are mutually supportive, multidimen- sional connectivity provides the following lessons: • ECA policies that build regional and global connectivity will stimulate robust growth. Being well connected in the global network of countries is important for long-run inclusive economic growth. • Multidimensional connectivity puts a spotlight on cross-border transfers of knowledge and ideas as the wellspring of sustained growth. An ongoing improvement in the stock of knowledge leads to increases in TFP that allow countries to get more output from the same amount of inputs. Knowledge flows that occur though connectivity are mostly “tacit”—that is, “learning by doing”—and not transmitted via books or blueprints. • As a result of the complementarity of the channels that link countries together, a balanced connectivity profile may be more important for knowledge spill- overs and growth than being well connected in a single connectivity dimension. Deep and comprehensive FTAs are a way to achieve this. • The number of connections a country’s economic partner has might be just as important as the type of connections. Not all partner countries are the same in this regard. Some partner countries have more connections than others, which makes them potentially better conduits for knowledge transfers. • In some cases, a country will be better off completing all channels of connectiv- ity to a poorly connected partner than building a single channel of connectivity to a well-connected country. • Connectivity’s being multidimensional implies that shocks in one dimension (say, migration) can have adverse effects in other dimensions (say, FDI and trade) as well. However, countries with the greatest connectivity are among those with the most resilience to shocks. The ECA region is a great laboratory for observing the role of multidimensional connectivity in action. Regional supply chains are strong, and links between coun- tries across the various forms of connectivity allow observation of how connectivity Overview ●  39 opens doors for the knowledge transfers that support sustained growth. The varia- tion in connectivity between countries can be exploited empirically to explore which forms of connectivity matter the most. Although the ECA region as a whole has moved toward greater connectivity, progress has been uneven across countries. Lower trade barriers have not always been associated with fewer restrictions on immigration or product markets, and some countries still rely heavily on other ECA partners for global connectivity. Most higher-income countries have pursued complementary policies in most areas of connectivity, but complementary policies have been less prevalent in lower-­ middle-income countries (e.g., lower tariffs are not uniform across partner coun- tries and lose effectiveness because of more restrictive domestic regulatory regimes). Moreover, infrastructure linkages remain quite poor in some parts of ECA, particularly Central Asia. ECA is a diverse region, and the appropriate policy mix to promote multidi- mensional connectivity will vary from one country to the next. This study supports some general observations about the direction of policy, how- ECA is a diverse ever. Most obviously, countries can maximize their exposure to interna- region, and the tional knowledge flows and their ability to exploit their comparative appropriate policy advantages by maintaining low barriers to international transactions, mix to promote including low tariff rates, minimal constraints on inward or outward multidimensional FDI, and efficient procedures for border transactions. Multidimensional connectivity will vary connectivity suggests countries should not try to rely on one or two from one country types of connections; rather, they should develop a wide variety of to the next. mutually supportive outside links, including migration. In addition, partici- pation in deep multilateral trade agreements that support integration of services markets and the reduction of differences in rules governing product mar- kets would increase the impact of low barriers to international transactions on connectivity. Broad improvements to the domestic business climate can make opening up an economy to the outside world more beneficial. A host of policies, desirable in themselves because they increase domestic economic efficiency, offer improved connectivity as a bonus: strengthening institutions, boosting financial sector devel- opment, and ensuring flexible labor markets. Adopting international best practice for standards governing product markets, worker protections, and the environ- ment also tends to encourage international connectivity, particularly by making it easier to participate in global and regional supply chains. Infrastructure investments that focus on improving the efficiency of logistics ser- vices are critical for most forms of international connectivity. Recognizing how some connections are more meaningful can give countries a framework to help evaluate the costs and benefits of infrastructure projects. In most economies, trade in final goods relative to trade in services is not as dominant as it once was. So improve- ments in telecommunications, including reducing the cost and increasing the effi- ciency of internet connections, would support international commercial transactions and improve contacts with diasporas and other foreign sources of knowledge. Greater connectivity has two sides: the sending and the receiving countries. A lack of pertinent market information could keep the two from building busi- ness connections. Countries seeking to improve connectivity—in particular, 40  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia those without a history of involvement with trade and FDI—may benefit from proactive policies that can help increase foreign investment and contacts with foreign firms. Investment promotion activities may be useful to encourage investors who lack sufficient information on domestic business opportunities or the policy regime and to reduce unnecessary or burdensome procedures for investment approvals. Policies should aim to boost the positive effects of connectivity at the firm level, which are associated with better performance. Encouraging skilled immigration may facilitate the introduction of the foreign management practices that increase productivity. Promoting linkages with foreign affiliates, both within ECA and out- side it, increases the kind of learning that improves the operation of local firms. Policies to help local firms acquire and absorb efficient global best practices could include increasing access to finance, management training, and supporting labor force mobility. Migration is a key element of multidimensional connectivity. Improving its benefits within the ECA will require policy reforms to better integrate migrants into the labor force and increased investment in education for all workers to cope with the ongoing transformation of work driven by technological change. From 2000 to 2010, the share of temporary migration rose in more than two-thirds of high-income ECA host countries. Temporary work has also been a more preva- lent feature of the labor market, reflecting greater connectivity and faster tech- nological change. Helping all workers benefit from a rapidly changing global economy involves moving away from social safety nets tied to long-term employ- ment toward general safety nets, allowing for more flexible contracts, invest- ments in education, and the removal of constraints on workers’ ability to move to find employment. Improvements in various types of connectivity are perhaps most critical for the lower-middle-income countries, particularly those in the South Caucasus and Central Asia. Because of both their geographic position and limited infrastructure, many of these countries are only weakly connected to other ECA countries and the global economy. The vast distances between Central Asia and Europe will remain an obstacle to connectivity. However, infrastructure investments and policies to improve integration through freer trade, infrastructure, and investment policies are likely to provide large growth benefits in Central Asia. The more diversified upper-middle-income countries, on the other hand, have a broader set of opportunities to improve connectivity. Participation in supply chains is strong in most of these countries, but balancing the currently uneven supply chain linkages, particularly low levels of imports of intermediate goods, versus already-high levels of exports of these goods, would support greater ben- efits from connectivity. A large set of policies—from removing barriers to trade and FDI to strengthening intellectual property protection and competitiveness reforms—are needed to improve participation in value chains and make the most of cross-border production opportunities. The high-income countries that recently entered the EU have established relatively open economies and strong domestic business climates in the context of deep integration within the EU. Nevertheless, further efforts are required to bring their domestic business climates to the level of the most advanced European countries. Most of the transition EU economies would benefit from Overview ●  41 reducing the excessive economy-wide restrictiveness in product markets that are relevant from a ­connectivity dimension. This calls for a review of the state’s role in the economy and the extent to which regulatory regimes impede new firms’ entry into sectors. If a country has relatively closed markets internally, its external connectivity will be affected by the reduction in its attractiveness for inward FDI and participation in global supply chains. The ECA region’s once-confident march toward greater connectivity has for the most part stalled in the past decade. Voices are currently casting doubt on the wisdom of opening to the outside world. The economic benefits of deeper and more diverse connectivity, however, are strong—most notably, the knowl- edge transfers from trade, FDI, and migration that deepen participation in mul- tinational supply chains and lead to faster growth. By exploring multidimensional ­ connectivity and its impact, Critical Connections provides a framework for under- standing the benefits of and concerns about globalization and helps provide information for policy discussions and actions that recognize how the various aspects of connectivity might work together to deliver resilient and faster growth. 42  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Annex OA.  Selected Indicators TABLE OA.1  Multidimensional Connectivity Indexes (on an Absolute Basis) Global ranking, from best to worst, in combined connectivity (lower rankings indicate better connectivity) Multidimensional Country connectivity Trade FDI Migration ICT Airlines Portfolio flows ECA Germany 2 1 5 3 4 3 3 United Kingdom 4 6 2 4 2 1 4 Netherlands 5 10 3 14 12 8 14 France 6 5 6 5 5 4 5 Belgium 7 7 7 18 9 13 18 Italy 8 8 13 7 6 7 6 Spain 10 12 12 10 7 6 9 Switzerland 13 15 10 17 8 10 17 Ireland 14 16 11 29 14 14 31 Sweden 15 17 14 19 13 12 19 Poland 17 28 25 20 27 28 23 Austria 18 18 22 21 18 15 20 Russian Federation 22 19 37 13 23 30 13 Czech Republic 23 21 28 39 35 25 38 Hungary 25 25 26 43 37 33 42 Denmark 28 26 23 25 20 11 24 Luxembourg 29 38 4 61 39 37 57 Finland 30 22 24 31 30 21 30 Turkey 31 32 39 16 33 27 16 Portugal 32 31 29 28 25 20 29 Norway 33 33 21 22 22 16 21 Slovak Republic 35 34 41 52 53 68 51 Ukraine 40 42 64 48 67 66 47 Greece 42 39 40 26 26 22 27 Bulgaria 48 47 59 62 58 62 62 Lithuania 50 49 58 68 80 49 65 Croatia 51 53 54 56 49 57 54 Estonia 52 50 57 84 59 61 78 Belarus 54 65 101 67 100 100 63 Latvia 58 55 63 79 73 54 76 Cyprus 60 71 52 74 45 51 70 Malta 62 64 60 94 70 63 95 Kazakhstan 63 62 83 53 95 88 53 Bosnia and Herzegovina 68 68 76 65 64 80 82 Macedonia, FYR 77 69 74 95 79 79 96 Albania 78 80 87 76 72 75 90 Moldova 86 89 100 102 89 92 102 Georgia 97 103 102 99 102 104 97 Armenia 98 105 103 103 103 103 101 Kyrgyz Republic 99 96 95 106 101 101 104 Tajikistan 100 106 106 107 105 106 105 Azerbaijan 101 102 107 87 104 102 84 Other countries United States 1 2 1 1 1 2 1 China 3 3 8 6 15 19 7 Canada 9 9 9 9 3 5 8 Mexico 11 11 20 8 10 9 11 Japan 12 4 17 2 19 18 2 Singapore 16 14 19 42 29 29 40 Brazil 19 29 18 11 24 32 10 Malaysia 20 13 31 36 31 38 37 continued Overview ●  43 TABLE OA.1  continued Multidimensional Country connectivity Trade FDI Migration ICT Airlines Portfolio flows India 21 24 38 12 17 24 12 Indonesia 26 23 35 24 48 36 25 Thailand 27 20 34 32 43 31 32 Hong Kong SAR, China 34 27 16 33 16 26 33 South Africa 36 35 30 27 38 50 26 Argentina 37 37 27 30 42 48 28 Chile 38 41 33 40 50 59 39 Israel 39 36 46 37 28 56 35 New Zealand 41 45 32 41 21 35 41 Morocco 43 46 51 45 46 40 52 Peru 44 52 45 49 52 60 48 United Arab Emirates 45 43 53 34 47 42 34 Saudi Arabia 46 40 43 23 36 53 22 Egypt, Arab Rep. 47 57 49 47 55 52 46 Colombia 49 58 44 35 41 45 36 Nigeria 53 67 48 46 32 73 45 Tunisia 55 44 71 59 51 44 60 Trinidad and Tobago 56 59 61 70 62 69 77 Costa Rica 57 54 65 72 63 43 68 Pakistan 59 48 66 44 66 76 43 Dominican Republic 61 51 67 54 40 34 58 Algeria 64 56 81 38 56 55 44 El Salvador 65 60 73 57 34 46 72 Panama 66 77 50 75 77 64 74 Guatemala 67 61 77 58 44 58 64 Qatar 69 70 56 60 57 47 56 Bangladesh 70 63 92 51 97 94 50 Uruguay 71 73 55 77 71 77 71 Bahamas, The 72 84 42 92 65 17 89 Jordan 73 66 68 82 61 67 80 Mauritius 74 74 47 93 83 83 94 Ecuador 75 85 70 55 60 65 55 Ghana 76 78 82 83 78 85 83 Jamaica 79 87 62 63 68 39 79 Sri Lanka 80 76 91 69 93 93 66 Oman 81 72 84 64 54 41 59 Kenya 82 93 72 71 84 71 69 Bahrain 83 91 85 80 99 82 75 Lebanon 84 79 78 73 76 78 67 Cameroon 85 90 79 78 92 87 73 Mozambique 87 97 69 89 90 98 91 Kuwait 88 81 89 50 69 95 49 Gabon 89 94 75 88 88 97 86 Barbados 90 99 36 97 75 70 100 Syrian Arab Republic 91 75 88 66 74 81 61 Namibia 92 86 90 96 86 89 92 Paraguay 93 95 80 86 85 86 88 Guyana 94 92 86 85 81 84 108 Botswana 95 83 94 90 91 96 85 Swaziland 96 82 93 104 94 99 103 Ethiopia 102 100 99 81 98 90 81 Brunei Darussalam 103 88 96 91 96 91 87 Benin 104 98 104 101 107 105 99 Belize 105 101 97 105 87 74 106 Antigua and Barbuda 106 104 98 108 82 72 107 Burkina Faso 107 108 108 100 108 108 98 Afghanistan 108 107 105 98 106 107 93 Note: ECA = Europe and Central Asia; FDI = foreign direct investment; ICT = information and communication technology. 44  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE OA.2  Multidimensional Connectivity Indexes (on a Per Capita Basis) Global ranking, from best to worst, in combined connectivity (lower rankings indicate better connectivity) Multidimensional Country connectivity Trade FDI Migration ICT Airlines Portfolio flows ECA Luxembourg 1 2 1 1 1 3 32 Ireland 2 3 6 4 3 5 28 Netherlands 3 6 5 21 15 13 9 Belgium 4 4 4 6 7 19 79 Switzerland 6 5 8 2 2 7 7 Malta 7 23 7 35 23 6 53 Sweden 9 8 13 8 10 12 84 Norway 11 13 12 10 13 8 15 Cyprus 12 37 3 16 6 11 14 United Kingdom 13 19 15 11 9 17 5 Denmark 14 11 17 15 12 4 11 Finland 15 9 16 31 25 16 16 Austria 16 7 19 9 17 14 70 Germany 18 10 20 12 19 26 19 France 19 17 18 13 22 30 18 Czech Republic 21 12 28 29 34 34 30 Hungary 22 18 25 51 37 36 26 Spain 23 25 24 17 29 23 48 Estonia 24 20 30 23 27 22 1 Portugal 26 27 29 24 30 24 21 Slovak Republic 27 14 32 41 43 56 66 Italy 31 22 31 32 32 33 12 Lithuania 35 28 40 45 77 35 2 Latvia 37 32 41 20 55 21 60 Poland 38 34 36 47 45 49 34 Croatia 41 44 37 59 38 39 56 Greece 42 45 42 26 31 31 20 Bulgaria 45 43 45 71 51 50 62 Turkey 56 55 57 39 76 62 85 Belarus 58 48 73 30 83 100 98 Bosnia and Herzegovina 60 60 63 88 49 63 82 Macedonia, FYR 61 52 58 83 65 58 69 Russian Federation 63 49 67 43 73 80 40 Ukraine 71 62 76 27 87 81 104 Albania 75 79 75 37 60 48 75 Kazakhstan 76 68 83 25 99 90 99 Georgia 84 102 65 63 98 99 90 Moldova 86 81 94 48 62 68 80 Armenia 87 90 70 77 89 79 81 Kyrgyz Republic 95 99 97 50 100 95 94 Tajikistan 102 109 101 67 107 98 103 Azerbaijan 106 106 107 87 103 93 106 Other countries Singapore 5 1 11 14 11 20 10 Hong Kong SAR China 8 15 10 22 5 29 4 Bahamas, The 10 41 9 36 14 1 8 Canada 17 16 22 7 4 18 22 Barbados 20 85 2 49 28 10 63 Australia 25 35 23 3 21 44 17 Qatar 28 33 21 42 18 9 23 New Zealand 29 39 26 5 8 27 13 United States 30 30 27 18 26 46 64 Mauritius 32 57 14 52 66 52 6 Trinidad and Tobago 33 31 33 66 33 28 72 continued Overview ●  45 TABLE OA.2  continued Multidimensional Country connectivity Trade FDI Migration ICT Airlines Portfolio flows United Arab Emirates 34 29 34 55 39 37 77 Malaysia 36 21 43 54 40 60 65 Israel 39 24 46 19 24 51 61 Chile 40 46 35 58 63 66 29 Japan 43 26 50 75 57 71 24 Panama 44 59 38 53 52 40 73 Bahrain 46 42 49 72 70 42 37 Uruguay 47 73 39 40 56 59 3 Mexico 48 40 51 56 42 53 107 Thailand 49 38 55 78 82 69 36 Costa Rica 50 36 60 70 54 38 83 Argentina 51 66 44 34 61 74 27 Saudi Arabia 52 51 54 44 41 65 42 Oman 53 56 56 61 35 32 41 South Africa 54 53 53 64 68 82 39 Brazil 55 69 48 69 85 87 49 Gabon 57 80 47 46 81 75 33 Brunei Darussalam 59 50 71 60 67 43 78 China 62 54 62 113 88 96 89 Peru 64 74 59 85 72 77 71 El Salvador 65 63 61 80 16 45 86 Jamaica 66 83 52 68 46 25 88 Morocco 67 75 66 106 79 61 44 Dominican Republic 68 61 64 62 36 41 91 Tunisia 69 47 79 90 59 55 25 Jordan 70 65 68 33 44 54 51 Indonesia 72 71 77 99 97 89 43 Guyana 73 64 80 81 53 47 76 Philippines 74 58 84 79 80 86 45 Colombia 77 84 72 86 74 78 50 Swaziland 78 67 89 82 91 83 93 Antigua and Barbuda 79 92 69 28 20 2 58 Lebanon 80 77 74 103 64 57 31 Namibia 81 70 93 74 75 67 92 Kuwait 82 76 86 104 48 73 55 Botswana 83 72 92 109 92 76 87 Egypt, Arab Rep. 85 86 78 97 95 85 67 Guatemala 88 78 88 84 58 64 96 Belize 89 93 87 38 47 15 74 Ecuador 90 96 85 65 69 70 101 Algeria 91 82 98 107 93 72 52 Ghana 92 87 95 100 50 94 105 Paraguay 93 100 96 57 94 84 46 Nigeria 94 101 90 101 71 104 68 Cameroon 96 105 91 91 102 97 35 India 97 88 99 95 84 101 47 Zimbabwe 98 89 100 76 90 91 54 Sri Lanka 99 91 102 92 105 102 59 Mozambique 100 107 82 73 104 103 108 Kenya 101 103 81 94 86 88 95 Pakistan 103 97 103 111 78 106 57 Sudan 104 98 105 105 112 113 113 Syrian Arab Republic 105 94 104 108 96 92 102 Bangladesh 107 95 108 96 101 109 38 Benin 108 108 109 89 111 111 100 Togo 109 104 113 93 108 105 97 Ethiopia 110 111 110 112 109 107 112 Afghanistan 111 113 106 110 106 110 110 Niger 112 112 111 98 113 112 111 46  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Notes 1. Transmission of shocks is not new. The bubonic plague of 542 CE, which decimated the Byzantine Empire, is thought to have arrived in Constantinople (today’s Istanbul) by way of the Silk Road. The spice trade was also accompanied by struggles for economic dominance as wars were fought, lands were colonized, and fortunes were made and lost. 2. This study is available electronically at http://www.worldbank.org/en/region/eca​ /­publication​/­critical-connections. 3. Note that the OECD average includes the European Union countries. 4. In a seminal contribution, Coe, Helpman, and Hoffmaister (1997) identify that by trad- ing with industrial countries with a large “stock of knowledge” accumulated through R&D activities, developing countries boosted their productivity by importing intermedi- ates and capital goods that embodied knowledge and information. Van Pottelsberghe de la Potterie and Lichtenberg (2001) identify that FDI, and in particular outward FDI, has also been a conduit for R&D spillovers for 13 industrial countries (including 11 EU member countries). References Bernstein, William J. 2008. A Splendid Exchange: How Trade Shaped the World. New York: Grove. Bloom, Nicholas, Kalina Manova, John Van Reenen, Stephen Teng Sun, and Zhihong Yu. 2018. “Managing Trade: Evidence from China and the US.” NBER Working Paper 24718, National Bureau of Economic Research, Cambridge, MA. http://www.nber.org​ /­papers​/­w24718. Coe, David T., Elhanan Helpman, and Alexander W. Hoffmaister. 1997. “North–South R&D Spillovers.” Economic Journal 107: 134–49. Parthesius, Robert. 2010. Dutch Ships in Tropical Waters: The Development of the Dutch East India Company (VOC) Shipping Network in Asia 1595–1660 . Amsterdam: Amsterdam University Press. Rodrik, Dani. 2018. “Populism and the Economics of Globalization.” Journal of International Business Policy 1 (1–2): 12–33. https://link.springer.com/article/10.1057/s42214​ -018-0001-4. Romer, Paul. M. 1990. “Endogenous Technological Change.” Journal of Political Economy 98: S71–S102. Starr, S. Frederick. 2015. Lost Enlightenment: Central Asia’s Golden Age from the Arab Conquest to Tamerlane. Princeton, NJ: Princeton University Press. van Pottelsberghe de la Potterie, Bruno, and Frank Lichtenberg, 2001. “Does Foreign Direct Investment Transfer Technology across Borders?” Review of Economics and Statistics 83 (3): 490–97. World Bank. 2012. Golden Growth: Restoring the Lustre of the European Economic Model. Washington, DC: World Bank. ———. 2018. Moving for Prosperity: Global Migration and Labor Markets. Policy Research Report. Washington, DC: World Bank. 1 Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia International connections through trade, foreign direct investment (FDI), migration, the internet, and other channels are critical for the transmission of knowledge and growth. But how much knowledge is transmitted to a country is not only the result of the overall level of connectivity, but also to whom a country is connected and how the connections complement each other. For example, being well connected to an economy with wide-­ reaching global connections is likely to be a stronger conduit for knowledge transfers than being connected to an isolated economy. Likewise, connec- tions are likely to complement each other. For example, e-commerce is often seen as a benefit of internet connectivity, but without transport con- nectivity, e-commerce may not amount to much. This broader definition of connectivity, referred to as multi­dimensional connectivity, is explored in this chapter for Europe and Central Asia (ECA) and forms the basis for examin- ing connectivity in various channels in subsequent chapters of this flagship report. ECA’s international multidimensional connectivity has expanded sharply over the past three decades, owing to greater global integration driven by lowering of costs to economic transactions, the breakup of the Soviet Union, and increasing integration within, and expansion of, the European Union (EU). Enhanced international connectivity generally has been asso- ciated with growth, through the transmission of technologies across borders. This transmission is most effective when deep connections exist ­ across different channels, and when countries are connected to other, 47 48  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia connected countries. This chapter discusses the multidimensional well-­ character of connectivity. It assesses the impact of improved connectivity on income and income distribution, and it addresses the question of whether the region is optimally connected to other economic poles in the world. The report will summarize the related policy options. Main Messages • Regional connections through trade, FDI, migration, telecommunications, and other channels over the past two decades have risen more rapidly than ECA’s connections outside the region, in part reflecting policies geared toward increasing regional integration and the rising importance of regional value chains. However, countries in ECA that are linked to strong ­ globally connected ECA countries, for example, Germany or the United Kingdom, may nevertheless have experienced substantial increases in global connectivity as well. Among connectivity channels, ECA’s intraregional trade links are stronger than its FDI links, while airline connections and labor mar- ket integration have increased sharply among European countries. • Network connectivity measures for trade, FDI, migration, information and communication technology, airline flights, and portfolio flows are all positively related to growth, and each is associated with higher growth over and above the influence of standard growth determinants. However, not all channels play an equally important role. Trade connectivity is perhaps the most impor- tant and is related to overall growth and the income growth of the bottom 40 percent of the income distribution. Increasing linkages in each form of connectivity are complements to one another, suggesting that a balanced connectivity profile along all dimensions of connectivity is more important than a large increase in one channel only. The growth impact of multidimen- sional connectivity is higher than the impact of each of the individual network indexes, suggesting that overall connectivity is more important than each of the individual channels separately. Thus, policies to promote connectivity across trade, migration, and FDI are likely more beneficial than focusing on enhancing only one channel, and reducing connectivity in one dimension may reduce the impact of growth from other channels. • Greater international connectivity can increase a country’s exposure to inter- national shocks, but may also mitigate shocks by enabling a country to increase its reliance on other links in its network. Both countries with low and countries with high levels of connectivity tend to be more resilient to shocks in the global network, the first because of the limited number of partners that may become a source of shocks, the second because well-diversified con- nections may provide alternative sources of, for example, finance or export demand. In contrast, countries in the “middle” of the connectivity spectrum, that is, countries that are highly dependent on a few well-connected coun- tries, appear to be most susceptible to shocks that originate from, or affect, these countries. Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  49 Introduction Globalization often means different things to different people. For some, it is the large number of imported goods seen on store shelves. To others it is a social phenomenon that includes everyday exposure to a wide variety of cul- tures, peoples, foods, products, and spoken languages. In major cities through- out ECA the change is perhaps most apparent, while in smaller towns or villages it may be less so. In Central Europe the look and feel of major cities is very different now than before the transition to market economies in the early 1990s. But even in the small towns in Northern, Western, and Southern Europe, integration brought by the European Union and greater global connectivity has changed the look and feel of everyday life and economic opportunities. Regardless of where one is physically located, or how one observes globaliza- tion, the interconnectedness of the world is increasingly touching us either directly by the people we encounter or indirectly through the items we pur- chase or foreign firms that employ us. Since the early 1990s, the countries of ECA have been radically trans- formed, as borders were opened and many hurdles impeding cross-border connectivity were lowered. The move toward the common market in the European Union and the fall of the Iron Curtain had impacts well beyond trade—including positive impacts on income and income distribution. This trend toward income convergence with developed countries occurred despite intermittent external shocks, suggesting, in turn, that economic connectivity to regional and global markets has likely been an important driver of growth and improved standards of living. Since 2008–09, the global financial crisis, deepening geopolitical tensions, the refugee influx, and sharp commodity price fluctuations have posed new challenges for the region, pointing, among other things, to the need to more fully understand the role that economic connectivity can play in preserving economic growth and incomes in times of political and economic flux. Much of the empirical work done to date has recognized the importance of openness for economic growth, including through trade, FDI, the internet (infor- mation and communication technology—ICT), migration, and other forms of connectivity (Dollar 1992; Ben-David 1993; Sachs and Warner 1995; Edwards 1998; Frankel and Romer 1999; and Javorcik 2004, among others). While there are many nuances to the empirical findings, and questions remain regarding causality between outcomes and policies (Rodriguez and Rodrik 2000), the asso- ciation appears to be strong and intuitively appealing. Technologies embodied in goods, investments, and people are likely to be transmitted across borders, as long as the source and host countries are open and have the capacity to absorb these innovations. In other words, in addition to the gains from specialization that openness brings through each layer of connectivity, knowledge ­ spillovers are also likely created. This leads not only to one-time increases in output, but also, in the context of endogenous growth theory, long-term increases in eco- nomic growth because the cost of acquiring new knowledge falls with an increas- ing stock of knowledge (Romer 1990; Helpman 2004). 50  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia To date, economic research has only examined one dimension at a time of partner connectivity and the relationship to economic growth. Empirical research is available on the relationship between various types of trade and economic growth, FDI and growth, the internet and growth, and migration and growth (Mountford 1997; Borensztein, De Gregorio, and Lee 1998; Alfaro et al. 2004; Czernich, Falck, and Kretschmer 2011). But empirical and theoretical work has yet to examine how the interplay between these various layers of connectivity comple- ment each other. For example, internet connectivity has various direct avenues for influencing economic growth, including providing individuals the ability to quickly research products available in foreign countries, take online courses, and transact in services remotely, as well as serving as the backbone for facilitating greater deepening of cross-border global supply chains. E-commerce has been greatly enhanced by the availability of broadband internet. Nonetheless, without trans- port connectivity through roads, rail, shipping, and air transport, the effects of broadband connectivity as a channel to stimulate growth via e-commerce would be greatly diminished. (See Spotlight 3, “Reaping Digital Dividends through Complementary Investments.”). More telling regarding the interplay between various forms of connectivity is the role of migration and international travel, be it for permanent migration, foreign study, or tourism. While Gould (1994) first identified the complementary relationship between migration and trade between the home and host countries of migrants, subsequent research has also identified migration’s importance in influencing FDI and its direct influence on growth through knowledge transfers (Onodera 2008). Consequently, migration may not only be important for growth by directly transferring knowledge between the host and home countries, but also by facilitating knowledge embodied in trade and FDI flows through bridging market information gaps. The Importance of Identifying Partners in Connectivity A fundamental prerequisite for identifying the complementarity between various forms of connectivity is the ability to identify the specific country links in the con- nectivity chain. For example, matching migration and trade flows between country partners is essential to identify the complementarity between migration and trade in enhancing economic growth. Knowing the size of overall trade and migration flows for a country is not sufficient to identify that trade from specific countries is facilitated by migration from those same countries. Mapping these direct connections between countries also brings to Being light the potential importance of indirect connections. While two similar connected countries may have the same number and size of connections, they to “well-­connected” may be connected to very different countries. Being connected to countries may provide “well-connected” countries may provide greater opportunities for greater opportunities knowledge transfers from partners of partner countries (Duernecker, for knowledge transfers Meyer, and Vega-Redondo 2014). For example, a dollar of trade from partners of between Algeria and Germany may provide greater knowledge spill- partner countries. overs than a dollar of trade between Algeria and Morocco, because Germany is much more connected to the global economy and is likely to be a source of advanced knowledge as well as a conduit for technol- ogy and knowledge from other countries it is connected to. Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  51 The importance of direct and indirect connections between countries, as well as the complementarity between various types of connections for knowledge transfers and economic growth, lends itself to the use of multilayer network ­analysis. Network analysis is simply a tool for studying the direct and indirect c ­ onnections between countries. Multilayer networks (Kivela et al. 2014) go beyond the notion of a single, one-dimensional isolated network, and provide a description of the interactions between various types of connections (layers) in a larger network. This chapter finds that being connected to well-connected countries matters for economic growth, but there is complementarity in the various types of con- nections that enhances growth as well. Countries can benefit from (a) multiple types of economic links (such as trade, investment, migration, modern telecom- munications, and transport) that underpin the movement of technologies and ideas, but also (b) the quality of connections in terms of knowledge spillovers and the indirect connections made through partners that are well connected. These are both aspects of interconnectedness that affect growth and growth spillovers. Trends in Economic Connectivity Economic links facilitate trade and the transfer of factors of production (capital, labor, and so on) and therefore has an impact on the overall level of production. However, these links are also likely to facilitate the flow of ideas and innovation, and, hence, long-run growth. But the strength of information flows is likely different for different economic connections. For example, the knowledge spillovers associated with merchandise trade are likely centered around the information embodied in the products traded (knowledge spillovers related to trade in processed food, for example, may be different from those related to trade in semiconductors) (Hidalgo and Hausmann 2009). FDI links are associated with a transfer of managerial, organizational, and corporate governance exper- tise (see chapter 2). And migration flows can facilitate the transfer of knowledge directly, but also support less tangible cultural exchange, increase exposure to foreign languages, and bridge gaps in trust in business dealings that cannot always be narrowed through explicit contracting, particularly with differences in governance and legal systems (see chapter 5). As mentioned earlier, country and regional connectivity has typically been viewed as the size of trade relative to gross domestic product (GDP) or global trade. ECA’s share of global trade has declined since the early 2000s (figure 1.1, panel a). Moreover, although intra- and extraregional trade have grown, intrare- gional trade has grown more rapidly (figure 1.1, panel b). Increased trade within ECA may reflect greater cooperation across institutional and regulatory dimen- sions with important implications for regional integration and convergence. Similar aggregate measures of connectivity (total FDI, migration, telecommunications, and others) suggest that while regional integration has increased substantially, growth of global connections may not have kept up. While past studies find evidence that greater trade openness and integration improves growth, these aggregate measures of integration may obscure the 52  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 1.1 a. ECA’s share of global trade has been falling Trends in intraregional trade 37 of total world trade (exports and imports) in ECA ECA intraregional trade as a percentage 35 33 31 29 27 25 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 b. Intraregional trade is growing faster than extraregional trade 10,000 Intra-ECA trade versus extra-ECA trade (billions of dollars) 8,000 6,000 4,000 2,000 0 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Intra-ECA Extra-ECA Source: Calculations based on data from the United Nations Conference on Trade and Development (UNCTAD) and World Bank, World Development Indicators. Note: ECA = Europe and Central Asia. underlying bilateral connections and the importance of partner country connec- tions. In other words, while ECA’s intraregional connectivity may be growing faster in the aggregate than extraregional connectivity, it is difficult to determine how well ECA countries are connecting to other ECA countries that are well con- nected globally. ECA global connectivity may actually be increasing for the aver- age ECA country, if countries in ECA are linking to strong globally connected ECA countries. The following section examines the pattern of connections between countries and how countries are connected to the broader network of countries. Subsequently, the chapter uses this wider network of connectivity information to analyze types of connectivity, how the various layers of connectiv- ity interact, and how connectivity might be associated with economic growth and shared prosperity. Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  53 Examining ECA Connectivity in the Global Context While there are potentially hundreds of ways countries can connect, with varying implications for the transfer of ideas, this chapter focuses on six types of economic connections: trade, FDI, migration, information and communication technologies (ICT), air transport, and portfolio financial flows. While other forms of connectivity may also be important for how knowledge transfers between countries and for economic growth, these data are the only ones available on a global and country- to-country basis. Subsequent chapters will drill down into additional layers of con- nectivity (for example, transport in ECA in chapter 5) as well as unique aspects of the layers of connectivity (such as firm foreign management in chapter 4). This chapter takes a broad macro view of the many layers of connectivity as a means to observe general trends in the strength of connections, how ECA countries are con- nected to each other and the rest of the world, and how these connections influ- ence growth. As an initial overview of ECA connectivity in the global context, we show graph- ically how countries in ECA and the rest of the world are connected in each layer of connectivity. We show data for the initial year and the last period available to see how connectivity in a particular dimension has changed over time. In figures 1.2–1.7, ECA countries are highlighted in shades of blue, and the rest of the world is shown in shades of orange. The size of each country node is pro- portional to the size of its total connectivity in each layer of connectivity described. Outward arrows point to the two strongest bilateral partners. The methodology for plotting the countries attempts to show clearly the connections between countries in the global network of countries.1 The largest country nodes are pulled to the outer boundaries of the graph, but the pull is counterbalanced by the number and strength of connections with partner countries. Consequently, country nodes will tend to be grouped together if they share common connections with well-­ connected countries. Trade Figure 1.2 is based on trade of manufactured goods between all the countries in the world in 2000 and 2014. The size of each country node reflects the total vol- ume of trade (exports and imports) for each country, and the two outward arrows are pointed to each country’s two top export destinations. One of the most dra- matic developments in global trade has been the emergence of China (CHN) since 2000, which has not only grown to one of the three largest traders, along with the United States (USA) and Germany (DEU), but also is pulled toward the center of the graph, with numerous connections to regional hubs in Europe, the Americas, and Asia. ECA’s relative dominance in trade has declined, along with other regions, as China and other Asian countries’ share of global trade has increased. Interestingly, however, ECA country nodes are much closer to each other in recent years than in the past, reflecting the higher degree of regional integration and value chain development in Europe. Germany is the primary hub for ECA’s manufacturing integration. The United Kingdom (GBR) is pulled equally between Europe and the United States, and, hence, is located almost equidistant between these two poles. 54  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 1.2  Exports of manufactured goods a. 2000 b. 2014 Source: Calculations based on data from the United Nations Conference on Trade and Development. Note: The size of each country node reflects the total volume of trade. Each node has two outgoing links, which point to the country’s two top export partners. Europe and Central Asia countries are shown in shades of blue. Foreign Direct Investment Figure 1.3 is based on total stocks of FDI (inward and outward) for all the countries in the world. Each country node shows two outward arrows that are pointed to each country’s two top FDI destinations. Unlike the international trade network, global FDI remains dominated by the developed countries in Europe and the United States, although developing countries have seen some modest increase. Much FDI moves between countries of similar levels of development, with rela- tively modest investments going from developed to developing countries and vice versa. China and the rest of Asia were the beneficiaries of significant incoming investment flows between 2000 and 2012. China, on the other hand, has focused its outgoing investments toward the United States and neighboring countries. For ECA, the distribution of FDI is also less regionally focused than trade. FDI appears to be connected to language and historical colonial linkages (for example, francophone African countries largely share in FDI flows with France and Belgium) as well as driven by corporate acquisitions for technology, transport, access to markets, and natural resource endowments. Financial centers (e.g., the United Kingdom and Switzerland) have also become increasingly important for attracting FDI. Interestingly, compared to trade, Germany’s and China’s participation in global FDI is small, but while China’s share has grown, Germany’s relative FDI stocks have fallen. Nonetheless, ECA’s overall participation in global FDI increased from 2002 to 2012, and has Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  55 FIGURE 1.3  Foreign direct investment a. 2002 b. 2012 Source: Calculations based on data from the Organisation for Economic Co-operation and Development and fDiMarkets.com. Note: The size of each country node reflects the country’s total foreign direct investment (FDI) stocks (incoming and outgoing). Each country has two outgoing links that point to its two main FDI destinations. Europe and Central Asia countries are shown in shades of blue. become more equally distributed as Spain (ESP), Belgium (BEL), the Russian Federation (RUS), and Sweden (SWE) have seen relative increases. International Migration Perhaps more than any other connectivity layer, international migration is domi- nated by the United States, which is the main recipient of migrants in the world (figure 1.4). Although the importance of China in this network increased between 2000 and 2010, China’s migration connectivity in the global network is significantly lower than its other connections. Russia is a particularly large center of migration in ECA, but this is primarily a legacy of the breakup of the Soviet Union. Individuals that were born in former Soviet Republics living in Russia are classified as foreign born, although at the time of birth they were nationals of the same country as Russian natives. Nonetheless, Russia remains an important destination for migrants from Central Asian countries, such as Tajikistan (TJK) and Uzbekistan (UZB). Remittance flows generated by these migrant workers living in Russia account for a substantial share of income for many Central Asia economies (in some cases, more than 30 percent, see chapter 4). The share of immigrant populations among European countries appears to have increased from 2000 to 2010. The region is more integrated in its labor market, as evidenced by the somewhat higher clustering of European countries in ­ the last year, reflecting easing of immigration rules in the EU under the Schengen Agreement. It is interesting to note that Germany was a large recipient of migra- tion flows from Russia and other communist bloc countries during the first decade after the breakup of the Soviet Union; however, in the figure for 2010 Germany is 56  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 1.4  Migration a. 2000 b. 2010 Source: Calculations based on data from the Organisation for Economic Co-operation and Development. Note: The size of each country node represents the total number of foreign-born individuals residing in the country plus the total number of native-born citizens living outside the country. Each country node has two outgoing links that represent the country’s two largest emigration destinations. Europe and Central Asia countries are shown in shades of blue. pulled much closer to the center of EU countries. Likewise, Poland after joining the EU has closer migration linkages to Germany and the United Kingdom, compared to its connections to Russia and the United States. In general, migration flows are strongly influenced by language similarities (e.g., Romanians living in Italy and Spain), proximity (e.g., the United States, Canada, and Mexico), and historical colonial ties (e.g., France and North Africa; the United Kingdom and Australia). Passenger Airline Connectivity The bilateral airline connectivity shown in figure 1.5 represents not simply the number of flights between countries, but the origin and final destination of pas- sengers, which requires information on passenger itineraries. Oftentimes passen- gers utilize hubs and transfer to other flights and airlines before reaching their destination, which, without data on itineraries, can overweight hubs as being the final passenger destination and underweight countries that connect to the global network of countries through hubs. These data were painstakingly estimated by the International Civil Aviation Organization, using flight and itinerary information to build a data set for passenger flight origins and final destinations. However, despite these efforts, private flights are not always included in available data and hubs may still be overrepresented as the final destination for air passengers. What appears from the data is that, similar to the migration network, the links among European countries increased substantially between 2000 and 2012 as shown by the increased clustering of ECA countries in the later period. Moreover, Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  57 FIGURE 1.5  Airline connectivity a. 2000 b. 2012 Source: Calculations based on International Civil Aviation Organization data. Note: The size of each country node represents the estimated total number of air passengers. Each country node has two outgoing links that point to the country’s two largest passenger flight destination countries. Europe and Central Asia countries are shown in shades of blue. ECA high-income countries, particularly Germany, the United Kingdom, France, Spain, Italy, and Turkey, have increased in importance in airline passenger origin and destination countries. Since 2007, Europe’s direct connectivity gains within the region have been driven by regional integration policies and the subsequent expansion of regional short-haul low-cost carriers. Meanwhile, full-service carriers have seen their regional direct connectivity drop with the greater competition. Low-cost carriers now account for nearly a third of Europe’s direct connectivity and are focused on linking airports within the intra-European market. The lion’s share of Europe’s direct connectivity to other world regions is still held by full-service carriers. Outside of ECA, rapid economic development made China an attractive desti- nation for international flyers; China overtook Japan’s role as the top airline flight destination in Asia in 2012. The development of popular Middle East Gulf carriers (e.g., Qatar and Emirate Airlines) may indicate a country bias in the data due to their importance as regional hubs, but it is also reasonable that they are attracting greater final destination air traffic due to broader investments to diversify their economies away from oil. International Internet and Communications Technology The international ICT global flows network is shown in figure 1.6. The data are constructed by using total country internet bandwidth (bilateral country data are not available) and allocating bilateral traffic in proportion to bilateral telephone calls for each country. ICT flows appear to be clustered in three groups: Europe, 58  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 1.6  Internet and communication technologies a. 2002 b. 2010 Source: Calculations based on TeleGeography data. Note: Each country node represents the combined value of the estimated incoming and outgoing information and communication technology communication and has two outgoing links that point to the country’s two main outgoing communication partners. Europe and Central Asia countries are shown in shades of blue. North America, and Asia. The United States and the United Kingdom are the major hubs in this network, with a notable increase in the importance of India in 2010 due to the back-office outsourcing of service jobs and call centers. It should be noted, however, that data on the network in the latter period are not as complete as in the early period and should be interpreted skeptically. Nonetheless, intuitive regional patterns persist with connections driven by language, supply-chain link- ages, and economic activity. Portfolio Financial Flows The portfolio financial flows are derived from the Bank for International Settlements Consolidated Banking Statistics.2 The Consolidated Banking Statistics capture the worldwide consolidated positions of internationally active banking groups headquar- tered in the Bank for International Settlements reporting countries. Portfolio financial flows appear to be driven by the largest financial centers, without a strong relationship to underlying trade or other economic relationships (­ figure 1.7). The six top centers include Germany, the United Kingdom, Japan, France, the United States, and Switzerland. Some relationships are economically intuitive, with many ECA countries having at least one top portfolio flow connection in the ECA region and the second in the United States. In 2010 Spain’s two top connections included Germany and the United States, reflecting integration within ECA and outside, while in 2000 it was Mexico and the United States. In other words, ECA ties became relatively stronger. Nonetheless, even within ECA, because of the agglomeration benefits of financial sectors and the practice of many companies to issue many companies’ practice of Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  59 FIGURE 1.7  Portfolio financial flows a. 2000 b. 2010 Source: Calculations based on Bank for International Settlements Consolidated Banking Statistics. Note: Each country node represents the combined value of portfolio inflows and outflows and has two outgoing links that point to the country’s two main recipients of portfolio financial flows. Europe and Central Asia countries are shown in shades of blue. issuing portfolio financial bond or equity instruments in well-established markets where market size and transparency help to stimulate supply and demand, financial flows do not particularly match the level of real economic relationships. The concentra- tion of portfolio flows centered in a few country nodes, however, provides some insight into how vulnerable portfolio flows may be to shocks in the central nodes. In summary, Europe’s integration policies have had a positive impact on internal European connectivity through most economic relationships, especially in trade, migration, and air passenger transport, but less so in FDI, ICT, and portfolio finan- cial flows. In the early 2000s, there were strong migration patterns between the transition economies and Northern, Western, and Southern Europe (particularly Russia and Germany), which then diverged into two regional blocks, one centered around Russia and the other the EU, with Germany as the primary country node. Transitional European countries trade and migrate intensively within Europe, but are increasingly creating linkages with the rest of the world. Established (advanced) European countries have had wider global connections with the United States and Asian countries, but regional connections are deepening. Overall, ECA’s relative importance as a central node for connectivity with the rest of the world has fallen as emerging economies (especially China) are growing and account for a larger share of global economic activity. This is true for most advanced countries, 60  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia including the United States, as emerging economies are increasing in economic size and wealth. Perhaps not surprisingly, airline and ICT connectivity have changed the most over the past decade because of deregulation and innovation, while portfolio financial flows have tended to be concentrated in a few dominant finan- cial sectors that have changed only slightly. Multidimensional connectivity network analysis adds to the previous research on economic relationships and their influence on growth by not looking at one network layer independently of others, but by examining the many layers of con- nectivity together. Not only do individual connections matter but so do their interdependence in economic relationships. Connectivity should be seen as a multidimensional concept including trade, migration, finance, transport, com- munications, and other factors. Greater connectivity in one area may be a complement to or substitute for connectivity in another area. ­ Connectivity and Income Growth According to traditional growth models, an increase in trade or other forms of connectivity will have no impact on long-run income growth. The level of income will increase due to gains from specialization, but this will not lead to sustained increases in growth unless it has an impact on improving technologi- cal accumulation over time (i.e., the endogenous growth model, Romer 1990). Thus, the main mechanism through which connections affect growth is the transfer of knowledge and innovative ideas and technologies. Innovations are continuously generated globally, and they travel the world through the net- work of countries. Greater multidimensional connectivity increases the proba- bility that an economy will absorb these new ideas and increase long-run growth. The empirical strategy we use for understanding how a country’s international connections and the interplay of these connections influence economic growth is threefold. First, we simply estimate a baseline growth model that includes stan- dard explanatory variables, including the initial GDP per capita level, schooling, size of government, inflation, quality of governance, and investment rates. Second, we include the traditional measure of connectivity, trade/GDP, that is used in the economic literature on openness. Our interest is not so much replicating previous research, but rather determining a benchmark against which to compare network effects of connectivity. Third, we develop network centrality measures for each type of connection indi- cator (for example, trade and FDI) based on a modified Google PageRank algo- rithm (Page et al. 1999). This algorithm gives a higher ranking to countries that have a larger number of connections to well-connected countries as well as connections to countries with a high “intrinsic value.” Intrinsic value in our ­ context means a high propensity to generate and disseminate knowledge. We use the size of the coun- try’s population and GDP per capita as proxies for this intrinsic value. We modify and expand the analysis in Duernecker, Meyer, and Vega-Redondo (2014) of the relationship between a network measure of trade and economic growth to other measures of connectivity (trade, FDI, migration, airline transport, Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  61 portfolio flows, and ICT). We compare our six individual network centrality results with the relationship between traditional measures of connectivity (for example, overall trade to GDP) and growth to determine whether network centrality mea- sures are any better at describing long-run growth than the standard, nonnetwork, measures. Finally, we develop a comprehensive measure of overall network centrality, referred to as multidimensional connectivity, that combines all six types of connectivity into a single network measure. This indicator takes into consider- ation the complementarity of the various forms of connectivity, as described in the introduction. Multidimensional connectivity is found to Multidimensional be significantly related to long-run growth, and provides a better connectivity . . . provides explanation of long-run growth than the individual connectivity a better explanation of long-run growth than channels. In other words, the whole is greater than the sum of its the individual connectiv- parts. As a robustness check, we develop an alternative index of ity channels. In other network centrality, multiplex connectivity, that describes the com- words, the whole is plete network but does not impose the restriction that each layer of greater than the sum of the network is a complement to other layers. This indicator has a its parts. similar, albeit less strong, relationship with growth than the multidimen- sional connectivity indicator. Network Centrality This section introduces a measure of centrality, or influence, based on the well- known Google PageRank algorithm, which was used to rank websites based on their links in the network. The algorithm was initially developed to rank websites in terms of their “importance” and “relevance” to a search query. Network analy- sis was a natural starting point for this problem because websites with more hyperlinks pointing at them were thought of as being of higher quality. In addi- tion to the number of the incoming links, having more links from higher-quality websites is yet another indicator of website quality. The innovation by Page et al. (1999) consisted in modifying the popular network eigenvector centrality mea- sure so that the centrality value of a website was proportional to the probability that a person clicking randomly on hyperlinks would land on that page.3 Or more precisely, the PageRank value reflects the share of visits to the website by a ran- dom web surfer over some period of time. We modified the PageRank algorithm so that its initial idea—capturing the probability that a random traveler in the network will arrive at a certain node— remains in place. In the economic network discussed in this chapter, the connectivity index is proportional to the probability that a random economic or technological innovation will reach the country. This probability reflects the likelihood that an innovation will be transmitted to that country through each form of connectivity (trade, FDI, and so on), based on the country’s links to other countries, those coun- tries’ links to other countries, and so on (the value of connections is progressively reduced by 15 percent at each link in the chain).4 The index also reflects the intrin- sic probability that each connection (country) will innovate and disseminate knowl- edge independently (proxied by population and GDP per capita).5 The formulation of the centrality measure is shown in annex 1D. 62  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Network Centrality Measures and Growth In this section, we estimate the importance for growth of the connectivity mea- sures described in the previous section. We first estimate a standard cross-country, long-run model over 2000–16, where growth depends on the initial levels of GDP per capita, education, investment rate, governance, government size, and infla- tion. We then add trade/GDP, the traditional measurement of openness, as an additional explanatory variable. Finally, we add each of our network measures of connectivity to determine whether network centrality measures are any better at describing long-run growth than the standard, nonnetwork measure. Two adjustments are required to the network connectivity measures described above before including them in the model. First, we scale the value by popula- tion to account for the fact that more populous countries are expected to depend less on being connected to the rest of the world for innovations and growth than smaller countries, which, due to their size, naturally rely more on connectivity (e.g., China vs. Singapore). This has the effect of transforming connectivity into per capita terms. Second, because the network connectivity measure includes the country’s own level of GDP per capita (as well as GDP per capita of partner countries), we subtract the country’s own level of GDP per capita from the con- nectivity measure because it is already included as an explanatory variable in describing country growth. This, in effect, eliminates double counting. The intrin- sic value of partner countries’ GDP per capita is still included in the network connectivity measure. The estimation of the relationship between growth and the variables typically used in the empirical literature (and included here) faces several key challenges. Perhaps the most difficult concerns endogeneity, often reflecting reverse causality, or the influence of the dependent variable (growth) on some of the independent variables (e.g., government size). Our main goal is to measure the contribution of our network connectivity measures to growth, after controlling for other variables (inflation and so) on thought important to growth. However, both our connectivity measures and these other variables may themselves be determined, in part, by growth (they may not be exogenous, as assumed in our estimation procedure). Thus, most researchers using cross-sections are only able to capture partial correla- tions instead of causality. Our identification strategy attempts to reduce problems from endogeneity, although it does not eliminate them. First, we calculate the right-hand side vari- ables by taking the earliest observation available in the data at the start of the growth period. This of course does not correct all potential endogeneity prob- lems, but it is indicative of a lack of reverse causality. Second, our measures of connectivity build on direct and indirect links for the various types of connectivity in the global network, and countries are only able to impact direct links and not indirect ones. By taking into consideration higher-order indirect links, our con- nectivity measures are at least to some extent exogenous, or unaffected by growth of the country being measured. Moreover, for robustness we also include geographic distance between countries as a separate layer of connectivity to account for geographic proximity that may affect growth and the strength of connectivity channels simultaneously.6 For a deeper treatment of endogenous Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  63 relationships in economic growth we refer the reader to the rich literature (see Frankel and Romer 1999; Rodriquez and Rodrik 2000; Beck 2008; Helpman 2004; Feyrer 2009; and Panizza and Presbitero 2013). This kind of estimation also may suffer from the existence of unobserved coun- try effects (which are potentially correlated with the independent variables used in the empirical model). Furthermore, most variables are measured with considerable error. Since developing countries represent a large fraction of our sample, results depend on the reliability of the data. Hence, measurement error can be a source for inconsistent coefficient estimates. We examine the growth effects of connectivity along each network layer sep- arately (trade, FDI, migration, ICT, airline connectivity and portfolio flows) in table 1.1. Compared to the base model and the standard measure of openness (trade/ GDP), nearly every network connectivity measure manages to increase the explan- atory power (Adj-R2) of the standard growth equation, although not every network layer is statistically significant at the minimum 10 percent level. Deeper integration along each individual dimension is associated with stronger per capita GDP growth over the subsequent 16-year period. Unlike the traditional measure of openness and connectivity (trade/GDP), the PageRank-based index, which was designed to capture the knowledge spillovers from connections, is associated with higher long- term growth in the case of international trade, FDI, migration, and airline connec- tivity. A one-standard-deviation increase in the trade connectivity of a country is associated with more than half a percentage point (0.6 percent) higher annual economic growth over the long term. The effect of FDI connectivity is similar (0.59 percent), and the effects of migration and airline connectivity are markedly lower (0.34 percent and 0.19 percent, respectively). TABLE 1.1  Connectivity Effects on Overall Income Growth (1) (2) (3) (4) (5) (6) (7) (8) GDP per capitat=0 −0.91*** −1.10*** −1.31*** −1.29*** −1.15*** −1.06*** −1.11*** −1.11** Years of schoolingt=0 2.46*** 2.4*** 1.7*** 1.87*** 1.60*** 2.06*** 1.99*** 2.04*** Government sizet=0 −9.24** −8.64** −5.81 −5.24 −4.91 −5.44 −5.21 −5.79 Inflationt=0 1.02 0.99 2.92 2.75 2.54 2.1 1.82 1.73 Governancet=0 1.18 2.13 1.2 0.99 1.04 1.43 1.32 1.63* Investment ratet=0 0.160** 0.170*** 0.190*** 0.21*** 0.20*** 0.20*** 0.20*** 0.20*** Baseline Standard Connectivity model  Trade/GDPt=0 0.28 Network effects (PageRank) Trade Connectivity per capitat=0 0.61***   FDI Connectivity per capitat=0 0.59***   Migration Connectivity per capitat=0 0.34*   ICT Connectivity per capitat=0 0.12   Portfolio Flows per capitat=0 0.17   Airline Connectivity per capitat=0           0.19* Adjusted R2 0.54 0.53 0.59 0.58 0.56 0.54 0.55 0.56 Note: The dependent variable in each model is the annualized income growth (in percent) between 2000 and 2016. All right-hand-side variables are transformed in logs, and the first available observation for the growth period is taken. There are 111 countries for which each version of the model can be estimated. The connectivity variables/PageRank are normalized using the standard normal distribution. Therefore, the size of the coefficient represents the growth impact of a one-standard-deviation change. All model specifications include an intercept, which is not reported in the table. All coefficients are estimated with ordinary least squares regression. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. 64  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Connectivity can also boost shared prosperity. Economic channels by which the poor and bottom 40 percent may directly benefit from greater connectivity include improved access to finance and markets, changes in Connectivity the return to capital or labor, exposure to technology and better gover- can also boost nance, and changes in the relative prices of goods and services. Trade, shared prosperity. . . . for example, may enhance resource allocation across countries leading The poor and to improved opportunities for asset use by the bottom 40 percent of bottom 40 percent the income distribution. Investment flows may generate new returns may directly for the bottom 40 percent. As production becomes more competitive, benefit from the poor may also experience a mix of welfare gains and losses from greater relative price changes. Migration may open new opportunities, but also connectivity. has implications for the labor market. Connectivity influences commerce and investment, but it also is a means for transferring ideas, technology, and institutional arrangements, which are all potential sources for spillovers to growth and may indirectly influence shared prosperity. Table 1.2 summarizes the estimated impact of connectivity on the income growth of the poorest 40 percent of the income distribution in each coun- try. Trade connectivity has the largest impact. In fact, the knowledge spillover effects from trade appear to be more important for the bottom 40 percent than for the top 60 percent. However, the other measures of connectivity do not appear to play a statistically significant role in bottom-40 growth. Multidimensional Connectivity: Interplay of Network Connections and Growth In this section, we develop two unique methods for combining each individual network layer (trade, FDI, migration, ICT, airline transport, and portfolio flows) into TABLE 1.2  Connectivity Effects on Bottom-40 Income Growth (1) (2) (3) (4) (5) (6) (7) (8) GDP per capitat=0 −0.77** −0.69* −1.19*** −1.08*** −0.69*** −0.71* −0.75** −1.77*** Years of schoolingt=0 2.1*** 2.1*** 2.32*** 2.39*** 2.15*** 1.91** 2.65*** 2.51*** Government sizet=0 −7.37 −7.38 −3.61 −9.05 −10.38 −6.4 −8.05 −5.72 Inflationt=0 2.61 2.62 5.69 6.21 5.08 5.23 4.01 0.96 Governancet=0 1.13 1.14 −0.74 0.75 1.37 1.5 2.18 0.3 Investment ratet=0 0.09* 0.08* 0.08* 0.13** 0.09* 0.12** 0.09* 0.09* Baseline Standard Connectivity model  Trade/GDPt=0 0.02 Network Effects (PageRank) Trade Connectivity per capitat=0 1.49**   FDI Connectivity per capitat=0 0.8   Migration Connectivity per capitat=0 0.18   ICT Connectivity per capitat=0 0.21   Portfolio Flows per capitat=0 −0.13   Airline Connectivity per capitat=0           0.11 Adjusted R2 0.24 0.24 0.28 0.26 0.25 0.23 0.23 0.23 Note: The dependent variable in each model is the annualized bottom-40 income growth (in percent) between 2000 and 2016. All right-hand-side variables are transformed in logs, and the first available observation for the growth period is taken. There are 88 countries for which each version of the model can be estimated. The connectivity variables/PageRank are normalized using the standard normal distribution. The size of the coefficient therefore represents the growth impact of a one-standard-deviation change. All model specifications include an intercept, which is not reported in the table. FDI = foreign direct investment; ICT = information and communication technology. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  65 TABLE 1.3  Correlation between Connectivity Layers Is High, Except in the Case of Portfolio Financial Flows Trade FDI Migration ICT Airline Portfolio flows Trade 1 FDI 0.9295* 1 Migration 0.7173* 0.7092* 1 ICT 0.7107* 0.7882* 0.6789* 1 Airline 0.8515* 0.9090* 0.6200* 0.8348* 1 Portfolio flows 0.2560* 0.2751* 0.2624* 0.2286* 0.2697* 1 Note: FDI = foreign direct investment; ICT = information and communication technology. Significance level: * = 10 percent or higher. a single total network measure of connectivity to address the complementarity between connectivity measures and their relationship to growth. Indeed, there appears to be a strong correlation between all measures of connectivity, with per- haps the exception of portfolio financial flows (table 1.3). FDI and trade are the most correlated across connectivity layers, airline transport and migration less so, while portfolio flows is highly idiosyncratic. Intuitively, interplay between various forms of connectivity can be seen most clearly in migration and international travel. Much research has found that migration and trade tend to be complements (greater migration between two countries is associated with greater trade between them), and subsequent research has also identified migration’s importance in influ- encing FDI and its direct influence on growth through knowledge transfers (see, for example, Gould 1994 and Onodera 2008). Thus, people-to-people contact may be important for growth by directly transferring knowledge between the host and home countries, as well as indirectly by facilitating knowledge embodied in trade and FDI flows through bridging market information gaps. The six network connectivity measures could be aggregated in a simple, ad hoc way (for example, taking averages of the network centrality measures). However, this is likely to result in a loss of important information and would not account for the interaction of various network layers and their effect on economic growth. For example, vastly different bilateral connectivity patterns in each dimension can result in similar average values (box 1.1). We therefore adopt a somewhat more intuitively appealing procedure for aggregating the connectivity measures. This includes calculating the weighted multiplicative average of the separate connectivity measures.7 Essentially, the six networks are collapsed into one network where each bilateral link is a function of each of the layers, as shown in figure 1.8. The functional form used has several desirable features that have been well documented in other economic contexts. First, it imposes decreasing returns to scale, that is, having a large amount of one type of connection provides the country with decreasing informational returns, with other forms of connectivity held constant. A balanced increase in connectivity along each dimension would have a stronger impact on the bilateral informational link than a rapid increase in the connectivity along one layer only. It is very likely that these different channels complement each other in terms of the information they transmit. For example, a foreign investor is likely to be more successful in transferring know-how in the host country if there already are deep links through migration and trade that can complement the information flows embedded in FDI. 66  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia BOX 1.1 A Better Way of Measuring Network Connectivity Figure B1.1.1 provides two examples of network and the overall network for country A are vastly dif- connectivity and the modified PageRank of country A. ferent between the two cases. It is easy to show In the left and right panels of figure B1.1.1, that using aggregation at the country level, the country A has the same centrality index calculated modified PageRank used in this study produces a using simple averages of modified PageRank cen- higher centrality for country A in the case in the trality across three types of networks, represented right panel compared to the one in the left panel. by the three types of arrows (line, dash, and dot). This is an intuitively more appealing method than It is clear, however, that the patterns of connections simple averaging. FIGURE B1.1.1  Examples of network connectivity and the modified PageRank B C B C A A D D Furthermore, the estimated weights on each of the network layers can be inter- preted as the efficiency or importance of each channel in transmitting information that facilitates long-term income growth. Each country’s aggregate connectivity index, representing the likelihood of a country adopting an innovation, is then calculated in a fashion similar to that used in calculating the individual connectivity indexes. That is, the aggregate index of connectivity is summed across partner countries and added to the likelihood of the country generating an innovation independently (represented by GDP).8 The impact of multidimensional centrality on growth and the indicator’s com- ponent weights for each network layer are estimated simultaneously using a maximum likelihood procedure. The estimated weights for each layer of the multidimensional connectivity indicator and the indicator’s impact on growth are shown in table 1.4. The growth impact of the multidimensional connectivity indi- cator is higher than each of the individual network indexes (shown in table 1.1). A one-standard-deviation increase in the multidimensional connectivity indicator is associated with 0.67 percent higher annualized growth. These results suggest that the overall connectivity profile of the country (one that combines all network layers) is more important than each of the individual layers separately. Moreover, in the combined network, trade has the highest importance, followed by FDI, Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  67 FIGURE 1.8 Multidimensional connectivity network Trade network FDI network N-network MDC network Note: FDI = foreign direct investment; MDC = multidimensional connectivity; N-network = other measured global networks. TABLE 1.4  Multidimensional Connectivity   Overall growth B40 growth Multidimensional connectivity impact 0.67*** 1.49*** Efficiency exponents/weights of connectivity channels Trade channel efficiency 0.532 1 FDI channel efficiency 0.37 0 Migration channel efficiency 0.1 0 ICT channel efficiency 0 0 Airline channel efficiency 0 0 Portfolio flows 0 0 Adjusted R2 0.61 0.28 Note: The dependent variable in each model is annual income growth (in percent). All right-hand-side variables are transformed in logs. The PageRank coefficient is standardized to represent the effect of a change of one standard deviation. The values of the exponent parameters (efficiency exponents/ weights) α, β, γ, and δ were estimated using the maximum likelihood procedure where the objective function was to maximize the goodness-of-fit measure (adjusted R2). B40 = bottom 40 percent of the income distribution. FDI = foreign direct investment; ICT = information and communication technology. Significance level: *** = 1 percent in an ordinary least squares regression. and then migration. Neither ICT, airline transport, nor portfolio flows add addi- tional information above these three connectivity channels. By contrast, the multi­ dimensional measure does not add new information above the single network measure of trade connectivity in explaining changes in the growth of the incomes of the bottom 40 percent of the distribution. 68  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Figure 1.9 is based on the values of each country’s multidimensional connectivity index in the overall growth model. As the figure indicates, multidimensional con- nectivity shows a much stronger cohesion between ECA countries than any single network connection and these connections grew from 2000 to 2014. Of all the ECA countries, the United Kingdom shows the strongest overall linkages within ECA and non-ECA countries. In contrast, Germany is the strongest overall connector between ECA countries, but has few strong links outside of ECA. Interestingly, while China has increased network linkages with the world, it is much smaller and less connected compared to only the trade network as indicated in figure 1.1; as a result, its impor- tance to the global network is about the same as Germany’s, but less than Japan’s. In terms of per capita levels in ECA subregional multidimensional connectivity (table 1.5), Western Europe has the highest global ranking, followed by Northern, Central, and Southern Europe, while Russia, Turkey, and Eastern Europe are in the middle range, and the Western Balkans, Central Asia, and the South Caucasus have the lowest levels of overall connectivity. Not surprisingly, levels of connectiv- ity are associated with higher levels of development. Interestingly, although Central Asia and the South Caucasus rank relatively low on overall connectivity, they also saw the greatest improvement from 2000 to 2014 (figure 1.10). The South Caucasus has seen connectivity increase by nearly 75 ­percent, while Central Asia has seen connectivity increase by more than 40 ­ percent. Eastern Europe and the Western Balkans, although also starting from relatively low levels, have not seen as rapid an increase, with connectivity increasing only 20 and 10 FIGURE 1.9  Multidimensional network connectivity a. 2000 b. 2014 Note: The size of the node represents the multidimensional connectivity index of each country. Each node has two outgoing links that point to the strongest two connections in the multidimensional network according to the overall growth model (table 1.4, column 1). Europe and Central Asia countries are shown in shades of blue. Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  69 percent, respectively. The key challenge for these regions is to find ways to improve balanced connectivity, particularly easing constraints and facilitating trade, FDI, and airline and ICT connectivity. For the ECA region as a whole, connectivity has improved more than global connectivity, reflecting the integration process of the EU as well as strides taken in transition economies. The analysis of multidimensional connectivity and its relationship to economic growth can be useful in evaluating where countries can benefit the most in terms TABLE 1.5  Multidimensional Connectivity Varies by ECA Subregion, with the Highest Connectivity in the Western Part of the Region and the Lowest in the Eastern Part Multidimensional ECA subregions connectivity Trade FDI Migration ICT Airline Portfolio flows High connectivity Western Europe 6 6 6 9 9 15 19 Northern Europe 12 12 17 26 21 23 22   of which Baltics 30 28 36 38 50 28 21 Southern Europe 25 24 26 21 28 23 22 Central Europe 31 27 34 36 41 46 46 Medium connectivity Russian Federation 55 53 61 28 63 64 83 Turkey 57 51 67 33 73 79 40 Eastern Europe 62 59 60 81 54 57 76 Low connectivity Western Balkans 88 75 97 45 88 86 99 Central Asia 94 99 93 101 101 103 101 South Caucasus 104 104 102 64 104 104 93 Note: The table shows global rankings, from best to worst, in combined per capita connectivity, with lower values indicating better connectivity. Subregion indicators are median values of the subregion’s countries ECA = Europe and Central Asia; FDI = foreign direct investment; ICT = information and communication technology. FIGURE 1.10  Europe and Central Asia's connectivity has grown, but there are wide variations across subregions Growth in connectivity, percent, 2000–14 80 70 60 50 40 30 20 10 0 Central Central Eastern Northern Russian South Southern Turkey Western Western Global Asia Europe Europe Europe Federation Caucasus Europe Balkans Europe Note: Subregional and global indicators are median regional averages. 70  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia of reducing barriers to entry and facilitating linkages with well-connected coun- tries. It can also help identify which connections are likely to have the largest impact on growth. For example, in China’s case, while the trade network is strong, the migration and FDI networks are weak in comparison. Likewise, taking the case of Kazakhstan, increasing FDI in Bulgaria would bring a higher increase in multidimensional connectivity than increasing FDI in Poland, despite Poland’s greater overall connectivity. This is due to the higher complementarity of Kazakhstan’s FDI with other preexisting connections with Bulgaria compared to Poland (see box 1.2). BOX 1.2 Example of Using Connectivity Measures for Investment Decisions Assume that a country like Kazakhstan would like to integrated in European global value chains, and in use its national sovereign wealth fund to invest particular into Germany’s manufacturing industries. US$100 million of its income from natural resources By virtue of its strong ties with the Western European in Central Europe. Assume also that the risk- economies, Poland has one of the highest overall adjusted rate of return in the region has been equal- connectivity indexes in ECA, and the benefits of ized by the market. Consequently, the g ­ overnment connecting with it are significant (table B1.2.2). decides to choose a strategic destination for its Somewhat counterintuitively, Kazakhstan achieves investment, which would create future knowledge the highest connectivity increase by investing in spillovers and innovation transfers. Table B1.2.1 lists Bulgaria and not in countries with better integration the potential markets and their connectivity indexes. in the global network such as Poland and Hungary. Not surprisingly the country with the highest There are two reasons for this result. First, having connectivity index is Poland. Poland is well a balanced connectivity portfolio is superior, TABLE B1.2.1  Potential Markets and Their Connectivity Indexes Potential country in which to place investment Multidimensional Connectivity Index Poland 0.29 Hungary 0.27 Czech Republic 0.26 Romania 0.25 Bulgaria 0.21 Slovenia 0.20 TABLE B1.2.2  Kazakhstan’s New Multidimensional Connectivity Index after Investing $100 Million Each in Various Markets Change in Kazakhstan’s Multidimensional Country Connectivity Index (percent) Bulgaria .00735 Poland .00525 Czech Republic .00523 Hungary .00519 Slovenia .00515 Romania .00510 continued Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  71 BOX 1.2 Example of Using Connectivity Measures for Investment Decisions continued in terms of knowledge spillovers, to being well con- economic growth. This translates into US$1.07 nected in only one dimension at the expense of the m ­ illion higher annual income. Although this amount others. For example, migration may bridge informa- is negligible in terms of overall income growth, it is tion between countries and stimulate other types of an additional 1.07 percent spillover return on the economic connections, such as external investment, original investment from better connectivity. trade, and communications linkages. Kazakhstan Examining this from another direction, figure has relatively stronger ties to Bulgaria in terms of B1.2.1 summarizes the improvement in Kazakhstan’s trade and migration but less so in terms of FDI. connectivity rank under several different scenarios. In Therefore, increasing its FDI generates more the first scenario Kazakhstan increases each of benefits because of the complementarity of the its connections in trade, FDI, and migration by channels. 20 percent with every country in the world and Second, although richer countries have more improves its centrality by four places in the global knowledge than poorer ones, that knowledge is ranking. However, if the same amount of increased more difficult to reach. A moderate investment in a trade, FDI, and migration is focused on bilateral con- small economy can tap into a greater share of the nections with Germany, for example, Kazakhstan’s potential of that economy than the same invest- overall connectivity ranking improves by eight places. ment in a large one. Likewise, its ranking increases with a focus on greater Choosing Bulgaria over Poland would generate connectivity with China and the Russian Federation over the long term 0.0021 percent greater as well, but by slightly less. FIGURE B1.2.1  Kazakhstan’s connectivity ranking change Ranking change resulting from 20 percent increase in connectivity vis-à-vis the world and selected countries 35 30 25 20 15 10 5 0 World Germany China Russian Federation Multidimensional Trade FDI Migration Note: FDI = foreign direct investment. As a robustness test of the multidimensional connectivity indicator used above, a second approach to calculating multidimensional connectivity is evaluated using the recent techniques in the study of multiplex networks, where the functional form of the relationships is unknown. These multiplex networks do not rely on col- lapsing the network into a single layer and do not restrict the functional form (i.e., they do not rely on the Cobb-Douglas functional form as used earlier). A ­ benefit is that the functional form need not be assumed; the cost is that 72  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia economic theory suggests a particular relationship (i.e., complementarity if ­ between network layers), this information is not used. Multiplex networks are observed in all types of complex systems, including economic, social, biological, infrastructure, and socio-technical systems. For example, the air transportation system is a socio-technical system that exhibits many layers, all of them contributing to and essential for the overall functioning of the system. Interdependence among different layers of the air transportation sys- tem arises naturally because different airlines use the same airports. As a conse- quence, if an airport is closed, all the flights coming into and out of it stop for all the airlines. Another aspect of interdependence arises because for a flight to take off, it needs both a crew and a plane. Similarly, banks are also connected (in often very complex ways) by their derivatives positions, by overlap of portfolio composi- tion, by joint exposure to the same creditors, and so on. In other words, a multi- layer network model of each system is essential to estimate the degree of resilience of the entire system to random events or attacks against some of its parts. The aim, then, would be to study potential contagion effects via multiple channels in first attempts at modeling multidimensional network structures. For example, a description of the financial system as a three-layer multilayer network, composed of layers representing financial activities for funding, collateral, and assets, has been recently proposed by Bookstaber and Kenett (2016). Similar to the measure of multidimensional connectivity described above, we consider the multilayer network of the individual flow networks. In this approach, we examine the multilayer network as a whole, and do not collapse the different flow layers one on another. Instead, we follow the approach developed by Rahmede et al. (2017) to calculate the Multiplex PageRank centrality (see annex 1C). This approach assigns a measure of centrality to each country based on its con- nectivity across all the layers put together. A country’s centrality is measured by assigning a score based on its connectivity in one flow-defined layer, and by assigning a score to the overall importance of a given economic flow-defined layer. These two scores are calculated simultaneously and are codependent. Following the approach described above, we repeat the regression analysis using the same dependent and independent variables in the standard growth model, but instead use the standardized Multiplex PageRank centrality measure. This results in a statistically significant (albeit smaller) coefficient of the multiplex connectivity measure of 0.39 (p-value = 0.02, adj. R2 = 0.534). This alternative methodology confirms the importance of combining the multiple ways countries can connect, rather than simply focusing on one connection layer at a time, par- ticularly for overall growth. Trade-Offs and Resilience to Shocks Although the long-run effects of connectivity on growth appear to be positive, connectivity can also expose an economy to shocks and exacerbate crises. For example, Kaminsky and Reinhart (2000), Kaminsky, Reinhart, and Végh (2003), and Bae, Karolyi, and Stulz (2003) show that financial sector linkages play an important role in propagating shocks. The 2005 commodity food price shock and the 2008 Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  73 global financial crisis also demonstrated the cascade effects that shocks in one market can have in other markets. Connectivity, however, may also mitigate shocks that originate in some country nodes in the network. For example, if a given country is well integrated in the network, then a shock to one of its partners can be ameliorated by leveraging its other links to the remainder of the network. This analysis provides supporting evidence for both of these phenomena. Countries with low levels of connectivity are more resilient to shocks in the global network because they have few partners and fewer connections that would transmit shocks. On the other hand, countries with high levels of connectivity also appear to be less affected by shocks to the network. This is likely due to well-diversified con- nections that can mitigate the severity of the shock. Countries in the “middle” of the connectivity spectrum appear to be most susceptible to international shocks, that is, they have low levels of diversified connectivity and are highly dependent on a few well-­ connected countries and connections (which boosts their overall con- nectivity). They are particularly susceptible to shocks affecting one of the well-­ connected countries where they derive access to global markets and connectivity. Figure 1.11 shows this pattern. A 10 percent simultaneous, negative shock is simulated to three connections (trade, migration, and FDI) in each of three “­central” and well-connected countries (Germany, the United States, and Russia). The countries with the largest declines in their initial connectivity are those that are strongly connected to the country experiencing the shock and do not have strong connections to other partner countries. These countries tend to be in the middle range of centrality and receive their connectivity through a few well-connected countries. A shock to one of these well-connected countries would do the largest damage to their global connectivity. As figure 1.11 shows, a 10 percent adverse shock to German trade, migration, and FDI has an important impact on most countries in the world because of Germany's high centrality. (The vertical axis shows the change in connectivity and the horizontal axis shows the initial level of connectivity.) However, not surprisingly, the most affected countries are the smaller countries for which Germany is the main partner country, including countries in ECA, the Middle East, and parts of Asia. The largest decrease in connectivity, caused by a 10 percent decline in German connectivity, is in Poland, Ukraine, and Sri Lanka, followed by Bangladesh, the former Yugoslav Republic of Macedonia, Croatia, and Turkey. However, because of the importance of German in the ECA network, even well-connected countries such as Switzerland and the Netherlands experience a significant decline in their centralities. The least affected countries are the small Latin American coun- tries and well-connected Asian economies, like Singapore. An adverse shock to US connectivity has an even stronger impact on most coun- tries in the world because of the high centrality of the United States (compare the range of the left axes in all the graphs). However, not surprisingly, the most affected countries are the smaller countries for which the United States is the main partner country. The largest decreases in connectivity, caused by a 10 percent decline in US connectivity, are in Jamaica and Belize, followed by Guatemala and the Dominican Republic. Because of the importance of the United States in the interna- tional global network, even well-connected large countries such as Japan, Mexico, 74  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia and Canada experience a significant decline in their centralities. The least affected countries are the small European countries whose main trading partners are Germany, the United Kingdom, or Russia. Thus, Luxembourg, Estonia, the Slovak Republic, and Lithuania barely experience any decline in their overall connectivity. A 10 percent shock originating in Russia would have a modest impact on global connectivity (left axis). The shock would most affect countries that are closely tied to Russia, such as the former Soviet Republics that are, in general, less connected to the global economy as a whole. In other words, they are highly reliant on Russia for connectivity to the world. FIGURE 1.11 a. Shock originating in Germany Simulated impact on 0 SGP URY NZL HKG individual countries’ PAN AUS CAN BLZ TTO KGZ PRY JAM SLV ARG MYS IRL connectivity measure GUY GTM BLR QAT THA MEX –1 ATG BRN SWZ BWAKWT OMN ECU JOR SAU NOR FIN SWE (modified PageRank) of a BEN GEO GAB ALB MUS CRI ARE LUX JPN GBR Change in level of connectivity (percent) AZE MOZ BHS DOM PER ISR BRA CHN 10 percent decline in trade, NAM CMR CYP NGA MAR CHL IDN BEL FRA AFG TGO KEN COL EST DNK USA BFA ETH TJK ARM LBN GHA DZA LVA EGY ESP foreign direct investment, –2 BRB TUN PRT IND CHE PAK LTU ZAF ITA and migration in Germany, BHR BIH BGR NLD RUS the Russian Federation, and MKD KAZ GRC the United States –3 SVK CZE HUN –4 MLT SYR HRV AUT MDA TUR BGD –5 LKA POL UKR –6 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Initial level of connectivity b. Shock originating in the United States 0 SVK LUX CYPEST LTU FIN CZE SWE BEL KGZ SWZ BHR BWA OMN MKD BIH LVAHRV BGR PRT DNKAUT ESP NLDDEU NAM ALBMUS MLT BLR NOR HUN BFA AZE LBN MDACMR URY DZA TUNMAR NZL UKR HKG POL CHE ITA FRA –2 AFG BEN GEO ARM SYR MOZLKA BGD KAZ GRC TUR RUS IRL GBR TGO BRN TJK GAB ARE ARGZAF AUS SGP SAU IDN Change in level of connectivity (percent) MYS CHN ETH PRY JOR THA BRA KEN QAT EGY CHL IND –4 BRB GHA NGA PAK PAN GUY JPN KWT DOM PER CAN –6 ATG COL BHS ISR –8 CRI ECU GTM MEX SLV –10 TTO JAM BLZ –12 –14 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Initial level of connectivity continued Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  75 c. Shock originating in the Russian Federation FIGURE 1.11 0 continued SGP –0.02 LUX NLD IRL ARE AUS CZE CHE Change in level of connectivity (percent) NOR DNK SWE CAN FRA DEU MLT ISR FIN JPN BEL –0.04 HRV HKG PRT AUT ESP GBR LTU ARG URY CRI TTO NZL SVK TUR ITA BGR HUN –0.06 BLZ QAT TUN ZAF PAN DOM EST UKR CHL BRA USA PRY BHS BLR SAU KGZ JAM CYP PER –0.08 ECU SLV GTM COL MYS THA BEN ARM LVA POL MEX GEO BHRMKD GRC IDN –0.1 ATG BRN BRB GUYKWTOMN LKA MUS BIH JOR IND CHN SWZ GAB NAM LBNALB KAZ MAR BWA MDA DZA EGY SYR GHA NGA BFA AZE CMR KEN BGD PAK –0.12 AFG TGO TJK MOZ ETH –0.14 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Initial level of connectivity This framework allows for a multitude of scenarios, including impacts to just one dimension and country (say, trade from China), or several dimensions across a subset of countries. However, a particularly pertinent one in recent times is a shock to the United Kingdom’s ties to the rest of the EU, the “Brexit” scenario. Brexit would affect the connectivity of ECA countries. Table 1.6 shows the effect on ECA's overall connectivity index from a 10 percent reduction in all flows from the United Kingdom to other EU27 countries. Even though the other EU27 countries are those affected directly by the shock, all of ECA is impacted by Brexit because of their indirect links to the United Kingdom and EU countries. Smaller, well-connected nations such as Malta, Ireland, Cyprus, and Luxembourg would be the most affected countries from this assumed Brexit scenario. Alternatively, the countries in Central Asia and the South Caucasus would be the least affected. Different regions in ECA have different exposures to types of connectivity shocks (trade, migration, FDI). For example, Western Europe is the most exposed to shocks in other Western European economies. Table 1.7 shows the largest two contributors to the decline in overall connectivity of each ECA subregion in response to a 10 percent shock in three network layers (trade, FDI, migration). Not surprisingly, the overall connectivity of Central Asia in terms of shocks to trade, FDI, and migration is affected most by Russia (trade, FDI, migration), but also by China (trade and FDI), and Germany (migration). The rest of ECA appears to be more sensitive to trade shocks in other ECA countries, particularly Germany, as well as the United States. Belgium and the Netherlands have the greatest impact on overall connectivity for the ECA region due to shocks to FDI, because of their large role in trade logistics and finance. Migration shocks are transmitted to various ECA subregions via countries in close proximity and with language similarities and historic ties. 76  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE 1.6  ECA Countries Most and Least Affected by Brexit Percent decrease in multidimensional connectivity Most affected Least affected United Kingdom −3.46864 Georgia −0.00105 Malta −1.35494 Kazakhstan −0.00109 Ireland −1.05116 Azerbaijan −0.00141 Cyprus −0.76504 Armenia −0.00153 Luxembourg −0.70449 Tajikistan −0.00194 Netherlands −0.65897 Kyrgyz Republic −0.00234 Belgium −0.57851 Albania −0.00456 Sweden −0.30127 Bulgaria −0.00459 Spain −0.30023 Macedonia, FYR −0.00623 Denmark −0.28094 Latvia −0.00796 Note: ECA = Europe and Central Asia. TABLE 1.7  Transmission of Trade, Migration, and FDI Shocks to ECA Subregions Largest origin countries of shocks due to a 10 percent shock in ECA region affected Trade FDI Migration Central Asia Russian Federation and China Russian Federation and China Russian Federation and Germany Central Europe Germany and Netherlands Germany and Austria Germany and Austria Western Balkans Italy and Germany Austria and Hungary Italy and Germany South Caucasus Turkey and United States Russian Federation and Russian Federation and Ukraine Kazakhstan Eastern Europe Russian Federation and Germany Russian Federation and Germany Russian Federation and Poland Russian Federation Germany and United States Germany and Switzerland Germany and Ukraine Turkey Germany and Italy Belgium and Netherlands Germany and Netherlands Southern Europe Germany and France Belgium and Netherlands United Kingdom and Poland Northern Europe Germany and Netherlands Belgium and Netherlands Finland and Norway Western Europe Germany and Netherlands Belgium and Netherlands Italy and United Kingdom Note: ECA = Europe and Central Asia; FDI = foreign direct investment. Conclusion While it has been well documented that globalization has long-term growth benefits through the technology and knowledge transferred via international connections, this is the first analysis to examine how the connections of the con- nections of partner countries matter for growth and how various types of con- nections interact with each other to influence economic growth. Economic interactions, aside from their direct benefits, also have indirect effects that can have lasting influence. Trade, migration, and FDI move the flow of ideas and innovation across borders. Each of these channels individually appears to be an important source of economic growth by facilitating the transmission of knowl- edge. Moreover, multidimensional connectivity is more important for growth than any individual type of connectivity by itself. The whole of the connectivity network is greater than the sum of its parts. Although there is certainly some level of substitutability between the various layers, when it comes to informa- tion flows, complementarity dominates. In fact, there might be a high degree of complementarity of the information flows that contribute to growth. Therefore, Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  77 policies to promote balanced connectivity in many dimensions—those that focus on trade, migration, and FDI—are more beneficial than focusing on a policy to enhance only one. Indeed, reducing connectivity in one dimension may have adverse impacts on growth derived from other dimensions. Proposals to reduce migrations flows, for example, may have adverse consequences for the growth-enhancing benefits of trade and FDI flows. Annex 1A. Data TABLE 1A.1  Long-Term Growth Determinants Indicator Description Coverage Initial GDP per capita Logarithm of initial value of GDP per capita for growth period in question 2000–16 (2000–16). Source: World Bank, World Development Indicators (WDI). Governance Index of quality of governance that takes into account corruption, rule of law, and 2000–16 quality of institutions. Source: WDI. Inflation Measure of consumer price index change. Source: WDI. 2000–16 Government size Total government expenditure as a share of GDP. Source: WDI. 2000–16 Years of schooling Average number of years of schooling. Source: www.barrolee.com. 2000–10 TABLE 1A.2  Network Country Data Indicator Description Coverage FDI Total bilateral FDI stocks. Source: Organisation for Economic Co-operation and 2002–13 Development (OECD). Trade Bilateral total trade flows for manufacturing goods. Source: United Nations 2000–15 Conference on Trade and Development. Migration Total migration stocks. Source: Individual countries’ census data; OECD and World 2000, 2010 Bank estimates (see Artuc et al. 2017). ICT Proxy for ICT flows; estimated by combining bilateral duration of phone conversations 2003–11 and bandwidth capacity between countries. Source: Derived from TeleGeography data. Airlines Estimated bilateral number of flights (end destination). Source: International Civil 2002–12 Aviation Organization. Portfolio flows Total bilateral portfolio flows. Source: Bank for International Settlements, 2000–14 Consolidated Banking Statistics. Note: FDI = foreign direct investment; ICT = information and communication technology. Annex 1B. Network Graph Methodology Country node placement utilizes Barnes-Hut algorithm (http://arborjs.org/docs​ barnes-hut). The algorithm attempts to place large country nodes closer to the /­ edges of the graph as a means of more clearly showing their numerous connec- tions to smaller country nodes. The repulsion of the country nodes away from the center of the graph is proportional to their size. The repulsion away from the center of the graph is counterbalanced by the attraction forces caused by how strongly each pair of countries are connected to one another. Once the forces of repulsion and attraction on the country nodes have been defined, the behavior of the entire graph under these forces may then be simu- lated as if it were a physical system. In such a simulation, the forces are applied to the country nodes, pulling them closer together or pushing them further apart. This is repeated iteratively until the system comes to a mechanical equilibrium 78  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia state; that is, their relative positions do not change from one iteration to the next. The positions of the country nodes in this equilibrium generate the graphical depiction of the network. Annex 1C. Multiplex PageRank Centrality Given the surge of interest in multiplex networks, methodologies have recently been proposed to assess the centrality of nodes in multiplex, and more generally multilayer, structures (Halu et al. 2013; Sola et al. 2013; Kenett, Perc, and Boccaletti 2015; De Domenico et al. 2015; Rahmede et al. 2017). Halu et al. (2013) and Iacovacci and Bianconi (2016) propose an algorithm that captures how the cen- trality of the nodes in a given layer of the multiplex can affect the centrality of the nodes in other layers. This effect is modeled by considering a PageRank algo- rithm based on the centrality of the nodes in the master layer. De Domenico et al. (2015) propose instead to rank simultaneously nodes and layers of the multiplex network based on any previous measure of centrality established for single-layer networks, including random walk processes that hop between nodes of the same layer and between nodes of different layers as well. The resulting centrality, called “versatility,” strongly awards nodes active (connected) in many layers; however, the description was not intended to weight layers in any specific way. Recently, Rahmede et al. (2017) proposed a different approach, in which they consider a random walk hopping through links of different layers with different probabilities determined by the centrality of the layers (influences). This is follow- ing the work of Sola et al. (2013) in which different measures for the centrality of the nodes given a set of influences of the layers have been proposed. Rahmede et al. (2017) propose a ranking algorithm, called MultiRank, that is specified by a coupled set of equations that simultaneously determine the centrality of the nodes and the influences of the layers of a multiplex network. The MultiRank algorithm applies to any type of multiplex network, including weighted and directed multi- plex network structures. Very generally, this algorithm proposes an extension to the classical PageRank centrality calculation by coupling the centrality of the node to the influence of the layer in which it is active. This is done by considering the node-layer interaction as a bipartite network. Such a coupling provides new insights into the centrality of a node across different connectivity dimensions. Annex 1D. Centrality Indicator More formally, the centrality value Θi is proportional to the probability that an innovation will be transmitted to a country: Θi = λ ∑A ki Θk + ( y i * Pi ). k The value Aki is a function of the links between countries k and i, l is an exog- enous parameter that captures the weight of decay placed on connections Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  79 (set to 0.85), yi is GDP per capita, and Pi is the population (the last two terms together equal aggregate GDP). The intrinsic value, proxied by GDP, plays an important part in determining the value of the index. For example, even a completely isolated country has a positive probability of innovating and growing based on its domestic resources only. Our choice of proxy for the intrinsic (internal) likelihood to innovate is based on two simple considerations. First, the greater the number of people in a country, the greater the knowledge (or new ideas) that could potentially be generated. Second, we assume that higher-income countries are closer to the technological frontier and thus have a higher probability of producing new knowledge. If a coun- try does not produce the knowledge intrinsically, it can learn from others through its connections. This mechanism is captured by the term λ ∑k Aki Θk. Thus, the probability that an economy has the knowledge to innovate is a sum of the likelihood of its intrinsic innovation (proxied by GDP) and a weighted aver- age of the connectivity of its partners where the weights (A) are a function of the connections. These weights reflect the strength of the informational link and ulti- mately the probability of successful transmission of ideas. Aki takes on the following set of values:   Tradeki FDI stockki Migration stockki ICT flow ki Flightski Portfolio flowski    ; ; ; ; ; .  GDPk  GDPk POPk POPk GDPk GDPk   Each connection (total bilateral trade, total FDI stock, bilateral migration stocks, ICT, airline transport, and portfolio flows) is divided by a proxy for the size of the country (GDP or population). In the original PageRank algorithm, this feature is introduced by dividing by the total number of outgoing links of the ­ partners. Therefore, the probability of getting from website A to website B by a random web surfer decreases as the number of outgoing links in A increases (there are more sites on which the surfer can land). In the case at hand, the prob- ability of an idea reaching a specific country decreases with the population of the sending country. Similar adjustments are necessary when one considers information flows between countries along the various networks. For example, conditional on an innovation being present in country A, the probability that a single migrant from A to B will carry this idea decreases with the size of the population of A. Although large countries are more likely to generate ideas domestically, they need greater flows and deeper links to transmit those ideas to their partners. This chapter argues that this measure is a good proxy for the probability of growth-relevant knowledge generation by each country (either through learning from its connections or developing knowledge domestically). Notes 1. The data and graphing methodology are described in annexes 1A and 1B. 2. https://www.bis.org/statistics/consstats.htm. 80  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia 3. Eigenvector centrality is a measure of the influence of a node in a network. A high eigenvector score means that a node (country in our case) is connected to many nodes that themselves have high scores. 4. This is the standard value of similar parameters used in most network analyses. 5. In the original search engine applications of PageRank this value captured the likeli- hood that the random surfer can type the URL of the website without relying on hyper- links to get to it. 6. Including a layer of network connectivity that was determined solely by geographic (capital to capital) distance between countries was not a significant determinant of growth, nor did it change the empirical results related to our empirical inferences related to the multidimensional connectivity index described later in the discussion. α β γ δ v η 7. The equation is Iki = xki fki mkiiki aki pki , in which Iki is the network information function and a,b,γ,d,n,h. are the estimated weights for each connectivity layer. The weights are calculated using the maximum likelihood procedure where the objective function was to maximize the goodness of fit of the growth equation (adjusted R-squared). 8. The functional form being Θµ = λ i ∑ k µ Iki Θk ( ) + y i *Pi . References Alfaro, Laura, Areendam Chanda, Sebnem Kalemli-Özcan, and Selin Sayek. 2004. “FDI and Economic Growth: The Role of Local Financial Markets.” Journal of International Economics 64 (1): 89–112. Artuc, Erhan, Frederic Docquier, Caglar Ozden, and Chris Parsons. 2017. “Global Skilled Migration: Structural Estimation of 2000–2010 Patterns.” Unpublished manuscript, Development Research Group, World Bank, Washington, DC. Bae, Kee-Hong, G. Andrew Karolyi, and Reneé M. Stulz. 2003. “A New Approach to Measuring Financial Contagion.” Review of Financial Studies 16 (3): 717–63. Beck, Thorsten. 2008. The Econometrics of Finance and Growth. Washington, DC: World Bank. Ben-David, Dan. 1993. “Equalizing Exchange: Trade Liberalization and Income Convergence.” Quarterly Journal of Economics 108 (3): 653–79. Bookstaber, R., and D. Y. Kenett. 2016. “Looking Deeper, Seeing More: A Multilayer Map of the Financial System.” Office of Financial Research Brief 16-06, US Department of the Treasury, Washington, DC. https://www.financialresearch.gov/briefs/files​ /­OFRbr_2016-06_Multilayer-Map.pdf. Borensztein, Eduardo, Jose De Gregorio, and Jong-Wha Lee. 1998. “How Does Foreign Direct Investment Affect Economic Growth?” Journal of International Economics 45 (1): 115–35. Czernich, Nina, Oliver Falck, and Tobias Kretschmer. 2011. “Broadband Infrastructure and Economic Growth.” Economic Journal 121 (552): 505–32. De Domenico, M., A. Sole-Ribalta, E. Omodei, S. Gomez, and A. Arenas. 2015. “Ranking in Interconnected Multilayer Networks Reveals Versatile Nodes.” Nature Communications 6: 6868. Dollar, David. 1992. “Outward-Oriented Developing Economies Really Do Grow More Rapidly: Evidence from 95 LDCs, 1976–1985.” Economic Development and Cultural Change 40 (3): 523–44. Duernecker, Georg, Moritz Meyer, and Fernando Vega-Redondo. 2014. “The Network Origins of Economic Growth.” No. 14–06, Working Paper Series, Department of Economics, University of Mannheim, Mannheim, Germany. Edwards, Sebastian. 1998. “Openness, Productivity and Growth: What Do We Really Know?” Economic Journal 108 (447): 383–98. Multidimensional Connectivity: Pathways to Growth and Shared Prosperity in Europe and Central Asia ●  81 Feyrer, J. 2009. “Trade and Income—Exploiting Time Series in Geography.” Working Paper 14910, National Bureau of Economic Research, Cambridge, MA. http://www​ .nber.org/papers/w14910. Frankel, Jeffrey A., and David Romer. 1999. “Does Trade Cause Growth?” American Economic Review 89 (3): 379–99. Gould, David M. 1994. “Immigrant Links to the Home Country: Empirical Implications for US Bilateral Trade Flows.” Review of Economics and Statistics 76 (2): 302–16. Halu, A., R. J. Mondragon, P. Panzarasa, and G. Bianconi. 2013. “Multiplex Pagerank.” PLoS One 8: e78293. Helpman, E. 2004. The Mystery of Economic Growth. Cambridge, MA: Harvard University Press. Hidalgo, C., and R. Hausmann. 2009. “The Building Blocks of Economic Complexity.” Proceedings of the National Academy of Sciences 106: 10570–75. Iacovacci, J., and G. Bianconi. 2016. “Extracting Information from Multiplex Networks.” Chaos 26: 065306. Javorcik, Beata Smarzynska. 2004, “Does Foreign Direct Investment Increase the Productivity of Domestic Firms? In Search of Spillovers through Backward Linkages.” American Economic Review 94 (3): 605–27. Kaminsky, Graciela, and Carmen Reinhart. 2000. “On Crises, Contagion, and Confusion.” Journal of International Economics 51 (1): 145–68. Kaminsky, Graciela, Carmen Reinhart, and Carlos A. Végh. 2003. “The Unholy Trinity of Financial Contagion.” Journal of Economic Perspectives 17 (4): 51–74. Kenett, Dror Y., Matjaž Perc, and Stefano Boccaletti. 2015. “Networks of Networks–An Introduction.” Chaos, Solitons & Fractals 80: 1–6. Kivelä, Mikko, Alex Arenas, Marc Barthelemy, James P. Gleeson, Yamir Moreno, and Mason A. Porter. 2014. “Multilayer Networks.” Journal of Complex Networks 2 (3): 203–71. Mountford, Andrew. 1997. “Can a Brain Drain Be Good for Growth in the Source Economy?” Journal of Development Economics 53 (2): 287–303. Onodera, Osamu. 2008. “Trade and Innovation Project: A Synthesis Paper.” OECD Trade Policy Papers, No. 72, Organisation for Economic Co-operation and Development, Paris. Page, Lawrence, Sergey Brin, Rajeev Motwani, and Terry Winograd. 1999. The PageRank Citation Ranking: Bringing Order to the Web. Technical Report. Stanford, CA: Stanford InfoLab, Stanford University. Panizza, Ugo, and Andrea Filippo Presbitero. 2013. “Public Debt and Economic Growth in Advanced Economies: A Survey.” Swiss Journal of Economics and Statistics 149 (2): 175–204. Rahmede, Christoph, Jacopo Iacovacci, Alex Arenas, and Ginestra Bianconi. 2017. “Centralities of Nodes and Influences of Layers in Large Multiplex Networks.” Journal of Complex Networks. https://doi.org/10.1093/comnet/cnx050. Rodriguez, Francisco, and Dani Rodrik. 2000. “Trade Policy and Economic Growth: A Skeptic’s Guide to the Cross-National Evidence.” In NBER Macroeconomics Annual, vol. 15, ed. Ben S. Bernanke and Kenneth Rogoff, 261–38. Cambridge, MA: MIT Press. Romer, P. M. 1990. “Endogenous Technological Change.” Journal of Political Economy 98: S71–102. Sachs, Jeffrey D., and Andrew Warner. 1995. “Economic Reform and the Process of Global Integration.” Brookings Papers on Economic Activity 1995 (1): 1–118. Sola, L., M. Romance, R. Criado, J. Flores, A. Garca del Amo, and S. Boccaletti. 2013. "Eigenvector Centrality of Nodes in Multiplex Networks.” Chaos 23: 033131. 82  ●  Critical Connections: Why Europe and Central Asia’s Connections Matter for Growth and Stability SPOTLIGHT 1 Trends in Foreign Direct Investment in Europe and ­Central Asia F oreign direct investment (FDI) flows have made an important contribution to the level of the Europe and Central Asia (ECA) region’s multi- outflows from the ECA region were consistently higher than FDI inflows, in terms both of values and of shares of world FDI flows (see figure S1.1). dimensional connectivity. This spotlight high- EU15 countries have been both the main source lights the trends and composition of FDI in ECA and main destination of FDI announcements. countries.1 Accounting for 80 percent of the region’s GDP, ECA has been a key player both as a destination advanced EU15 countries (Western, Northern, and and as a source of FDI. Accounting for about Southern Europe) have generated more than percent of world’s GDP, the ECA region hosted, 45 ­ 80 ­percent of FDI outflows from the region to the on average, about 28 percent of the world’s inward world and have received more than 40 percent of FDI. At the same time, it was the source of total FDI inflows from the world. The share of EU13 40 ­percent of outward FDI.2 Focusing on FDI member countries and non-EU ECA countries of FDI announcements, during 2014, US$155 billion in FDI inflows from the world is 10 times larger than their projects had ECA countries as destinations, while share of FDI outflows to the world (see figure S1.2). FDI projects for US$256 billion were originated in All ECA countries, apart from the original core of ECA countries. Between 2003 and 2014 FDI the EU (EU15), receive FDI well above what would FIGURE S1.1  The relevance of ECA as both a destination and an origin of FDI has fallen since 2008 ECA FDI patterns over time: Share of world FDI flows, 2003–14 50 40 30 20 10 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Inflows Outflows Source: Calculations based on fDiMarkets.com data set. Note: ECA = Europe and Central Asia; FDI = foreign direct investment. ­ entral Asia Trends in Foreign Direct Investment in Europe and C ●  83 FIGURE S1.2  World FDI inflows into ECA are relatively more diversified by ECA destination than ECA FDI outflows to the world World FDI flows over time, by subgroup of countries a. ECA FDI outflows to the world (share by subregion) b. ECA FDI inflows from the world (share by subregion) 100 100 80 80 60 60 40 40 20 20 0 0 03 04 05 06 07 08 09 10 11 12 13 14 03 04 05 06 07 08 09 10 11 12 13 14 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 EU15 EU13 Other ECA Source: Calculations based on fDiMarkets.com data set. Note: ECA = Europe and Central Asia; EU = European Union; FDI = foreign direct investment. be expected given their size. The share of global FDI countries reflects the importance of deep interna- flows received by the core EU countries is less than tional agreements as determinants of FDI attraction their contribution to global GDP—that is, the FDI (see spotlight 2). However, FDI levels differed within intensity index was well below 1 during the period each of these groups. Resource-rich countries, such 2003–14 (figure S1.3). By contrast, the 13 EU mem- as Azerbaijan (or even Kazakhstan), have secured ber states that joined the EU in or after 2004 (EU13) particularly high levels of FDI. Similarly, countries and the rest of ECA have been important recipients that are attractive for financial investments, such as of FDI, given their size. An intuition behind this pat- Montenegro, show high levels of FDI, particularly tern is that capital should flow from capital-abundant because of low tax regimes. In addition to these economies (EU15), where returns are expected to exceptions, other countries have managed to attract be relatively low, to capital-scarce countries (rest of high levels of FDI within ECA. These include ECA), where returns are expected to be high. Hungary, Bulgaria, and Estonia among the EU13 Patterns of FDI inflows vary substantially across group; Georgia, Moldova, Turkmenistan, Serbia, countries. Figure S1.4 shows the average level of and Albania have also attracted above-­ average lev- FDI inflows as a percentage of GDP per capita for all els of FDI inflows—Albania and Serbia on the back countries between 2003 and 2014. It essentially con- of high integration into European global value firms the results described above by country group chains (see chapter 6).3 (unconditional on size): on average, FDI inflows have FDI in natural resources is more prevalent in non- been the greatest among EU13 countries, followed EU ECA members. While almost 90 percent of FDI by EU15, and then by the rest of ECA. The fact that inflows to EU (EU13 and EU15) countries goes to the intensity is greater for EU13 than for other ECA manufacturing and services sectors, this share is 84  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia SPOTLIGHT 1 continued FIGURE S1.3  ECA’s share of world FDI inflows is greater than its share of world GDP FDI intensity index (share of FDI/share of GDP), 2003–14 3 2 1 0 EU15 EU13 Other ECA Source: Calculations based on fDiMarkets.com data set. Note: ECA = Europe and Central Asia; EU = European Union; FDI = foreign direct investment. FIGURE S1.4  FDI attraction patterns increase with development levels but vary by country Average FDI inflows as a percentage of GDP per capita, 2003–14 30 20 AZE MNE IRL BEL BGR HUN 10 EST GEO TKM KAZ MDA ALB SRB KGZ ARM BIH AUT TJK HRV LVA CZE UKR POL SVK PRTESP GBR SWE UZB MKD BLR LTU RUS SVN FRA FIN TUR GRC ITA DEU 0 DNK 6 8 10 12 Log of GDP per capita (average, at purchasing power parity) Source: Calculations based on fDiMarkets.com data set. Note: Countries shown in red are in the EU15, those in blue are in the EU13 group, those in yellow are in the rest of ECA, and those in green are outside of ECA. ECA = Europe and Central Asia; EU = European Union; FDI = foreign direct investment. ­ entral Asia Trends in Foreign Direct Investment in Europe and C ●  85 closer to 80 percent for other ECA countries (non- from 63 percent in 2003 to 48.6 percent in 2014. EU), with natural resources playing a more impor- This approximately 14 percentage point decline tant role (see figure S1.5). over the period was compensated for by Increasingly, the rest of the world is becoming increases in FDI from East Asia and Pacific, the a crucial source of FDI into ECA. The share of FDI Middle East and North Africa, and other regions inflows into ECA that originated in the region fell (see table S1.1). FIGURE S1.5  Services and manufacturing dominate FDI inflow patterns across ECA Average percentage of FDI inflows by sector, 2003–14 100 80 60 40 20 0 EU15 EU13 Other ECA Resources based Manufacturing Services Source: Calculations based on fDiMarkets.com data set. Note: ECA = Europe and Central Asia; EU = European Union; FDI = foreign direct investment. TABLE S1.1  ECA Is the Main Investor in ECA Source of announced ECA foreign direct investment Share (percent) Investment (US$ million) Region 2003 2014 2014 Europe and Central Asia 63.0 48.6 75,102 North America 25.2 23.4 36,140 East Asia and Pacific 6.9 18.8 29,053 European Free Trade 2.1 3.7 5,648 Association Middle East and North Africa 0.9 3.5 5,483 Other 1.8 2.1 3,215 Total 154,642 Source: Calculations based on fDiMarkets.com data set. Note: ECA = Europe and Central Asia. 86  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia SPOTLIGHT 1 continued FIGURE S1.6  Germany and the United States dominate EU investment; France and China lead elsewhere Main investors in ECA, by subregion, 2014 a. EU15 b. EU13 c. Other ECA USA DEU CHN DEU USA DEU CHN FRA USA GBR CHN ITA FRA AUT RUS ESP GBR FRA ARE JPN JPN LUX SWE GBR CHE IND SWE JPN ITA HKG 0 10 20 30 0 10 20 30 0 10 20 30 Percentage of FDI inflows Source: Calculations based on fDiMarkets.com data set. Note: ECA = Europe and Central Asia; EU = European Union; FDI = foreign direct investment. The country source of FDI inflows is similar across 3. Serbia is integrated as a first-tier supplier in the auto- ECA subregions. The United States, Germany, and motive value chain. Albania is, for example, closely integrated with Italy in garments and footwear. China are the principal investors in EU15, EU13, and other ECA countries, representing 44 percent of total FDI inflows into the ECA region. The United States represents almost 30 percent of EU15 FDI Reference inflows, while Germany accounts for 19 percent of Laget, E., N. Rocha, and G. Varela. 2018. “FDI and EU13 FDI inflows. However, China is the main Deep Preferential Trade Agreements: An Empirical source of foreign investment in other ECA countries Investigation.” Unpublished, World Bank, such as Russia and Turkey (see figure S1.6). Washington, DC. Notes 1. This spotlight draws from Laget, Rocha, and Varela (2018). 2. Both the share of ECA in the world’s GDP and the share of ECA’s inward and outward FDI are calculated as an average for the period 2003–14. 2 Knowledge Transfers from International Openness in Trade and Investment: The European Case This chapter reviews the role of international trade, foreign investment, and global value chains (GVCs) in transferring knowledge that helps to improve productivity in Europe and Central Asia (ECA) countries. This discussion complements the microanalysis in chapter 3 on the impact on domestic firms of foreign ownership and management. This discussion is particularly rele- vant for the European continent, where connectivity through trade, invest- ment, and production sharing is high and has greatly increased in the past two decades. Two main questions are considered: what type of knowledge and innovation are being created in Europe and how knowledge diffusion takes place across the continent and to what extent firm-to-firm connectivity within Europe contributes to productivity growth through learning and knowledge transfer. The first section discusses knowledge creation in Europe, and the second section reviews the literature on the link between openness and learning, and how importing, exporting, exposure to foreign firms, and participation in GVCs leads to technological catchup across bor- ders. The third section investigates technology diffusion across and within national borders. The fourth section concludes. Main Messages • Learning is the principal source of productivity growth for most countries, largely through the absorption of existing innovations rather than own research. 87 88  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia The most advanced firms in Europe tend to be larger and more capital inten- sive, have greater investment in intangibles, and have a greater level of human capital than other firms. Europe compares well to the global frontier in manu- facturing, but lags behind the global frontier in services and in some innovation- based growth industries. • Firms can learn through importing as a result of exposure to more diverse and sophisticated inputs; through exporting from opportunities to achieve econo- mies of scale, upgrade workers’ skills, and learn techniques to improve product appeal; and from foreign direct investment (FDI) through technology transfers and exposure to high-skilled workers. These effects are significant, for countries in general and in ECA in particular, once the impact of cultural or geographical distance, level of development, and the quality of domestic and international institutions is taken into account. • The typical channel of technology transfer is from the most advanced global firms to the most advanced national firms (which tend to be strongly involved in GVCs), and then to other domestic firms. This process is confirmed by econo- metric evidence for Europe, where a rise in total factor productivity (TFP) growth of national frontier firms leads to a similar increase in TFP by other domestic firms, but an increase in TFP by GVC frontier firms has little direct impact on other domestic firms. • The reduction in trade from the global economic crisis reduced firms’ propensity to expand participation in GVCs, leading to a sharp drop in productivity growth compared to the precrisis period. Calls to limit exposure to foreign volatility in light of the deep global recession following the crisis would limit firms’ ability to benefit from foreign technology, particularly depressing growth in less advanced economies where firms are further away from the productivity frontier. Knowledge Creation in Europe TFP Growth and Knowledge Flows Learning is the main source of productivity gains for most firms. Increases in knowledge are typically ascribed to two main sources: investment in new knowl- edge (e.g., through research and development [R&D]) and use of existing knowl- edge (past discoveries and knowledge sharing) (Griliches 1979). However, only a small set of companies invest in R&D and patents or introduce any radically new products or processes (Cirera and Maloney 2017).1 Most firms opt instead for learning from existing knowledge, which originates from many possible sources: universities, clients, suppliers, competitors, or other entities within the same cor- poration. In the United Kingdom, out of a sample of 804 firms surveyed through the UK Community Innovation Survey,2 51 percent reported learning from com- petitors, 65 percent from ­ suppliers, 68 percent from clients, 49 percent from other entities in the same c ­ orporation, and only 19 percent from universities (Crespi et al. 2008). Controlling for the impact of other factors, a UK firm learning from competitors, suppliers, and other entities in the same group enjoys a 4.7 percent faster growth in TFP than a firm without such learning (Crespi et al. 2008). This differential explains nearly 50 percent of the productivity gap between Knowledge Transfers from International Openness in Trade and Investment: The European Case ●  89 the top and the bottom performers (i.e., the firms in the top productivity quartile and the firms in the bottom quartile, respectively). It is very hard to measure knowledge flows. Direct measurement based on patent citations has two shortcomings. First, research on both US companies (Jaffe and Trajtenberg 2005) and European companies (Criscuolo and Verspagen 2008) shows that patents are a very noisy measure of information flows. As much as 91 percent of patent citations in Europe and 50 percent in the United States are entered by examining officers rather than the inventors themselves. More impor- tantly, only a small share of knowledge is patentable or patented. Most knowledge flows involve nonpatenting firms. For example, knowledge that is transferred between multinational enterprises and their affiliates or between suppliers and customers is usually not patented. An indirect method of measuring knowledge flows is based on TFP. By look- ing at the relationship between TFP growth and factors thought to be potentially causing information flows, a number of papers have established a statistically robust relationship between TFP differences and information flows (e.g., Crespi et al. 2008). The distance-to-frontier literature (e.g., Griffith, Redding, and Van Reenen 2004) postulates that knowledge from inside the firm can be measured by TFP levels (or TFP growth rates) in the firm itself, while learning from the fron- tier can be measured by differences in TFP levels (or in TFP growth rates) between the firm and a nominated frontier firm or set of firms. The indirect method of measuring knowledge flows as gaps in TFP growth between frontier and non- frontier firms has the advantage of implicitly looking at nonpatentable innova- tions and makes it easier to draw links with policies. The Innovators in Europe How do the most productive firms differ from less productive firms? Having estab- lished that there is a relationship between TFP differences and differences in infor- mation flows, the first step of the analysis is to assess the key characteristics of European Union (EU) innovators. Following the distance-to-frontier literature, we assume that the frontier in Europe consists of the top 100 productive firms in each two-digit industry/year, for 13 countries in the Amadeus (BvD) database over the period 2010–13.3 The frontier firms in each narrowly defined sector (across Europe) have several differences with the other, less productive firms that may be related to their higher productivity. Frontier firms stand out in terms of size, capital intensity, investment in intangi- bles, and human capital, but the importance of these characteristics differs by sec- tor (figure 2.1). In information technology (IT)–intensive manufacturing, frontier firms are 130 percent larger, 50 percent more capital intensive, and invest 49 ­percent more in intangibles. They pay a 31 percent wage premium to their workers com- pared to less productive peers in the same two-digit industry, suggesting that the quality of human capital in these firms is also higher. The size premium is even higher for traditional manufacturing (270 percent). In IT-intensive services, frontier firms are twice as capital intensive (102 percent) than laggards, invest 74 percent more in intangibles, and offer a wage premium of 49 percent, but their size is no different from other firms. Non-IT-intensive service firms also pay a high wage premium, but capital intensity and investment in intangibles are not outsized. 90  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 2.1  Differences 300 between frontier and laggard firms vary 250 across sectors Frontier firms as a 200 percentage of laggard firms, average, 2010–13 150 100 50 0 Size Capital Investment in Wage premium/ intensity intangibles human capital IT-intensive manufacturing IT-intensive services Construction Other manufacturing Other services Source: Calculations based on Amadeus data. Note: Regression at the two-digit industry level in which each variable is regressed on a dummy that equals 1 (frontier firm) or 0 (laggard). Country, year, and sector fixed effects are controlled for. IT = information technology. Finally, frontier firms in construction post a premium in capital intensity, while intangibles and size do not play a significant role. The European Versus the Global Frontier Europe compares well to the global frontier in manufacturing, but lags behind the global frontier in services and in some innovation-based growth industries. Technology creation in European manufacturing is very similar to that in other advanced economies, as measured by the TFP growth gap between frontier firms in Europe and in the Organisation for Economic Co-operation and Development (OECD) (figure 2.2).4 However, labor productivity growth in European services firms is lower than in firms at the OECD frontier. These numbers suggest that the continent could be served well by pursuing better connectivity to the global fron- tiers in the services sector. Direct measures of innovation based on R&D investment confirm that European technology is strong in key manufacturing industries, but lags behind in services and high-growth technology areas. The Innovation Union Scoreboard for the EU is an instrument developed by the European Commission under the Lisbon Strategy to compare the innovation performance of the EU member states. It indicates that the EU is almost as innovative as the United States, overall. Yet there is a gap in services sectors and in some strategically important manufacturing industries, where R&D intensity is above the overall manufacturing average and where lead- ing innovators are predominantly young companies. Compared with the United States, Europe posts a gap in the so-called Revealed Technology Advantage in the internet industries, computer hardware and services, and biotechnology, as well as in semiconductors, software, and health care equipment and services (table 2.1). By contrast, the European Revealed Technology Advantage is strongest in indus- tries including industrial machinery, electrical components and equipment, fixed Knowledge Transfers from International Openness in Trade and Investment: The European Case ●  91 FIGURE 2.2  Europe lags 3.0 behind the frontier in services 2.5 Average annual labor productivity growth, in percent, 2010–13 2.0 1.5 1.0 0.5 0 Manufacturing Services European frontier OECD frontier Source: Calculations based on data from the Organisation for Economic Co-operation and Development (OECD) and Amadeus. Note: Sample is based on firms with more than 20 employees. The European frontier is among the EU15 (that is, the original core countries of the European Union). The technology gap is proxied by the difference in productivity growth between frontier firms and other firms (laggards) in the same sector and year. TABLE 2.1  Europe Specializes in Several Sectors with Below-Average R&D Intensity and Growth Revealed technology advantage, European Union and United States RTA European Union United States Industrial machinery 1.84 0.24 Electrical components and equipment 1.56 0.18 Fixed and mobile telecommunications 1.53 0.20 Aerospace and defense 1.50 1.13 Telecommunication equipment 1.38 1.09 Chemicals 1.31 0.64 Pharmaceuticals 1.27 1.16 Auto and parts 1.26 0.58 Industrial metals 1.00 0.30 Health care equipment and services 0.70 1.86 Software 0.51 2.05 Semiconductors 0.50 1.72 Biotechnology 0.32 2.20 Computer hardware and computer services 0.08 1.39 Internet 0 2.54 Sources: Bruegel and the World Bank, based on the Institute for Prospective Technological Studies R&D Scoreboard. Note: Table depicts the revealed technology advantage (RTA) for the period from 2010 to 2015. The RTA is calculated as a region’s share in total sectoral research and development (R&D) relative to its share in total R&D. An RTA greater than 1 reflects a region’s specialization in a given sector. 92  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia and mobile telecom, aerospace and defense, telecommunications equipment, chemicals, pharmaceuticals, and autos and auto parts. Knowledge and Learning from Trade, Investment, and GVCs: Insights from the Economic Literature Neo-Schumpeterian models assume that a country-sector’s productivity growth depends on exposure to the global frontier and distance to the frontier (Aghion and Howitt 2006; Saia, Andrews, and Albrizio 2015). Having assessed that European technology is close to the global frontier in manufacturing but a bit more removed in services sectors, we next focus on how technology flows within the continent and the role played by firm-to-firm connectivity in helping technology spread. The mech- anism we track in this chapter is predominantly cross-border in nature and focuses on productivity convergence as a measure of technology convergence: variations in productivity growth differentials between different groups of firms are used to gain insights into how technology is transferred from the European frontier to firms located within the domestic economies of the countries in the rest of the continent. Openness facilitates learning, upgrading, and innovation through various channels. In the next paragraphs we focus on how different types of openness enable firms to acquire knowledge and valuable capabilities.5 We define capabili- ties as in Cirera and Maloney (2017): “firm features that are mainly internal to the firm, firm specific, knowledge-based, and not easily replicable … and that manifest themselves in routines, management practices, and assets that are adopted or acquired by the firm, internally or externally, that can be measured, and that are the result of learning and accumulation over time.” Imports: Direct and Indirect Knowledge Spillovers from International Suppliers of Inputs Importers are larger and more productive than firms that do not trade, with productivity gains generated mostly from access to more differentiated imported inputs. Earlier studies for developed countries conclude that import- ers are larger and more productive, a finding recently confirmed for develop- ing countries as well ( S¸ eker 2012). Importing is found to raise productivity (Amiti and Konings 2007), with the largest productivity gains due, not to a competition effect, but rather to improved access to inputs. Access to a more differentiated variety of inputs (Goldberg et al. 2010; Halpern, Koren, and Szeidl 2015) seems to matter more than the direct benefits from lower prices or higher-quality foreign inputs. For example, Goldberg et al. (2010) find that the increase in product scope by Indian firms following India’s comprehensive trade reforms in the 1990s was due to an increase in the number of imported varieties rather than a decrease in prices. Likewise, Halpern, Koren, and Szeidl (2015) find that Hungarian firms enjoyed a 22 percent rise in productivity from importing inputs over 1993–2002, with about half of the effect due to imper- fect substitution between foreign and domestic inputs. The firms that benefit most from importing are multinational firms: they use imports more effectively Knowledge Transfers from International Openness in Trade and Investment: The European Case ●  93 and benefit from scale economies that reduce the fixed costs of importing. Importing by these firms is also found to have beneficial effects on other domestic firms via supply chain linkages. Imports also trigger learning effects and feedback loops. Importing gives rise to tacit knowledge, which materializes in intangible assets (MacGarvie 2006; Koren and Csillag 2011). Sophisticated machinery and capital goods imports require highly trained operators. Using data from Hungary for the period 1994–2004, Koren and Csillag (2011) construct a measure of exposure to imported machines, combining data on workers’ occupations with information on Imports also imported products. They find that, other things equal, the average trigger learning wage of workers increases by about 3 percent after a firm acquires effects and the imported machinery. MacGarvie (2006) finds a positive associa- feedback loops. tion between the nature of technology imported and the subse- Importing gives rise quent patents by French firms, based on propensity score matching6 to tacit knowledge, applied to patent citations. which materializes Importing better inputs or connecting to better (more productive) in intangible firms also leads to upgrading for direct importers and other firms indi- assets. rectly connected via supply chains. Javorcik (2004) shows that the pres- ence of multinational corporations in a country and industry increases the productivity of firms in industries that are their suppliers. Kee (2015) highlights through interviews and empirically that firm-level connections induce productivity improvements in domestic suppliers. Kee (2015) further shows, using a sample of Bangladeshi garment firms, that local intermediate inputs may also enhance per- formance of other domestic firms through what she calls the “shared supplier spill- overs” of FDI firms. She finds that after EU firms expanded FDI in Bangladesh, domestic firms that shared the same suppliers with the foreign investors expanded their product scope by 25 percent and enjoyed productivity gains of 33 percent. Finally, there are complementarities in capabilities, with the return to importing and innovation activities increasing in the intensity of one another. For example, Bøler, Moxnes, and Ulltveit-Moe (2015) find that firms involved in foreign sourcing are more innovative, and innovation increases the profitability from the interna- tional sourcing. Specifically, they find that returns to lowering R&D costs are higher for importers. Cost complementarities between R&D and international sourcing are crucial to explaining productivity gains: 25 percent of the productivity gains are due to international sourcing and 75 percent to the complementary R&D investments. Exports: Knowledge Spillovers from Competitors and Clients Exporting generates opportunities for learning for firms and its workers (De Loecker 2013). Bustos (2011) finds that Argentinean firms increased their demand for skills after the creation of Mercosur (Bustos 2011).7 Evidence from Taiwanese firms shows that there is complementarity between exporting, R&D, and workers’ train- ing (Aw, Roberts, and Winston 2007). Exporting requires firms to acquire new capabilities to perform a complex set of activities, including manufacturing tasks, ­ ervices, even marketing, distribution, foreign trade finance activities, and exporting s when the product exported is non-skill-intensive (Feenstra and Hanson 1996; 94  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Matsuyama 2007; Grossman and Rossi-Hansberg 2008; Verhoogen 2008). Moreover, exporters tend to be larger and more productive and pay higher wages, inducing a greater need for more complex management structures, which in turn increases the demand for skills.8 Management is a key factor regulating ­ complexity in firms’ operations. Systematic, large-scale evidence from data on management practices, balance sheets, and comprehensive trading activity for the United States and China establishes a clear link between managerial competence and export performance (Bloom et al. 2018). Demand factors also play an important role in inducing positive knowledge spill- overs from exporting. Exporting enables firms to learn about more sophisticated consumers and more competitive markets. Investments in capability building to improve product appeal and demand is an important reason for the phenomenal success of Chinese manufacturing firms in world trade over the past two decades (Sutton 2007; Brandt, Rawski, and Sutton 2008; Schott 2008). Searching for consum- ers is tightly linked to marketing (Eslava et al. 2015), investments in the customer base (Fitzgerald, Haller, and Yedid-Levi 2015), and branding as a way to signal quality by building reputation, as buyers often do not observe quality directly (Cagé and Rouzet 2015). Coe and Helpman (1995) show that the extent to which a country benefits from other countries’ R&D efforts depends on how much it trades with them. The characteristics of destination markets affect the impact of exporting on learn- ing. Features such as income, quality valuation, distance, and transport costs affect firms’ decisions on upgrading, and hence on both the propensity to undergo the costs of acquiring new knowledge and the type of knowledge acquired (Hummels and Skiba 2004; Verhoogen 2008; Bastos and Silva 2010; Manova and Zhang 2012; Martin 2012). Firms that export to high-income countries more intensely use higher levels of skills and pay higher wages than domestic firms or exporters that are spe- cialized in middle- or low-income countries (Brambilla et al. 2012). Mexican compa- nies exporting to the US market upgrade quality, as measured by ISO 9000 certification, and raise the wages of white- and blue-collar workers more than do firms that export to lower-income countries or that do not export (Verhoogen 2008). Complex Engagements through GVCs: Learning from Import-to-Export Activity and Engagement with Multinational Corporations Exporters that import their intermediate inputs are more productive than compa- nies that only import or only export (Kasahara and Lapham 2013). Firms’ productiv- ity affects their decisions to be in international markets, and importing inputs affects productivity. An implication of the complementarities between importing and exporting is that imposing import restrictions can reduce exports and affect most negatively the domestic frontier firms. The productivity of firms at the domestic technology frontier matters since these firms tend to induce their suppliers to also upgrade technology and inno- vate. Producing higher-technology goods tends to require more skill-intensive and higher-technology inputs. The innovating firm will demand better inputs, thus inducing its suppliers to adopt newer technologies. Also, with economies of scale, the price of advanced inputs produced by the innovating firm will decline, Knowledge Transfers from International Openness in Trade and Investment: The European Case ●  95 increasing the incentives for other firms to use the same inputs to upgrade their own technology (see Kee 2015 for Bangladesh and Fieler, Eslava, and Xu 2018 for Colombia). Thus, if production exhibits internal or external economies of scale, increased demand for higher-technology goods will increase technology adoption in the aggregate. Innovation tends to flow faster and more easily within GVCs. Such production arrangements link together multiple firms, usually located in different countries, in ways similar to intragroup investment and trade. In so doing, they offer a high degree of exposure to, and learning from, the fast-evolving, technology-enabled business models that characterize GVCs, even without the need for participating firms to engage in ownership arrangements.9 At the same time, countries and firms with weaker connections to GVCs may reduce their innovation because of competition from countries and firms more strongly embedded in GVCs. For example, Mexican suppliers to US companies reduced their investment in innovation as demand fell because of their buyers’ increased GVC integration with third parties in China. And other buyers (e.g., German buyers) of the same Chinese suppliers also experienced adverse effects because their production costs in China rose as a consequence of the increased demand from the United States (Arkolakis and Muendler 2010). The power relationship between companies in GVCs also matters for innova- tion. Some suppliers within GVCs are heavily dependent on the buyer’s decisions, private standards, and technological requirements for production (referred to as a “captive” relationship). Success for these suppliers is largely determined by the ability to assimilate new and improved technologies developed and transferred by the large global buyers. FDI and Technology Spillovers Foreign direct investment (FDI) and multinational corporations can provide impor- tant learning opportunities. The cost of transferring technology is reduced within integrated companies, and foreign-owned companies tend to be better managed. Moreover, these companies are responsible for the largest share of trade and invest- ment globally (Bernard, Jensen, and Schott 2005; Yeaple 2013). The cost of transferring technology to subsidiaries encourages multinationals to acquire the most productive (or domestic frontier) firms (Arnold and FDI can Javorcik 2009; Criscuolo and Martin 2009; Ramondo 2009). A foreign- influence the acquired firm appears to be 57 percent more likely to have under- domestic productivity taken a process of innovation while f ­oreign owned, relative to a firm of firms in the same that stays domestic (Guadalupe, Kuzmina, and Thomas 2012). sector as the investment, FDI can influence the domestic productivity of firms in the same upstream sectors, sector as the investment, upstream sectors, and downstream sec- and downstream tors. Empirical studies that examine the impact of FDI on domestic sectors. firms’ productivity identify three types of spillovers. Horizontal spill- overs—the effect of FDI in a given sector on the productivity of domes- tic (or other foreign) firms operating in the same sector—can materialize through increased competition, technology and knowledge transfers, and workers’ circula- tion.10 Vertical spillovers through forward linkages—the effect of FDI in upstream 96  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia (inputs) sectors on the productivity of domestic (or other foreign) firms operating in downstream sectors (final product)—can materialize through improved provi- sion of inputs (more varied, cheaper, or better quality), and through technology and knowledge transfers via, for example, training of clients (Arnold et al. 2010; Fernandes and Paunov 2012; Duggan, Rahardja, and Varela 2013). Finally, vertical spillovers through backward linkages—the effect of FDI in downstream (final prod- uct) sectors on the productivity of domestic (or other foreign) firms operating in upstream (input) sectors—can materialize through training of suppliers and through the sophistication of inputs demanded (Javorcik 2004). Both inward and outward FDI can act as channels of spillovers. Most of the firm- level research on spillovers has focused on inward FDI.11 However, as van Pottelsberghe de la Potterie and Lichtenberg (2001) argue, foreign investment out- flows can also result in transfers of knowledge and technology leading to productiv- ity increases. If foreign firms imitate domestic counterparts, or if they source the domestic knowledge base, their country of origin may also benefit from potential spillovers. The authors make an analogy with language learners. One can learn a foreign language by bringing home a speaker of the foreign language, or by fully immersing oneself in the foreign country. While learning can happen in both cases, it is likely that the “full immersion” strategy provides greater scope for it. Indeed, there is evidence that firms engage in “technology sourcing” practices, targeting countries with substantial technological and scientific capabilities.12 Van Pottelsberghe de la Potterie and Lichtenberg (2001) consider three alternative chan- nels for international technology (or R&D) spillovers: (a) trade, (b) inward FDI, and (c) outward FDI. Their findings suggest that there are technology transfers only through outward FDI. However, evidence on the existence, magnitude, and channels of knowledge spillovers is mixed. In their meta-analysis of the literature, Iršová and Havránek (2013) conclude that horizontal spillovers are on average zero, and that the sign and size of these spillovers depend on both the characteristics of the domestic firms and domestic environment and the characteristics of the foreign investors as well.13 By contrast, the evidence on vertical spillovers (through both backward and forward linkages) tends to be more conclusive, suggesting a positive effect of FDI.14 Yet overall these studies suggest these positive spillover effects are neither inevitable nor automatic. Domestic technology investments, and in general, the building up of absorptive capabilities are necessary. One explanation for this lack of conclusive evidence is that the ability of countries and firms to benefit from spillovers from foreign firms may depend on other mediating factors. Indeed, one strand of literature identifies three factors that may be necessary for knowledge spillovers to materialize: distance, the level of economic development of partners, and the quality of domestic and international institutions. First, distance between partners may matter. A well-established body of work has shown that spatial proximity to the source of knowledge is an important condi- tion allowing spillovers to take place (Jaffe, Trajtenberg, and Henderson 1993; Audretsch and Feldman 1996). Take exposure through FDI. Geographic distance can be a deterrent for circulation of experts between the parent and subsidiary company, which tends to be a mechanism that facilitates knowledge transfers (Girma and Wakelin 2007). Although new technologies have reduced the cost of moving ideas across distant locations, the costs of moving people remain high, and Knowledge Transfers from International Openness in Trade and Investment: The European Case ●  97 they could reduce the scope of spillovers (Javorcik, Saggi, and Spatareanu 2004). Alternative measures of distance that take into account language barriers or cultural differences may also play a role by increasing the costs of communicating ideas. Second, the level of development of both partners plays a role in international spillovers. Firm-level empirical evidence that looks at FDI spillovers, for example, demonstrates that absorptive capabilities matter for the extent of the spillovers (Griffith, Redding, and Simpson 2002; Blalock and Gertler 2008; Benli 2016). At the country level, this argument means that for international spillovers to materialize, it may not be enough to trade or invest in a country that actively invests in R&D. Capabilities at home matter too, and these are linked to a country’s level of develop- ment. Thus, for example, the scope and scale of FDI knowledge spillovers may differ between industrial and industrializing countries, showing the importance of host country characteristics in exploiting benefits from FDI (Silajdzic and Mehic 2015). In addition, the capabilities of the partner are likely to matter too, given its level of investment in R&D, if these capabilities affect the effectiveness of that investment.15 Third, the quality of institutions can reduce the costs of transferring knowledge. Institutions have been identified as key determinants of economic growth. They could potentially also affect the degree to which domestic and foreign R&D investment affects TFP. Countries with better institutions may be more efficient at ­ investing in R&D (Coe, Dicken, and Hess 2008). A better doing business environ- ment, for example, may encourage more entrepreneurial R&D that results in larger quality improvements for a given R&D effort. Also, better infrastructure of interna- tional agreements—another dimension of institutional quality—could improve intellectual property protection and stimulate the circulation of proprietary infor- mation between parent and subsidiary companies. Summary of Lessons from the Literature Imports and exports between firms generate learning and technology transfer through a rich variety of push and pull factors. Importing activity induces technology transfer through four main channels. First, access to more diversified varieties and complementarities between imported inputs and domestic products generate gains in product scope and in productivity. Second, using more sophisticated imported technology leads to learning effects and feedback loops in tacit knowledge for workers. Third, self-reinforcing complementarities The positive between importing and innovation capabilities generate greater effects are not returns to both activities. Finally, domestic linkages of GVCs ensure limited to direct importers, but spread that the positive effects are not limited to direct importers but spread to other domestic to other domestic firms through shared supplier networks. Exporting firms through activity is also associated with learning and innovation, but spillovers shared supplier from exports are lower than the effect induced by imports, which prop- networks. agate to other domestic companies via shared supplier networks. The link between exporting and absorbing new technologies rests on compe- tition and both supply and demand factors. On the supply side, the main effect of exports is on skill upgrading. This is induced by learning from competitors as well as from the activity of exporting: exporting in itself is complex and skill inten- sive and leads to the need to expand the organizational structure and the organiza- tional capital of the company, as a result of the fact that exporters tend to become 98  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia larger. On the demand side, learning originates from a push to improve product appeal, including through better marketing, branding, and customer search and retention. Finally, destination market characteristics matter: exports to more sophis- ticated markets induce greater product and process upgrading. FDI can improve productivity through learning and exposure to competition. FDI can influence the domestic productivity of firms in the same sector as the investment and in upstream sectors and downstream sectors through the transfer of technology and knowledge (e.g., training of clients or suppliers or exposure to more sophisticated workers), through competition, which forces some firms to improve and weak firms to exit, and through improved availability of inputs. Evidence of the size of these spillovers is mixed, however, perhaps because the ability of countries and firms depends on other characteristics, referred to as medi- ating factors, including the geographic distance between investing and receiving countries, the level of development of each country, and the quality of domestic and international institutions, that can facilitate or impede knowledge transfer. Knowledge Diffusion in Europe: The Two-Stage Process of Technology Transfer The impacts of trade, foreign investment, and GVC participation on innovation discussed in the literature review indicate a two-stage process of technology transfer between firms. Technology spreads first from global frontier firms to national frontier firms through trade and investment linkages, and then to the rest of the firms in the domestic economy through domestic linkages.16 Sources of learning and innovation focus differ across firm types, as exemplified in figure 2.3, which shows sources of learning for different types of firms (left column) and FIGURE 2.3  Technology transfer tends to follow a Global frontier firms (including European frontier firms) typical sequence Firms predominantly specialized in research Learning predominantly through own radical and development; often information and communication innovation, research and development, and patenting technology and services intensive. They tend to activity and from foreign demand and as a strategy to undertake innovative and risky activities and may deal with growing complexity. produce disruptive technologies or business models. National frontier firms Learning predominantly through engagement with Innovation focuses on adapative improvements and suppliers and clients and intragroup automation. Firms tend to maximize absorptive capacity on tasks of high value. of know-how, quality, and new technologies. Rest of firms’ ecosystem: Technology diffusion through domestic firm networks Learning predominantly through domestic Learning is mainly focused on improving business networks and competitors or through practices or routine adaptive, replicative activities irregular engagement in and assimilation of standards and technology global value chains. that are becoming dominant. Knowledge Transfers from International Openness in Trade and Investment: The European Case ●  99 ­nnovation focus (right column). Firms at the global frontier of innovation learn i predominantly through own radical innovation, R&D, and patenting activity, and from addressing untapped needs of sophisticated customers, usually at a global scale. These firms tend to specialize in R&D and high-value information and com- munication technology and service activities. They undertake risky activity and may produce disruptive technologies or business models. National frontier firms tend to be the most connected to global frontier firms through complex trade and investment arrangements, as discussed in the subsection “Complex Engagements through GVCs.” Openness to trade and investment is critical to acquiring new technology, learning new processes, and achieving technological upgrading. The innovation in national frontier firms focuses predominantly on adapting own pro- cesses and products to the quality, scale, and efficiency required by the interna- tional markets, including through automating production. Learning for these firms is mostly focused on creating the mechanisms for absorbing foreign buyers’ know- how and of world-class standards and technology. Finally, the remainder of firms (i.e., firms with productivity, technology, and skills below the national frontier) tend to learn predominantly through domestic channels, including domestic supplier networks and competitors. Their direct engagement with global frontier firms is either nonexistent or irregular, and the learning content of the exposure to inter- national counterparts limited (more on this in the next paragraph). Learning for the average nonfrontier firm is mainly focused on improving their business prac- tices and routine adaptive and replicative activities, and through assimilation of standards and ­ technology that are already widespread or dominant. The two-stage process of technology transmission within Europe can be mea- sured. Using CompNet Data,17 we consider that national frontier firms in a given macrosector18 are those in the top 20 percent in terms of TFP. The middle firms are those between the 30th and the 70th percentiles in the productivity distribu- tion. And laggards are the bottom 20 percent of firms in terms of TFP. In line with the discussion in the earlier sections of this chapter, we test the assumption that technology flows from European frontier firms with GVC links with the countries of Central and Eastern Europe (CEE) to the CEE firms following a two-stage process of technology transmission (figure 2.4).19 At stage 1, knowledge flows from European technology frontier firms, which we assume are the GVC lead firms, to the national frontier firms. National frontier firms are suppliers that are deeply integrated in the production process of the GVC lead: they carry out core produc- tion processes for the lead firm. At stage 2, knowledge flows from the national frontier firms to the rest of the domestic firms, predominantly through domestic firm-to-firm and firm-to-worker linkages. Domestic networks are fueled by the out- sourcing of noncore activities of the value chain. There is also a direct channel from foreign companies to the medium-to-low productivity firms in peripheral economies, which, however, grants lower technol- ogy transfer. The engagement most likely involves noncore functions, requires capabilities that are basic in nature, and is based on arm’s length trade relation- ships. This is a critically different type of engagement from the one the GVC lead entertains with the national frontier firms, which is based on deep relationships that involve ownership or licensing, franchising, joint ventures, strategic alliances, or other forms of nonequity modes of investment. It is the deep nature of 100  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 2.4  How technology flows from European frontier firms (global value chain lead firms) to the remaining European firms STAGE 1: National Remaining GVC Integration of frontier: firms, • Predominantly through frontier: national firms in core STAGE 2: Top-productivity exposed to domestic networks Global lead production processes Outsourcing of firms technology firms in of GVC leads noncore • Some direct access to participating from non-CEE capabilities foreign multinationals Exposure to directly in national countries learning GVCs frontier Basic capabilities Irregular engagement of FDI with nonnational frontier firms on noncore GVC activities requiring noncore activities Note: CEE = Central and Eastern Europe; FDI = foreign direct investment; GVC = global value chain. engagement that allows the faster, more sizable, and knowledge-intensive transfer of capabilities between countries. Convergence to the technological frontier and participation in GVCs are correlated. Technology transfers are more intense in the context of complex firm ­ engagements, and both the degree of connectivity of firms at the receiving end of technology transfer and the degree of sophistication of the trade partner matter. The pecking order is summarized in figure 2.5. Firms far away from the technology frontier tend to have no engagement with GVCs or intermittent engagement through executing noncore, often low-value-added activities using basic capabili- ties requirements. Engaging in GVCs for these firms requires facing entry costs to align their goals and processes with those of the buyer. These firms tend to be most active in domestic production and sales, and they mainly compete on price. More dynamic domestic firms, both in terms of productivity and learning, tend to have a more stable engagement in GVCs. These firms need to manage greater size and complexity, and pay greater attention to the quality of their products. The more technologically sophisticated they are, the more likely they are to have a strong interdependence with their parent firms, as such collaboration requires sharing valuable, proprietary intellectual property. For such firms learning and absorbing know-how becomes fundamental. They also benefit from feedback loops from exposure to new technologies, skills, and processes. As companies upgrade toward the technological frontier (national, regional, or global) they increasingly focus on organizational capital, innovation, and high-value-added activities as core competences. This, and a strong focus on R&D and quality, allows these firms to create disruptive technologies and to achieve intersectoral upgrading. GVC participation is high in the EU, and particularly high in CEE countries (figure 2.6). A similar indicator shows that import intensity is exceptionally high for this region. The intensity of imports is measured as the ratio of all GVC- related imports to the value of final products, weighted by final output. Knowledge Transfers from International Openness in Trade and Investment: The European Case ●  101 FIGURE 2.5  Firms’ international connectivity and technology transfer follow three stages Mature engagement • Focus on organizational capital, innovation, and high-value-added services activities as core competences Upgrading • Outsourcing of noncore competences • Stable engagement • Strong focus on research and development • Need to manage greater size and complexity and quality • Greater attention to the quality of products • Top-productivity in the sector • Medium- to high-productivity firms in the sector Connection • Strong interdependence with parent firm • Irregular engagement • Facing entry costs (to align goals and • --> Feedback loops due to exposure to processes with buyer) new technologies, skills, and processes • Few basic capabilities required • --> Learning and absorbing know-how • Low- to medium-productivity firms in becomes fundamental a given sector • --> Role in domestic production networks 0.7 FIGURE 2.6  GVC participation is particularly high in Central and Eastern Europe 0.6 GVC participation, 2000–14 0.5 0.4 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 CEECs EU28 World Source: World Input-Output Database (WIOD). Note: Central and Eastern European countries (CEECs) include those for which data are available from the WIOD, including Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, the Slovak Republic, and Slovenia. Global value chain (GVC) participation is measured as the share (not in percent) in gross exports of the sum of two measures: domestic value added in third- country exports (forward GVC participation) and foreign value added in own exports (backward GVC participation). EU28 includes all EU member countries. While import intensity flattened out in 2008 at the global level, it continued growing in the EU until 2012. Import intensity of production in CEE countries has also stagnated since 2012, but overall it has increased dramatically (figure 2.7). The decline in trade with the global economic crisis may have contributed to the slowdown in productivity growth by reducing firms’ opportunities to learn through engagement with GVCs. Total factor productivity growth in the CEE members of the EU was 8.2 percentage points lower during 2008–14 compared to 2000–07 (figure 2.8). As documented in Chiacchio, Gradeva, and Lopez-Garcia (2018), econometric estimates of TFP growth in these countries, controlling for country-sector fixed effects, confirm that TFP growth was lower 102  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 2.7  Import intensity 0.15 varies over time for Central and Eastern European EU countries 0.10 Import intensity growth relative to 2000 0.05 0 –0.05 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Source: World Input-Output Database (WIOD). Note: Central and Eastern European countries (CEECs) include those for which data are available from the WIOD, including Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, the Slovak Republic, and Slovenia. Import intensity is measured by coefficients of variation, relative to 2000, calculated on the basis of regressions that control for country-sector fixed effects, in which sectors are relative to the end product. The data points represent the estimated coefficients for the year dummies, and the bars represent the 95 percent confidence intervals for the year dummy regression coefficients. EU = European Union. FIGURE 2.8  Productivity growth was lower in Central and Eastern Europe during the crisis Difference between labor productivity growth in Central European EU countries and that in Eastern European EU countries, 2000–07 versus 2008–14 2 0 –2 –4 –6 –8 –10 Latvia Estonia Lithuania Slovenia Romania Czech Slovak Bulgaria Croatia Hungary Non-CEE Poland Republic Republic EU Capital deepening Labor quality Total factor productivity Labor productivity Source: Calculations based on Conference Board data. Note: “Non-CEE EU” refers to the unweighted average for Austria, Belgium, Cyprus, Denmark, Finland, France, Germany, Ireland, Italy, Luxembourg, Malta, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom. CEE = Central and Eastern Europe; EU = European Union. Knowledge Transfers from International Openness in Trade and Investment: The European Case ●  103 during the crisis (2008–10) and postcrisis period (2010–12) compared to the period 2000–07 by 8.2 percentage points and 2.3 percentage points, on aver- age. The assumption explored in that paper, which is the basis for the current analysis, is that the slowdown in integration through global production net- works may have slowed the transmission of technology diffusion within the EU, thus slowing productivity growth, particularly in the less-advanced EU countries. TFP growth in the CEE members of the EU is strongly correlated with GVC activity. TFP growth at the sectoral level in CEE-EU countries over the period 2000–12 is explained by technology creation at the GVC frontier and by the lagged gap in TFP with the frontier, consistently with neo-Schumpeterian mod- els. A 10 percent increase in TFP growth at the EU GVC frontier increases TFP growth of national frontier firms by 4.8 ­ percent, and a 10 percent gap in TFP explains another 5.2 percent of overall sectoral TFP growth (Chiacchio, Gradeva, and Lopez-Garcia 2018).20 Estimates of the ­ determinants of TFP growth by national frontier firms confirm these more aggregate results. Furthermore, the capacity to learn from parent firms, that is, the absorptive capacity of host firms, decreased by about 10 percent in the crisis period (2008–10) and postcrisis period (2010–12). The slowdown in TFP growth at the GVC frontier affected growth of national frontier firms only after 2010, but it did so severely, with half of the precrisis impact lost. Sectors with higher GVC growth were more resilient to the crisis and postcrisis slowdown. This indicates that strengthening GVC connectivity may have important growth and eco- nomic convergence effects. Econometric evidence also confirms the two-stage process of knowledge transmission (see above). TFP growth by laggard firms is affected much more by their exposure to national frontier firms than by exposure to GVCs. A 10 percent increase in TFP growth of national frontier firms leads to a 9.2 percent increase in TFP growth of laggards, while a 10 percent increase in TFP growth for GVC frontier firms only generates 1.5 percent additional growth in TFP for laggard firms. This latter effect is additional to the indirect effect that GVC frontier TFP growth has on laggards through boosting productivity growth of national frontier firms. Qualitatively similar results hold for middle-productiv- ity firms. All results are robust to a battery of tests on the presence of specific year outliers, base effects linked to the GVC level, or the choice of the GVC indicator. What is driving the decline in productivity growth in the Central and Eastern part of the EU since 2008? Firms’ ability to innovate depends on their own investment in innovation, R&D, and human capital (Cohen and Levinthal 1989, 1990, 1994; Griffith, Redding, and Van Reenen 2004; Lopez-Garcia and Montero 2012). The decline in connections with GVCs as a result of the crisis reduced the return on such investment in intangibles by CEE-EU frontier firms. This led to a fall in R&D spending as a share of GDP and in firms’ propensity to introduce new products or processes, as shown in the World Bank’s Enterprise Surveys. Econometric evidence at the sectoral level confirms this hypothesis, finding that the drop in investment in intangibles is limited to R&D-intensive sectors only. 104  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Conclusion Trade and FDI offer significant opportunities for firms to increase their productivity through technology transfers. The economic literature provides an extensive dis- cussion of the theoretical benefits for developing-country firms of international interactions. The empirical evidence on such effects is mixed, in part because technology transfers from partner countries may depend on distance, the level of development, and the quality of institutions. Our empirical results suggest, both globally and for ECA, that trade with, and FDI from, countries that do more R&D are associated with increased productivity. However, the impact can be reduced by the extent of geographical or cultural distance between the two countries, and developing countries benefit from spillovers only from FDI originating in developed countries. An important technology-diffusion channel in the EU is through production sharing through GVCs, which can encompass both trade and FDI. Frontier and laggard firms in host countries benefit from GVC participation in different ways. Frontier firms benefit from learning that ranges from managerial practices to orga- nization of the supply chain to access to advanced cutting-edge research. Laggards benefit from contact with national frontier firms and also, to a lesser extent, from direct contact with parent companies. Learning focuses more on efficient identifi- cation of the firm’s own niche, efficient use of inputs, and adaptation of own prod- uct and processes to fit more smoothly in advanced production. The global economic crisis reduced international trade, thus slowing down EU firms’ engagement in GVCs. The benefit EU firms derived from cross-border knowledge flows therefore fell, which made an important contribution to the decline in productivity growth in the region. This underlines the importance of maintaining open policies toward foreign trade and investment to support the productivity growth of firms. Notes 1. In most developing countries well below 5 percent of firms do any patenting (Cirera and Maloney 2017). While the number grows as we approach the technological frontier, it remains below 25 percent for most countries, with the exception of Japan and Germany, where more than 40 percent of firms sampled report patenting. Even Australia, Singapore, the United Kingdom, and the United States appear to have relatively few firms patenting (about 10 percent). Bloom and Van Reenen (2002) report that 72 percent of all patents originated from as few as 12 firms, out of their sample of 59,919 UK firms. 2. The UK Community Innovation Survey is an official survey of businesses on innovation outputs, innovation inputs, and sources of knowledge for innovation efforts. 3. Sample of firms with at least 20 employees. 4. Note that the OECD average includes the European Union countries. 5. Openness as a source of knowledge spillovers and growth has long been acknowl- edged in the literature. Grossman and Helpman (1991) study the growth performance of a small country in which scientific and technological knowledge flows from abroad and these flows are related to its extent of foreign trade. In this environment, trade generates an externality that coexists with the externality of domestic innovation. They show that domestic innovation produces both positive and negative externalities, and Knowledge Transfers from International Openness in Trade and Investment: The European Case ●  105 the existence of negative externalities leads to an undersupply of innovation overall. Policies that reduce the extent of international trade contribute to the undersupply of innovation. By contrast, some trade-promoting policies accelerate growth and raise national welfare because they offset some of the harmful negative externalities from domestic innovation. 6. Propensity score matching is a popular approach to estimating causal treatment effects. It applies for all situations in which one has a treatment, a group of treated individuals, and a group of untreated individuals. This technique attempts to estimate the effect of a treatment, policy, or other intervention by accounting for the covariates that predict receiving the treatment. 7. The paper provides an explanation for the relationship between trade opening, export- ing, and the increased demand for skills. Exporters have higher innovation and skill intensity for both production and nonproduction occupations. New technologies reduce the variable costs of production but induce higher fixed costs. As a consequence, a pars- ing between firms takes place: the more skill-intensive new technologies are adopted by the more productive producers who also tend to export. On the other hand, middle- productivity firms export but do not adopt the advanced technologies. Meanwhile, mar- ket reallocation effects induce the least productive firms to downgrade skills. 8. A firm’s productivity depends on how production is organized (Caliendo and Rossi- Hansberg 2012). Heterogeneity in demand, which is likely to increase when one firm serves more than one market, leads to heterogeneity in productivity and expanding skill composition. Moreover, a larger firm uses more than one layer of managers: the higher they are in the hierarchy, the less-common problems they tackle, and the higher their skills. Adding an extra layer can be thought of as reducing the marginal cost of the firm in exchange for increasing the fixed cost of acquiring and communicating knowledge. 9. Despite the opportunities and incentives, in many cases spillovers do not occur because domestic firms lack the complementary capabilities that would allow them to accumulate knowledge. 10. While studies focusing on developing countries produce mixed results concerning the presence of positive spillovers (e.g., Djankov and Hoekman 2000; Konings 2001), a more optimistic picture emerges from studies on industrial nations (e.g., Haskel, Pereira, and Slaughter 2002; Keller and Yeaple 2003). For a review of this literature, see Lipsey (2004). 11. There are exceptions to this. One example is Tang and Altshuler (2015), who examine home effects of outward FDI from the United States and find evidence of positive and significant spillovers flowing from multinational customers to their domestic suppliers. 12. Kogut and Chang (1991) and Yamawaki (1993), for example, show that Japanese firms tend to enter the United States and European markets by acquiring domestic firms when Japanese parent companies are at a technological disadvantage relative to US and European firms. 13. They find that spillovers are higher when foreign investors form joint ventures, and when the technological gap of the investor with respect to the host economy is rela- tively narrow (suggesting that domestic firms will have the absorptive capabilities to learn from the multinationals) (see Iršová and Havránek 2013). 14. In the context of Indonesia, for example, Blalock and Gertler (2008) find evidence of positive vertical spillovers from increased FDI in downstream activities of the manufac- turing sector. Focusing on Lithuania, Javorcik (2004) provides evidence of positive pro- ductivity spillovers from FDI taking place through interactions between foreign affiliates and their local suppliers in upstream sectors. Most recent literature on vertical spillovers through forward linkages has concentrated on how FDI in upstream services sectors affects the productivity of downstream manufacturers. In the Czech Republic, for example, Arnold, Javorcik, and Mattoo (2007) find sizable effects on productivity of ­ increased foreign entry into upstream services. 106  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia 15. Amann and Virmani (2014), for example, show that developing countries benefit more when technology-rich countries invest in them than the other way around. 16. Such sequencing is in line with neo-Schumpeterian models (Aghion and Howitt 2006; Saia, Andrews, and Albrizio 2015) and models of technology diffusion in multiple stages (Bartelsman, Haltiwanger, and Scarpetta 2013; van der Wiel et al. 2008; Iacovone and Crespi 2010). 17. CompNet is a research network originally created in 2012 within the European System of Central Banks and is devoted to the analysis of competitiveness from a multidimensional perspective. The CompNet database is based mainly on administrative data from firm registries and is constructed following a micro-distributed approach due to the confiden- tial nature of firm-level information in most countries (Bartelsman, Haltiwanger, and Scarpetta 2004). The database provides harmonized cross-country information on all deciles of the distribution of a number of variables related to firm performance and com- petitiveness, including productivity, in a given country, sector, and year. In total, CompNet covers about 18 EU countries, 9 macrosectors including manufacturing and construction, and 7 service sectors for the period 2001–13. For more information on the data set and coverage, please refer to Lopez-Garcia, di Mauro, and the CompNet Taskforce (2015). 18. Defined roughly as one-digit sectors according to the NACE (Nomenclature des Activités Économiques dans la Communauté Européenne, the European Classification of Economic Activities) rev. 2 classification system. 19. The mapping between non-CEE EU countries and CEE countries is done according to the relative importance of each non-CEE EU country in total intermediate imports of CEE countries by sector and year. 20. See Chiacchio, Gradeva, and Lopez-Garcia (2018) for the detailed results. References Aghion, P., and P. Howitt. 2006. “Joseph Schumpeter Lecture: Appropriate Growth Policy; A Unifying Framework.” Journal of the European Economic Association 4 (2–3): 269–314. Amann, E., and S. Virmani. 2014. “Foreign Direct Investment and Reverse Technology Spillovers: The Effect on Total Factor Productivity.” OECD Journal: Economic Studies 2014: 129–53. Amiti, M., and J. Konings. 2007. “Trade Liberalization, Intermediate Inputs, and Productivity: Evidence from Indonesia.” American Economic Review 97 (5): 1611–38. Arkolakis, C., and M. Muendler. 2010. “The Extensive Margin of Exporting Products: A Firm-Level Analysis.” NBER Working Paper 16641, National Bureau of Economic .edu/~ka265/research/MultiProduct/ Research, Cambridge, MA. http://aida.econ.yale​ Arkolakis_Muendler_products.pdf. Arnold, M. J., and B. S. Javorcik. 2009. “Gifted Kids or Pushy Parents? Foreign Direct Investment and Plant Productivity in Indonesia.” Journal of International Economics 79 (1): 42–53. Arnold, J. M., B. S. Javorcik, M. Lipscomb, and A. Mattoo. 2010. “Services Reform and Manufacturing Performance: Evidence from India.” Discussion Paper DP8011, Centre for Economic Policy Research, London. Arnold, M. J., B. S. Javorcik, and A. Mattoo. 2007. “Does Services Liberalization Benefit Manufacturing Firms? Evidence from the Czech Republic.” Policy Research Working Paper 4109, World Bank, Washington, DC. https://openknowledge.worldbank.org​ /­handle/10986/6882. Audretsch, D., and M. P. Feldman. 1996. “R&D Spillovers and the Geography of Innovation and Production.” American Economic Review 86 (3): 630–40. Knowledge Transfers from International Openness in Trade and Investment: The European Case ●  107 Aw, B. Y., M. J. Roberts, and T. Winston. 2007. “Export Market Participation, Investments in R&D and Worker Training, and the Evolution of Firm Productivity.” World Economy 30 (1): 83–104. http://doi.wiley.com/10.1111/j.1467-9701.2007.00873.x. Bartelsman, E. J., J. Haltiwanger, and S. Scarpetta. 2004. “Microeconomic Evidence of Creative Destruction in Industrial and Developing Countries.” Discussion Paper 1374, Institute of Labor Economics (IZA), Bonn, Germany. ———. 2013. “Cross-Country Differences in Productivity: The Role of Allocation and Selection.” American Economic Review 103 (1): 305–34. Bastos, P., and J. Silva. 2010. “The Quality of a Firm’s Exports: Where You Export to Matters.” Journal of International Economics 82 (2): 99–111. Benli, M. 2016. “Productivity Spillovers from FDI in Turkey: Evidence from Quantile Regressions.” Theoretical and Applied Economics 3 (608): 177–96. Bernard, A. B., J. B. Jensen, and P. K. Schott. 2005. “Importers, Exporters, and Multinationals: A Portrait of Firms in the U.S. that Trade Goods.” NBER Working Paper 11404, National Bureau of Economic Research, Cambridge, MA. http://www.nber.org​ /papers/w11404. Blalock, G., and P. J. Gertler. 2008. “Welfare Gains from Foreign Direct Investment through Technology Transfer to Local Suppliers.” Journal of International Economics 74 (2): 402–21. Bloom, N., K. Manova, J. Van Reenen, S. Sun, and Z. Yu. 2018. “Managing Trade: Evidence from China and the US.” NBER Working Paper 24718, National Bureau of Economic Research, Cambridge, MA. Bloom, N. and J. Van Reenen. 2002. “Patents, Real Options and Firm Performance.” Economic Journal 112: C97–C116. Bøler, E. A., A. Moxnes, and K. H. Ulltveit-Moe. 2015. “R&D, International Sourcing and the Joint Impact on Firm Performance.” American Economic Review 105 (12): 3660–703. Brambilla, I., R. Dix-Carneiro, D. Lederman, and G. Porto. 2012. “Skills, Exports and the Wages of Seven Million Latin American Workers.” World Bank Economic Review 26 (1): 36–60. Brandt, L., G. T. Rawski, and J. Sutton. 2008. “China’s Industrial Development.” In China’s Great Economic Transformation, ed. L. Brandt and G. T. Rawski, 593–604. Cambridge, UK: Cambridge University Press. http://kczx​ .xhu.edu.cn/G2S/eWebEditor/uploadfile​ /20120427135908_684449439144.pdf. Bustos, P. 2011. “Trade Liberalization, Exports, and Technology Upgrading: Evidence on the Impact of MERCOSUR on Argentinian Firms.” American Economic Review 101 (1): 304–40. Cagé, J., and D. Rouzet. 2015. “Improving ‘National Brands’: Reputation for Quality and Export Promotion Strategies.” Journal of International Economics 95 (2): 274–90. Caliendo, L., F. Monte, and E. Rossi-Hansberg. 2015. “The Anatomy of French Production Hierarchies.” Journal of Political Economy 123 (4): 809–52. Caliendo, L., and E. Rossi-Hansberg. 2012. “The Impact of Trade on Organization and Productivity.” Quarterly Journal of Economics 127 (3): 1393–1467. http://qje.oxford​ journals.org/content/early/2012/04/13/qje.qjs016.abstract. Chiacchio, F., K. Gradeva, and P. Lopez-Garcia. 2018. “The Post-crisis TFP Growth Slowdown in CEE Countries: Exploring the Role of Global Value Chains.” Working Paper 2143, European Central Bank, Frankfurt, Germany. Cirera, Xavier, and William F. Maloney. 2017. The Innovation Paradox: Developing-Country Capabilities and the Unrealized Promise of Technological Catch-Up. Washington, DC: World Bank. Coe, N. M., P. Dicken, and M. Hess. 2008. “Global Production Networks: Realizing the Potential.” Journal of Economic Geography 8 (3): 271–95. Coe, D. T., and E. Helpman. 1995. “International R&D Spillovers.” European Economic Review 39 (5): 859–87. 108  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Cohen, W. M., and D. A. Levinthal. 1989. “Innovation and Learning: The Two Faces of R&D.” Economic Journal 99: 569–96. ———. 1990. “Absorptive Capacity: A New Perspective on Learning and Innovation.” Administrative Science Quarterly 35: 128–52. ———. 1994. “Innovation Fortune Favours the Prepared Firm.” Management Science 40 (2): 227–51. Crespi, Gustavo, Chiara Criscuolo, Jonathan E. Haskel, and Matthew Slaughter. 2008. “Productivity Growth, Knowledge Flows, and Spillovers.” NBER Working Paper 13959, National Bureau of Economic Research, Cambridge, MA. Criscuolo, C., and R. Martin. 2009. “Multinationals and U.S. Productivity Leadership: Evidence from Great Britain.” Review of Economics and Statistics 91 (2): 263–81. Criscuolo, Paola, and Bart Verspagen. 2008. “Does It Matter Where Patent Citations Come From? Inventor vs. Examiner Citations in European Patents.” Research Policy 37 (10): 1892–908. De Loecker, J. 2013. “A Note on Detecting Learning by Exporting.” American Economic Journal: Microeconomics 5 (3): 1–21. https://www.aeaweb.org/articles.php​ ? doi​ =​10.1257/mic.5.3.1. Djankov, S., and B. Hoekman. 2000. “Foreign Investment and Productivity Growth in Czech Enterprises.” World Bank Economic Review 14 (1): 49–64. Duggan, V., S. Rahardja, and G. Varela. 2013. “Service Sector Reform and Manufacturing Productivity: Evidence from Indonesia.” Policy Research Working Paper 6349, World Bank, Washington, DC. https://ssrn.com/abstract=2210300. Eslava, M., J. Tybout, D. Jinkins, C. Krizan, and J. Eaton. 2015. “A Search and Learning Model of Export Dynamics.” Paper presented at 2015 Annual Meeting, Society for Economic Dynamics, Warsaw, June 25–27. Feenstra, R. C., and G. H. Hanson. 1996. “Globalization, Outsourcing, and Wage Inequality.” American Economic Review 86 (2): 240–45. Fernandes, A., and C. Paunov. 2012. “Foreign Direct Investment in Services and Manufacturing Productivity: Evidence for Chile.” Journal of Development Economics 97 (2): 305–21. Fieler, A., M. Eslava, and D. Y. Xu. 2018. “Trade, Technology and Input Linkages: A Theory with Evidence from Colombia.” American Economic Review. 108 (1): 109–46 Fitzgerald, D., S. Haller, and Y. Yedid-Levi. 2015. How Exporters Grow. Staff Report 524, Federal Reserve Bank of Minneapolis, Minneapolis, MN. Girma, S., and K. Wakelin. 2007. “Local Productivity Spillovers from Foreign Direct Investment in the U.K. Electronics Industry.” Regional Science and Urban Economics 37 (3): 399–412. Goldberg, P., A. M. Khandelwal, N. Pavcnikh, and P. Topalova. 2010. “Imported Intermediate Inputs and Domestic Product Growth: Evidence from India.” Quarterly Journal of Economics 125 (4): 1727–67. Griffith, R., S. J. Redding, and H. Simpson. 2002. “Productivity Convergence and Foreign Ownership at the Establishment Level.” Discussion Paper 3765, Centre for Economic Policy Research, London. https://ssrn.com/abstract=388802. Griffith, R., S. J. Redding, and J. Van Reenen. 2004. “Mapping the Two Faces of R&D: Productivity Growth in a Panel of OECD Countries.” Review of Economics and Statistics 86 (4): 883–95. Griliches, Zvi. 1979. “Issues in Assessing the Contribution of Research and Development to Productivity Growth.” Bell Journal of Economics 10 (1): 92–116. Grossman, Gene M., and Elhanan Helpman. 1991. “Trade, Knowledge Spillovers, and Growth.” European Economic Review 35 (2–3): 517–26. Knowledge Transfers from International Openness in Trade and Investment: The European Case ●  109 Grossman, Gene M., and E. Rossi-Hansberg. 2008. “Trading Tasks: A Simple Theory of Offshoring.” American Economic Review 98 (5): 1978–97. Guadalupe, M., O. Kuzmina, and C. Thomas. 2012. “Innovation and Foreign Ownership.” American Economic Review 102 (7): 3594–27. http://ideas.repec.org/a/aea/aecrev​ v102y2012i7p3594-3627.html [Accessed November 14, 2015]. /­ Halpern, L., M. Koren, and A. Szeidl. 2015. “Imported Inputs and Productivity.” American Economic Review 105 (12): 3660–703. Haskel, J., S. Pereira, and M. J. Slaughter. 2007. “Does Inward Foreign Direct Investment Boost the Productivity of Domestic Firms?” Review of Economics and Statistics 89 (3): 482–96. Hummels, D., and A. Skiba. 2004. “Shipping the Good Apples Out? An Empirical Confirmation of the Alchian-Allen Conjecture.” Journal of Political Economy 112 (6): 1384–402. Iacovone, L., and G. A. Crespi. 2010. “Catching Up with the Technological Frontier: Micro- level Evidence on Growth and Convergence.” Industrial and Corporate Change 19 (6): 2073–96. Iršová, Z., and T. Havránek. 2013. “Determinants of Horizontal Spillovers from FDI: Evidence from a Large Meta-analysis.” World Development 42 (C): 1–15. Jaffe, Adam B., and M. Trajtenberg. 2005. Patents, Citations, and Innovations: A Window on the Knowledge Economy. Cambridge, MA: MIT Press. Jaffe, A., M. Trajtenberg, and R. Henderson. 1993. “Geographic Localization of Knowledge Spillovers as Evidenced by Patent Citations.” Quarterly Journal of Economics 108 (3): 577–98. Javorcik, B. S. 2004. “Does Foreign Direct Investment Increase the Productivity of Domestic Firms? In Search of Spillovers through Backward Linkages.” American Economic Review 94 (3): 605–27. Javorcik, B. S., K. Saggi, and M. Spatareanu. 2004. “Does It Matter Where You Come From? Vertical Spillovers from Foreign Direct Investment and the Nationality of Investors.” Policy Research Working Paper 3449, World Bank, Washington, DC. https:// ssrn.com/abstract=625327. Kasahara, H., and B. Lapham. 2013. “Productivity and the Decision to Import and Export: Theory and Evidence.” Journal of International Economics 89 (2): 297–316. Kee, H. L. 2015. “Local Intermediate Inputs and the Shared Supplier Spillovers of Foreign Direct Investment.” Journal of Development Economics 112 (C): 56–71. Keller, W., and S. R. Yeaple. 2009. “Multinational Enterprises, International Trade, and Productivity Growth: Firm-Level Evidence from the United States.” Review of Economics and Statistics 91 (4): 821–31. Kogut, B., and S. J. Chang. 1991. “Technological Capabilities and Japanese Foreign Direct Investment in the United States.” Review of Economics and Statistics 73 (3): 401–13. Konings, J. 2001. “The Effects of Foreign Direct Investment on Domestic Firms.” Economics of Transition 9 (3): 619–33. Koren, M., and M. Csillag. 2011. “Machines and Machinists: Capital-Skill Complementarity from an International Trade Perspective.” IEHAS Discussion Paper MT-DP - 2011/14. Institute of Economics, Hungarian Academy of Sciences. http://ideas.repec.org/p/has​ /­discpr/1114.html. Lipsey, R. E. 2004. “Home and Host Country Effects of FDI.” In Challenges to Globalization, ed. Robert E. Baldwin and L. Alan Winters. Chicago: University of Chicago Press. Lopez-Garcia, P., F. di Mauro, and the CompNet Taskforce. 2015. “Assessing European Competitiveness: The New CompNet Microbased Database.” Working Paper 1764, European Central Bank, Frankfurt. 110  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Lopez-Garcia, P., and J. M. Montero. 2012. “Spillovers and Absorptive Capacity in the Decision to Innovate of Spanish Firms: The Role of Human Capital.” Economics of Innovation and New Technology 21 (7): 589–612. MacGarvie, M. 2006. “Do Firms Learn from International Trade?” Review of Economics and Statistics 88 (1): 46–60. Manova, K., and Z. Zhang. 2012. “Export Prices across Firms and Destinations.” Quarterly Journal of Economics 127 (1): 379–436. Martin, J. 2012. “Markups, Quality, and Transport Costs.” European Economic Review 56 (4): 777–91. Matsuyama, K. 2007. “Beyond Icebergs: Towards a Theory of Biased Globalization.” Review of Economic Studies 74 (1): 237–53. Ramondo, N. 2009. “Foreign Plants and Industry Productivity: Evidence from Chile.” Scandinavian Journal of Economics 111 (4): 789–809. Saia, A., D. Andrews, and S. Albrizio. 2015. “Productivity Spillovers from the Global Frontier and Public Policy: Industry-Level Evidence.” Economics Department Working Paper 1238, Organisation for Economic Co-operation and Development, Paris. Schott, P. K. 2008. “The Relative Sophistication of Chinese Exports.” Economic Policy 23: 5–49. ¸ eker, M. 2012. “Importing, Exporting, and Innovation in Developing Countries.” Review S of International Economics 20 (2): 299–314. Silajdzic, S., and E. Mehic. 2015. “Knowledge Spillovers, Absorptive Capacities and the Impact of FDI on Economic Growth: Empirical Evidence from Transition Economies.” Procedia—Social and Behavioral Sciences 195 (3): 614–23. Sutton, J. 2007. “Quality, Trade and the Moving Window: The Globalisation Process.” Economic Journal 117 (524): F469–98. Tang, J., and R. Altshuler. 2015. “The Spillover Effects of Outward Foreign Direct Investment on Home Countries: Evidence from the United States.” Working Paper 1503, Oxford University Centre for Business Taxation. http://dx.doi.org/10.2139​ /­ ssrn.2545129. van der Wiel, H., H. Creusen, G. van Leeuwen, and E. van der Pijll. 2008. “Cross Your Border and Look Around.” Paper presented at the Dynamics, Economic Growth, and International Trade (DEGIT) Conference, Manila. van Pottelsberghe de la Potterie, B., and F. Lichtenberg. 2001. “Does Foreign Direct Investment Transfer Technology across Borders?” Review of Economics and Statistics 83 (3): 490–97. Verhoogen, E. A. 2008. “Trade, Quality Upgrading, and Wage Inequality in the Mexican Manufacturing Sector.” Quarterly Journal of Economics 123 (2): 489–530. Yamawaki, H. 1993. “Location Decisions of Japanese Multinational Firms in European Manufacturing Industries.” In European Competitiveness, ed. K. Hughes. Cambridge, UK: Cambridge University Press. Yeaple, S. R. 2013. “The Multinational Firm.” Annual Review of Economics 5 (1): 193–217. Attracting Foreign Direct Investment: The Role of Deep Preferential Trade Agreements ●  111 SPOTLIGHT 2 Attracting Foreign Direct Investment: The Role of Deep Preferential Trade Agreements I nternational trade agreements can be an important means of encouraging foreign direct investment (FDI) flows.1 Chapter 2 shows that FDI is often associ- in more distant regions that are otherwise less con- nected with the EU15 than for investments in its immediate vicinity. ated with transfers of knowledge that contribute to growth, underlining the importance of policies that attract FDI flows. This spotlight provides an empirical Deep PTAs in ECA: A Snapshot analysis of how deep preferential trade agreements (PTAs)—those that go beyond the reduction of bor- PTAs signed by countries in Europe and Central Asia der restrictions to trade to reduce behind-the-border (ECA) represent 40 percent of total active agree- barriers as well as harmonize regulations and ments. Countries around the world have increased ­ standards—encourage greater FDI flows. This is an their participation in PTAs, especially in the past two important illustration of how policies that encourage decades. From the 1950s onward, the number of one connectivity channel (trade) can affect another active PTAs increased more or less continuously to channel (FDI). almost 70 in 1990. Thereafter, PTA activity acceler- The main results suggest that PTAs going ated noticeably, with the number of PTAs more than beyond tariff liberalization and including disciplines doubling over the next five years and more than qua- in the area of trade, investment, and the business drupling until 2010 to reach close to 280 PTAs pres- environment, among others, are important for FDI ently in force. ECA countries are considered the most attraction. Each provision added to an agreement integrated in terms of the number of agreements between a pair of countries is associated with an signed. A total of 111 agreements have been signed average 3 percent increase in FDI flows between by ECA members, either between them or with other that pair, underlining the importance of participa- countries (see map S2.1). tion in PTAs (participation in PTAs is discussed in ECA agreements have become deeper over chapter 7). This positive impact is mainly driven by time. Agreements signed before 1991 included on policies governing competition, investment, move- average 9 provisions, whereas agreements signed ment of capital, and intellectual property rights, between 2005 and 2015 included on average which are key drivers of FDI flows. The impact of 15 provisions. Analysis based on the new World Bank deep agreements on FDI matters for manufacturing data set on the content of PTAs (see annex S2A) and services, but not for natural resources. Deep shows that the treaty establishing the Europen agreements seem to be more helpful in attracting Community and the EU enlargements are the deep- FDI from more culturally distant destinations within est agreements that have been signed and incorpo- the manufacturing sector, emphasizing the role of rate a total of 44 legally enforceable provisions, agreements in facilitating learning. For the ECA including all provisions that fall under the current region, the relevance of deep agreements in stimu- mandate of the World Trade Organization (WTO) lating FDI depends on the origin and the destina- (we refer to these provisions as “WTO”) and 30 dis- tion of the flows: for FDI originating in EU15 ciplines that go beyond the current WTO mandate countries, agreements matter more for investments (referred to as “WTO+”). PTAs signed by non-EU 112  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia SPOTLIGHT 2 continued MAP S2.1   The European Union and North America show the deepest forms of integration Number of agreements by country, 2015 ] 31–44 ] 21–30 11–20 5–10 2–5 0–1 Source: Calculations based on World Trade Organization, Regional Trade Agreement data set. FIGURE S2.1  The European Union shows the greatest depth of agreements among ECA country groups PTA coverage, by subgroups 50 80 45 70 69 40 60 Number of agreements Number of provisions 35 30 50 43 43 25 40 20 30 15 26 20 10 5 10 0 0 DCFTA EU13 EU15 Other ECA Maximum depth Average depth Number of agreements subgroups, see annex S2A. Source: Calculations based on World Bank data set on the content of PTAs. For definitions of the ­ Note: DCFTA = deep and comprehensive free trade agreement; ECA = Europe and Central Asia; EU = European Union; PTA = preferential trade agreement. members tend to be shallower, and include on aver- have been very important in attracting FDI and facili- age 9 provisions. Within non-EU members, the tating technology transfers across countries (see deepest agreement has been signed with EU coun- results below). In the case of non-EU non-DCFTA tries and includes 44 disciplines. Deep and compre- countries, the maximum number of disciplines hensive free trade agreements with the EU (DCFTAs) that has been included in an agreement is 22 Attracting Foreign Direct Investment: The Role of Deep Preferential Trade Agreements ●  113 FIGURE S2.2  Sectoral and customs-related provisions are the most frequent WTO provisions in ECA PTAs Percentage of ECA PTAs including WTO provisions 120 100 80 60 40 20 0 al e ms s g id de s gh f y t s n su ed Ri s o en xe re ar ise ur pin er t o tri eA ra ea lat sto su nit ult em ts es T ices s Ta rty ect n S en us pr um re oT t M -Re at ea sa ric Cu er Ind ur e i em rt op Asp St v tid st gM po yto nt Ag oc en de ad re A An er gE Ex r tm ra Ph al ed cP Tr l Ag FT A ilin rri e din FT tu lat Ba nd bli rva ra ra ec Re Pr ya Pu al ne te eT nic ell e ar Ge un Int Trad at nit Inv ch Co St Sa Te Included Legally enforceable Source: Calculations based on World Bank data set on PTA content. Note: ECA = Europe and Central Asia; FTA = free trade agreement; PTA = preferential trade agreement; WTO = World Trade Organization. (see ­figure S2.1 and table S2A.1 for the country agreements signed by non-EU members include classification). them (see table S2A.2). PTAs signed by ECA countries cover policy Only a few WTO+ provisions, such as competi- areas that fall under the current mandate of the tion policy, movement of capital, and intellectual WTO and go beyond tariff reductions. More than property rights are included and legally enforceable 50 percent of the agreements signed by ECA in a relevant number of ECA trade agreements (see countries include WTO provisions such as export table S2A.3 for a list of WTO and WTO+ policy areas taxes, customs, state-owned enterprises, anti- in PTAs). Competition policy is covered and enforce- dumping, countervailing measures, and Trade- able in more than 80 percent of ECA agreements, Related Aspects of Intellectual Property Rights. followed by movement of capital and intellectual Disciplines such as the General Agreement on property rights, which are covered in 40 percent of Trade in Services, public procurement, Technical ECA agreements (see ­ figure S2.3). However, while Barriers to Trade, and Sanitary and Phytosanitary these two provisions are included in more than half are included in only 20–40 percent of the agree- of the agreements that are signed by EU members, ments (see figure S2.2). The coverage of WTO they are only covered by less than 30 percent of provisions is in general larger in PTAs signed by the agreements signed by non-EU members EU members compared with non-EU members. (see table S2A.4). Other provisions such as labor While more than 40 percent of agreements market regulation and investment, which are pres- signed by EU members include all WTO disci- ent in almost 20 percent of ECA agreements, are plines except for Trade Related Investment legally enforceable in, respectively, 17 percent and Measures (TRIMS), only 10 percent of the 15 percent of agreements, on average. 114  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia SPOTLIGHT 2 continued FIGURE S2.3  Among WTO+ provisions, Competition Policy, Movement of Capital, and Intellectual Property Rights are the most frequent WTO+ provisions included in ECA PTAs Percentage of ECA PTAs including WTO+ provisions 100 90 80 Share of agreements 70 60 50 40 30 20 10 0 op C cy ar Inv Rig l t R stm s u t Ind viro Ag atis n on st e cu s se ic Co l L e ch lic er s d T ia on hn ue gio oc E logy ox a C a y at ial pe ers Inf n of ssis tion tio isla e o n at Co ax ty Po an pera on Ille litica Tra tion Vi Imm ialog g an gra e sy n d M Mon er P He m ey rot alth -S Il unde ion d E it D ng Hu nter rugs ta R es Au rote hts V n rro al bli uc M ism dm r S ing Inn Ant nistr fety at rru on vil o n ot es n r M erty pita eg en ke e ht Ec u nm ri tic ar Po op aw pr Fin al l M erg St latio Re om rial nta ltur ma g nc n S tio n sa i u d A tio dio ctio Ci ion P ptio tio uc ral T cie lu Te isu Pr of oli Da man pris Pr lici an y D ati ec log i ga l D ini ize lic ri ov ico ati im nc oo tt ion o at La ect P ig c A lea in or Le ta r io A ra i a ec o n a al t P tu en on d ec em iti i a ell ov et l um Int M omp Pu N Re S n Ed ltu ns C En Cu ium Co bo ed La Ap an all Sm Included Legally enforceable Source: Calculations based on World Bank data set on PTA content. Note: ECA = Europe and Central Asia; PTA = preferential trade agreement; WTO = World Trade Organization. Linking Deep Agreements with Linking Deep Agreements with FDI: Empirical Strategy FDI: Results The relationship between deep agreements and The empirical results show that deeper agreements FDI flows is estimated using a structural gravity tend to encourage higher FDI flows: equation. An augmented gravity equation is esti- mated for 170 ­countries, using data from 2003 to • Pairs of countries that have signed deeper 2014 (table S2.5), to investigate the relationship agreements—that is, incorporating additional between the depth of an agreement and FDI (see legally enforceable provisions—have higher FDI the discussion of methodology in box S2A.1). The flows than those that have not signed them. depth of an agreement is captured by the number In particular, each provision added to an agree- of legally enforceable provisions that it includes. ment between a pair of countries is associated This methodology has been extensively used by with an average 3 percent increase in FDI flows economists to test empirically the determinants of between that pair2 (see figure S2.4, panel a, and trade flows and to estimate the effect of preferen- table S2A.6, ­column 1).3 tial trade opening on trade flows. The results • The effect of deep PTAs on FDI attraction from should be viewed as conditional correlations rather ECA countries is not different than the average than causal effects, as causation likely runs both global effect (see table S2A.7).4 ways: deep PTAs can encourage FDI by reducing • The impact of deep agreements on FDI flows is the costs involved, while pressure from firms significant for both the manufacturing and services involved in FDI is one reason that countries enter sectors, but not for natural resources (see into deeper trade agreements. ­figure S2.4, panel a, and table S2A.6, columns 2–4). Attracting Foreign Direct Investment: The Role of Deep Preferential Trade Agreements  ●  115 FIGURE S2.4  The impact of deep agreements on FDI a. Deeper agreements attract more FDI in all sectors but natural resources Change in FDI flows from each additional provision in a PTA 0.03 –0.01812 Number of provisions 0.03639 0.04249 –0.1 –0.05 0 0.05 Change in FDI flows All sectors Natural resources Goods Services b. Inclusion of WTO+ provisions in deep agreements is more systematically associated with increased FDI Foreign direct investment and depth, by group of provisions 0.016 0.05 Number of WTO provisions 0.053 0.036 0.023 Number of WTO+ provisions 0.036 –0.05 0 0.05 0.1 Change in FDI flows All sectors Goods Services Note: The bullets in the figure represent the estimated coefficients from the regressions, and the lines represent the 90 percent confidence intervals, which in turn represent the range of values that describe the uncertainty surrounding an estimate. Confidence intervals are one way to represent how “good” an estimate is. The larger a confidence interval for an estimate, the more caution required when using the estimate. FDI = foreign direct investment; PTA = preferential trade agreement; WTO = World Trade Organization. 116  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia SPOTLIGHT 2 continued The average impact of deep agreements is slightly One of the reasons deep agreements stimulate larger for manufacturing sectors with lower levels FDI is because they mitigate the costs of cultural of technological intensity, although the difference distance, thus facilitating learning. Distance—be it is not statistically significant (see table S2A.9).5 geographic, cultural, or institutional—reduces • While the inclusion of one additional WTO+ pro- home effects associated with outward FDI vision is associated with a 3.6 percent increase in (Driffield, Love, and Yang 2016). Indeed, cultural FDI flows, the impact of WTO provisions is not distance has been identified by multinationals as a statistically different from zero (see figure S2.4, key obstacle for knowledge transfers. Signing panel b, and table S2A.8). WTO+ provisions can bilateral investment treaties, by providing clearer encourage FDI in different ways: competition rules, reduces the costs of investing in markets that policy, by preventing the abuse of market power, are unfamiliar to investors (Gomez-Mera et al. enables multinational firms to optimally fragment 2014). However, deep agreements seem to be production internationally to take full advantage more helpful in facilitating FDI in manufacturing in of cross-country cost differentials; and provisions more culturally distant destinations, while the governing investment or intellectual property opposite is shown for services6 (see ­ figure S2.5 protect firm-specific assets such as human capital and table S2A.10). and intellectual property in which international The relevance of deep agreements in stimulat- firms may have a competitive advantage. ing FDI depends on the origin and the FIGURE S2.5  Deep agreements are more helpful in facilitating FDI in culturally distant destinations for manufacturing, while the opposite is true for services Foreign direct investment and cultural distance –0.002 0.0078 Depth × Distance –0.011 –0.02 –0.01 0 0.01 0.02 Change in FDI flows All sectors Goods Services Note: Cultural distance is measured by comparing country pairs on four “cultural dimensions.” According to Hofstede (2011), these dimensions describe typical characteristics of cultures: power distance index, individualism versus collectivism, masculinity versus femininity, and uncertainty avoidance. The bullets in the figure represent the estimated coefficients from the regressions, and the lines represent the 90 ­percent confidence intervals, which in turn represent the range of values that describe the uncertainty surrounding an estimate. Confidence intervals are one way to represent how “good” an estimate is. The larger a confidence interval for an estimate, the more caution required when using the estimate. FDI = foreign direct i ­nvestment. Attracting Foreign Direct Investment: The Role of Deep Preferential Trade Agreements ●  117 destination of the flows. Agreements matter more EU15 includes the 15 EU member states that for FDI from EU15 countries to more distant joined before 2004; these countries are also referred regions that are otherwise less connected with to as “EU core.” EU13 includes the 13 EU member the EU15 than for FDI to countries in its immedi- states that joined in or after 2004; they can also be ate vicinity. For example, the depth of agree- referred to as “EU noncore.” The non-EU members ments matters for cross-border investments of are separated into two subgroups: DCFTA, which EU15 companies in the world, but it does not includes the 3 countries with DCFTA with the EU, matter for those investments in ECA countries and Other ECA, which refers to the 16 countries (table S2A.11, column 1). In fact, the effect of that do not have a DCFTA with the EU. depth on the investment of EU15 in non-ECA countries is about twice the size of the average PTA Content Data Set effect of depth on all cross-­ border investments (table S2A.11, column 2). Interestingly, results The new World Bank data set on content of PTAs also differ across ECA countries. Depth does not is an extension of Horn, Mavroidis, and Sapir matter for investments from the EU15 into the (2010) and WTO (2011) data sets and contains EU28, while it does matter for investments from 2 8 0 P TA s s i g n e d b y 1 8 0 c o u n t r i e s the EU15 into non-EU ECA countries (table between 1980 and 2015 (Hofmann, Osnago, and S2A.11, columns 3 and 4). These results further Ruta 2017). point to the role of cultural and institutional dis- The methodology of Horn, Mavroidis, and tance in facilitating FDI, beyond the effect of Sapir (2010) is followed to define the content and agreements. the legal enforceability of PTAs (see box S2A.1). As a first step, a set of 51 policy areas covered in PTAs is identified. These areas can be classified Annex S2A. Definition of Country into two different groups. The first group is repre- Groups and Methodology sented by WTO provisions that fall under the cur- rent mandate of the WTO and are already subject This spotlight covers a total of 47 countries referred to some form of commitment in WTO agree- to as Europe and Central Asia (ECA). These coun- ments. The second group of policy areas, which is tries are divided into four groups: EU15, EU13, denoted as WTO+ provisions, includes those DCFTA, and other ECA, as shown in table S2A.1. obligations that are outside the current mandate TABLE S2A.1  Country Groups and Subgroups ECA EU28 Non-EU EU15 EU13 DCFTA non-DCFTA (Other ECA) Austria Italy Bulgaria Lithuania Georgia Albania FYR Macedonia Belgium Luxembourg Croatia Poland Moldova Armenia Montenegro Denmark Netherlands Cyprus Romania Ukraine Azerbaijan Russian Federation Finland Portugal Czech Slovak Republic Belarus Serbia France Spain Republic Slovenia Bosnia and Herzegovina Tajikistan Germany Sweden Estonia Malta Kazakhstan Turkey Greece United Kingdom Hungary Kosovo Turkmenistan Ireland Latvia Kyrgyz Republic Uzbekistan Note: DCFTA = deep and comprehensive free trade agreement; ECA = Europe and Central Asia; EU = European Union. 118  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia SPOTLIGHT 2 continued BOX S2A.1 Methodology for the Estimation of the Impact of Deep Integration on FDI Flows Gravity equations are derived from models that seek The following structural gravity regression is to explain or predict the relationship between a estimated for a set of 160 countriesa using Poisson (dependent) variable (in this case bilateral foreign pseudo–maximum likelihood (see Piermartini and direct investment [FDI]) and a set of other indepen- Yotov 2016): dent or explanatory variables whose values can be FDIijt = exp {b1Depthijt + dij + dit + djt}+eijt , estimated (in this case, elements of deep integration). Endogeneity occurs when both the variable being in which FDIijt is a measure of FDI between coun- explained (the left-hand-side variable in the equation) try i and j at time t. Depthijt is a measure of the and the explanatory variable (the right-hand-side depth of preferential trade agreements. A statisti- variable in the equation) may be determined by a cally significant and positive coefficient b1 implies third factor not in the model. For example, firms that that signing a deeper agreement is associated want to invest in a country may also lobby for with greater FDI. This variable is calculated as the free trade agreements. Consequently, a free trade number of enforceable provisions that are agreement may not increase FDI, but both FDI and included in a certain agreement. The d s are a free trade agreements may come about as a result of series of fixed effects: i for importer, j for exporter, perceived economic benefits of firms and their politi- and t is year from 2003 to 2014. Finally, eijt is the cal lobbying efforts. error term. a. See table S2A.5 for the list of countries that are included in the regression. TABLE S2A.2  Percentage of ECA PTAs Including WTO Provisions, by Subgroup EU15 EU13 DCFTA Other ECA Number of agreements 43 43 27 70 FTA Industrial 100 100 100 99 FTA Agriculture 100 100 100 99 Export Taxes 93 93 100 93 Customs 95 95 59 76 Antidumping 98 98 26 46 State Aid 79 79 37 60 Countervailing Measures 86 86 22 37 Trade-Related Aspects of Intellectual Property Rights 72 72 30 44 State Trading Enterprises 67 67 19 39 Public Procurement 49 49 22 21 Technical Barriers to Trade 40 40 22 34 Sanitary and Phytosanitary 40 40 15 26 General Agreement on Trade in Services 42 42 22 10 Trade-Related Investment Measures 19 19 7 3 Source: Calculations based on World Bank data set on PTA content. Note: ECA = Europe and Central Asia; FTA = free trade agreement; PTA = preferential trade agreement; WTO = World Trade Organization. Attracting Foreign Direct Investment: The Role of Deep Preferential Trade Agreements ●  119 TABLE S2A.3  WTO and WTO+ Policy Areas in PTAs Areas covered by the WTO Areas beyond the WTO (WTO+) Tariffs: Industrial Goods Anticorruption Health Tariffs: Agricultural Goods Competition Policy Human Rights Customs Administration Environmental Laws Illegal Immigration Export Taxes Intellectual Property Rights Illicit Drugs Sanitary and Phytosanitary Measures Investment Measures Industrial Cooperation State Trading Enterprises Labor Market Regulation Information Society Technical Barriers to Trade Measures Movement of Capital Mining Countervailing Measures Consumer Protection Money Laundering Antidumping Data Protection Nuclear Safety State Aid Approximation of Legislation Political Dialogue Public Procurement Agriculture Public Administration Trade-Related Investment Measures Audiovisual Regional Cooperation General Agreement on Trade in Services Civil Protection Research and Technology Trade-Related Aspects of Intellectual Innovation Policies Small and Medium-Sized Enterprises Property Rights Cultural Cooperation Social Matters Economic Policy Dialogue Statistics Education and Training Taxation Energy Terrorism Financial Assistance Visa and Asylum Source: World Bank data set on PTA content. Note: PTA = preferential trade agreement; WTO = World Trade Organization. TABLE S2A.4  Percentage of ECA PTAs Including WTO+ Provisions, by Subgroup DCFTA Other ECA EU15 EU13 (non-EU) (non-EU) Number of agreements 43 43 27 70 Competition Policy 84 84 93 86 Movement of Capital 63 63 19 21 Intellectual Property Rights 58 58 19 29 Statistics 19 19 48 27 Social Matters 37 37 4 3 Labor Market Regulation 30 30 4 9 Investment 28 28 7 7 Approximation of Legislation 28 28 7 4 Environmental Laws 30 30 4 3 Illegal Immigration 28 28 4 3 Visa and Asylum 26 26 7 1 Cultural Cooperation 28 28 4 0 Financial Assistance 26 26 7 0 Health 26 26 4 0 Energy 21 21 7 3 Industrial Cooperation 23 23 4 1 Education and Training 23 23 4 1 Research and Technology 23 23 4 1 Data Protection 21 21 7 1 Economic Policy Dialogue 23 23 4 0 Agriculture 19 19 4 9 continued 120  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia SPOTLIGHT 2 continued TABLE S2A.4  continued DCFTA Other ECA EU15 EU13 (non-EU) (non-EU) Number of agreements 43 43 27 70 Terrorism 21 21 4 1 Consumer Protection 14 14 4 3 Nuclear Safety 14 14 0 0 Regional Cooperation 9 9 4 0 Audiovisual 9 9 4 0 Taxation 7 7 4 3 Mining 7 7 4 0 Anticorruption 5 5 4 0 Civil Protection 5 5 4 0 Public Administration 5 5 4 0 Money Laundering 2 2 4 3 Illicit Drugs 2 2 4 1 Small and Medium-Sized 5 5 0 0 Enterprises Information Society 2 2 4 0 Political Dialogue 2 2 4 0 Innovation Policies 0 0 0 0 Human Rights 0 0 0 0 Source: Calculations based on World Bank data set on PTA content. Note: ECA = Europe and Central Asia; PTA = preferential trade agreement; WTO = World Trade Organization. TABLE S2A.5  Countries and Economies Included in the Estimations Albania Cambodia France Korea, Rep. Algeria Cameroon Gabon Kuwait Andorra Canada Gambia, The Kyrgyz Republic Angola Chile Georgia Lao PDR Antigua and Barbuda China Germany Latvia Argentina Colombia Ghana Lebanon Armenia Congo, Dem. Rep. Greece Libya Australia Congo, Rep. Guatemala Liechtenstein Austria Costa Rica Guyana Lithuania Azerbaijan Côte d’Ivoire Haiti Luxembourg Bahamas, The Croatia Honduras Macao SAR, China Bahrain Cyprus Hong Kong SAR, China Macedonia, FYR Bangladesh Czech Republic Hungary Madagascar Barbados Denmark Iceland Malawi Belarus Djibouti India Malaysia Belgium Dominican Republic Indonesia Mali Belize Ecuador Iraq Malta Bolivia Egypt, Arab Rep. Ireland Mauritius Bosnia and Herzegovina El Salvador Israel Mexico Botswana Equatorial Guinea Italy Micronesia, Fed. Sts. Brazil Eritrea Jamaica Moldova Brunei Estonia Japan Montenegro Bulgaria Ethiopia Jordan Morocco Burkina Faso Fiji Kazakhstan Mozambique Burundi Finland Kenya Myanmar continued Attracting Foreign Direct Investment: The Role of Deep Preferential Trade Agreements ●  121 TABLE S2A.5  continued Namibia Romania Sri Lanka Ukraine Nepal Russian Federation St. Lucia United Arab Emirates Netherlands Rwanda Sudan United Kingdom New Zealand Samoa Sweden United States Nicaragua San Marino Switzerland Uruguay Nigeria Saudi Arabia Syrian Arab Republic Uzbekistan Norway Senegal Taiwan, China Vanuatu Oman Serbia Tajikistan Venezuela, RB Pakistan Seychelles Tanzania Vietnam Panama Sierra Leone Thailand Yemen, Rep. Papua New Guinea Singapore Togo Zambia Peru Slovak Republic Trinidad and Tobago Zimbabwe Philippines Slovenia Tunisia Poland Solomon Islands Turkey Portugal South Africa Turkmenistan Qatar Spain Uganda Note: All countries in the list are included in the estimations on the impact of deep agreements on foreign direct investment (FDI) flows. The countries in blue are included in the estimation on FDI spillovers. TABLE S2A.6  Regression Results: Deep Agreements and Foreign Direct Investment (1) (2) (3) (4) Natural Total resources Goods Services Depth 0.0296*** −0.0183 0.0357*** 0.0416***   (0.00699) (0.0371) (0.00956) (0.00822) Number of observations 106,635 16,214 57,351 66,020 R2 0.835 0.784 0.839 0.818 Note: Poisson pseudo–maximum likelihood estimations. All specifications include bilateral fixed effects and country-time fixed effects. Robust standard errors, clustered by country pair, are in parentheses. Significance level: *** = 1 percent. TABLE S2A.7  Foreign Direct Investment and Depth: Interactions with ECA (1) Total Depth 0.0314***   (0.00794) Depth × ECA receiving −0.00287   (0.00522) Number of observations 85,271 R2 0.795 Note: Poisson pseudo–maximum likelihood estimations. All specifications include bilateral fixed effects and country-time fixed effects. Robust standard errors, clustered by country pair, are in parentheses. ECA = Europe and Central Asia. Significance level: *** = 1 percent. 122  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia SPOTLIGHT 2 continued TABLE S2A.8  Foreign Direct Investment and Depth, by Group of Provisions (1) (2) (3) Total Goods Services WTO 0.0162 0.0490 0.0513 (0.0242) (0.0347) (0.0333) WTO+ 0.0352*** 0.0223 0.0357** (0.0105) (0.0149) (0.0141) Number of observations 106,635 43,871 50,261 R2 0.835 0.851 0.849 Note: Poisson pseudo–maximum likelihood estimations. All specifications include bilateral fixed effects and country-time fixed effects. Robust standard errors, clustered by country pair, are in parentheses. WTO = World Trade Organization. Significance level: ** = 5 percent, *** = 1 percent. TABLE S2A.9  Foreign Direct Investment and Depth, by Technology Level (OECD Rev. 3) (1) (2) (3) (4) (5) (6) Total goods Low technology High technology Depth 0.0295*** 0.0347*** 0.0296**   (0.00998) (0.0126) (0.0148) WTO 0.0490 0.0305 0.025   (0.0347) (0.0499) (0.0464) WTO+ 0.0223 0.0364* 0.022   (0.0149) (0.0216) (0.0185) Number of 43,871 43,871 34,004 34,004 28,674 28,674 observations R2 0.851 0.851 0.782 0.782 0.864 0.863 Note: Poisson pseudo–maximum likelihood estimations. All specifications include bilateral fixed effects and country-time fixed effects. Robust standard errors, clustered by country pair, are in parentheses. WTO = World Trade Organization. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. TABLE S2A.10  Foreign Direct Investment, Depth, and Distance (1) (2) (3) (4) (5) (6) Total Goods Services Total Goods Services Depth 0.0461** 0.0522* 0.0209 0.0419** −0.00706 0.0844***   (0.0229) (0.0305) (0.0351) (0.0186) (0.0210) (0.0305) Depth × Geographical −0.00115 −0.00275 0.00283   −0.00312 (0.00389) (0.00474) Depth × Cultural −0.00201 0.00778* −0.0111*   (0.00402) (0.00447) (0.00673) Number of observations 91,881 42,693 49,188 65,567 31,225 34,342 R2 0.849 0.851 0.850 0.840 0.858 0.818 Note: Poisson pseudo–maximum likelihood estimations. All specifications include bilateral fixed effects, country-time fixed effects, and industry-country-time fixed effects. Robust standard errors, clustered by country pair, are in parentheses. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. Attracting Foreign Direct Investment: The Role of Deep Preferential Trade Agreements ●  123 TABLE S2A.11  Foreign Direct Investment and Depth: Triple Interactions (1) (2) (3) (4) Variables Investment Investment Investment Investment Depth 0.0363*** 0.0254*** 0.0361*** 0.0253*** (0.00935) (0.00835) (0.00940) (0.00834) Depth × ECA destination −0.0109 (0.00857) Depth × EU15 source 0.0217 −0.0214* 0.0221 −0.0214* (0.0169) (0.0121) (0.0169) (0.0121) Depth × EU15 source × ECA destination −0.0431** (0.0197) Depth × non-ECA destination 0.0109 0.0106 (0.00857) (0.00863) Depth × EU15 source × non-ECA destination 0.0431** 0.0432** (0.0197) (0.0197) Depth × EU28 destination −0.0104 (0.00859) Depth × EU15 source × EU28 destination −0.0439** (0.0196) Depth × ECA non-EU destination −0.0155 (0.0336) Depth × EU15 source × ECA non-EU destination 0.154 (0.116) Number of observations 65,118 65,118 65,118 65,118 R2 0.813 0.813 0.813 0.813 Note: Poisson pseudo–maximum likelihood estimations. All specifications include bilateral fixed effects, country-time fixed effects, and industry-country-time fixed effects. Robust standard errors, clustered by country pair, are in parentheses. ECA = Europe and Central Asia; EU = European Union. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. of the WTO. Table S2A.3 lists the 51 policy areas Notes that are identified. The legal enforceability of the PTA obligations 1. This spotlight draws on Laget, Rocha, and Varela is established according to the language used in (2018). the text of the agreements. In other words, it is 2. For simplicity, the variable depth that is used in this analysis considers that all the provisions included in assumed that commitments expressed with clear, an agreement have the same weight, and therefore specific, and imperative legal language can more are equally important for FDI. Analysis not presented successfully be invoked by a complainant in a dis- in this spotlight and that uses alternative measures to pute settlement proceeding, and therefore are capture the depth of an agreement (e.g., depth con- more likely to be legally enforceable. In contrast, structed using principal component analysis) also confirms a positive relationship between deeper unclear legal language might be related to policy agreements and FDI flows (Osnago, Rocha, and areas that are covered but that might not be Ruta 2015a, 2015b). legally enforceable. 124  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia SPOTLIGHT 2 continued 3. These results are in line with the results from Osnago, Gomez-Mera, L., T. Kenyon, J. G. Reis, and G. Varela. Rocha, and Ruta (2015a, 2015b) showing that an addi- 2014. New Voices in Investment: A Survey of Investors tional provision in deep PTAs increases vertical FDI from Emerging Economies. Washington, DC: World flows by approximately 2 percent. Bank Group. 4. The existence of a differential effect is tested by aug- Hofmann, Claudia, Alberto Osnago, and Michele Ruta. menting the cross-country gravity equation with an 2017. “Horizontal Depth: A New Database on the interaction of depth with a dummy that identifies Content of Preferential Trade Agreements.” Policy observations for ECA countries as destinations of FDI Research Working Paper 65837, World Bank, flows. ECA countries and subgroup definitions can Washington, DC. be found in table S2A.1. Hofstede, Geert. 2011. “Dimensionalizing Cultures: The 5. Adding one extra provision in an agreement is associ- Hofstede Model in Context.” Online Readings in ated with increases in FDI flows of low- and high- Psychology and Culture 2 (1):1–26. intensity products of approximately 3.5 and 3 percent, respectively. Only the WTO+ provisions, Horn, H., P. C. Mavroidis, and A. Sapir. 2010. “Beyond the and not the WTO provisions, are significant for sec- WTO? An Anatomy of EU and US Preferential Trade tors with lower levels of technological intensity Agreements.” World Economy 33 (11): 1565–88. (see table S2A.9). The classification of sectors by Laget, E., N. Rocha, and G. Varela. 2018. “FDI and Deep technology intensity is taken from the OECD Rev.3 Preferential Trade Agreements: An Empirical classification. This classification is only available for Investigation.” Unpublished, World Bank, the manufacturing sector. Washington, DC. 6. The interaction between physical distance and depth Osnago, Alberto, Nadia Rocha, and Michele Ruta. 2015a. is not significant, suggesting that the impact of deep “Deep Trade Agreements and Vertical FDI: The Devil agreements is the same in distant countries as in non- Is in the Details.” Policy Research Working Paper distant ones (see annex S2A). 7464, World Bank, Washington, DC. ———. 2015b. “Do Deep Trade Agreements Boost Vertical FDI?” World Bank Economic Review 30 (1): 119–25. References Piermartini, Roberta, and Yoto Yotov. 2016. “Estimating Trade Policy Effects with Structural Gravity.” School of Driffield, Nigel, James Love, and Yong Yang. 2016. Economics Working Paper Series 2016-10, LeBow “Reverse International Knowledge Transfer in the College of Business, Drexel University, Philadelphia. MNE: (Where) Does Affiliate Performance Boost WTO (World Trade Organization). 2011. The WTO and Parent Performance?” Research Policy 45 (2): Preferential Trade Agreements: From Coexistence to 491–506. Coherence. Geneva: WTO. 3 Connectivity and Firms Ownership and management links with foreign firms enable domestic firms to perform better than firms lacking such connections. This chapter first examines the prevalence of firm connectivity in Europe and Central Asia (ECA) through o ­ wnership or management ties, and then discusses why these connections are important. Then evidence of the spillover ben- efits of foreign firms on local firms is reviewed, followed by a discussion of policy recommendations. Main Messages • Many firms in ECA have foreign connections, although the extent of foreign ownership varies greatly across the region. More than half of foreign-owned firms in ECA also have a predominantly foreign management. Firms that are foreign owned and foreign managed have higher growth in operating reve- nues, jobs, and average wages than firms lacking these connections, due to technology transfers and better management practices. Employment in ­ foreign-connected firms is less procyclical, likely because of access to finance and resources from the parent firm economy. • Domestic firms without foreign connections also benefit from the presence of foreign-owned firms. Competition from connected firms can force domestic firms to become more efficient, although competition from better-performing 125 126  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia foreign-connected firms also may force domestic firms to downsize or leave the market. In ECA, small and young domestic firms are particularly at risk. • Governments can implement policies to boost the positive effects of connectiv- ity while minimizing the risks: ◦ General improvements to the business environment ◦ Removing barriers to foreign direct investment (FDI) and carrying out investment-promotion activities to reduce information asymmetries and ­ burdensome regulatory procedures ◦ Promotion of skilled migration ◦ Efforts to strengthen firm linkages (for example, to encourage innovation transfers between domestic and foreign firms) ◦ Steps to help domestic firms compete (e.g., improved access to finance for small and young firms, educational programs to help strengthen local ­ management, and the easing of labor market regulations that restrict the ability of firms to manage workers efficiently) ◦ Efforts to smooth workers’ adjustment to unemployment (e.g., facilitating geographic mobility, improving education and training, and providing social insurance in ways that do not distort labor market decisions). Firm Connectivity in ECA Characteristics of Firm Connectivity in ECA There is a high incidence of connections between firms in ECA and foreign owners, ownership especially among larger firms. This section analyzes the extent of foreign ­ in ECA (a discussion of overall trends in FDI in ECA countries from a macro perspec- tive can be found in spotlight 1). We exclude firms owned by individuals or compa- nies located in countries considered to be tax havens, as the country of ownership may be a result of tax incentives and regime and not actually capture the economic impacts typically associated with FDI. In addition, firms that are owned by tax haven countries may not truly represent the characteristics of their source country as the parent company may be located in the tax haven country for tax purposes but the operational headquarters, where management decisions are made, are in a third country. Annex 3A presents the coverage of the firm-level data used in the descrip- tive statistics and analysis throughout this chapter. Many firms in ECA are owned by foreigners (see figure 3.1). At one extreme, more than 32 percent of all firms in the United Kingdom and Ireland are owned by people or firms in another country. Foreign ownership is also present in other ECA countries to a lesser extent. On average, about 5 percent or more of firms in most of Central Europe, the Western Balkans, Latvia, and Lithuania are foreign owned. At the other extreme, foreign ownership is negligible in Hungary, Bulgaria, Ukraine, the Russian Federation, Belgium and most Southern European countries. The shares of foreign ownership exclude firms owned by parent companies located in tax havens, which is discussed below. A restrictive FDI regime may explain the low presence of foreign companies in some of these economies, such as Ukraine, Connectivity and Firms ●  127 FIGURE 3.1  The presence of foreign firms varies substantially across ECA countries Foreign-owned and foreign-managed firms in ECA, 2013 35 30 25 Share of all firms (percent) 20 15 10 5 0 SVK CZE ROU SVN POL HRV BGR HUN DNK LVA EST FIN SWE LTU ISL UKR RUS GRC ITA ESP PRT BIH SRB GBR IRL AUT DEU FRA NLD BEL Central Europe Northern Europe Southern Europe Western Western Europe Other Eastern Federation Russian Balkans Europe Foreign-owned firms Foreign-managed firms Foreign-owned and foreign-managed firms Source: Calculations using Orbis data. Note: Sample excludes firms with owners in tax haven countries. ECA = Europe and Central Asia. Belgium, and Italy, which rank poorly in the Services Trade Restrictiveness Index (Borchert, Gootiiz, and Mattoo 2012). Some of the efficiency effects of foreign companies take place through the transfer of soft technologies, such as management skills, since their capital invest- ment is often accompanied by the hiring of foreign managers (Djankov and Hoekman 2000; Blalock and Gertler 2008). In ECA, hard and soft investments from abroad are highly correlated, as countries with a higher share of foreign-owned firms also tend to have a high share of foreign managers. Indeed, a significant share of the foreign-owned firms in ECA are also managed by foreigners. On aver- age, more than half of foreign companies in ECA also have foreign managers. Large firms in ECA are more likely than their smaller peers to be foreign owned. As seen in figure 3.2, while at least 15 percent of firms with 250 employees or more are foreign owned in Central, Northern, Southern, and Western Europe, that figure is less than 4 percent for small firms. While the extent of foreign ownership differs across the region, the ratios of the shares of small and large foreign firms are simi- lar: for every firm in the country there are three times more large foreign firms than small ones. As discussed in the following sections, these patterns could be partially explained by foreign-owned firms exhibiting higher growth than those with domestic owners. However, it may also reflect the fact that foreign investors are more likely to invest in firms that have achieved a large enough size. 128  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 3.2  Large firms are more likely to be foreign owned in ECA Share of foreign-owned firms by number of employees, 2013 20 18 18.0 17.7 17.2 16 14.8 14 Share of firms (percent) 12 11.7 11.2 10 8.6 8.5 8 6.7 6.9 6 4.6 3.8 4.1 3.8 4 2 2.0 1.6 1.8 2.0 2.3 0.6 0 0.0 0.1 0.2 0.3 0.0 0.1 0.1 0.5 Central Europe Northern Europe Ukraine Russian Federation Southern Europe Western Balkans Western Europe 1–9 employees 10–49 employees 50–249 employees 250+ employees Source: Calculations using Orbis data. Note: Sample excludes firms with owners in tax haven countries. ECA = Europe and Central Asia. In most subregions, older firms are not significantly more likely than younger ones to be foreign owned (figure 3.3). For instance, old and young firms in Central Europe, Ukraine, and the Western Balkans are equally likely to be foreign owned. This is an interesting finding considering that firms age 30 years or older in 2013 were founded before the transition among the former socialist economies. In con- trast, older firms in Northern, Western, and Southern Europe as well as Russia are slightly more likely to be foreign owned than their younger peers. This may indicate that for this group of countries, FDI tends to be attracted by firms with a longer pres- ence in the market, or that foreign firms are more likely to survive. The comparison between the age of foreign-owned firms in former socialist economies suggests that there are fewer entry barriers to foreign startups in the Western Balkans and Central Europe. In these countries, there are equal shares of young and old ­ foreign-owned firms compared with Russia, where foreign-owned firms are older. Most foreign owners of firms in ECA are persons or companies in Germany and the United States (see table 3.1). Given the size of these two economies and their level of economic development, it is expected that both countries would have a strong presence in the arena of multinational companies. However, in addition to the characteristics of the investor country, other factors such as geographic prox- imity, common language, and cultural heritage seem to be important determi- nants of owning a company in another nation. This is consistent, for example, with Nordic countries being among the most common foreign owners of companies in Northern Europe; or with Croatia and Slovenia being among the most common owners of foreign companies in the Western Balkans. Connectivity and Firms ●  129 FIGURE 3.3  There is no clear relationship between a firm’s age and the likelihood of its being foreign owned Share of foreign-owned firms by age of firm, 2013 8 7.0 7 6 5.9 5.6 5.2 5 4.9 Share or firms (percent) 4.2 4 3.5 3.4 3.5 3.0 3 2.9 2.8 2.6 2.7 2.1 2.1 2 1.6 1.4 1.2 1.3 1 0.7 0.1 0.1 0.0 0.1 0.0 0.1 0.1 0 Central Europe Northern Europe Ukraine Russian Federation Southern Europe Western Balkans Western Europe 1–4 years 5–9 years 10–29 years 30+ years Source: Calculations using Orbis data. Note: Sample excludes firms with owners in tax haven countries. Determinants of Firm Connectivity in ECA We shed light on the drivers of foreign ownership by estimating a gravity model that examines the bilateral relationships between the source country, which owns the foreign affiliates or sends foreign managers, and the host country, where the foreign affiliate is located. The share of foreign ownership and management are measured in three ways: (a) the number of foreign affiliates owned by the source country and the number of firms primarily managed by foreigners from the source country as a share of all firms; (b) the sales of; and (c) the operating revenue of the foreign affiliate owned by and companies primarily managed by the source country as a share of all firms.1 Foreign ownership and management are likely to be driven by the usual patterns observed in FDI and international trade flows. Table 3.2 explores how foreign ownership and management are determined by linkages between the source and host country through a shared cultural and historical ­heritage, political relationships or union, and geographic proximity or shared borders.2 Geographic and economic linkages, and to a lesser extent historical and ­cultural linkages, are important determinants of foreign ownership and management. 130  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE 3.1  Most Foreign-Owned Firms in ECA Are Owned by Germany and the United States Most common global ultimate owner Others (from left to right, top to bottom): Denmark, Norway, Russian Federation, United Belgium, Croatia, Slovenia, Region Germany United States Kingdom Netherlands Austria France Italy Finland Sweden and Japan Central Europe Northern Europe Ukraine Russian Federation Southern Europe Western Balkans Western Europe Source: Calculations using Orbis data. Note: Sample excludes firms with owners in tax haven countries. Each row in the table shows the five (or more, if there is a tie) most common countries of ownership, among the top ten countries of ownership, for each country or region at left. For the Russian Federation and Ukraine, the rows show the five countries with the largest ownership shares. ECA = Europe and Central Asia. Connectivity and Firms ●  131 TABLE 3.2  Determinants of Foreign Ownership and Management   Log (Foreign Log (Foreign Log (Foreign Log (Foreign Log (Foreign manager, firms, simple Log (Foreign firms, operating manager, manager, operating count) firms, sales) revenue) simple count) sales) revenue) Log (Distance) −0.369*** −0.766** −0.717** −0.505*** −1.689*** −1.483***   (0.0771) (0.307) (0.291) (0.0732) (0.327) (0.303) Log (Immigrants) 0.105*** 0.188*** 0.165*** 0.147*** 0.132** 0.0990**   (0.0120) (0.0557) (0.0457) (0.0116) (0.0574) (0.0483) = 1 if countries were or are 0.536*** 0.749 1.425** 0.429*** 0.580 1.173*   same country   (0.180) (0.652) (0.669) (0.165) (0.689) (0.673) = 1 if contiguous 0.416*** 0.699 0.543 0.413*** 0.240 0.589   (0.121) (0.473) (0.451) (0.113) (0.501) (0.464) = 1 if common language 0.243** 0.367 0.0269 0.580*** 2.404*** 2.169***   (0.115) (0.523) (0.432) (0.111) (0.580) (0.463) Log (Exports) 0.271*** 0.597*** 0.680*** 0.228*** 0.683*** 0.694***   (0.0291) (0.130) (0.110) (0.0283) (0.135) (0.119) Log (Imports) 0.0969*** 0.293*** 0.168* 0.112*** 0.151 0.145*   (0.0225) (0.0914) (0.0869) (0.0206) (0.0926) (0.0873) Tax haven 2.994*** 2.504 10.93** 0.356 2.402 1.106   (1.128) (3.659) (4.816) (0.667) (3.563) (2.681) Number of observations 1,747 1,362 1,637 1,880 1,457 1,748 R2 0.853 0.730 0.716 0.831 0.737 0.727 Note: The regressions are ordinary least squares regressions using 2013 data and including source and host country fixed effects. Robust standard errors are in parentheses. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. Geographically proximate countries and countries that share b ­ orders are more managers. A 10 ­ likely to have higher shares of foreign affiliates and foreign ­ percent decrease in the bilateral distance between source and host country is correlated with a 3.7 percent increase in the share of foreign affiliates and a 5.1 percent increase in the share of foreign managers in the host country. Economic linkages matter too: when the host country exports more to and imports more from the source country, the share of foreign ownership and management from the source country is higher. Similarly, when there are more ­ country, immigrants from the source ­ foreign ownership and management by that country increases. Sharing a common colonial history and language is correlated with a higher share of foreign owner- ship and management but only in terms of the number of foreign affiliates and not the sales and revenue share of these foreign affiliates. These effects of historical and cultural linkages may be absorbed by including bilateral imports and exports in the econometric model. Firms in Tax Havens Some foreign affiliates are owned by persons or companies located in tax havens, which are countries with low tax regimes.3 Examining the linkages between these foreign affiliates and their ownership may be misleading as the owners are located in the tax haven country but the actual headquarters that has control and gives operational directions may be in another country. The incidence of foreign affiliates with owners in a tax haven country is low in many ECA countries, where fewer than 20 percent of the foreign affiliates have owners in a tax haven country (figure 3.4). 132  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 3.4  Foreign affiliates owned by tax haven countries are small and medium sized 100 90 80 70 Share of firms (percent) 60 50 40 30 20 10 0 RUS UKR BGR GRC ITA DEU POL GBR FRA AUT PRT IRL NLD CZE BEL SRB ROU ESP LVA HUN LTU SVN DNK SVK BIH EST FIN SWE HRV 1–9 employees 10–49 employees 50–249 employees 250+ employees Not owned by tax haven Source: Calculations using Orbis data. Note: Foreign affiliates are further disaggregated by firm size: micro-sized firms with 1 to 9 employees, small-sized firms with 10 to 49 employees, medium-sized firms with 50 to 249 employees, and large-sized firms with more than 250 employees. However, two countries stand out: 59 percent of foreign affiliates in Russia and 44 percent in Ukraine are owned by an individual or company located in a tax haven country. One important detail is that many foreign affiliates with owners in a tax haven country are micro- and small-sized with fewer than 50 employees, sug- gesting that these foreign affiliates may not benefit as much from their foreign connections. Profits and assets in these foreign companies may be transferred to the parent companies located in the tax haven to take advantage of the low tax rate. As a result, these firms remain small as they are not retaining their profits for investment to expand their operations. Importance of Firm Connectivity FDI and Transfer of Technology Multinational enterprises (MNEs) bring technology and know-how that can ben- efit local firms. The linkages between MNEs and the companies they own in the foreign market, or foreign affiliates, are usually established through direct invest- ments in existing local companies or greenfield investments. These MNEs are often the most productive firms in their domestic market and can transplant Connectivity and Firms ●  133 ­ntangible inputs, such as know-how and management practices, as well as capi- i tal and technology to their foreign affiliates. Atalay, Hortaçsu, and Syverson (2014) show that the vertical ownership structures between US firms and their subsidiaries do not constitute input-output linkages; in fact, there is very little shipment of physical goods from upstream firms to downstream firms. Instead, the ownership structures are in place to transfer intangible inputs efficiently between firms. MNEs can contribute significantly to ECA countries, and most empirical studies find that foreign-owned firms tend to be larger (Haddad and Harrison 1993) and more productive (Girma, Greenaway, and Wakelin 2001; Conyon et al. 2002; Vahter and Masso 2007)4 and pay higher wages (Girma, Greenaway, and Wakelin 2001; Lipsey and Sjöholm 2004; Conyon et al. 2002) than their local counterparts. The performance gap between foreign-and domestically-owned firms can be explained by a selection effect,5 but multinational firms may also benefit from knowledge assets such as technological, managerial, and foreign-market-related knowledge that domestic firms do not have. The existence of such knowledge assets is supported by studies finding that multinational firms invest more in new technologies and training of their employees (Djankov and Hoekman 2000; Görg, Strobl, and Walsh 2007),6 are better managed (Bloom and Van Reenen 2010), and export more (Aitken, Hanson, and Harrison 1997). Knowledge transfers to foreign affiliates not only benefit the affiliates but the presence of foreign firms can also benefit domestic firms indirectly through knowledge spillovers. Competition from foreign-owned firms may also induce domestic firms to increase productivity to maintain their market shares. Empirical studies support, to some extent, the suggested positive spillovers from knowl- edge transfers and competition. Firm- and plant-level studies by Keller and Yeaple (2009) for the United States, Dries and Swinnen (2004) for Poland,7 and Haskel, Pereira, and Slaughter (2007)8 for the United Kingdom are examples of the studies that find evidence of productivity spillovers from multinational firms. Keller and Yeaple (2009) argue that the spillovers stem from technological trans- fers, and Javorcik (2004) likewise shows that contacts between partially foreign- owned firms and their local suppliers in Lithuania facilitate positive productivity spillovers. Moreover, Kokko, Tansini, and Zejan (2001) for Uruguay and Aitken, Hanson, and Harrison (1997) for Mexico find that the presence of foreign firms increases the likelihood that domestic firms will export. These results may sug- gest that some ­ foreign-market-related knowledge is being transferred to domes- tic firms. Although the effect from knowledge spillovers is difficult to distinguish from the potential positive effects from foreign competition, a paper by Bao and Chen (2016) presents evidence of the latter. They show that merely the prospect of foreign firms entering the local market induces local firms to be more produc- tive in various countries. Although these studies provide empirical support of positive spillovers, the ­ overall empirical evidence on spillovers from multinational to domestic firms is ambiguous (Görg and Greenaway 2004). For example, studies by Haddad and Harrison (1993) on Morocco, Aitken and Harrison (1999) on República Bolivariana de Venezuela, Djankov and Hoekman (2000) on the Czech Republic, and Konings (2001) on three Central and Eastern European countries question the findings of 134  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia positive productivity spillovers. Konings (2001), for instance, uses firm-level data for the mid-1990s and finds evidence of negative productivity spillovers to domestic firms in Bulgaria and Romania and no spillovers in Poland. The same ambiguous picture emerges for wage spillovers. Lipsey and Sjöholm (2001) find evidence of positive wage spillovers in Indonesia; Girma, Greenaway, and Wakelin (2001) find no evidence of wage spillovers in the United Kingdom; and the find- ings of Aitken, Harrison, and Lipsey (1996) suggest no or negative effects in Mexico and República Bolivariana de Venezuela and positive wage spillovers in the United States.9 Although the evidence of positive spillovers from multina- tional to domestic firms is inconclusive, it seems to be stronger in developed countries. If spillovers stem from knowledge transfers, it is plausible that the esti- mated effect of spillovers is influenced by the absorptive capacity of the domestic firms (Görg and Greenaway 2004). Kokko, Tansini, and Zejan (1996), for instance, examine intra-industry spillovers in Uruguay in 1988 and find positive spillovers only to domestic firms with a moderate technology gap with respect to foreign firms. They interpret their findings as evidence of the importance of the absorp- tive nature of domestic firms. The presence of positive knowledge spillovers is supported theoretically and to some extent empirically. However, the findings of nonpositive spillovers may ­ suggest that other opposing effects from multinational firms are in play. The displacement of production, jobs, and tax revenue can be a concern when it comes to FDI, and while increased competition from foreign firms may raise the productivity of some firms, other firms may lose market share or be pushed out of the market (Aitken and Harrison 1999). These crowding-out effects are documented empirically by Kosová (2010), who finds that initial foreign entry in the Czech Republic market increases the exit rates of domestic firms. Accordingly, Aitken and Harrison (1999) find stronger negative productivity spillovers from multinational to small domestic firms in República Bolivariana de Venezuela10 and suggest that small firms were less able to compete with multinational firms. Aitken, Harrison, and Lipsey (1996) also suggests that the observed negative wage spillovers in República Bolivariana de Venezuela may partly be due to increased competition for workers resulting in foreign firms attracting the best workers. So while foreign firms contribute to a productivity increase for some firms, empirical evidence also suggests that less competitive incumbent firms are at risk of losing market share and being driven out of business. While these negative effects may create some tensions in the short term, they also contribute to the process of creative destruction, where old and traditional sectors shrink and new and more productive ones emerge. As less productive firms exit the market, resources are shifted to more productive incumbents and the foreign entrants. As a result, the aggregate productivity of the country increases. The speed by which factors of production move from the former to the latter is crucial to maxi- mize efficiency gains (Hollweg et al. 2014). Transfer of Management Practices One channel by which MNEs can increase the productivity and performance of their foreign affiliates is through the transfer of management practices. Similar to the discussions on information and communication technology (ICT) and the internet, management practices can be thought of as a technology (Bloom et al. 2016) Connectivity and Firms ●  135 that can be transplanted from headquarters to foreign affiliates. Different quality of management practices from headquarters can account for large differences in the productivity of firms. Heyman, Norback, and Hammarberg (2014) show that about 30 percent of the difference in the labor productivity of foreign firms in Sweden can be explained by variations in management practices at headquarters. The manage- ment practices adopted by foreign affiliates can also determine their adoption and use of technology. The US-owned firms in the UK have more ICT capital and higher ICT intensity than domestic firms and firms with owners in other countries (Bloom, Sadun, and Van Reenen 2012). The main reason for the difference in ICT capital stock and intensity of use is that the US-owned firms have management practices that are centered on merit-based promotion, reward, and hiring and firing, which are management practices prevalent in US-owned firms in Europe. The transmission of management practices between MNEs and their foreign affiliates provides an avenue for countries to improve their aggregate ­productivity.11 Studies show that management practices can account for the heterogeneity in total factor productivity (TFP), among firms. The transmission Even within narrowly defined sectors, Syverson (2004) estimates that of management a plant is four times more productive at the 90th percentile than at practices between the 10th percentile of the productivity distribution. Bloom, Sadun, MNEs and their foreign and Van Reenen (2016) find that management practices can account affiliates provides an for, on average, 30 percent of the differences in cross-country TFP. avenue for countries to Better management practices are also beneficial to workers: using improve their aggregate employer-employee-linked data, Bender et al. (2016) show that bet- productivity. ter-managed German firms pay higher wages. Understanding the determinants of firm productivity is key to promoting economic growth and reducing income inequality between countries. Examining how management practices affect firm performance and productiv- ity requires some measure of management practices. The measurement of management practices, especially across a large set of countries, is still a nascent area ­ of research. Bloom, Sadun, and Van Reenen (2016) have created a data set of management practices—the World Management Survey (WMS)—by surveying ­ companies in operations management, performance monitoring and talent ­ ­ management, and target setting and leadership management. The WMS contains the most comprehensive measures of management practices. Because it is a resource-­ intensive task, the data set is currently available for only 34 countries and it does not cover all the countries of foreign owners and managers in the Orbis database. A proxy for management practices can be created from the World Economic Forum (WEF) Global Competitiveness survey, which measures the quality of national business schools and the reliance on professional management. A man- agement index is calculated from the average scores of the two measures, or Average Management Index (AMI). There is a strong positive correlation between the WMS and the AMI, which is presented in figure 3.5.12 Unlike the WMS, which has the management scores of individual firms, the AMI is available at the country level and is used in this chapter as a proxy for the management quality of the aver- age owner and manager from that country. Foreign affiliates tend to be from countries that have better management practices than the host country. The average AMI of the foreign affiliates is higher ­ 136  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 3.5  The Average 6.0 CAN Management Index is SGP SWE USA Average Management Index based on WEF NZL AUS strongly correlated with the IRL FRA 5.5 GBR WMS Management Index IND CHL ESP DEU 5.0 ARG POL JPN KEN BRA PRT 4.5 GHA CAN ZMB MEX NGA COL 4.0 TUR ITA GRC TZA NIC 3.5 ETH MOZ 3.0 2.0 2.5 3.0 3.5 WMS Management Index Source: Calculations using data from the World Economic Forum (WEF) Global Competitiveness Survey and the World Management Survey (WMS) (Bloom et al. 2016). than the AMI of local firms in Bosnia and Herzegovina, Central Europe, and many Southern European countries (figure 3.6). In contrast, the foreign affiliates in Western European and Northern European countries tend to have slightly lower AMI than the local firms. The difference may reflect FDI patterns where Western European firms (such as German, French, and British firms) with better manage- ment practices invest in Eastern European countries while American firms with higher AMI invest in Western and Northern Europe. Similar patterns emerge when analyzing the percentage of foreign firms with better AMI than local firms (see figure 3.7). The differences are stark for some ECA countries: almost 100 percent of foreign affiliates in Bosnia and Herzegovina, Bulgaria, Greece, Italy, and Ukraine are owned by companies in countries that have a higher AMI than the local AMI (figure 3.7). In contrast, given the highly rated management practices in Sweden and Denmark, only 4 percent of foreign affiliates in Sweden and 26 percent in Denmark are owned by countries with better AMI. Transmission of Economic Volatility: Do Foreign Firms Import Volatility in ECA? Foreign-owned firms may also affect the domestic economy by allowing shocks to be transmitted across countries. On the one hand, if foreign-owned companies are more affected by the unpredictability of international finance or are subject to policy changes from the parent country, they may bring volatility to their country of location. At the same time, by being more connected to the other country, they may be less exposed to local shocks if, for example, they have better access to finance or talent in the parent country or if they rely more on demand from abroad. The extent to which foreign companies attenuate or exacerbate the effect of the volatility of the local economy also depends on the extent to which local shocks are correlated with those experienced by the parent country. Trade is one channel through which the presence of multinational firms may affect the host economy. While FDI and exports can be alternative ways to enter Connectivity and Firms ●  137 FIGURE 3.6  Foreign 6 affiliates tend to have better IRL FIN management practices than SWE ITA DEU BEL DNK local firms LTU PRT AUT GBR HUN GRC POL FRA NLD CZE ESP HRV SVK AMI of foreign-owned firms 5 BGR RUS SVN ROU BIH 4 3 3 4 5 6 AMI of locally owned firms Western Europe Western Balkans Southern Europe Russian Federation Central Europe Other Eastern Europe Northern Europe 45-degree line Source: Calculations using data from the World Economic Forum Global Competitiveness Survey and Orbis. Note: AMI = Average Management Index (see chapter text for construction). FIGURE 3.7  More foreign affiliates are owned by countries with better management indexes 100 90 80 Foreign firms coming from countries with better 70 AMI than the local AMI (percent) 60 50 40 30 20 10 0 SWE DNK NLD AUT BEL FRA FIN ESP ROU GBR SVN CZE IRL DEU SVK POL HUN PRT RUS LTU HRV ITA UKR BGR GRC BIH Source: Calculations using data from the World Economic Forum Global Competitiveness Survey, the World Management Survey (Bloom et al. 2016b), and Orbis. Note: AMI = Average Management Index (see chapter text for construction). 138  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia a market, FDI can also spur imports if foreign affiliates tend to import intermediate goods from their country of origin. Most empirical work finds a positive relation- ship between exports and FDI (Blomström, Lipsey and Kulchycky 1988;13 Clausing 2000; Head and Ries 2001), and studies suggest that this relationship is partly driven by increased demand for intermediate goods from the country of origin (Lipsey and Weiss 1984; Head and Ries 2001). The possible increase in trade induced by multinational firms may, as mentioned, raise concerns about the vola- tility of the connected economies, and empirical evidence suggests that countries connected through trade and FDI are indeed more exposed to demand and sup- ply shocks of their partnering countries. For example, an early study by Frankel and Rose (1998) examining a panel of 20 industrial countries shows that ­ bilateral trade intensity is strongly associated with correlated business cycles. Kleinert, Martin, and Toubal (2015) examine the prevalence of comovements of gross domestic product (GDP) growth in France and connected countries. They find that the regional share of foreign affiliates’ employment is associated with significantly stronger comovements in regional GDP growth and GDP growth of the ownership country. Using the Orbis database Cravino and Levchenko (2016) likewise show that there is strong comovement in sales growth of the h ­ eadquarters and sales growth of the affiliate. They also find that shocks in the source country are transmit- ted to the host country and estimate that the combined shock of all ­ multinationals accounts for 10 percent of the aggregate productivity shock. Another recent study by Boehm, Flaaen, and Pandalai-Nayar (2016) finds that international production networks of multinational firms affect aggregate volatility of an economy. More specifically they show that multinational firms and affiliates in Japan transmitted the shock of the 2011 Tohoku earthquake to their US parent firms because of low short-run input substitution of multinational firms. Other studies also suggest that rent sharing between parent and affiliate firms takes place. Budd, Konings, and Slaughter (2005), for instance, show that affiliate wages of multinationals respond to both parent and affiliate profitability and that parent firms’ profit may explain more than 20 percent of observed variation in affiliate wages. In sum- The volatility mary, the empirical evidence seems to be in favor of bilateral firm con- of the domestic nections imposing an increased exposure to supply and demand economy and shocks of the ­partnering countries. labor market The volatility of the domestic economy and labor market may may also be affected also be affected by the potential footloose nature of multinational by the potential firms. Navaretti, Checchi, and Turrini (2003) examine employment footloose nature of adjustments of firms in 11 Northern and Western European countries multinational firms. and find that multinational affiliates adjust employment significantly faster to shocks than do their domestic counterparts. However, they also find that the extent of the adjustment is more limited and that for a given wage increase, multinational affiliates decrease employment by less than domestic firms. Buch and Lipponer (2010) find that multinational affiliates in Germany do not adjust employment systematically more in response to wages and output than do domestic firms. Table 3B.4 shows how sales, employment, and average wages of firms in ECA respond to local and foreign business cycles. Interestingly, once a number of variables are controlled for, the performance of an average firm in ECA is not Connectivity and Firms ●  139 FIGURE 3.8  Foreign firms’ 10 employment decisions are 8 Foreign firms’ employment growth relative to local firms’ less procyclical than those of GDP growth and difference between foreign-firm and their domestic peers 6 Local GDP growth local-firm employment growth (percent) 4 2 0 –2 –4 –6 –8 –10 t1 t4 t7 t10 t13 t16 t19 t22 t25 t28 t31 t34 t37 t40 t43 Time period Note: See table 3B.4 for full regression results. ­ ignificantly correlated with local economic growth. In contrast, foreign firms are s much more responsive to aggregate economic conditions. As seen in the second column of the table, foreign firms create more jobs when the foreign country of ­ ownership is growing. This could reflect increasing demand for exports or better access to finance during an economic upswing in the parent country. In contrast, foreign companies’ employment decisions seem to be more countercyclical with respect to the domestic economy than those of domestic companies, as the former tend to create fewer jobs when the local economy expands (see f ­ igure 3.8). Likewise, this also means that foreign companies tend to destroy fewer jobs than do domestic firms when the domestic economy experiences a recession, possibly reflecting access to external factors that allow them to buffer the impact of the decline in local economic activity. In other words, while foreign firms in ECA seem to contribute less to job creation than their local counterparts when the local economy is g ­ rowing, they seem to bring more stability to the labor market during a downturn in economic activity because they lay off workers to a lesser extent than local companies do. Effects of Foreign Firms on Local Economy Performance of Foreign Firms Versus Local Firms If MNEs transplant their management practices to foreign affiliates, the better management practices will enable foreign affiliates to perform better than local firms. Figure 3.9 explores whether foreign-owned and foreign-managed firms perform better than local firms. The figure depicts the estimation results that ­ are presented in table 3B.1 and the estimation distinguishes between the combinations of foreign ownership and foreign management. Firms included in ­ 140  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 3.9  Foreign-owned 30 and -managed firms perform better than local firms 25 20 Percent 15 10 5 0 Growth in operating revenues Growth in jobs Growth in average wages Locally owned, foreign managed Foreign owned, locally managed Foreign owned and managed Note: Each bar in the figure represents the difference in growth (of the type labeled) between the type of firm depicted in that bar and that of firms that are both locally managed and locally owned. Table 3B.1 presents full regression results. All underlying coefficients are statistically significant. the estimation can be firms owned and managed by a local firm or person, foreign affiliates managed by a local national, a local firm managed by a foreigner, and a foreign affiliate managed by a foreigner. The estimation examines the different performance of firms owned, managed, or both owned and managed by foreign- ers and does not consider the different quality of management practices. Compared with local firms, foreign affiliates perform better and the benefits of foreign ownership are present even with local management. Foreign affili- ates with local managers have 25.5 percent higher growth in operating reve- nue, 14.9 percent higher job growth, and 12.4 percent higher growth in wages over the 2010–13 period. Foreign managers have a smaller but still positive effect on firm performance. Importantly, it is the combination of being f ­ oreign owned and having a foreign manager that has the largest effect on firm perfor- mance. These firms have 28.3 percent higher growth in operating revenue, 19.6 percent higher job growth, and 16.8 percent higher wage growth than local firms. The performance of foreign firms is heterogeneous and can depend on the quality of management practices that are transmitted from headquarters. Table 3.3 examines how better management practices affect the growth of operating reve- nue, jobs, and average wages. The econometric model examines how firm perfor- mance is affected by the quality of management practices controlling for factors in the parent country that can affect firm performance through more exports (per capita income and population), better financial access (market capitalization), and cultural links (immigrant stock). Factors in the source country that may affect firm performance such as governance and the business environment are captured by country fixed effects. Foreign affiliates that are owned by countries with better management practices perform better than other foreign affiliates, even after controlling for ­ income levels, financial development, population, and immigrant stock in the source country. Source countries that are classified as tax havens may not Connectivity and Firms ●  141 TABLE 3.3  Better-Managed Foreign Affiliates Perform Better Log (Operating Log (Number Log (Average revenue), change, of employees), wage), change, 2010–13 change, 2010–13 2010–13 Log management index 0.232*** 0.121*** 0.0242   (0.0473) (0.0441) (0.0307) Log GDP per capita 0.0573*** 0.0452*** 0.00458   (0.0169) (0.00865) (0.00455) Log market capitalization of GDP −0.0243** −0.00713 −0.00375   (0.0110) (0.00732) (0.00485) Log population 0.0108** 0.0132*** 0.00669***   (0.00541) (0.00338) (0.00165) Log immigrant stock −0.00476 −0.00414** −0.00282   (0.00311) (0.00204) (0.00189) Number of observations 92,320 114,837 74,858 R2 0.055 0.114 0.067 Note: The estimation is restricted to foreign affiliates. The independent variables in the regression capture the characteristics of the country of ownership. The regression includes the initial values of the dependent variables to account for mean reversion, and country, sector, size, and age fixed effects. The sector is equivalent to the 1-digit NACE rev. 2 code. Age is grouped into intervals of 0–4 years, 5–14 years, 15–30 years, and 30+ years, and size is grouped into intervals of 1–2 employees, 3–7 employees, 8–49 employees, 50–249 employees, and 250+ employees. Robust standard errors (in parentheses) are clustered around country of ownership. Significance level: ** = 5 percent, *** = 1 percent. transplant their management practices to the foreign affiliates so the relationship between management practices and firm performance for these firms may not hold. Excluding these firms from the estimation does not change the relation- ships, and indeed the effects of better management practices on firm perfor- mance are slightly stronger. Better management practices in the source country are related to better firm performance. A higher AMI in the source country is correlated with higher growth of operating revenue and number of employees between 2010 and 2013. This reinforces the findings in Bloom, Sadun, and Van Reenen (2016) that better-­ managed firms perform better than poorly managed firms. The economic size of the source country, measured by income levels and population size, is also posi- tively related to growth in operating revenue and number of employees. While information about firms’ export activities is not available, it is likely that foreign affiliates will export to their source country so it is not surprising that foreign affili- ates owned by large countries will perform better. The quality of m ­ anagement practices is, however, not related to growth in average wages in foreign affiliates. While workers in foreign firms generally have higher wage growth, better firm performance from better management practices is not accompanied by higher wage growth. Similar relationships are present between better management prac- tices and firms managed by foreign managers, and the results are presented in annex 3B. Selection bias makes it difficult to establish a causal relationship between the man- agement advantages of foreign ownership and the performance of foreign affiliates. Foreign companies can choose the most productive region and sector in a country for greenfield investments or the best local firms to directly invest in. As a result, bet- ter firm performance may not be a result of foreign owners’ management practices. 142  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Indeed, many foreign affiliates are located in regions and sectors with well-­functioning local firms that have higher operating revenues and more employees and pay higher wages (table 3.4). More ­ foreign affiliates with better management practices are located in regions with good local firms, but fewer foreign affiliates are located in the well-functioning sectors. Selection bias is a persistent issue in many studies of FDI and it is difficult to resolve (Javorcik 2015). This issue is even more difficult to tackle with the data set used in this section, as foreign ownership and management are recorded only for the latest year. In other words, it is not ­possible to identify changes in owner- ship or management for a company over time with the present data. Spillover Effects of Foreign Firms on Local Firms The location of foreign affiliates may have spillover effects on local firms. Local firms can learn from foreign affiliates through demonstration effects, where they obtain new information about management practices and knowledge through observation, and labor mobility where local firms hire workers that are trained in foreign affiliates. Evidence of spillovers across industries is mixed but some studies do show positive spillover effects.14 For example, Haskel, Pereira, and Slaughter (2007) find that there is a positive correlation between the TFP of domestic firms in the UK and the share of foreign firms in the industry. A 10 percentage point increase in foreign presence in the industry raises the TFP of domestic plants by 0.5 percent. ­ A higher presence of foreign firms with good management practices in ECA could have both displacement effects, driving local firms out of the market, and spillover effects, if increased competitive pressures create i ­ncentives for local firms to become more efficient. Disentangling these effects is difficult since, as shown in the previous section, foreign firms do not choose where to locate randomly within ECA, and in fact they are more likely to locate in regions or sectors with firms that TABLE 3.4  Foreign Firms Locate in Regions and Sectors with Larger Local Firms   Share of foreign AMI of foreign Share of foreign AMI of foreign firms in region firms in region firms in sector firms in sector Mean operating revenue of local firms in region 0.00348*** 0.0237**       (0.000550) (0.00964)     Mean number of employees of local firms in region 0.0191*** 0.268***       (0.00408) (0.0997)     Mean average wages of local firms in region 0.00420*** 0.0356**       (0.00117) (0.0168)     Mean operating revenue of local firms in sector     0.00153*** −0.0123       (0.000575) (0.0102) Mean number of employees of local firms in sector     0.0150*** −0.00814       (0.00327) (0.0516) Mean average wages of local firms in sector     0.00144 −5.26E-05       (0.00152) (0.0203)           Country fixed effects Yes Yes Yes Yes Sector fixed effects     Yes Yes Cluster standard errors Region Region Sector Sector Note: The estimation is performed between each dependent variable and independent variable separately. For example, the share of foreign firms in a region is regressed on the mean operating revenue of local firms in the region. The sector is equivalent to the 1-digit NACE rev. 2 code. The region is defined at the NUTS-3 level of aggregation, or at a more aggregated level if NUTS-3 is not available. Robust standard errors are in parentheses. Significance level: ** = 5 percent, *** = 1 percent. Connectivity and Firms ●  143 either have higher sales or employment or pay higher wages. In other words, for- eign firms may locate in regions with better access to public services or infrastruc- ture or a skilled labor force. Accordingly, foreign investors may choose to own firms in sectors that are growing for reasons such as better relative prices or technologi- cal change. Thereby, a positive correlation between a higher presence of foreign firms and domestic firms’ performance may capture not only an effect of the for- mer on the latter, but also the impact of a host of other factors. With these caveats in mind, table 3.5 explores how a higher presence of foreign-owned firms may affect the performance of firms with local owners. The first ­ two rows indicate that domestic firms perform better, in terms of sales growth, in regions and sectors with a higher presence of foreign firms. Accordingly, domestic firms in regions or sectors with a higher fraction of foreign firms experience a higher rate of average wage growth. In contrast, a higher presence of foreign firms in a region is not significantly correlated with higher employment growth of domestic firms. In other words, if these coefficients are biased upward, these results may suggest that while a higher presence of foreign firms in an area may generate some displacement effects (in the sense that domestic firms lay off employees), some domestic firms are able to become more efficient and thereby increase their sales and wages paid.15 Not only the presence of foreign firms matters, but also the quality of their man- agement practices. As seen in the third row of table 3.5, domestic firms grow faster in terms of sales, employment, and wages when foreign firms in the region are from countries with better management practices. For instance, domestic firms’ sales and employment grow by approximately an additional 3 percentage points in regions where the AMI score of the foreign firms in the region is 6 (a score almost as high as Finland’s) instead of 4 (a score slightly lower than Bosnia and Herzegovina’s). In contrast, better-managed foreign firms in the same sector seem to be associated with lower wage growth among domestic firms. Since changes in the measure of TABLE 3.5  Domestic Firms in ECA Grow Faster in Regions with Well-Managed Foreign Firms Log(Operating revenue), Log(Number of employees), Log(Average wage), change, 2010–13 change, 2010–13 change, 2010–13 Share of foreign firms in the region 0.500*** 0.0405 0.466***   (0.112) (0.0593) (0.0685) Share of foreign firms in the sector 0.762*** 0.271*** 0.728***   (0.181) (0.100) (0.180) AMI score of foreign firms in the region 0.0133*** 0.0158*** 0.00557** (0.00396) (0.00308) (0.00233) AMI score of foreign firms in the sector 0.00263 0.0122 −0.00850* (0.00930) (0.00772) (0.00499) Number of observations 2,273,884 2,891,026 1,741,141 R2 0.062 0.177 0.122 Log of dependent variable in level, 2010 Included Included Included Country fixed effects Yes Yes Yes Sector fixed effects Yes Yes Yes Size fixed effects Yes Yes Yes Age fixed effects Yes Yes Yes Note: Sample includes firms with local owners. The region is defined at the NUTS-3 level of aggregation, or at a more aggregated level if NUTS-3 is not available. The sector is equivalent to the 1-digit NACE rev. 2 code. The Average Management Index (AMI) score is that of the foreign country of ownership, from 1 (worst) to 7 (best). The log of the dependent variable in level in 2010 is included to control for regression toward the mean. Robust standard errors are in parentheses. ECA = Europe and Central Asia. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. 144  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia wage growth used here are driven by both changes in the level of wages and the composition of employment, this result is consistent with foreign firms in the sector attracting high-wage employees from domestic firms. The horizontal lines in figure 3.10 show the spillover effects by region and sector (from table 3.5) for the three indicators of firm performance. Bigger and older firms seem to capture the positive spillover effects of having well-managed foreign firms in their sector or region to a larger extent than small and young ones do. As seen in figure 3.10, the total spillover effects for small and young firms (i.e., those four years old or younger and with fewer than 50 employees in 2007) are negative, whether we consider the regional or sectoral dimensions, ­ suggesting that small and young firms are more likely to suffer the displacement effects of an increased prevalence of well-managed foreign companies. In c ­ ontrast, the correlation between performance and the presence of foreign companies is positive and large for big and older domestic companies (i.e., those with 50 employ- ees or more, or older than four years), suggesting that they may actually be better able to cope with the additional competition and thereby experience a larger increase in sales, average size, and wages than their smaller counterparts. How does a higher presence of foreign companies translate into shared prosperity? While the data do not allow providing a definitive answer to this ques- ­ tion, the estimated relationships between the share and the performance of ­ foreign companies, and the performance of domestic ones provide some guidance. On the one hand, the results from the previous section suggest that foreign companies’ size (their number of employees) and average wages grow faster than those of their domestic counterparts. At the same time, the presence of foreign companies from countries with good management practices is positively correlated with the perfor- mance of domestic firms, especially bigger and older ones. On the other hand, the results suggest the existence of some displacement effects of well-managed foreign companies for smaller and younger domestic firms. The sign of the net effects will depend on the extent to which the displaced workers from small and young compa- nies can be absorbed by other domestic firms or by foreign companies. Connected Firms: A Bridge to Economic Development? The international connections of firms in ECA have historical origins based on language and geographic proximity, but they were also shaped by evolving economic forces. Countries that have strong ties through international trade or ­ migration are also more likely to have businesses that are owned or managed by citizens from their partner countries. A virtuous circle exists in which one type of connection reaffirms and even magnifies the other ones. When firms in ECA have owners or managers from more advanced economies, they tend to grow faster than the rest. Part of this successful performance may be due to technology trans- fers but also to the transfer of soft technologies, as foreign companies can trans- plant their more efficient management practices to developing countries where business skills are scarcer. In ECA, even local businesses may benefit from the presence of well-managed foreign companies, as competitive pressures may force them to become more efficient to stay in the market. Another advantage of having country connected firms is that they may be more resilient to negative shocks in the ­ of location. In ECA, foreign firms tend to smooth the impact of the local business Connectivity and Firms ●  145 FIGURE 3.10  The positive spillovers of well-managed foreign firms seem weaker for small and young firms a. Regional spillover effects 60 2 Changes due to a one-unit increase in the Average Management Index All-firm impact (right axis) score of foreign firms in the region (percentage points) 1.5 40 1 20 0.5 0 0 –0.5 –20 –1 –40 –1.5 –60 –2 Mature and Small and Mature and Small and Mature and Small and larger firms young firms larger firms young firms larger firms young firms Change in operating revenue Change in number of employees Change in average wage b. Sectoral spillover effects 20 1.5 Changes due to a one-unit increase in the Average Management Index All-firm impact (right axis) 15 1.0 score of foreign firms in the sector (percentage points) 10 0.5 5 0 0.0 –5 –0.5 –10 –15 –1.0 –20 –1.5 –25 –2.0 –30 –35 –2.5 Mature and Small and Mature and Small and Mature and Small and larger firms young firms larger firms young firms larger firms young firms Change in operating revenue Change in number of employees Change in average wage Note: Table 3B.3 presents full regression results. Small firms are those with 49 employees or fewer; young firms are those four years old or younger. Each bar represents the effect of increasing by one point the Average Management Index scores of foreign firms. The bars in each panel show the baseline effect (mature and larger firms), the baseline effect plus the interaction term associated with size, and the baseline effect plus the interaction terms associated with size and age simultaneously, according to table 3B.3. The horizontal lines show these spillover effects for all local firms as a benchmark, by region and sector (from table 3.5), for the three indicators of firm performance. 146  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia cycle, by creating more employment than local firms during a downturn and grow- ing more slowly during a boom. However, improving firms’ connection to the outside world also comes with risks. Domestic firms that are unable to compete with their more efficient foreign coun- terparts may have no choice but to downsize or leave the market. In ECA, small and young firms seem the most at risk when they have to compete against Improving the foreign firms in the region. The associated job destruction could have a connection of firms to the negative impact on households’ welfare and shared prosperity, since outside world also comes jobs and labor income are a crucial component of households’ total with risks. Domestic firms income. Accordingly, while foreign firms are more resilient to local that are unable to shocks, they can also make the local economy more exposed to compete with their more efficient foreign counter- external shocks, as the performance of foreign firms in ECA is highly parts may have no choice linked to the level of economic activity of the owners’ and managers’ but to downsize or leave country of origin. the market. Given this evidence, policies should aim to boost the positive effects of ­ connectivity while minimizing the risks. Improving the busi- ness environment and removing barriers to FDI and trade would strengthen the incentives for foreign investors to invest in new or existing local firms (see, for example, Antras, Desai, and Foley 2009). Investment promotion has also been an effective tool for fostering FDI by reducing information asymmetries and unneces- sary procedures in developing countries,16 which can be a significant obstacle to capital flows across b ­ orders (Harding and Javorcik 2011).17 Policies to promote skilled migration may also be an effective tool for fostering the transfer of manage- ment practices from abroad. Finally, governments should also employ policies to promote firm linkages, especially those between foreign affiliates and domestic firms, to increase the spillover benefits of foreign ownership. These policies can focus on more innovation transfers between foreign and domestic firms.18 Policies to level the playing field among firms can help local businesses catch up with their more efficient foreign competitors. For example, improving access to finance for small and young firms can facilitate productive investments and expan- sion. Policies to improve the quality of local management are also crucial. More business education could contribute to improving management, especially in developing economies (Bloom and Van Reenen 2010). Labor market regulations ­ that make it difficult to hire, lay off, pay, and promote employees may also restrict the ability of firms to manage workers efficiently. When foreign firms lead their local counterparts to downsize or go out of business, the final impacts on workers will depend on the extent to which they can ­ cope with the shock of losing their jobs. On the one hand, the extent to which workers can adjust to the shock by moving to another location, or find a job in another sector or occupation, will be crucial to mitigating the final economic impact. If workers can easily relocate, or have skills that are transferable to a job in another sector or occupation, the impact of displacement on their well-being will likely be smaller than in a case in which workers cannot quickly find another job. Policies to facilitate geographic mobility or to improve education (in terms of quality, but also in terms of fostering lifelong learning) can contribute to smoothing ­ the impact of displacement. At the same time, a social insurance system that pro- tects workers in the short term and does not distort the decision to work or to look for a job could mitigate the negative short-term impacts of unemployment. Connectivity and Firms ●  147 Multinational firms could serve as a bridge to economic development by ­acilitating the cross-border transfer of factors that are key to explaining the suc- f cess of richer countries. However, higher connectivity also entails more exposure to new risks from abroad. Policies to promote capital and skilled labor inflows can set countries on the path of long-term economic growth. Policies to smooth the short-term disruptive impacts of higher connectivity are also crucial, not only to protect vulnerable workers but also for the political economy of implementing long-term economic reforms. Annex 3A. Coverage of Orbis Data The data for this analysis come from the Orbis database, which covers 170 million private (99 percent) and public firms from around the world. The data are collected by Bureau van Dijk (BvD), which also maintains the quality and the accuracy of the database and ensures standardized formats, making cross-border comparisons possible (http://www.bvdinfo.com/). The database contains detailed firm-level information on balance sheet accounts, profit and loss accounts, sectors, location, ownership structures, and management, and relies on firms’ legal obligation to file accounts. The coverage varies substantially across countries, and for many coun- tries, information for only a few firms is usable. The data are processed per the recommendations by Kalemli-Özcan et al. (2015) to ensure national representa- ­ tiveness. In the estimation on foreign versus local and the management regres- sions (table 3B.1 and table 3B.2), we look at 30 European countries,19 and in the estimation on spillovers (table 3B.3 and table 3B.4), we include 26 out of the 30 countries.20 The most recent data are from 2013 and while the balance sheet information is available at a yearly frequency, information such as ownership and management is available only for the most recent year. Ownership and management are therefore approximated in all years by information in 2013. Moreover, as opposed to the global ultimate owner, there may be many people with different nationalities listed as management personnel and the primary nationality of the management per- sonnel is used to classify the nationality of the managers in each firm.21 The data sets used for the analysis include only unconsolidated financial statements to avoid double counting of firms,22 and all financial variables are ­ adjusted to 2010 prices. Observations lacking information essential to the analysis or obviously incorrect registrations are also dropped from the sample. This includes firms with no record of an opening or a closing date, balance sheet information prior to the registered opening date of the firm (negative age), or no industry speci- fication. Observations are also dropped from the sample if information on total assets, operating turnover, sales, and number of employees is missing; the ­ currency of the financial variables is misstated;23 or material costs, operating revenue, and total assets are either missing or negative. Moreover, all firm observations are dropped from the sample if total assets, sales, or tangible fixed assets are negative in any year or if the number of employees is negative or larger than 2 ­ million in any year. Some firms have registered multiple financial accounts within the same year. This may be explained by quarterly reports (Kalemli-Özcan et al. 2015), and to prevent double counting, only the observation with the highest operating revenue 148  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia is kept in each year. Moreover, firms reporting their financial variables in different units over time are dropped from the sample if the change in unit is accompanied by a change in total assets of more than 70 percent. Firms registered with zero employees are also dropped, thereby making the sample more comparable across countries as well as in an attempt to exclude dormant firms. The classification levels of sectors and regions across countries are largely the same when constructing the spillover variables (NACE Rev. 2). However, the level of detail in the registration of the regional location of firms varies considerably across countries. Therefore, firms are classified according to the NUTS-3 classification if possible. When this level of disaggregation is not feasible, the NUTS-2 level is used. Annex 3B. Additional Tables TABLE 3B.1  Foreign-Owned and Foreign-Managed Firms Perform Better Than Local Firms   Log (Operating revenue), Log (Number of employees), Log (Average wage), change, change, 2010–13 change, 2010–13 2010–13 Foreign owned, locally managed 0.227*** 0.139*** 0.117***   (0.0459) (0.0435) (0.0375) Foreign managed, locally owned 0.136*** 0.110*** 0.0551***   (0.0213) (0.0306) (0.0191) Foreign owned and managed 0.249*** 0.179*** 0.155***   (0.0353) (0.0576) (0.0307) Number of observations 2,482,453 3,131,653 1,886,679 R2 0.063 0.171 0.111 Note: The independent variables in the regression are dummy variables that equal 1 if the firm is foreign owned, foreign managed, or both. As the dependent variable is in logs, the coefficients of the dummy variables have to be calculated according to the formula: 100 × [exp(b)−1]. The regression includes the initial values of the dependent variables to account for mean reversion and country, sector, size, and age fixed effects. The sector is equivalent to the 1-digit NACE rev. 2 code. Age is grouped into intervals of 0–4 years, 5–14 years, 15–30 years, and 30+ years, and size is grouped into intervals of 1–2 employees, 3–7 employees, 8–49 employees, 50–249 employees, and 250+ employees. Robust standard errors (in parentheses) are clustered around country of ownership. Significance level: *** = 1 percent. TABLE 3B.2  Firms with Better Foreign Managers Perform Better   Log (Operating revenue), Log (Number of employees), Log (Average wage), change, change, 2010–13 change, 2010–13 2010–13 Log management index 0.192*** 0.166*** −0.0427   (0.0541) (0.0323) (0.0683) Log GDP per capita 0.0313*** 0.0195*** 0.00732   (0.0114) (0.00451) (0.00580) Log market capitalization of GDP −0.0147 −0.00984* 0.00391   (0.00889) (0.00513) (0.0117) Log population 0.00878** 0.0104*** 0.00693***   (0.00423) (0.00226) (0.00230) Log immigrant stock −0.00583** −0.00402*** −0.000567   (0.00229) (0.00143) (0.00211) Number of observations 98,208 120,305 89,521 R2 0.046 0.073 0.065 Note: The estimation is restricted to foreign affiliates. The independent variables in the regression capture the characteristics of the country of ownership. The regression includes the initial values of the dependent variables to account for mean reversion and country, sector, size, and age fixed effects. The sector is equivalent to the 1-digit NACE rev. 2 code. Age is grouped into intervals of 0–4 years, 5–14 years, 15–30 years, and 30+ years; and size is grouped into intervals of 1–2 employees, 3–7 employees, 8–49 employees, 50–249 employees, and 250+ employees. Robust standard errors (in parentheses) are clustered around country of ownership. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. Connectivity and Firms ●  149 TABLE 3B.3  Spillover Effects of Foreign-Owned Firms on Domestic Firms Log(Operating Log(Number of Log(Average revenue), change, employees), change, wage), change, Variables 2010–13 2010–13 2010–13 Share of foreign firms in the region 1.158*** 0.576*** 1.707***   (0.184) (0.137) (0.202) Share of foreign firms in the sector 0.806*** 0.437*** 1.052***   (0.195) (0.167) (0.222) AMI score of foreign firms in the region 0.145*** 0.368*** 0.138***   (0.0221) (0.0464) (0.0224) AMI score of foreign firms in the sector 0.0531** 0.126** 0.0368*   (0.0250) (0.0506) (0.0221) AMI score of foreign firms in the region × Small firm −0.696*** −0.555*** −1.297***   (0.160) (0.135) (0.199) Share of foreign firms in the region × Small firm −0.0921 −0.211 −0.529***   (0.142) (0.151) (0.149) AMI score of foreign firms in the sector × Small firm −0.120*** −0.350*** −0.133***   (0.0213) (0.0445) (0.0228) Share of foreign firms in the sector × Small firm −0.0503** −0.116** −0.0448**   (0.0246) (0.0502) (0.0224) AMI score of foreign firms in the region × Young firm −0.135 −0.0715 −0.346**   (0.145) (0.115) (0.167) Share of foreign firms in the region × Young firm −0.191 −0.315 −0.150   (0.186) (0.220) (0.149) AMI score of foreign firms in the sector × Young firm −0.125*** −0.159*** −0.0242***   (0.0185) (0.0337) (0.00917) Share of foreign firms in the sector × Young firm −0.0147 −0.0237* −0.0117   (0.0190) (0.0126) (0.00796)   2.933*** 0.350*** 3.321***   (0.214) (0.0923) (0.311) Number of observations 2,273,882 2,891,023 1,741,140 R2 0.063 0.182 0.124 Initial value Included Included Included Country fixed effects Yes Yes Yes Sector fixed effects Yes Yes Yes Size fixed effects Yes Yes Yes Age fixed effects Yes Yes Yes Cluster Sector country Sector country Sector country Outliers Included Included Included Note: Sample includes firms with local owners. The region is defined at the NUTS-3 level of aggregation or at a more aggregated level if NUTS-3 is not available. The sector is equivalent to the 1-digit NACE rev. 2 code. The Average Management Index (AMI) score is that of the foreign country of ownership, from 1 (worst) to 7 (best). The log of the dependent variable in level in 2010 is included to control for regression toward the mean. Large firms are those with 50 employees or more; young firms are those four years old or younger. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. 150  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE 3B.4  Firm Growth over the Business Cycle   Log(Operating Log(Number of Log(Average revenue), annual employees), annual wage), annual   change change change Foreign owned, Locally managed 0.0669*** 0.0410*** 0.0369***   (0.0148) (0.0116) (0.0117) Foreign managed, Locally owned 0.0422*** 0.0366*** 0.0175***   (0.00700) (0.0102) (0.00613) Interaction between Foreign managed and Foreign owned 0.0823*** 0.0678*** 0.0470***   (0.0120) (0.0219) (0.0120) Local GDP growth 2.435 −0.0576 −1.151   (1.551) (0.405) (0.824) GDP growth of country of ownership 0.150 0.272** −0.0147   (0.106) (0.102) (0.0782) Foreign-owned × Local GDP growth −0.851** −1.170** 0.0223   (0.390) (0.512) (0.473) Number of observations 7,097,750 9,108,130 5,421,604 R2 0.019 0.073 0.024 Initial value Included Included Included Country fixed effects Yes Yes Yes Sector fixed effects Yes Yes Yes Size fixed effects Yes Yes Yes Age fixed effects Yes Yes Yes Cluster Country Country Country Year fixed effects Yes Yes Yes Note: Sample includes all firms with annual data between 2010 and 2013. Robust standard errors are in parentheses. Significance level: ** = 5 percent, *** = 1 percent. Notes 1. As the dependent variable is in logs, the results do not change if the foreign firms are taken as a share of all firms or all foreign firms in the country. 2. The estimation of the gravity model follows Anderson and van Wincoop (2003). As with most trade and FDI data, there are source countries that do not have any foreign ­ affiliates or foreign managers in the host country. These zero shares may represent either a true absence of a linkage or missing data, which can bias the results. The results in table 3.2 presents the ordinary least squares (OLS) coefficients. To account for the bias, a Poisson pseudo–maximum likelihood (PPML) estimation is also per- formed following Silva and Tenreyro (2006). The results from the PPML estimation are not presented but are s ­ imilar to the OLS results. 3. There are lists of tax haven countries from different sources (for example, OECD 2000; Dharmapala and Hines 2009). We use a combined list produced by Gravelle (2015), which combines the various lists and classifies these countries as tax havens: Andorra; Anguilla; Antigua and Barbuda; Aruba; The Bahamas; Bahrain; Barbados; Belize; Bermuda; British Virgin Islands; Cayman Islands; Cook Islands; Costa Rica; Cyprus; Dominica; Gibraltar; Grenada; Guernsey; Hong Kong SAR, China; Ireland; Isle of Man; Jersey; Jordan; Lebanon; Liberia; Liechtenstein; Luxembourg; Macau SAR, China; Maldives; Malta; Marshall Islands; Mauritius; Monaco; Montserrat; Nauru; the Netherlands Antilles; Niue; Panama; Samoa; San Marino; Seychelles; Singapore; St. Kitts and Nevis; St. Lucia; St. Vincent and the Grenadines; Switzerland; Tonga; Turks and Caicos; US Virgin Islands; and Vanuatu. 4. Konings (2001), on the other hand, looks at firm performance in Bulgaria, Romania, and Poland and find that only in Poland do foreign firms on average perform better than domestic firms. Connectivity and Firms ●  151 5. Where foreign firms acquire firms that are more productive or more promising than the average firm on the market as well as locate in more productive sectors and regions (Aitken and Harrison 1999). 6. More specifically, the findings of Görg, Strobl, and Walsh (2007) suggest that the training of employees provided in foreign-owned firms is more productive than that in ­ domestically owned firms. 7. Dries and Swinnen (2004) examine the Polish dairy sector. 8. Haskel, Pereira, and Slaughter (2007) find evidence of intra-industry spillovers but not regional spillovers. 9. Görg and Greenaway (2004) note that the two analyses supporting the idea of positive wage spillovers should be treated with caution because they are based on cross-­ sectional data. 10. The analysis is conducted with data from between 1976 and 1989. 11. Management practices can also be improved through business training programs. Bloom et al. (2013) provided management training to Indian textile firms, and firms that received free consulting services and adopted the new management practices increased their productivity by 17 percent. These firms were also able to grow faster, compared with firms that did not receive the consulting services. 12. The correlation between the WMS management index and the WEF management index is 0.73. 13. Blomström, Lipsey, and Kulchycky (1988) look at the relationship for both Sweden and the United States. In the case of Sweden, they find a positive relationship whereas the evidence is mixed for the United States. 14. See Gorg and Greenaway (2004) and Smeets (2008) for a discussion. 15. Given that the omitted factors seem to be positively correlated with both the presence of foreign firms and the performance of domestic ones, the estimated OLS coefficients would be biased upward. 16. The authors find a positive effect of investment promotion on FDI only for ­ developing countries, not industrial economies. 17. Harding and Javorcik (2012) also find that higher-quality investment promotion ­agencies (measured by more professional service standards and better websites) are able to attract more FDI. 18. Sánchez-Martín et al. (2015) find that foreign firms have more backward linkages with (i.e., purchase inputs from) domestic firms if (a) these foreign firms entered the country to serve the market (“market-seeking”) compared with those that are producing locally and exporting their goods back to the home country (“export oriented”); (b) the foreign firms are not using foreign-licensed technology. 19. Austria, Belgium, Bosnia and Herzegovina, Bulgaria, Croatia, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, the Netherlands, Poland, Portugal, Romania, Russia, Serbia, the Slovak Republic, Slovenia, Spain, Sweden, Ukraine, and the United Kingdom. 20. Austria, Belgium, Bosnia and Herzegovina, Bulgaria, Croatia, the Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Lithuania, the Netherlands, Poland, Portugal, Romania, Russia, the Slovak Republic, Slovenia, Spain, Sweden, Ukraine, and the United Kingdom. 21. If there are an equal number of local managers and managers from the same foreign ­ country, then the foreign country is assigned as the primary nationality of the manage- ment personnel in the firm. If there are an equal number of managers from different foreign ­ countries, then the primary nationality of the management personnel of the firm is chosen randomly. 22. We do not include firms with limited financial variables in the samples. Information on firms with limited financial variables is often based on rounded figures or class levels officially available. For most of these firms only information on the number of employ- ­ ees and the operating revenue is available (BvD online, Orbis—User Guide). 23. We assume that the currency is misstated if it is not the local currency, former local ­currencies, the US dollar, the euro, the British pound, the Chinese yuan, or the Japanese yen. 152  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia References Aitken, B. J., G. H. Hanson, and A. E. Harrison. 1997. “Spillovers, Foreign Investment, and Export Behavior.” Journal of International Economics 43 (1): 103–32. Aitken, B. J. and A. E. Harrison. 1999. “Do Domestic Firms Benefit from Direct Foreign Investment? Evidence from Venezuela.” American Economic Review 89 (3): 605–18. Anderson, J. E. and E. Van Wincoop. 2003. “Gravity with Gravitas: A Solution to the Border Puzzle.” American Economic Review 93 (1): 170–92. Antras, Pol, Mihir A. Desai, and C. Fritz Foley. 2009. “Multinational Firms, FDI Flows, and Imperfect Capital Markets.” Quarterly Journal of Economics 124 (3): 1171–219. Atalay, E., A. Hortaçsu, and C. Syverson. 2014. “Vertical Integration and Input Flows.” American Economic Review 104 (4): 1120–48. Bao, Cathy Ge, and Maggie X. Chen. 2016. “Foreign Rivals Are Coming to Town: Responding to the Threat of Foreign Multinational Entry.” Working Paper 2016-22, Institute for International Economic Policy, George Washington University, Washington, DC. Bender, S., N. Bloom, D. Card, J. Van Reenen, and S. Wolter. 2016. “Management Practices, Workforce Selection and Productivity.” Working Paper 22101, National Bureau of Economic Research, Cambridge, MA. Blalock, Garrick, and Paul J. Gertler. 2008. “Welfare Gains from Foreign Direct Investment through Technology Transfer to Local Suppliers.” Journal of International Economics 74 (2): 402–21. Blomström, M., R. Lipsey, and K. Kulchycky. 1988. “U.S. and Swedish Direct Investment and Exporters.” In Trade Policy Issues and Empirical Analysis, edited by R. E. Baldwin, 259–97. Chicago, IL: University of Chicago Press. Bloom, N., E. Brynjolfsson, L. Foster, R. Jarmin, M. Patnaik, I. Saporta-Eksten, and J. Van Reenen. 2016. “What Drives Differences in Management?” Unpublished manu- script, Stanford University, Stanford, CA. Bloom, N., B. Eifert, A. Mahajan, D. McKenzie, and J. Roberts. 2013. “Does Management Matter? Evidence from India.” Quarterly Journal of Economics 128 (1): 1–51. Bloom, N., R. Sadun, and J. Van Reenen. 2012. “Americans Do IT Better: US Multinationals and the Productivity Miracle.” American Economic Review 102 (1): 167–201. ———. 2016. “Management as a Technology?” Working Paper 22327, National Bureau of Economic Research, Cambridge, MA. Bloom, N., and J. Van Reenen. 2010. “Why Do Management Practices Differ across Firms and Countries?” Journal of Economic Perspectives 24 (1): 203–24. Boehm, C. E., A. Flaaen, and N. Pandalai-Nayar. 2016. The Role of Global Supply Chains in the Transmission of Shocks: Firm-Level Evidence from the 2011 To¯ hoku Earthquake. Washington, DC: Board of Governors of the Federal Reserve System. Borchert, I., B. Gootiiz, and A. Mattoo. 2012. “Guide to the Services Trade Restrictions Database.” Policy Research Working Paper 6108, World Bank, Washington, DC. ———. 2014. “Policy Barriers to International Trade in Services: Evidence from a New Database.” World Bank Economic Review 28 (1): 162–88. Buch, C. M., and A. Lipponer. 2010. “Volatile Multinationals? Evidence from the Labor Demand of German Firms.” Labour Economics 17 (2): 345–53. Budd, J. W., J. Konings, and M. J. Slaughter. 2005. “Wages and International Rent Sharing in Multinational Firms.” Review of Economics and Statistics 87 (1): 73–84. Clausing, K. A. 2000. “Does Multinational Activity Displace Trade?” Economic Inquiry 38 (2): 190–205. Conyon, M. J., S. Girma, S. Thompson, and P. W. Wright. 2002. “The Productivity and Wage Effects of Foreign Acquisition in the United Kingdom.” Journal of Industrial Economics 50 (1): 85–102. Connectivity and Firms ●  153 Cravino, J., and A. A. Levchenko. 2016. “Multinational Firms and International Business Cycle Transmission.” Quarterly Journal of Economics 132 (2): 921–62. Dharmapala, D., and J. R. Hines. 2009. “Which Countries Become Tax Havens?” Journal of Public Economics 93: 1058–68. Djankov, Simeon, and Bernard Hoekman. 2000. “Foreign Investment and Productivity Growth in Czech Enterprises.” World Bank Economic Review 14 (1): 49–64. Dries, L., and J. F. M. Swinnen. 2004. “Foreign Direct Investment, Vertical Integration, and Local Suppliers: Evidence from the Polish Dairy Sector.” World Development 32 (9): 1525–44. Frankel, J., and A. Rose. 1998. “The Endogeneity of the Optimum Currency Area Criteria.” Economic Journal 108: 1009–25. Girma, S., D. Greenaway, and K. Wakelin. 2001. “Who Benefits from Foreign Direct Investment in the UK?” Scottish Journal of Political Economy 48 (2): 119–33. Görg, H., and D. Greenaway. 2004. “Much Ado about Nothing? Do Domestic Firms Really Benefit from Foreign Direct Investment?” World Bank Research Observer 19 (2): 171–97. Görg, H., E. Strobl, and F. Walsh. 2007. “Why Do Foreign-Owned Firms Pay More? The Role of On-the-Job Training.” Review of World Economics 143 (3): 464–82. Gravelle, Jane G. 2015. “Tax Havens: International Tax Avoidance and Evasion.” Washington, DC: Congressional Research Service. Haddad, M., and A. Harrison. 1993. “Are There Positive Spillovers from Direct Foreign Investment? Evidence from Panel Data for Morocco.” Journal of Development Economics 42: 51–74. Harding, Torfinn, and Beata S. Javorcik. 2011. “Roll Out the Red Carpet and They Will Come: Investment Promotion and FDI Inflows.” Economic Journal 121 (557): 1445–76. ———. 2012. “Investment Promotion and FDI Inflows: Quality Matters.” CESifo Economic Studies 59 (2): 337–59. Haskel, J. E., S. C. Pereira, and M. J. Slaughter. 2007. “Does Inward Foreign Direct Investment Boost the Productivity of Domestic Firms?” Review of Economics and Statistics 89 (3): 482–96. Head, K., and J. Ries. 2001. “Overseas Investment and Firm Exports.” Review of International Economics 9 (1): 108–22. Heyman, F., P. J. Norback, and R. Hammarberg. 2014. “Foreign Direct Investment, Source Country Heterogeneity and Management Practices.” Working Paper 1041, Research Institute of Industrial Economics (IFN), Stockholm. Hollweg, C. H., D. Lederman, D. Rojas, and E. R. Bulmer. 2014. Sticky Feet: How Labor Market Frictions Shape the Impact of International Trade on Jobs and Wages. Washington, DC: World Bank. Javorcik, B. S. 2004. “Does Foreign Direct Investment Increase the Productivity of Domestic Firms? In Search of Spillovers through Backward Linkages.” American Economic Review 94 (3): 605–27. ———. 2015. “Does FDI Bring Good Jobs to Host Countries?” World Bank Research Observer 30 (1): 74–94. Kalemli-Özcan, S., B. Sorensen, C. Villegas-Sanchez, V. Volosovych, and S. Yesiltas. 2015. “How to Construct Nationally Representative Firm Level Data from the ORBIS Global Database.” Working Paper 21558, National Bureau of Economic Research, Cambridge, MA. Keller, W., and S. R. Yeaple. 2009. “Multinational Enterprises, International Trade, and Productivity Growth: Firm-Level Evidence from the United States.” Review of Economics and Statistics 91 (4): 821–31. Kleinert, J., J. Martin, and F. Toubal. 2015. “The Few Leading the Many: Foreign Affiliates and Business Cycle Comovement.” American Economic Journal: Macroeconomics 7 (4): 134–59. 154  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Kokko, A., R. Tansini, and M. C. Zejan. 1996. “Local Technological Capability and Productivity Spillovers from FDI in the Uruguayan Manufacturing Sector.” Journal of Development Studies 32: 602–11. ———. 2001. “Trade Regimes and Spillover Effects of FDI: Evidence from Uruguay.” Review of World Economics 137 (1): 124–49. Konings, J. 2001. “The Effects of Foreign Direct Investment on Domestic Firms: Evidence from Firm Level Panel Data in Emerging Economies.” Economics of Transition 9 (3): 619–33. Kosová, R. 2010. “Do Foreign Firms Crowd Out Domestic Firms? The Evidence from the Czech Republic.” Review of Economics and Statistics 92 (4): 861–81. Lipsey, R. E., and F. Sjöholm. 2001. “Foreign Direct Investment and Wages in Indonesian Manufacturing.” Working Paper 8299, National Bureau of Economic Research, Cambridge, MA. ———. 2004. “Foreign Direct Investment, Education and Wages in Indonesian Manufacturing.” Journal of Development Economics 73 (1): 415–22. Lipsey, R. E., and M. Y. Weiss. 1984. “Foreign Production and Exports of Individual Firms.” Review of Economics and Statistics 66 (2): 304–8. Navaretti, G. B., D. Checchi, and A. Turrini. 2003. “Adjusting Labor Demand: Multinational Versus National Firms: A Cross-European Analysis.” Journal of the European Economic Association 1 (2–3): 708–19. OECD (Organisation for Economic Co-operation and Development). 2000. Towards Global Tax Competition. Paris: OECD. Sánchez-Martín, Miguel Eduardo, Jaime De Piniés, and Kássia Antoine. 2015. “Measuring the Determinants of Backward Linkages from FDI in Developing Economies: Is It a Matter of Size?” Policy Research Working Paper 7185, World Bank, Washington, DC. Silva, J. S., and S. Tenreyro. 2006. “The Log of Gravity.” Review of Economics and Statistics 88 (4): 641–58. Smeets, R. 2008. “Collecting the Pieces of the FDI Knowledge Spillovers Puzzle.” World Bank Research Observer 23 (2): 107–38. Syverson, C. 2004. “Product Substitutability and Productivity Dispersion.” Review of Economics and Statistics 86 (2): 534–50. Vahter, P., and J. Masso. 2007. “Home versus Host Country Effects of FDI: Searching for New Evidence of Productivity Spillovers.” Applied Economics Quarterly 53: 165–96. ­nvestments Reaping Digital Dividends through Complementary I ●  155 SPOTLIGHT 3 Reaping Digital Dividends through Complementary ­Investments T he growth of supply chains, which has been driven by improvements in both telecommuni- cations and transport infrastructure, is one example foreign products into the US market finds that the probability of product entry increases 0.65 percent when there are 10 additional internet users per 100 of the interdependence of different connectivity people in the foreign country, but the probability channels in contributing to growth.1 Similarly, inter- increases 1.18 percent when these 10 additional net technology has tremendous potential to trans- internet users are in a country with highly efficient form ­ economic systems in Europe and Central Asa export logistics (Riker 2015). Finally, an efficient (ECA). But in addition to establishing the necessary postal system can reduce the costs for e-­ commerce infrastructure and regulatory environment for the companies to make the last-mile parcel delivery to ­ internet, reaping the full benefits from internet con- their consumers. nectivity requires a host of complementary activi- E-commerce also requires the availability of ties. Efficient transport infrastructure, financial electronic payment instruments, such as debit systems, education, and institutions that support or credit cards. For example, one reason that market competition are all important to enable indi- commerce is more limited in the European Union e-­ viduals to exploit internet technology and avoid (EU) than in Japan or the United States, despite the potentially undesirable consequences from this EU’s greater affordability and quality of internet technology. access, is that the EU is well behind these countries Successful e-commerce requires strong trade in the use of credit cards. Similarly, countries in infrastructure to guarantee timely and reliable deliv- Central and Eastern Europe underperform the ery of goods. Improvements in trade facilitation, par- economies of East Asia in this regard. In Central ticularly streamlining procedures and investing in Asia, the South Caucasus, and the Western Balkans, the infrastructure required to ­ speedily transit imports the use of credit cards is almost negligible. On through ports, could greatly expand the potential average, only 15 percent of individuals in many ­ for e-commerce in many ECA countries. For exam- ECA countries have credit cards, compared with ple, eBay introduced a global shipping program that about 50 percent in Western Europe.2 Together, the facilitates shipping and customs clearance for its quality of payment systems and logistics systems is sellers. Sellers selected for the program had strongly associated with the share of firms with 2.7 ­percent more exports than those not selected. online sales (figure S3.1). Conversely, countries can leverage information and Adequate levels of education are necessary for communication technology (ICT) to modernize individuals and firms to use the internet effectively. customs agencies and procedures. For example, ­ For instance, more than 50 percent of 15-year-olds Albania reduced its customs clearance time and in Central Asia are functionally illiterate (that is, they increased trade significantly from 2007 to 2012 by know how to read and write, but cannot make infer- implementing the Automated System for Customs ences or understand forms of indirect meaning).3 Data to improve its risk management and inspection Given this deficit, better internet provision may not processes. Efficient logistics and internet use can necessarily complement their skills and translate reinforce each other. A study examining the entry of them into wage gains. At the same time, 156  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia SPOTLIGHT 3 continued FIGURE S3.1  Firms’ online sales rise with more efficient logistics and payment systems 50 DNK Share of firms that sell online (percent) UKR SVN ISL SWE 40 BEL NLD CZE DEU RUS IRL NOR GBR 30 EST FRA FIN ARM HRV PRT MDA SRB MNE AZE LTU AUT ESP 20 HUN SVK ALB MLT KAZ POL LVA LUX GEO ITA KGZ ROU BGR TUR 10 BIH TJK CYP GRC MKD 0 2 3 4 5 6 Quality of logistics infrastructure and payment systems Emerging Transitioning Transforming Fitted values Source: Calculations based on data from the World Bank Enterprise Surveys, Eurostat, and World Economic Forum 2015. Note: Data for firms selling are calculated from the Enterprise Survey and Eurostat and obtained for the latest available years. The quality of logistics infrastructure and payment systems is calculated as an average of the response from executives in the World Eco- nomic Forum (WEF) Competitiveness Survey. The questions concern the quality of roads, railroad infrastructure, port infrastructure, and air transport infrastructure and the affordability and availability of financial services, with the answers ranging from 1 (worst) to 7 (best). For each country, the year of the data from the WEF Competitiveness Survey ­ corresponds to the year of the data on firms selling online. educational and training systems need to provide fluency in languages that are widely used on the the skills needed for the new economy, such as internet would increase the benefits of more socioemotional or high cognitive skills. If not, firms widespread internet access. interested in technological upgrades may have dif- The shortage of advanced computer skills in ficulty recruiting workers with the skills that comple- ECA, reflected in the high level of vacancies in com- ment the new technologies. puter jobs, also constrains internet adoption. These A World Bank assessment of skills shortages shortages could intensify in the older, rapidly aging in the region finds that the quality of upper-­ ECA economies. In many countries, the lower level secondary and tertiary education in many coun- of skills in older workers appears to be driven by a tries is not keeping up with the changing demand mixture of cohort effects and a deterioration of skills for skills (Sondergaard et al. 2012). For instance, with age. While the share of ICT specialists in total the highly specialized and compartmentalized employment has increased in most countries with nature of tertiary education in the Russian adequate data, most studies find that a shortage of Federation, often closely affiliated with specific ICT specialists creates bottlenecks even in the sectors of the economy, fails to deliver the flexi- developed European economies (Falk 2002). bility needed by workers in the internet economy A study by the European Commission (Attström (OECD 2013). In many ECA countries, improving et al. 2014) finds that ICT specialists are one of the ­nvestments Reaping Digital Dividends through Complementary I ●  157 top three occupations with the largest skills bottle- Competitive, well-regulated markets are necks in Europe. required to encourage internet adoption and to The more developed countries also can satisfy exploit the resulting gains. Poor competition poli- their demand for ICT specialists through immigra- cies can mean that firms fail to have the incentives tion and encouraging greater participation by to incorporate the internet into their production women in science, technology, engineering, and processes. Countries with poor competitive frame- mathematics specialties. For instance, immigrant works in ECA often have lower-than-average shares workers in Canada are overrepresented among of firms that use websites (figure S3.2). For exam- information technology occupations in Canada ple, import tariffs or subsidies may reduce the com- when compared with their share of total employ- petitive pressures necessary to force firms to adopt ment (OECD 2004). And in most European coun- new technology. Reducing tariffs on ICT equipment tries, only about a third or less of ICT specialists are can improve firms’ ability to use the internet. Many women, and the gender gap has risen in almost ECA countries have committed to removing tariffs every country during the past 10 years.4 These gen- on ICT equipment (under the International der gaps start early in life, as girls perform worse Technology Agreement of the World Trade than boys in mathematics (OECD 2015). Organization), but eight ECA countries still have Encouraging parents and teachers to become more import tariffs on ICT products. The lack of consumer aware of these gender gaps could increase the par- protection legislation can make consumers particu- ticipation of women in the internet economy and larly reluctant to purchase from foreign sellers, as it reduce bottlenecks in ICT skills. may be difficult to obtain refunds or protection FIGURE S3.2  Internet use by firms is associated with the intensity of local competition 100 FIN DNK SWE AUT NLD ISL SVN NOR DEU GBR Share of firms that use websites (percent) 80 LUX SVK ARM IRL LTU EST CZE BEL MLT SRB HRV ITA CYP ESP TUR RUS POL FRA BIH MDA GRC 60 PRT HUN KGZ UKR LVA MKD ALB GEO BGR KAZ ROU 40 TJK MNE AZE 20 0 3.5 4.0 4.5 5.0 5.5 6.0 Intensity of local competition Source: Calculations based on data from the World Bank Enterprise Surveys, Eurostat, and World Economic Forum 2015. Note: Data for most countries are from 2013; exceptions are the Russian Federation, for which Internet use data are from 2012, and Tajikistan, for which competition data are from 2014. 158  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia SPOTLIGHT 3 continued against fraud. And absent regulation to ensure 52 percent relied on mixed agreements. 5 competition, the economies of scale generated by Freelancers often face barriers in participating in digital technologies can lead to market concentra- social protection systems—pensions, health insur- tion and the reemergence of monopolies, and thus ance, unemployment insurance—that are often to less future innovation. provided as part of permanent employment. This Improvements in regulation in the more advanced new mode of employment will require that social economies are necessary to adapt competition poli- protection benefits become increasingly attached cies to the new market structures that the internet is to the worker and not to the job. creating, such as multisided markets. These markets Without appropriate incentives to be in the for- or platforms are different from the usual markets, as mal economy and access to insurance against job there are often two (or more) distinct groups of cus- loss, sickness, and disability, the rise of digital tech- tomers, network effects within and across these nologies will not improve inclusion, and might groups, and an intermediary that brings these two increase inequality. The rise in the informal econ- customer groups together (such as eBay). In stan- omy facilitated by digital technologies could also dard economic models, if prices are higher than reduce tax receipts, undermine pension systems, marginal costs, firms have some form of market and increase the burden on the budget. For power. With a multisided market, prices do not nec- ­ example, if online freelancers do not make pension essarily equal marginal costs, as it may be efficient contributions, they are at risk of falling into poverty (to maximize the size of the market) for the platform in old age and becoming eligible for social assis- to charge certain customer groups higher prices and tance. Cooperation between governments and others lower prices or even to offer the product for online platforms to facilitate tax and social contribu- free. Thereby, the competition agency cannot con- tion payments would nudge workers out of the sider price without considering the effects on the shadow economy and provide them with some other parts of the market (OECD 2009). employment protection. Appropriate institutions also are necessary to Some of the complementary activities required to avoid undesirable labor market effects of the wide- capitalize on the adoption of internet technology are spread adoption of internet technologies. By allow- largely under the control of firms. Strong manage- ing traditional tasks to be broken into smaller and ment can ensure that new technologies are incorpo- more specialized tasks, the internet is driving the rated effectively into business processes. rise of the “gig” economy, in which traditional Organizational change, as well as a new strategy and employment—defined as a full-time, permanent vision for the company, is often needed. A study of salaried job—shrinks, and alternative work arrange- UK firms finds that information technology invest- ments grow. In fact, countries that were early adopt- ments have a positive impact on firm productivity, ers of telecommunications reforms aimed at but the effects are larger when this investment is improving the availability and affordability of the complemented by organizational change (Crespi, internet have experienced more dramatic increases Criscuolo, and Haskel 2007). Similarly, the combina- in the incidence of alternative work arrangements tion of skilled labor and firm reorganization explains among ICT-intensive sectors. Moreover, many the returns to ICT investment in Italian manufactur- online freelancers operate in the informal economy. ing firms (Bugamelli and Pagano 2004). The extent For example, in a sample composed mostly of and type of organizational change will depend on Russians and Ukrainians, only about 12 percent of what type of technology the firm is implementing. online freelancers had a full legal contract with their For example, introducing a website to provide cus- clients, while 34 percent relied on fully informal and tomer service might require modifying the existing ­nvestments Reaping Digital Dividends through Complementary I ●  159 customer service functions, while introducing more Bloom, N., R. Sadun, and J. Van Reenen. 2013. collaboration through cloud-based office software “Management as a Technology.” Unpublished, London School of Economics, London. requires large changes to processes, procedures, and workflows. A study of French firms shows that Bocquet, R., O. Brossard, and M. Sabatier. 2007. “Complementarities in Organizational Design and adopting an enterprise resource planning system the Diffusion of Information Technologies: An requires redesigning the organization to focus on Empirical Analysis.” Research Policy 36 (3): 367–86. core competencies, quality improvements, and a Bugamelli, M., and P. Pagano. 2004. “Barriers to decentralized decision-making structure (Bocquet, Investment in ICT.” Applied Economics 36 (20): Brossard, and Sabatier 2007). 2275–86. Governments also can help firms improve their Crespi, G., C. Criscuolo, and J. Haskel. 2007. “Information ability to adopt new technologies. Programs that Technology, Organisational Change, and Productivity provide free consulting services to improve man- Growth: Evidence from UK Firms.” Discussion Paper 783, Centre for Economic Performance, London agement practices have been shown to improve School of Economics. productivity and quality among firms. A study that Falk, M. 2002. “What Drives the Vacancy Rate for randomly assigned Indian firms to receive consult- Information Technology Workers?” Jahrbucher fur ing services shows that such programs can increase Nationalokonomie und Statistik 222 (4): 401–20. firm productivity 17 percent and that these Kelly, Tim, Aleksandra Liaplina, Shawn W. Tan, and firms grow faster than firms in the control group Hernan Winkler. 2017. Reaping Digital Dividends: (Bloom, Sadun, and Van Reenen 2013). But man- Leveraging the Internet for Development in Europe agement programs need to be tailored for each and Central Asia. Washington, DC: World Bank. firm and can be expensive to implement. OECD (Organisation for Economic Co-operation and Development). 2004. Information Technology Outlook, 2004. Paris: OECD. Notes ———. 2009. Two-Sided Markets: Policy Roundtable. Paris: OECD. 1. This spotlight is based on findings from Kelly et al. ———. 2013. “Russia: Modernising the Economy.” (2017). Better Policies Series, OECD, Paris. 2. This is based on 2014 data from the World Bank’s ———. 2015. “The ABC of Gender Equality in Education: Findex database Aptitude, Behavior, and Confidence.” OECD, Paris. 3. Based on data from PISA, latest available data points Riker, D. 2015. “The Internet and Product-Level Entry (2015, 2012, and 2009). Available at http://www. into the U.S. Market.” Research Note 2015-05B, oecd.org/pisa/data/. Office of Economics, U.S. International Trade 4. The data are from Community Statistics for Commission, Washington, DC. Information Society (CSIS) Eurostat. 5. The source is Shevchuk and Strebkov (2012), based on Shevchuk, A., and D. Strebkov. 2012. “Freelance a sample of online freelancers from Russia (70 percent), Contracting in the Digital Age: Informality, Virtuality, Ukraine (11 percent), Belarus (3 percent), and other and Social Ties.” Research Paper BRP 12, Higher countries (16 percent). School of Economics, National Research University, Moscow. Sondergaard, L., M. Murthi, D. Abu-Ghaida, References C. Bodewig, and J. Rutkowski. 2012. Skills, Not Just Diplomas: Managing Education for Results in Eastern Europe and Central Asia. Washington, DC: Attström, K., S. Niedlich, K. Sandvliet, H.-M. Kuhn, and World Bank E. Beavor. 2014. “Mapping and Analysing Bottleneck Vacancies on the EU Labour Markets.” European World Economic Forum. 2015. Global Competitiveness Commission, Brussels. Report 2015–2016. Geneva: World Economic Forum. 4 Migration and Connectivity Migration is an integral part of supporting connectivity between countries. Migration facilitates connectivity by narrowing market information gaps between countries and can lead to greater cross-border investments and trade between the migrant host and origin countries. Migration also directly increases sharing of technology and knowledge between countries through schooling and language skills attained abroad. Migrants, whether by returning home and swaying policies based on experiences gained abroad, or by influencing their friends and family at home, can also have an impact on home and host country institutions as they shape perspectives on governance and influence expectations of what type of government works best. While migration has often been thought of as simply an increase in the supply of labor, with consequent distributional wage impacts, openness to migration also helps many countries gain the skills, technology, and resources required to improve efficiency and compete in an increasingly complex globalized world. Main Messages • Migration has, on balance, benefited Europe and Central Asia (ECA). Both emi- gration and immigration rates in many ECA countries continue to be higher than the global average, mostly driven by European Union (EU) integration and flows after the opening up of Eastern bloc countries. There is considerable empirical and anecdotal evidence that diaspora investments and trade and knowledge transfer have benefited ECA economies. The region’s 161 162  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia disproportionately high flows of skilled workers have provided a conduit for the transfer of technology between countries. And the increasing share of migrants going to the United States and Northern, Western, and Southern Europe may have contributed to improving institutions in ECA transition economies. • But some ECA countries have failed to reap all of the potential gains from migra- tion. Problems affecting the institutional and policy environment have reduced the attractiveness of some ECA countries for immigration. Countries in Central Europe, Eastern Europe, and Central Asia attract relatively few immigrants from outside the region. The lack of adequate programs to facilitate the integration of immigrants has increased concerns about the economic benefits of migration, but policy actions to improve economic participation may alleviate potential problems. High-income ECA countries tend to have relatively liberal visa regimes compared with the United States, but the policy environment to facilitate access to labor markets is more burdensome. • Work and migration patterns are changing, and ECA countries need to evolve in response. To fully reap the benefits from cross-border mobility of people, policy reforms should help both migrants and native-born residents cope with increased and unavoidable challenges in the new and dynamic econ- omy, in which lifetime employment is rare and the rate of technological change has increased. Successful reforms would include increased portability of benefits, greater income security for workers with flexible contracts, invest- ments in education to ensure that workers can compete in this globalized environment, and better integration of migrants in host countries. Policies in origin countries could strengthen ongoing engagement with the diaspora and reduce challenges to return migration by improving the institutional and economic environment at home. Migration Patterns in Europe and Central Asia Emigration plays an important role in ECA. While emigration rates for most coun- tries in the world are low (map 4.1), many countries in the ECA region have higher emigration rates than the global average. This is partly due to mobility within the EU. In addition, there is significant mobility between former Soviet Republics, par- ticularly from Central Asia to the Russian Federation. However, some of these migrants are people who moved within the former Soviet Union and technically became international migrants after the breakup of the country following 1989. Overall, the level of emigration from ECA countries (excluding the EU15+)1 slightly increased from 2000 to 2010.2 This is partly due to the financial crisis, which reduced demand in labor markets across Europe. The largest growth in emigration between 2000 and 2010 is observed for Romania (129 percent) and Bulgaria (66 percent), countries that entered the EU in 2007. By contrast, some of the 2004 EU accession countries experienced substantial declines in emigration from 2000 to 2010 (the Czech Republic by 52 percent and Poland by 30 percent), and others experienced small declines (except Slovenia’s increase of 32 percent). Immigration rates are also high in ECA. Immigration rates in Central and Eastern European countries tend to be above the global average, and are on par with some Migration and Connectivity ●  163 MAP 4.1  Emigration shares have seen the highest increase in ECA a. Global emigration shares, 2000 IBRD 43817 | JULY 2018 Emigrants as a percentage of the population, 2000 20–100 10–20 5–10 1–5 0–1 No data b. Global emigration shares, 2010 IBRD 43819 | JULY 2018 Emigrants as a percentage of the population, 2010 20–100 10–20 5–10 1–5 0–1 No data Source: World Bank 2018. of the EU15+ countries (map 4.2). The number of emigrants from Central and Eastern Europe exceeded that of immigrants by about 14 million people in both 2000 and 2010 (including estimates for countries with missing data—see annex 4A for methodology).3 Thus, a significant share of these countries’ emigrants go either to the EU15+ countries or outside the region. (Their lack of attractiveness for 164  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia MAP 4.2  Immigration shares are significant in many ECA countries a. Immigration shares in 2000 IBRD 43818 | JULY 2018 Immigrants as a percentage of the population, 2000 20–100 10–20 5–10 1–5 0–1 No data b. Immigration shares in 2010 IBRD 43820 | JULY 2018 Immigrants as a percentage of the population, 2010 20–100 10–20 5–10 1–5 0–1 No data Source: World Bank 2018. immigration from outside the region is discussed in Artuc et al. 2015, and the OECD DIOC-E data set provides available data.) The only major exception is Russia, which hosts large flows of migrants from Central Asian countries. Per capita income, proximity, and regional integration are major determinants of emigrant patterns from ECA. Emigrants from the EU15+ countries go largely to other high-income countries (figure 4.1, panel a). Since income differences are the Migration and Connectivity ●  165 main determinants of migration flows, migrants from high-income countries are unlikely to migrate to lower-income countries. Seven of the top ten destinations are in Europe, indicating the important role of physical proximity as well as the removal of mobility barriers within the European Union. Six of the top ten destina- tions for emigrants from Eastern and Central Europe are also high-income coun- tries—Germany, Italy, Spain, the United Kingdom, the United States, and Israel (figure 4.1, panel b). The United States is the largest destination country in the world, and Israel has special status because of Jewish migration, especially after FIGURE 4.1  Top a. From EU15+ countries destinations of emigrants United States 3,213,734 and share of total who have completed tertiary France 1,766,284 education, 2010 Australia 1,619,747 Canada 1,457,805 Germany 1,392,754 United Kingdom 1,194,836 Switzerland 969,956 Spain 940,960 Italy 626,335 Belgium 526,987 0 20 40 60 80 100 Share of emigrants (percent) b. From Central and Eastern European countries Germany 5,273,544 Russian 3,112,693 Federation Ukraine 1,882,058 United States 1,740,797 Italy 1,530,950 United Kingdom 987,012 Kazakhstan 846,317 Spain 775,010 Israel 768,711 Belarus 695,878 0 20 40 60 80 100 Share of emigrants (percent) Tertiary education not completed Tertiary education completed continued 166  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 4.1  continued c. From Central Asia and the South Caucasus Russian 5,750,890 Federation Germany 750,685 Ukraine 659,008 Kazakhstan 416,099 Uzbekistan 157,925 United States 149,068 Israel 146,238 Armenia 137,757 Greece 119,490 Belarus 113,190 0 20 40 60 80 100 Share of emigrants (percent) Tertiary education not completed Tertiary education completed Note: The numbers on the bars are the total number of migrants in each destination country. the collapse of the Soviet Union where emigration was restricted. Finally, four countries—Belarus, Kazakhstan, Russia, and Ukraine—are on the list as leading examples of intraregional migration flows, again, mainly after the breakup of the Soviet Union. The role of regional ties and distance are more prominent in shaping emigration patterns from Central Asian countries (figure 4.1, panel b). There are two Western and Southern European destinations among the top ten—Germany and Greece—and only two non-European destinations—the United States and Israel. All the other destinations—Armenia, Belarus, Kazakhstan, Russia, Ukraine, and Uzbekistan—reflect regional mobility flows, distance, and cultural and political ties. The destination of highly educated emigrants reflects income levels, p ­ olicies, 4 distance, and language. The share of the highly educated among emigrants from the EU15+ countries reaches more than 40 percent in several high-income destinations, such as the United States, the United Kingdom, and Canada. This is due to immigration policies in these destination coun- The destination tries that favor the high-skilled, the skill premium in these labor mar- of highly educated kets, the ability of the tertiary educated to migrate further distances emigrants reflects than the less educated, and perhaps the use of English, which many income levels, ­policies, educated people learn as a second language (language fluency is distance, and language. an important determinant of the returns to migration for highly skilled workers). Similarly, more than 50 percent of emigrants from Eastern and Central Europe to the United States, the United Kingdom, and Israel are highly educated, while emigrants to other regional destina- educated ratios of less than 25 percent (figure 4.1, panel b). tions have tertiary-­ And ­ emigrants from Central Asia to the United States and Israel tend to be highly Migration and Connectivity ●  167 educated, with emigrants to regional destinations significantly less educated figure 4.1, panel c). (­ The share of tertiary-educated individuals among emigrants varies consider- ably within subregions. For Northern and Western European countries, the share of the high skilled among emigrants exceeds 40 percent in Sweden, Norway, Iceland, Denmark, France, Belgium, and the United Kingdom, but is only slightly more than 20 percent among the Mediterranean and Southern European coun- tries such as Spain, Greece, Italy, and Portugal. Patterns are even more diverse among Eastern European countries. The share of the tertiary educated is near or more than 40 percent in Ukraine, Latvia, Lithuania, Hungary, Belarus, and Estonia. On the other hand, in countries with lower levels of income and education, such as Turkey, Albania, the Former Yugoslav Republic of Macedonia and Bosnia, the level is about 15 percent. Similarly, in Central Asian countries the share of the ter- tiary educated in 2010 ranges from 19 percent in Kazakhstan to more than 50 percent in Turkmenistan (figure 4B.3). While the share of the tertiary educated among emigrants from the EU15+ declined from 2000 to 2010, this share rose from Central and Eastern Europe. One potential explanation is that Central and Eastern Europe’s rapid increase in education levels and integration into the global economy increased the pool of younger and better-educated workers with globally marketable skills. Differences in the age distribution between sending and receiving countries play a limited role in migration in some ECA subregions. Globally, differences in the relative sizes of the working-age populations (25–65 age group) are important determinants of migration flows (see World Bank 2018). Working-age individuals may move from developing countries with young and rapidly growing popula- tions, often suffering from youth unemployment and underemployment and related social problems, to high-income countries that have completed their demographic transitions and have aging populations. Thus, the 25–64 age group comprises slightly more than 60 percent of the native-born population of the EU15+ but 75 percent of the foreign-born population, while immigrants’ share among the elderly (older than age 65) is only half that of the native born (­figure 4.2, panel a). However, many Central and Eastern European countries have relatively old and rapidly aging populations. Thus, differences in the age composition of populations are less important in shaping immigration to Central and Eastern Europe, where the share of working-age individuals is slightly less among immi- grants than among natives. The role of demography in regional migration flows can be seen more clearly when comparing the age composition of native-born and emigrant populations from sending countries (figure 4.2, panel b). Emigrants from each region are concentrated in the 25–64 age group, especially in Central Asia, where their share in this group is 11 percentage points higher than that of those who did not emigrate. These data confirm that labor market concerns are important motivations for migration, and are consistent with the view that most migrants leave their countries after they complete their education (and enter the 25–64 age group). Women make up the majority of emigrants from ECA. Female migrants made up about 48 percent of the global migrant stock in 2015 (World Bank 2018), about the same level in 2000 and slightly lower than the share of 50 percent in 1980. 168  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 4.2  Age a. In EU15+ and Central and Eastern European countries composition of native-born 80 and immigrant Share of population (%) populations, 2010 60 40 20 0 15–24 25–64 65+ 15–24 25–64 65+ EU15+ Central and Eastern Europe Native Foreign born b. In ECA 80 Share of population (%) 60 40 20 0 15–24 25–64 65+ 15–24 25–64 65+ 15–24 25–64 65+ EU15+ Central and Eastern Europe Central Asia and South Caucasus Origin population Emigrants However, women accounted for more than 50 percent of emigrants from almost every country in Northern, Western, and Southern Europe, except for Greece, Italy, the Netherlands, and Portugal. In Sweden and Finland women make up percent about 60 percent of emigration (figure 4.3, panel a). Similarly, more than 50 ­ of emigrants from Central and Eastern Europe (figure 4.3, panel b) and Central Asia and the South Caucasus (figure 4.3, panel c) are women. Despite the common perception that most migrants from Central Asian countries, especially to Russia, are men, the majority of emigrants in most of the countries are women, reaching 60 percent among many of the largest sending countries, such as Russia, Ukraine and Moldova. The main exceptions are the lower-income countries such as Albania and Turkey, but there is significant increase over time in these cases as well. The reasons for the overrepresentation of women in emigration are unclear. One potential explanation is that women in many low- or middle-income countries make up at least half of the tertiary educated, but face various forms of discrimina- tion and restrictions in the labor market. Thus, they prefer to migrate to higher- income countries where potential career opportunities tend to be superior and discrimination lower. Many of these empirical observations on migration in ECA are confirmed by an analysis based on global migration patterns. A gravity model is used to estimate the global relationship between several variables and migration for 2000 and 2010 (annex 4B and table 4B.1 provide a description of the model and estimation results), which is then applied to the ECA context (see Anderson 2011 and Beine, Bertoli, and Moraga 2014 for a review of gravity models and estimation). Migration Migration and Connectivity ●  169 FIGURE 4.3  Percentage a. In EU15+ countries 65 of women among emigrants, 2010 60 55 50 45 40 35 AUT BEL CHE DEU DNK ESP FIN FRA GBR GRC IRL ITA NLD NOR PRT SWE b. In Central and Eastern European countries 65 60 55 50 45 40 35 ALB BGR BIH BLR CZE EST HRV HUN LTU LVA MDA MKD POL ROU RUS SVK SVN TUR UKR YUG c. In Central Asia and the South Caucasus 65 60 55 50 45 40 35 ARM AZE GEO KAZ KGZ TJK TKM UZB 2000 2010 Note: The figure presents the number of female emigrants as a percentage of the total number of emigrants. YUG = Serbia and Montenegro. tends to be higher the smaller the distance between countries. This effect is mar- ginally stronger for the unskilled (relative to the skilled) and men (relative to women). At the extreme, sharing a border has a strong positive effect on migra- tion. Almost half of all unskilled migrants in the world and 20 percent of skilled migrants move to a neighboring country (World Bank 2018). The stock of unskilled 170  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia migrants almost doubles for both men and women in both years if two countries are neighbors (assuming all other bilateral relationships are kept identical). The effect on high-skilled migrants, on the other hand, is negative, especially in 2010. In other words, high-skilled migrants prefer to move to nonneighboring countries. This is consistent with the fact that most high-skilled migrants move to high-income Organisation for Economic Co-operation and Development (OECD) member countries, especially the English-speaking countries. Migrants from Central Europe, Eastern Europe, and Central Asia (that is, all ECA countries except the EU15+) are likely to go to other countries within this region.5 The effect is very similar regardless of the education level or the gender of the migrants. This is in stark contrast to other pairs of countries in other regions where the effect is almost zero (since this analysis controls for distance and contiguity). In 2010, any pair of non-EU15+ ECA countries (except the cases where both are former Soviet Union countries) have between 450 and 800 ­percent more migration between them compared with other country pairs. Furthermore, the effect has increased since 2000 when it was between 300 and 400 percent. The preference for moving within this group of countries likely reflects remaining ties (business, transport, language, and personal) from when most were members of the Communist bloc. The regional effect for EU15+ countries is positive for the high skilled but negative for the low skilled. This ­ implies that low skilled EU15+ migrants are more likely to leave the region, while the high skilled tend to stay. However, the likelihood of Central and Eastern Europe and Central Asia and the South Caucasus migrants moving to EU15+ is also very strong, especially in 2010. The migration from Central and Eastern Europe and Central Asia and the South Caucasus to EU15+ is about 200 percent (for skilled females) to 275 percent (for unskilled men and women) higher than that of between other country pairs. The lowest is for skilled men, at about 140 percent. Mobility between former Soviet Union countries is also high for skilled migrants. In addition to many similar economic and academic institutions that are the legacy of the Soviet Union, the economies of these countries are closely integrated. These effects are very high and positive for the high skilled but almost zero for the low skilled. In 2010, high-skilled migrants from Mobility between former former Soviet Union countries are almost 400 percent more likely to Soviet Union countries is move to other former Soviet Union countries, compared with a ran- high for skilled migrants... dom pair of countries. This figure is less than those for the Central and Eastern Europe and Central Asia and the South Caucasus countries, but much higher than that estimated for migration to Northern, Western, and Southern Europe. The existence of a diaspora tends to increase the stock of migrants, as it reduces a number of costs associated with migration. Diasporas provide valu- able information to migrants regarding labor markets, housing, education, and various other social norms. They can help with the financing of migration costs and may be a source of insurance in case of negative shocks. In an extensive economet- ric analysis of the role of diasporas in shaping migration patterns, Beine, Docquier, and Özden (2011) find that the elasticity of a diaspora with respect to migrant stocks is slightly less than 0.5 for the unskilled in 2010 and 0.2 for the skilled for both Migration and Connectivity ●  171 gender groups. These levels are lower than those in 2000, especially for skilled migrants. Sharing a similar language also has a positive effect on migration patterns, particularly for skilled migrants. The effect is large and positive for the high skilled, slightly positive for unskilled women, and insignificant for unskilled men. More specifically, if two countries share a similar language, skilled migration for both men and women increases by almost 175 percent in 2010. The almost-zero effect for the unskilled reflects the lesser importance of strong language skills in their work compared with that of skilled migrants, where facility with the local language interacts with skills to boost returns to human capital (see Borjas 1995 for overall impacts of migration). Migration Patterns in ECA Are Likely to Change Migration will continue to play an important, but changing, role in the economic and social development of the region. Differences in income and unemployment rates, as well as demand for skilled labor from the region’s economic power- houses, will remain key drivers of voluntary migration. While near-term political debate about benefits of migration may slow more open polices toward migra- tion, the trend toward regional economic integration is expected to continue. Improvements in transport and communications have greatly increased the inte- gration of labor markets, in part through the rise of global value chains, and general technological improvements have intensified global competition for high-skilled workers. These developments will boost cross-border connectivity in many dimensions, which will facilitate migration. Some aspects of technological progress, for example, ECA’s significant participation in internet-based platforms that connect workers and employers across the world,6 increase services trade rather than migration. However, in general, technological advancement is com- plementary to global movements of skilled workers (Kerr et al. 2016). Indeed, the level of high-skilled migration to ECA countries that are OECD members increased more than that of less-skilled migration during 2000–10 (figure 4.4), and ECA countries are increasingly trying to raise the number of high-skilled immigrants (see Castles, de Haas, and Miller 2013 for global migration patterns and their future). Technological progress and global integration are also reducing the duration of migration. More than two-thirds of OECD host countries for which migration data are available witnessed a rise in the share of temporary migration between 2000 and 2010. In part this reflects the weakening of skilled workers’ ties to a location or national identity and increased global perspectives and connections, which also promotes more circular migration. It also reflects the rise in the share of temporary employment in most ECA subregions from 2002 to 2016 (figure 4.5), in part driven by technological progress, as countries with larger shares of temporary employ- ment tend to have larger shares of temporary immigration (figure 4.6).7 Higher education is increasingly globalized (box 4.1), which generates signifi- cant benefits. International students gain wider access to education and employ- ment opportunities abroad, while the receiving countries capture a broader range 172  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 4.4  The share of 100 3 3 2 2 high-skilled immigrants to 90 high-income ECA OECD 29 Percent of total immigrants 80 39 37 member countries increased 46 70 between 2000 and 2010 60 50 36 33 34 40 32 30 20 27 33 10 19 25 0 2000 2010 2000 2010 ECA OECD Non–ECA OECD High skilled Skilled Low skilled Unknown Source: Organisation for Economic Co-operation and Development (OECD), Database on Immigrants in OECD Countries. Note: “Low skilled” refers to people with no more than lower-secondary education; “Skilled” refers to people with upper-secondary to postsecondary nontertiary education; “High skilled” refers to people with tertiary education. “ECA OECD” includes Austria, Belgium, the Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, the Netherlands, Poland, Portugal, the Slovak Republic, Spain, Sweden, Turkey, and the United Kingdom. “Non-ECA OECD” includes Australia, Canada, Japan, Mexico, New Zealand, Norway, Switzerland, and the United States. ECA = Europe and Central Asia. FIGURE 4.5  The share of 25 temporary employment increased in Europe and Central Asia between 2002 20 Percent of employees ages 15–65 and 2016 15 10 5 0 Southern Europe Northern Europe Turkey Western Europe Central Europe 2002 2016 Source: World Bank 2016b. Note: Shares for Turkey are based on 2006 and 2016 data. of skills (Tse 2012). Student mobility can be an important source of longer-term immigration. Globally, between 1970 and 2000, a 10 percent increase in interna- tional students (including ECA students) increased the stock of tertiary-educated workers in host countries by 0.9 percent (Felbermayr and Reczkowski 2012). Foreign students who join the host country workforce have a positive impact on host countries. An internationally diverse workforce tends to improve the perfor- mance of research and development (see Chellaraj, Maskus, and Mattoo 2008 and Kerr and Lincoln 2010 for the United States and Niebuhr 2010 for Germany). Migration and Connectivity ●  173 FIGURE 4.6  The share of temporary migration is positively related to the share of temporary employment 60 50 Cyprus Share of temporary migration (percent) 40 Norway Finland United Kingdom Chile 30 Spain Belgium Sweden Denmark Italy Slovenia Bulgaria Ireland Switzerland 20 Greece Luxembourg Australia Russian Federation Austria Canada Portugal Czech Republic Hungary 10 Malta France Netherlands Romania Estonia Iceland Germany Slovak Republic Poland Lithuania Latvia 0 0 5 10 15 20 25 30 35 Share of temporary employment (percent) Source: Organisation for Economic Co-operation and Development (OECD) data for 2010. BOX 4.1 The Globalization of Education While students coming to study in foreign coun- Federation witnessed remarkable increases in the tries have been an important channel for knowl- number of international undergraduate and gradu- edge transfer within Europe and Central Asia ate students, of 80 percent and 145 percent, (ECA) since the ancient Greeks, the movement of respectively (figure B4.1.4). students within ECA and between ECA and other During a meeting in Bologna in 1999, European regions increased following the collapse of the officials proposed harmonizing their postsecondary Soviet Union (Ackers and Gill 2008). ECA has educational systems and offering programs in experienced a further rise in the number of inter- English, with the aim of increasing interest in and national students over the past decade, facilitated recognition of their degrees globally. This harmoni- by technological progress and rising incomes in zation has resulted in a common three-cycle system source countries. The number of international stu- for tertiary education (the bachelor’s, master’s, and dents hosted by the top 10 ECA destination coun- doctorate degrees). The European Higher tries increased significantly between 2004 and Education Area and Bologna Process has 48 full 2014, except in Germany (­ figure B4.1.1). In 2014, members. The Commonwealth of Independent apart from China and India, most of the top 10 States (CIS) signed an agreement on the mutual sources of foreign students in ECA were other recognition of education credentials for secondary ECA countries (figure B4.1.2). and vocational education in 2004, effective begin- ECA countries hosted about half of all foreign ning in September 2005. Russia also has bilateral students globally (figure B4.1.3). Among the top 10 agreements on mutual recognition with other CIS corridors of international tertiary student flows, countries, including Azerbaijan, Moldova, China–United Kingdom and Kazakhstan–Russian Turkmenistan, and Ukraine. continued 174  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia BOX 4.1 The Globalization of Education continued FIGURE B4.1.1  Most top ECA destinations attracted more international tertiary students in 2014 than in 2004 450 400 350 300 Thousands 250 200 150 100 50 0 m e y n ly ia ine ds y lic an rke nc tio Ita do str lan ub ra Fra rm ra Tu ing Au ep Uk er de Ge hR dK th Fe Ne ec ite ian Cz Un ss Ru 2004 2014 Source: United Nations Educational, Scientific, and Cultural Organization (UNESCO) Institute for Statistics data set on international student mobility in tertiary education, available at http://data.uis​ .­unesco.org/. FIGURE B4.1.2  Most source countries of international tertiary students are in ECA 200 180 160 140 120 Thousands 100 80 60 40 20 0 a y n an ia ine e ion ly co an in ta nc Ita Ind oc ist Ch ra at hs rm Fra or en Uk er ak Ge M ed rkm z Ka nF Tu s sia Ru 2004 2014 Source: United Nations Educational, Scientific, and Cultural Organization (UNESCO) Institute for Statistics data set on international student mobility in tertiary education, available at http://data.uis​ .unesco.org/. Note: Data for 2014 are not available for Georgia, Greece, and Spain. Data for 2004 are not available for Azerbaijan, Bosnia and Herzegovina, Croatia, Luxembourg, Serbia, Turkmenistan, Ukraine, and Uzbekistan. continued Migration and Connectivity ●  175 BOX 4.1 The Globalization of Education continued FIGURE B4.1.3  ECA hosted half of the world’s tertiary students in 2014 1.2% 2.0% 1.1% 5.8% 16.5% 49.8% 23.7% Europe and Central Asia East Asia and Pacific Sub-Saharan Africa South Asia North America Middle East and North Africa Latin America and the Caribbean Source: United Nations Educational, Scientific, and Cultural Organization (UNESCO) Institute for Statistics data set on .­ international student mobility in tertiary education, available at http://data.uis​unesco.org/. Note: Regional groupings follow the World Bank classification, with the following exceptions: (a) Europe and Central Asia (ECA) is adjusted to include the countries monitored by the World Bank’s ECA regional office; (b) countries not included in the World Bank list are classified using the UN Population Division’s regional grouping; and (c) countries not included in either list are assigned to a region on the basis of geographic location, using World Bank region names. Table 4B.6 lists the countries for which data are available. FIGURE B4.1.4  Top 10 corridors of international tertiary students with ECA hosts Number of tertiary students (thousands) a. 2004 b. 2014 China–United Kingdom China–United Kingdom Morocco–France Kazakhstan–Russian Federation Turkey–Germany Germany–Austria China–Germany China–France Greece–United Kingdom Morocco–France Algeria–France Belarus–Russian Federation Kazakhstan–Russian Federation Germany–Netherlands Poland–Germany Slovak Republic–Czech Republic Ireland–United Kingdom China–Germany India–United Kingdom India–United Kingdom 0 10 20 30 40 50 60 0 20 40 60 80 100 Source: United Nations Educational, Scientific, and Cultural Organization (UNESCO) Institute for Statistics data set on international student mobility in tertiary education, available at http://data.uis.unesco.org/. 176  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia In general, students who stay contribute to the cultural enrichment of host coun- tries (Van Mol 2014). Collaboration between academics also is an important exam- ple of the benefits of connectivity. For example, scientists from Western European countries have played a key role in stimulating international academic collabora- tion by engaging in research projects with Eastern European scholars (Teodorescu and Andrei 2011). Support for migration should be an integral part of the growth agenda for coun- tries in ECA. Individuals, employers, and countries may be more successful if they can find out how best to navigate these new, more integrated global labor markets, considering their own regulatory constraints. Seizing the opportunities of new tech- nologies will require an institutional and policy framework that welcomes workers, encourages circular migration, and reaps the benefits of the diaspora community. As markets become increasingly integrated, countries should help people, migrants and native born alike, navigate new competitive forces and try to prevent growing inequalities. One aspect of these efforts, which could benefit both individual work- ers and countries, should involve increased investment in education, so that a coun- try’s citizens can better participate in the global labor market. Increased competition from immigrants is not a major threat to natives’ employ- ment prospects. Competition for high-quality jobs is largely driven by Both the positive the broader rise in connectivity, rather than direct migration. Competition and negative effects of occurs almost irrespective of where competing workers are located. increased connectivity Indeed, empirical evidence suggests that migration has only a small depend on the flexibil- and temporary impact on average domestic workers’ wages and ity of labor markets and employment (see, for instance, Longhi, Nijkamp, and Poot 2005; the complementarity National Academies of Sciences, Engineering, and Medicine 2017), between the skills of although close substitutes may lose, and complements win, espe- native-born workers cially in the short run. Both the positive and negative effects of and migrants. increased connectivity depend on the flexibility of labor markets and the complementarity between the skills of native-born workers and migrants. The workers who lose out are often migrants who arrived previously. Policies Should Aim to Improve the Integration of Migrants Improving the integration of migrants in ECA host countries would help maximize the gains from international migration for both origin and host countries (see Eurostat 2017). In most countries in ECA, unemployment rates are higher among the foreign-born than the native-born population (figure 4.7). In part, this is because when jobs are scarce, many native-born workers may leave the labor force and rely on their assets, while foreign-born workers are less likely to have such resources to fall back on and thus continue to search for work. While natives tend to have higher secondary education rates than immigrants, tertiary education rates are nearly the same between natives and nonnatives (figure 4.8). A lack of integra- tion of migrants into the employed workforce weakens the economic benefits of migration and its contribution to host countries’ economies. Migration and Connectivity ●  177 FIGURE 4.7  Unemployment Sweden 11.0 Belgium 9.5 rates are higher for foreign- Finland 8.8 born than for native-born Greece 7.7 workers in most countries in Spain 7.7 Europe and Central Asia France 7.5 Austria 6.7 Denmark 5.9 Norway 5.8 Netherlands 5.1 Switzerland 5.0 Luxembourg 4.1 Poland 4.0 Italy 3.5 Slovenia 3.4 Germany 3.2 Estonia 2.3 Portugal 2.0 Czech Republic 2.0 Ireland 1.6 Turkey 1.6 Iceland 1.1 United Kingdom 0.7 Canada 0.7 Hungary 0.6 Mexico 0.5 Australia 0.1 New Zealand –0.5 United States –0.8 Israel –1.2 Slovak Republic –3.5 –4 –2 0 2 4 6 8 10 12 Percentage point difference between unemployment rates for foreign born and native born .­ Source: Organisation for Economic Co-operation and Development, available at http://www​oecd.org​ /els/mig/keystat.htm. Note: Data are for 2016. FIGURE 4.8  Tertiary 60 education rates in the 50 European Union are about 40 the same among native- and foreign-born working-age 30 populations (ages 25–54) 20 10 0 Preprimary, primary, and Upper-secondary and First and second stage of lower-secondary education postsecondary nontertiary tertiary education (levels 0–2) education (levels 3–4) (levels 5–8) Native born Foreign born EU born Non–EU born Source: Eurostat. Note: Data are for 2017. “EU born” does not include the reporting country. EU = European Union. 178  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Immigration is most likely to be complementary to native workers in host coun- tries, but this depends on the labor market response. Bussolo, Koettl, and Sinnott (2015) show that people who immigrated to Northern, Western, and Southern Europe between 1990 and 2000 had complementary skills to natives and contrib- uted to increasing wages and reducing inequality among natives. The effect of immi- gration also depends on the work responses of natives. Cattaneo, Fiorio, and Peri (2015) find that native workers in Europe are more likely to move to occupations associated with higher skills and status when they are faced with a large inflow of migrants into the labor market. Foged and Peri (2015) find that in Denmark the pres- ence of low-skilled migrants was associated with upward wage and skill mobility of low-skilled native workers. ECA policies could be more supportive of efforts to integrate migrants. The Migrant Integration Policy Index (MIPEX), which measures policies that affect inte- gration of immigrants into host economies (the scale is 0–100, and higher values denote more favorable integration policies), indicates that the average ECA country has a lower degree of migrant integration for labor markets than other ­ regions (Huddleston et al. 2015).8 Comparing ECA to benchmark countries (Canada, Japan, New Zealand, the Republic of Korea, and the United States), ECA has the lowest degree of overall integration except for Japan and the lowest degree of integration for labor market mobility. Moreover, ECA has made little or no policy progress in either labor integration or political participation over time (both are components of the overall MIPEX and have a range of 0–100; figure 4.9). The labor market dimension captures the ease of access to public or private sector as well as self-employment. It also accounts for access to general and targeted support for the worker (state facilitation of recognition of workers’ qualifications) and for workers’ rights. The political participation dimension instead captures elec- toral rights and political liberties such as the right to association. FIGURE 4.9  Migrant 90 Integration Policy Index overall, labor market 80 integration, and political participation scores in ECA 70 and selected countries 0–100 scale 60 50 40 30 2010 2014 ECA Overall score Labor market mobility Political participation United States Overall score Labor market mobility Political participation Canada Overall score Labor market mobility Political participation New Zealand Overall score Labor market mobility Political participation Japan Overall score Labor market mobility Political participation Korea Overall score Labor market mobility Political participation Migration and Connectivity ●  179 Policy efforts to support the integration of migrants in the labor market vary considerably across countries. More comprehensive programs are more com- mon in the EU15 countries than in the rest of the region (table 4.1). Integration policies are weak in Turkey and Central Europe. Western, Southern, and Northern Europe perform almost as well as the best performers outside ECA. Progress is apparent only in Northern and Central Europe (see also Bamieh, Fiorini, and Hoekman 2017). TABLE 4.1  The Availability of Programs Designed to Integrate Migrants Varied in ECA, 2015   Policy on migrant integration, 2015   Language skills Transfer of Protection training professional against credentials discrimination EU15 Austria Yes Yes Yes Belgium Yes Yes Yes Denmark Yes Yes No Finland Yes Yes Yes France Yes Yes Yes Germany Yes Yes Yes Greece Yes Yes Yes Ireland Yes Yes Yes Italy Yes Yes Yes Luxembourg Yes Yes Yes Netherlands Yes Yes No Portugal Yes Yes Yes Spain Yes Yes Yes Sweden Yes Yes Yes United Kingdom Yes Yes Yes EU13 Bulgaria Yes Yes Yes Croatia Yes Yes Yes Cyprus Yes Yes Yes Czech Republic Yes No Yes Estonia Yes Yes Yes Hungary Yes Yes Yes Latvia Yes No No Lithuania No Yes Yes Malta Yes Yes Yes Poland Yes No Yes Romania Yes Yes Yes Slovak Republic Yes Yes Yes Slovenia Yes Yes Yes Western Balkans Albania No No No Bosnia and Herzegovina No No No Montenegro No No No Serbia No No No South Caucasus Armenia Yes Yes Yes Azerbaijan No No Yes Georgia No Yes Yes Central Asia Kazakhstan No No Yes Tajikistan No No Yes Turkmenistan No No No Uzbekistan n.a. n.a. n.a. Russian Federation Yes No Yes Turkey No No Yes Other Eastern Belarus Yes No Yes Europe Ukraine Yes Yes Yes Source: United Nations Department of Economic and Social Affairs (UNDESA) Population Division, World Population Policies Database. Note: The table shows official responses about policies or combinations of policies aimed at integrating immigrants into the host society. ECA = Europe and Central Asia; n.a. = not available. 180  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Support for integration is even more important in the context of refugees. Evidence from the 2008 EU Labor Force Survey shows that refugees take six years to achieve the labor force participation rates of migrants who moved for family reasons and more than 15 years to catch up with migrants who came for work or education (OECD 2016). The recent influx of refugees accentuates the need for strong integration programs. Emigration Generates Net Benefits in ECA Origin Countries While emigration does have distributional impacts in origin countries, with losers and winners, emigration is often a safety valve for jobs during economic down- turns or crises, with net positive economic effects. Emigrants themselves benefit, as the choice to migrate is made based on weighing the expected benefits and costs to moving relative to those from staying. Workers in the origin countries that stay and who are close substitutes are likely to benefit, while people with comple- mentary skills may not benefit or may even lose (see Elsner 2013 for Lithuania; Bouton, Paul, and Tiongson 2011 for Moldova; and Dustmann, Frattini, and Rosso 2015 for Poland). For some ECA countries, remittances from workers living aboard are large relative to gross domestic product (GDP) and an important source of income in the region (figure 4.10). They have a mildly positive impact on long-term economic growth in emigration countries in ECA and a positive impact on poverty reduction for the poorest households (Mansoor and Quillin 2006). They can also improve access to international capital markets. There is considerable anecdotal evidence on diaspora investments and the promotion of trade and knowledge transfer. The return of migrants to their home country can support economic development, particularly when they bring capital and knowledge with them and the origin country provides the framework condi- tions to help them make use of their skills and investments. The development of the wine industry in Argentina provides a striking example of the impact of tech- nology transfer as a result of migration (box 4.2). The return of Albanian migrants FIGURE 4.10  Many 35 countries in Europe and Remittances as a share of GDP 30 Central Asia depend on 25 remittances (percent) 20 15 10 5 0 c an a nia ia nia ia ine n ria bli ov ta rg rb ist lga me ba ra kis pu old o Se jik Ge Uk Al Bu Re be Ar M Ta Uz yz rg Ky Source: World Bank, World Development Indicators database. Note: Data are for 2016. Migration and Connectivity ●  181 BOX 4.2 Nicolas Catena Zapata and the Malbec: Technology Transfer through Migration Nicolas Catena Zapata, one of the most important methodology, adapted to Argentina, to achieve innovators in the international wine industry, devel- international quality levels. oped an entire new industry of quality wine in Catena resurrected the Malbec as a varietal in Argentina based on technology learned abroad.a the 1995 harvest. The new Malbec was produced Catena’s approach to wine making illustrates what in traditional areas of Mendoza, but was produced Galenson (2007) refers to as an experimental inno- with what Catena calls his Californian-French style vation, which develops by a process of trial and or methodology. This was a significant change from error. Experimental innovators proceed tentatively, what Catena calls the “ancient Italian” style tradi- building their skills gradually, and tend to make tionally used in Argentina. Catena’s Malbec had their greatest contributions late in their careers. By unique characteristics, and its appearance was a contrast, conceptual innovations tend to be dra- milestone in the world of wine. matic, often something completely different that Despite his success, Catena found that his breaks the conventional rules of a discipline or wines, particularly the Cabernet Sauvignon, were activity. Generally, conceptual innovators have pre- of a lower quality than European and Californian cise goals, which allows them to plan their work wines. He believed that the problem was the high and execute it decisively. Their most radical new temperature in the vineyards, and began planting ideas, and consequently their greatest innovations, in locations at higher altitudes and in the south, tend to occur early in their careers. While the with lower temperatures. This was a risky experi- breakthrough ideas of conceptual innovators are ment, because colder areas can be subject to an easy to communicate among people in the same early and a late frost, and there was some potential field, experimental innovations are hard to commu- that the grapes would not ripen. nicate and have to be experienced to understand. The results were excellent. Catena undertook Catena’s presence in Napa Valley during the trials with Chardonnay, Cabernet Sauvignon, early 1980s was critical for importing the new Malbec, and Pinot Noir. Surprisingly, the Malbec California winemaking technique to Argentina. He not only behaved better with the cold, but pro- was born into a family of Argentinean wine produc- duced something original. These experiments led ers, but became an economics professor. While a to the birth of the high-altitude Malbec, Catena’s visiting professor at UC Berkeley in the early 1980s, second experimental innovation. he discovered that the techniques used to achieve Without Catena’s incidental exposure to foreign international-level quality in California were far in wine technology through his temporary emigration advance of those used in Argentina. He subse- to California, the wine industry in Argentina likely quently began a project to use the California would have been substantially less productive. a This box was written by Julio Elias based on an interview with Nicolas Catena in 2016. The author thanks Mr. Catena for his great generosity with  his time. with the Greek crisis—which increased Albania’s labor force by 5 percent between 2011 and 2014 alone—had positive effects on the wages of low-skilled nonmi- grants and overall positive effects on employment of those who stayed (Hausmann and Nedelkoska 2017). Return migrants are also more often self-employed than workers who never left, potentially contributing to employment generation and economic growth. The majority of ECA countries have developed policies to encourage the return of their nationals. Between 2000 and 2015, the number of EU13 countries with return policies increased significantly (table 4.2). 182  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE 4.2  Nearly All EU13 Countries Have Developed Policies to Encourage the Return of Their Citizens   Policy to encourage the return of citizens?   2005 2015 EU15 Austria Yes Yes Belgium No No Denmark No No Finland No Yes France No No Germany No No Greece Yes Yes Ireland Yes Yes Italy No Yes Portugal No Yes Spain Yes Yes Sweden No No United Kingdom No No EU13 Bulgaria n.a. Yes Croatia Yes Yes Cyprus Yes n.a. Czech Republic No Yes Estonia Yes Yes Hungary No Yes Latvia Yes Yes Lithuania No Yes Malta No No Romania No Yes Slovenia Yes Yes Western Balkans Albania Yes Yes Bosnia and Herzegovina Yes Yes Montenegro n.a. Yes Serbia n.a. Yes South Caucasus Armenia Yes Yes Azerbaijan Yes Yes Georgia n.a. Yes Central Asia Kazakhstan Yes Yes Tajikistan Yes No Turkmenistan n.a. n.a. Uzbekistan n.a. n.a. Russian Federation n.a. Yes Turkey No No Other Eastern Europe Belarus Yes Yes Ukraine No Yes Source: United Nations Department of Economic and Social Affairs (UNDESA) Population Division, World Population Policies Database. Note: Table shows official responses regarding whether the government has adopted any policies or programs to encourage the return of its citizens living abroad. n.a. = not available. Emigration can help improve institutions in the origin country, although the destination matters. Emigration can improve institutions by increasing the home country population’s exposure to the values and norms of the host countries. For example, Docquier et al. (2010) present cross-country evidence that unskilled emi- gration from a large sample of developing countries to OECD countries over 1975–2000 improved institutional quality in origin countries, as measured by the value of political rights, civil liberties, and openness of political institutions.9 Beine and Sekkat (2014) find, however, that emigration to economically or politically Migration and Connectivity ●  183 powerful countries has a positive impact on the quality of home country institu- tions, but no effect is found when the destination is former colonizers. Emigration also may lead to worse institutions and values if dissatisfied people with the moti- vation to change them leave. There is evidence indicating that emigration helped relax domestic pressure to reform autocratic regimes in Mexico (Hansen 1988), Cuba (Colomer 2000; Hoffman 2005), and Haiti (Ferguson 2003). Furthermore, remittance inflows may relieve the governments of public finance accountability, similar to the effect of large natural resource flows, and have adverse effects on domestic institutional quality (Abdih et al. 2012). Econometric estimates show that emigration to more democratic countries has a positive impact on political institutions, but emigration to less democratic countries does not (box 4.3). BOX 4.3 Emigration Can Improve Political Institutions in the Home Country Econometrics can identify the effect of the dias- and its extension, which also includes non-OECD pora on the institutional quality of the sending receiving countries) are only available for most country through the following model: countries for every 10 years (from 1960 to 2010). To allow for the effect of emigration, the lag period Ii ,t = α + βEmigranti , j ,t −1 + γIi ,t −1 + ΦXi ,t + ∂ yeart + εi , j ,t chosen is 5–10 years. For example, the average institutional value of 1995–2000 is regressed (B4.3.1) against the emigrant stock of 1990, and so on. Values of institutional quality come from Polity IV, a where Iit is the institutional quality of origin country global data set that consists of multiple dimen- i in time t, Emigranti,j,t−1 is the lagged number of sions of governance and political systems. individuals born in country i and living in country j Education data are derived from the World Bank’s as a share of the origin country’s population, and World Development Indicators (WDI) database, Xi,j is a vector of control variables. The regression the Barro and Lee (2001) database, and the over the pooled cross-section also includes a year ILOStats database of the International Labour dummy yeart to control for the time period. The Organization. Population and GDP data come control variables include time-varying confounders from the WDI. Since GDP in purchasing power par- that likely have an effect on institutional quality, ity terms is only available since the 1980s, the data including GDP per capita (in purchasing power par- are limited to four rounds of cross-section: 1980, ity terms), the share of tertiary-educated popula- 1990, 2000, and 2010. tion among the population age 25 and older, and To examine the impact of the host country’s val- age composition of the population (the share of ues on the home country, the regressions distin- population age 0–14 and share of population age guish three destinations: all countries, countries 65 and older). The inclusion of the lagged value of with higher institutional quality (according to the institutional quality enables us to control for time- composite measure polity2 by the Polity IV proj- invariant characteristics of the origin country linked ect), and those with worse institutional quality. to the quality of institutions. The results are summarized in table B4.3.1. Migration stock data (from World Bank 2016b, The quality of fit is high (60–70 percent). When all which includes new bilateral data on migration destinations are considered, the size of the dias- stocks; the Global Bilateral Migration Database pora (as a share of the population of the home 2013 by the World Bank; Özden et al. 2011; and country) has no relationship with institutional qual- the 2010 OECD International Migration Database ity. However, emigration to more democratic continued 184  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia BOX 4.3 Emigration Can Improve Political Institutions in the Home Country continued TABLE B4.3.1  Impact of Emigration on Institutions of Origin Countries, by Type of Destination Institutionalized Institutionalized Executive Executive Political Polity democracy autocracy recruitment constraints competition a. All emigration Emigrants (% of 0.005 −0.003 0.005 0.004 0.000 0.007 home population) (0.009) (0.008) (0.006) (0.006) (0.008) (0.015) Lagged institution 0.736 0.684 0.731 0.654 0.712 0.726 value (0.002)** (0.002)** (0.002)** (0.002)** (0.002)** (0.002)** R2 0.77 0.71 0.72 0.70 0.74 0.76 Number of 99,189 99,189 99,189 99,189 99,189 100,522 observations b. Emigration to more democratic countries Emigrants (% of 0.056 -0.030 0.032 0.032 0.030 0.080 home population) (0.015)** (0.012)* (0.009)** (0.009)** (0.011)** (0.024)** Lagged institution 0.734 0.718 0.768 0.650 0.716 0.740 value (0.005)** (0.004)** (0.004)** (0.005)** (0.005)** (0.004)** R2 0.62 0.71 0.70 0.58 0.65 0.69 Number of 28,275 28,275 28,275 28,275 28,275 28,882 observations c. Emigration to similar or less democratic countries Emigrants (% of −0.090 0.068 −0.064 −0.044 −0.050 −0.147 home population) (0.020)** (0.017)** (0.014)** (0.012)** (0.020)* (0.032)** Lagged institution 0.556 0.425 0.499 0.470 0.528 0.510 value (0.005)** (0.005)** (0.006)** (0.005)** (0.005)** (0.005)** R2 0.69 0.50 0.54 0.60 0.64 0.65 Number of 35,598 35,598 35,598 35,598 35,598 35,931 observations  Note: Robust standard errors are in parentheses. *Significant at 5 percent level, **significant at 1 percent level. countries is significantly associated with improve- Institutionalized Autocracy is an additive ments in institutionalized democracy, as well as 11-point scale (0–10) indicator derived from coding associated aspects such as open political compe- of the competitiveness of political participation, tition, regulated executive recruitment, and con- the regulation of participation, the openness and straints on executive power. The reverse is true for competitiveness of executive recruitment, and the emigration to countries that are equally or less existence of institutionalized constraints on the ­ democratic. exercise of power by the executive. Institutionalized Democracy is an additive Executive Recruitment combines information 11-point scale (0–10) indicator derived from coding presented in three component variables: Regulation of the competitiveness of political participation, of Chief Executive Recruitment (whether there are the openness and competitiveness of executive any established modes at all by which chief execu- recruitment, and the existence of institutionalized tives are selected); Competitiveness of Executive constraints on the exercise of power by the Recruitment (the extent that prevailing modes of executive. advancement give subordinates equal opportunities continued Migration and Connectivity ●  185 BOX 4.3 Emigration Can Improve Political Institutions in the Home Country continued to rise in the firm hierarchy); and Openness of participation (the extent to which alternative prefer- Executive Recruitment (the extent to which all the ences for policy and leadership can be pursued in politically active population has an opportunity, in the political arena) and the regulation of principle, to attain the position through a regular- participation (participation is regulated to the extent ­ ized process). It uses an 8-point scale (1–8). that there are binding rules on when, whether, and Executive Constraints (using a 7-point scale) how political preferences are expressed). It uses a refers to the extent of institutionalized constraints 10-point scale (1–10). on the decision-making powers of chief executives, The polity score is computed by subtracting the whether individuals or collectivities. autocracy score from the democracy score; the result- Political Competition combines information pre- ing unified polity scale ranges from +10 (strongly sented in two components: the competitiveness of democratic) to −10 (strongly autocratic). Source: Adapted from Nguyen 2017. Former Warsaw pact countries may have benefited from increased emigra- tion to countries with more democratic institutions. The breakup of the Soviet Union led to an increase in the share of ECA emigration going to the United States and Northern, Western, and Southern Europe as opposed to Russia, resulting in a steady rise in the stock of emigrants from ECA to countries that rank higher on various indicators of political institutions by the Polity IV project (Center for Systemic Peace 2015).10 In Moldova, for example, communities with higher prevalence of emigration to Northern, Western, and Southern Europe are more likely to vote for western-style democratic parties than those with higher migration to Russia, and westward migration significantly contrib- uted to putting an end to the Communists’ rule in the 2009 election (Barsbai et al. 2017). The significant benefits from international migration in ECA are reflected in high-income countries’ relatively low restrictions on mobility. For example, the restrictiveness of visa regimes in ECA high-income countries declined moder- ately from 2006 to 2012, and remains well below that of the United States figure 4.11, panel a). These lower restrictions in part reflect intra-EU mobility, but (­ much of the difference is driven by fewer mobility barriers imposed on middle- and low-income countries. The United States is stricter than ECA high-income countries toward all countries, particularly countries in Latin America and the Caribbean (figure 4.11, panel b). Both the United States and high-income ECA apply no visa restrictions on North American nationals, which includes the United States, but the United States still applies restrictions on nationals of high-income ECA countries. 186  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 4.11  High-income ECA countries are much more permissive toward international mobility than the United States Mobility restrictions imposed by high-income ECA countries and the United States a. Restrictions imposed on nationals of other countries 2.2 Mobility Barrier Index strictness 2.0 1.8 1.6 1.4 1.2 06 07 08 09 10 11 12 20 20 20 20 20 20 20 b. Restrictions imposed on nationals of nine country groups High-income ECA Lower-middle-income ECA Upper-middle-income ECA 3 3 3 2 2 2 1 1 1 0 0 0 06 07 08 09 10 11 12 06 07 08 09 10 11 12 06 07 08 09 10 11 12 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Mobility Barrier Index strictness East Asia and Pacific Latin America and the Caribbean Middle East and North Africa 3 3 3 2 2 2 1 1 1 0 0 0 06 07 08 09 10 11 12 06 07 08 09 10 11 12 06 07 08 09 10 11 12 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 North America South Asia Sub-Saharan Africa 3 3 3 2 2 2 1 1 1 0 0 0 06 07 08 09 10 11 2 6 7 08 09 10 11 12 06 07 08 09 10 1 12 1 0 0 1 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 High-income ECA receiver US receiver Source: Bamieh, Fiorini, and Hoekman 2017. Note: The figure presents data covering 199 countries. The Mobility Barrier Index, compiled as part of The European Visa Database project, measures visa requirements, visa-issuing practices, and consular services (for more information see http://www.mogenshobolth.dk). It has an ordinal scale from 0 to 3 (0 = no barriers, 1 = low barriers, 2 = medium barriers, 3 = high barriers) and is constructed as follows: if no visa requirement is in force, a score of 0 is assigned. If a receiving state does not provide visa-related consular services in a sending state, a score of 2 is assigned. If a receiving state relies on the consular services of another for visa issuance, the two states are assumed to have similar practices. If the visa refusal rate is below 3 percent, a score of 1 is assigned; between 3 percent and 20 percent, a score of 2; and above 20 percent, a score of 3. This grouping is based on a quantitative analysis of the total data set: group 1 is approximately the first interquartile range, group 2 the second and third, and group 3 the fourth and last. If the number of visa applications is low (below 20 percent of a modeling estimate) compared with the population of the sending and receiving country—and the travel distance between them—the score is increased by 1 to take into consideration that receiving states can put into place barriers that prevent people from lodging applications (see http://www​ .lse.ac.uk/government/research​ /­resgroups/MSU/documents/workingPapers/WP_2012_03.pdf). ECA = Europe and Central Asia. Migration and Connectivity ●  187 Conclusion Migration has played a key role in enhancing connectivity and improving econo- mies and institutions in the ECA region for centuries. Maintaining supportive poli- cies toward migration would make a critical contribution to prosperity in the region. Although the recent flows of refugees have certainly dominated the news and are a concern to policy makers in the region, a broader, longer-term perspec- tive is critical to appreciating the economic gains from migration. Measures to increase the integration of migrants in destination countries and greater support for migrant education would improve productivity in the host and home countries. Increasing the flexibility of labor market institutions and improving skills would help promote employment; increasing the portability of benefits and improving income security could reduce fears over the economic impact of immigrants on native workers as well as facilitate circular migration. Reaching international agree- ment on migration in a multilateral framework could enhance the benefits of migration for both origin and destination countries. Annex 4A. Gravity Model The standard gravity equation for migration mij from origin i to destination j is k ,t expressed as = exp  c k + Bk ,t X  + et ij i j ij ij mk + dk , ,t  ,t ,t  i where c k ,t is the origin fixed effect (the push variable) for origin country i and educa- j tion group k at time t, dk ,t is the destination fixed effect (the pull variable) for destina- ij tion country j, X is the set of bilateral variables between i and j, Bk,t is the coefficient ij of the bilateral variables, and et is the regression residual. The bilateral variables include language similarity, diaspora size in 1960, colonial links, distance, and regional dummies, such as migration within ECA, migration from Eastern and Central Europe to Northern, Western, and Southern Europe, and so on. Since we do not have detailed data on skilled and unskilled migration for many destination countries, we cannot include skill-specific destination fixed effects in the gravity regressions. We assume that we can express the skill-­ specific destination fixed effect in terms of a general (not specific to skill) destination fixed effect plus some explanatory variables, such as education level, GDP, military service, and female labor force participation. Then the skill-specific destination fixed effect becomes j dk ,t = dtj + Ak ,t Yt j , where dtj is the general destination fixed effect (the pull variable) for country j, Ytij is the set of bilateral explanatory variables, Ak,t is the coefficient of the explan- atory variables. Then we can combine the two equations to get the final gravity regression equation = exp  c k + dtj + Ak ,t Yt j + Bk ,t X  + et ij i ij ij mk . ,t  ,t  The gravity model must be estimated on a global scale using a full set of origin and destination countries to identify push, pull, and gravity forces correctly. 188  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Omitting a destination or an origin country can affect the multilateral resistance terms and bias the estimates as pointed out by Anderson and van Wincoop (2003). However, unlike the bilateral international trade data, there are no comprehensive bilateral migration data sets with different skill and education levels available. In other words, we lack the full square data matrix to estimate the migration model on a global scale using a traditional gravity model. One approach to solve the missing data problem is presented in a recent paper by Artuc et al. (2015). This paper is based on this novel approach that obtains unbi- ased estimates of the gravity parameters in the case of missing corridors. The esti- mation strategy is based on two statistical algorithms: (a) Poisson pseudo–maximum likelihood to estimate gravity parameters introduced by Santos Silva and Tenreyro (2006) and (b) an Expectation Maximization algorithm, which was originally used in the genetics literature by Dempster, Laird, and Rubin (1977), followed by economics research such as Hamilton (1990) and Arcidiacono and Bailey (2003). The estimation strategy consists of two recursive steps: (a) Maximization step: This step estimates determinants of migration using a detailed gravity model that includes both skilled and unskilled migrants, with origin fixed effects, destination fixed effects, and con- nectivity parameters such as existing diaspora links, distance, and common lan- guage; (b) Expectation step: Using the estimated gravity parameters, the procedure predicts and fills in missing migration data. After the expectation step, the algorithm returns to the maximization step, and continues going back and forth between the steps recursively until it converges to a solution. The estimation procedure is based on the Expectation-Maximization algorithm and consists of two iterative steps, similar to Arcidiacono and Bailey (2003). The first, “Expectation,” step fills in the missing cells based on the theoretical gravity model. The second, “Maximization,” step updates the coefficient estimates using the actual and simulated data. Then these two steps are repeated back and forth a few hundred times until the coefficients converge. Expectation Step ij ˆ ij , For a moment, let us assume that we have estimates of mk ,t , expressed as m k ,t ij for every skill level k. Then we can use the aggregate data, mt , (which is not skill specific) to impute the number of type k migrants, because the aggerate data should be equal to the sum of different skill groups. For example, we observe the total number of Polish immigrants in Russia, but we do not know the number of college graduate Polish immigrants in Russia. However, if we add college graduate Polish immigrants in Russia and non-college-graduate Polish immigrants in Russia we should get the total Polish immigrants in Russia. Thus, we can use this restric- tion and the data on aggregate migration for identification. Therefore, ∑m ij ij mt = , k ,t . k The expectation step equation is ij ij m k ,t m k ,t = ij mtij , ∑m l l ,t ij where m k ,t is the “imputed” migration data that goes into the maximization step. Migration and Connectivity ●  189 After imputing the number of skilled and unskilled immigrants for each corridor using estimates of skilled and unskilled immigrants and the data on total immi- grants, we move on to the maximization step. Maximization Step ij We estimate the gravity equation using the imputed migration data, mk ,t , when the migration corridor is missing with the following gravity regression = exp  c k + dtj + Ak ,t Yt j + Bk ,t X  + et ij i ij ij mk , ,t  ,t  ij ij where mk ,t is equal to mk ,t when the skill-specific migration data are available, ij and equal to mk ,t when skill-specific migration data are not available. After the gravity regression, we calculate the migrant estimates using the regression equation and estimates of the coefficients, c ˆi , dˆ j, A ˆ ˆ : and B k ,t t k ,t k ,t m ij ˆ = exp c  i ˆ +d ˆ +A j ˆ Y +B j ˆ X . ij  k ,t  k ,t t k ,t t k ,t  Then we go back to the expectation step and continue moving between the expectation and maximization steps recursively until the estimates converge. Annex 4B. Additional Tables and Figures TABLE 4B.1  Gravity Regression Results Male Female Unskilled Skilled Unskilled Skilled a. 2010 Diaspora 0.67 0.57 0.63 0.50 (72.86) (49.50) (71.48) (41.46) Distance −0.30 −0.40 −0.30 −0.40 (−11.18) (−10.71) (−11.41) (−10.38) Colony 0.06 0.08 0.11 0.36 (1.22) (1.20) (2.36) (5.85) Contiguity 0.40 −0.78 0.51 −0.80 (8.33) (−10.64) (10.93) (−10.61) Language similarity 0.22 0.43 0.33 0.61 (4.71) (7.19) (6.97) (9.49) Within EU15+ −0.11 0.27 −0.18 0.45 (−1.11) (2.12) (−1.92) (3.53) Within non-EU15+ 0.56 1.32 0.79 1.63 (3.09) (5.83) (4.52) (7.41) Non-EU15+ to EU15+ 1.05 0.60 1.10 0.86 (9.40) (4.51) (10.61) (7.05) EU15+ to non-EU15+ 0.16 0.79 0.11 1.28 (0.69) (2.28) (0.49) (3.74) Rest of world within regions −0.17 −0.98 0.05 −0.55 (−2.68) (−11.33) (0.74) (−6.11) Former Soviet Union −1.19 0.58 −1.09 0.63 (−7.49) (2.74) (−7.38) (3.17) b. 2000 Diaspora 0.53 0.44 0.52 0.38 (70.03) (37.06) (67.53) (30.03) Distance −0.26 −0.05 −0.22 −0.10 (−10.55) (−1.38) (−9.50) (−2.59) continued 190  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE 4B.1  continued Male Female Unskilled Skilled Unskilled Skilled Colony 0.27 −0.06 0.17 0.01 (5.45) (−0.86) (3.65) (0.15) Contiguity 0.81 0.14 0.83 −0.10 (19.28) (1.80) (20.01) (−1.29) Language Similarity 0.05 0.83 0.35 0.89 (1.12) (11.62) (8.04) (10.97) Within EU15+ 0.09 0.34 0.01 0.44 (1.09) (2.66) (0.10) (3.39) Within non-EU15+ 1.55 1.42 1.80 1.59 (9.88) (6.92) (11.61) (7.47) Non-EU15+ to EU15+ 0.72 0.15 0.60 0.22 (7.16) (1.01) (6.38) (1.57) EU15+ to non-EU15+ 0.41 1.12 0.67 1.25 (2.01) (4.08) (3.51) (4.40) Rest of world within regions 0.47 −0.14 0.51 0.06 (8.28) (−1.56) (8.99) (0.57) Former Soviet Union 0.08 1.02 −0.02 1.37 (0.55) (5.26) (−0.13) (7.25) Note: Estimation using Poisson pseudo–maximum likelihood methodology. t-statistics are in parentheses. All t-statistics greater than 2 are significant at the 5 percent level or higher. TABLE 4B.2  Emigration in ECA Countries (excluding EU15+), 2000 Male Female Total Unskilled Skilled Total Unskilled Skilled Albania 553,074 172,993 14,664 472,082 104,386 11,338 Armenia 460,824 224,521 72,749 383,759 161,846 58,998 Azerbaijan 728,953 336,907 105,207 776,845 343,218 115,961 Belarus 760,391 447,020 185,458 1,105,949 749,091 222,883 Bosnia and Herzegovina 807,098 576,014 95,336 863,830 668,324 63,213 Bulgaria 376,007 211,479 58,526 403,931 232,267 41,307 Croatia 462,978 352,168 64,958 501,449 396,553 58,627 Czech Republic 435,443 61,911 46,715 505,619 100,658 49,419 Estonia 102,553 23,109 18,837 133,550 35,144 27,780 Georgia 520,417 210,683 110,741 603,456 251,875 120,511 Hungary 220,989 85,309 74,643 251,699 108,198 72,097 Kazakhstan 1,451,504 752,767 228,507 1,911,536 1,027,953 346,063 Kyrgyz Republic 317,079 117,318 151,764 376,412 90,279 243,139 Latvia 158,502 37,222 33,372 193,339 51,805 47,821 Lithuania 256,909 69,472 40,771 317,752 120,129 50,375 Moldova 290,522 154,216 47,154 367,904 184,459 63,064 Poland 2,443,641 337,078 269,635 2,733,042 504,030 338,718 Romania 632,140 183,424 90,652 694,834 217,507 100,846 Russian Federation 4,587,448 1,717,500 1,323,930 6,223,586 2,882,230 1,488,465 Serbia and Montenegro 989,318 302,838 101,023 927,636 302,611 59,040 Slovak Republic 267,240 156,897 30,947 309,978 202,288 27,452 Slovenia 63,424 43,217 14,991 74,310 59,002 10,242 Tajikistan 276,557 148,773 87,153 277,286 80,834 152,889 Turkey 1,687,246 1,004,105 102,098 1,414,001 789,980 47,649 Turkmenistan 162,461 64,904 47,428 173,258 64,671 64,184 Ukraine 2,532,632 1,277,681 719,933 3,690,195 2,088,707 947,751 Uzbekistan 782,593 346,369 130,359 875,333 370,056 183,670 Migration and Connectivity ●  191 TABLE 4B.3  Emigration in ECA Countries (excluding EU15+), 2010 Male Female Total Unskilled Skilled Total Unskilled Skilled Albania 686,254 489,005 26,744 580,727 306,411 48,507 Armenia 433,146 59,005 358,475 359,428 69,830 249,374 Azerbaijan 697,067 109,370 313,891 601,591 215,612 49,915 Belarus 698,438 211,611 202,025 936,088 413,037 259,463 Bosnia and Herzegovina 811,913 494,037 122,814 844,606 579,000 82,800 Bulgaria 628,499 266,694 81,222 653,981 291,891 101,381 Croatia 423,617 207,428 70,405 487,444 362,669 51,400 Czech Republic 186,061 67,503 64,776 259,596 138,559 73,115 Estonia 79,545 32,633 22,808 97,178 32,117 26,069 Georgia 414,859 139,993 233,122 396,486 237,164 90,953 Hungary 244,367 101,364 89,033 265,290 144,362 91,940 Kazakhstan 1,823,123 1,019,518 776,016 2,084,622 1,764,949 271,431 Kyrgyz Republic 374,380 8,080 357,691 402,790 60,532 316,507 Latvia 145,379 65,245 56,265 165,765 61,726 60,879 Lithuania 226,364 117,139 75,759 285,839 124,438 100,884 Macedonia, FYR 262,770 156,234 22,685 263,437 176,375 20,673 Moldova 412,838 161,159 179,818 445,820 204,669 155,733 Poland 1,639,333 976,854 571,161 1,880,192 976,757 743,278 Romania 1,432,015 1,072,317 218,177 1,617,985 895,175 319,485 Russian Federation 4,833,306 1,444,919 1,280,908 6,103,269 2,334,833 1,232,999 Serbia and Montenegro 766,434 384,644 60,569 713,213 279,555 52,486 Slovak Republic 235,286 155,537 46,104 291,857 201,430 65,385 Slovenia 75,508 27,000 13,969 110,160 73,444 15,593 Tajikistan 351,935 2,749 336,162 258,065 18,623 189,075 Turkey 1,622,419 1,259,307 192,738 1,472,566 1,130,708 142,303 Turkmenistan 172,289 51,872 109,178 181,282 126,979 47,318 Ukraine 2,537,323 944,572 918,172 3,111,315 1,539,198 1,015,666 Uzbekistan 1,048,511 281,092 740,620 939,362 441,072 441,231 TABLE 4B.4  Immigration in ECA Countries (excluding EU15+), 2000 Male Female Total Unskilled Skilled Total Unskilled Skilled Albania 34,929 21,052 1,877 39,507 27,800 975 Armenia 121,835 54,361 17,395 177,492 84,668 25,787 Azerbaijan 102,159 51,247 20,371 134,417 69,522 28,435 Belarus 570,469 247,946 157,737 662,863 339,789 175,335 Bosnia and Herzegovina 21,166 6,279 2,710 22,647 9,051 2,955 Bulgaria 54,645 26,211 8,299 66,682 33,956 10,668 Croatia 292,472 227,649 33,716 329,881 280,473 20,229 Czech Republic 206,350 132,093 24,580 245,551 174,097 21,799 Estonia 101,027 48,303 23,772 148,804 73,693 35,721 Georgia 107,619 49,966 21,284 121,723 63,173 24,636 Hungary 136,400 73,328 13,150 174,012 98,106 12,705 Kazakhstan 1,313,270 513,796 496,568 1,540,344 773,235 461,300 Kyrgyz Republic 151,446 59,928 46,892 211,444 92,642 62,550 Latvia 241,596 132,802 51,697 364,690 218,375 78,790 Lithuania 92,038 50,256 20,190 120,029 69,249 27,606 Macedonia, FYR 45,414 19,622 6,823 63,563 31,332 7,811 continued 192  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE 4B.4  continued Male Female Total Unskilled Skilled Total Unskilled Skilled Moldova 208,461 92,191 47,463 265,901 137,377 61,207 Poland 336,829 156,444 38,635 485,144 291,692 38,260 Romania 93,493 48,061 11,534 76,278 47,985 6,169 Russian Federation 5,356,263 2,921,347 1,236,481 6,916,092 3,852,371 1,641,808 Serbia and Montenegro 462,001 285,747 51,065 590,930 424,932 43,262 Slovak Republic 51,947 22,836 9,976 66,007 37,422 9,787 Slovenia 91,621 65,158 12,823 80,082 59,842 8,983 Tajikistan 138,039 48,706 44,480 182,308 70,315 51,675 Turkey 600,760 289,293 82,763 648,474 300,052 39,512 Turkmenistan 96,811 31,699 39,191 127,863 41,091 51,338 Ukraine 2,362,068 1,119,696 512,934 3,151,215 1,769,083 586,758 Uzbekistan 580,500 198,380 224,781 766,716 276,013 279,472 TABLE 4B.5  Immigration in ECA Countries (excluding EU15+), 2010 Male Female Total Unskilled Skilled Total Unskilled Skilled Albania 42,170 11,620 2,218 46,939 16,702 2,229 Armenia 142,812 31,817 5,895 176,430 55,825 9,055 Azerbaijan 134,549 32,280 24,171 155,524 47,485 27,377 Belarus 519,123 133,307 114,168 606,195 208,895 160,349 Bosnia and Herzegovina 12,821 3,499 352 12,930 5,843 382 Bulgaria 33,830 13,075 3,815 42,617 21,810 7,547 Croatia 346,346 214,112 22,873 387,281 264,718 20,092 Czech Republic 340,920 192,963 67,938 309,463 208,124 55,425 Estonia 89,989 17,970 30,874 131,098 31,203 43,425 Georgia 87,595 20,385 16,871 100,369 29,558 18,427 Hungary 208,835 100,789 31,446 230,718 124,933 34,879 Kazakhstan 1,753,698 520,151 294,058 1,847,960 666,760 318,856 Kyrgyz Republic 106,038 27,052 31,556 131,612 40,487 43,316 Latvia 140,009 33,450 37,497 203,169 64,132 52,549 Lithuania 89,522 20,183 22,356 116,085 34,214 29,441 Macedonia, FYR 54,814 19,279 5,313 76,533 34,472 6,607 Moldova 184,216 61,629 36,865 212,611 89,634 41,195 Poland 278,302 150,322 30,226 399,062 255,551 38,254 Romania 80,043 37,056 3,748 83,233 51,938 3,882 Russian Federation 5,513,127 2,517,076 1,717,419 5,679,115 2,958,227 1,739,533 Serbia and Montenegro 314,766 90,028 4,814 410,972 224,430 6,871 Slovak Republic 76,309 24,263 11,218 87,577 42,186 15,143 Slovenia 131,073 73,887 31,985 97,714 59,696 21,219 Tajikistan 122,586 27,898 27,648 167,666 43,960 37,109 Turkey 879,406 278,401 115,422 869,667 269,961 120,188 Turkmenistan 99,888 19,943 55,318 118,715 23,834 68,911 Ukraine 2,123,859 721,228 303,278 2,590,436 1,180,337 336,223 Uzbekistan 532,901 155,159 202,368 638,141 207,455 245,718 Migration and Connectivity ●  193 TABLE 4B.6  Economies Included in Figure B4.1.3 Region Economies East Asia and Pacific Australia; Brunei Darussalam; China; Hong Kong SAR, China; Japan; Korea, Rep.; Lao PDR; Macao SAR, China; Malaysia; Mongolia; New Zealand; Thailand; Vietnam Europe and Central Asia Albania; Armenia; Austria; Azerbaijan; Belarus; Belgium; Bosnia and Herzegovina; Bulgaria; Croatia; Cyprus; Czech Republic; Denmark; Estonia; Finland; France; Germany; Hungary; Ireland; Italy; Kazakhstan; Kyrgyz Republic; Latvia; Liechtenstein; Lithuania; Luxembourg; former Yugoslav Republic of Macedonia; Malta; Netherlands; Norway; Poland; Portugal; Moldova; Romania; Russian Federation; Serbia; Slovak Republic; Slovenia; Sweden; Switzerland; Tajikistan Latin America and the Caribbean Aruba; Brazil; Chile; Colombia; Dominican Republic; Ecuador; El Salvador; Honduras; St. Kitts and Nevis; St. Lucia; Sint Maarten (Dutch part) Middle East and North Africa Bahrain; Egypt, Arab Rep.; Iran, Islamic Rep.; Israel; Morocco; Oman; Qatar; Saudi Arabia; United Arab Emirates North America Bermuda; United States Sub-Saharan Africa Botswana; Burundi; Cabo Verde; Côte d’Ivoire; Ghana; Lesotho; Madagascar; Mauritius; Mozambique; Namibia; Rwanda; South Africa South Asia India; Sri Lanka FIGURE 4B.1  Share of 60 tertiary educated among 50 emigrants from EU15+ 40 countries, 2000 and 2010 30 20 10 0 AUT BEL CHE DEU DNK ESP FIN FRA GBR GRC IRL ITA NLD NOR PRT SWE 2000 2010 Note: Share is calculated as the number of tertiary-educated emigrants as a percentage of the total number of emigrants. “Tertiary” is defined, for 2000, as at least some tertiary and for 2010, as completed tertiary. FIGURE 4B.2  Share of 50 tertiary educated among 40 emigrants from Central and 30 Eastern European countries, 2000 and 2010 20 10 0 ALB BGR BIH BLR CZE EST HRV HUN LTU LVA MDA MKD POL ROU RUS SVK SVN TUR UKR YUG 2000 2010 Note: Share is calculated as the number of tertiary-educated emigrants as a percentage of the total number of emigrants. “Tertiary” is defined, for 2000, as at least some tertiary and for 2010, as completed tertiary. YUG = Serbia and Montenegro. 194  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 4B.3  Share of 90 tertiary educated among emigrants in Central Asia 80 and the South Caucasus, 70 2000 and 2010 60 50 40 30 20 10 0 ARM AZE GEO KAZ KGZ TJK TKM UZB 2000 2010 Note: Share is calculated as the number of tertiary-educated emigrants as a percentage of the total number of emigrants. “Tertiary” is defined, for 2000, as at least some tertiary and for 2010, as completed tertiary. Notes 1. This group includes the 15 countries that were members of the EU before the 2004 accession countries joined, plus Norway and Switzerland. 2. These figures include predicted migration stocks for the missing corridors so that the data can be compared over time. See annex 4A for methodology and data. 3. Reported data for Central Asia for 2000 or 2010 are particularly spotty and, hence, we rely heavily on the estimation methodology. 4. In the migration context, the most widely used proxy for human capital and skills is the education level of the workers. Even though there are numerous complications, this is the most widely and consistently available metric in most countries. Thus, “high skilled” and “tertiary educated” are used interchangeably here. 5. This analysis uses dummy variables for the following intraregional migration corridors: (a) within Central and Eastern Europe and Central Asia and the South Caucasus: takes the value of 1 if both countries (origin and destination) are in the ECA region (except EU15+); (b) within EU15+: takes the value of 1 if both countries are in Western Europe; (c) Central and Eastern Europe and Central Asia and the South Caucasus to EU15+: takes the value of 1 if the origin country is in Central and Eastern Europe and Central Asia and the South Caucasus and destination in EU15+; (d) EU15+ to Central and Eastern Europe and Central Asia and the South Caucasus: takes the value of 1 if the origin country is in EU15+ and destination is in Central and Eastern Europe and Central Asia and the South Caucasus; (e) rest of world within regions: takes the value of 1 if both countries are in the same World Bank regions (except ECA); and (f) former Soviet Union: takes the value of 1 if both countries are former Soviet Union countries. These six vari- ables capture all regional relationships; the excluded relationship is if two countries are not in the same region. 6. Russia and Ukraine are the fifth- and sixth-largest suppliers, respectively, of contract labor to the US market, and Ukraine had the third-largest cumulative online worker wage bill through 2014 (Horton, Kerr, and Stanton 2017). 7. Temporary migrants are defined as foreign-born residents who have been in the host country for less than five years. They could stay longer. Migration and Connectivity ●  195 8. MIPEX is sourced from the MIPEX database of CIDOB (Barcelona) and MPG (Brussels); see http://www.mipex.eu/. For the definitions of MIPEX indicators, see http://www​ .mipex.eu/sites/default/files/downloads/Definitions_of_Who_Benefits_Outcome_and​ _Beneficiaries_Indicators.pdf. 9. Skilled emigrants, however, had an ambiguous effect. A positive relationship between international emigration and the demand for political accountability is found by Batista and Vicente (2011) for Cape Verde, and between emigration and political stability and voice and accountability by Li and McHale (2009) based on cross-country data. Spilimbergo (2009) shows that foreign education acquired in democratic countries seems to promote democracy in home countries. Li, McHale, and Zhou (2013) find that skilled migrants have positive effects on the home country’s political institutions, mea- sured by voice and accountability, political stability and absence of violence, govern- ment effectiveness, regulatory quality, rule of law, and control of corruption. There is evidence for Mexico (Pérez-Armendáriz and Crow 2010) and Mali (Chauvet and Mercier 2013) that individuals in migrant households are more likely to vote. 10. On the scale from +10 (strongly democratic) to −10 (strongly autocratic) per the Polity IV project, Russia’s rank has fluctuated between 3.5 and 4.7 over the years while the United States, Great Britain, France, Germany, Italy, and Spain have been ranked consistently at 10 since 1980. ­ References Abdih, Yasser, Ralph Chami, Jihad Daghe, and Peter Montiel. 2012. “Remittances and Institutions: Are Remittances a Curse?” World Development 40 (4): 657–66. Ackers, Louise, and Bryony Gill. 2008. Moving People and Knowledge: Scientific Mobility in an Enlarging European Union. Northampton, MA: Elgar. Anderson, James E. 2011. “The Gravity Model.” Annual Review of Economics 3 (1): 133–60. Anderson, James E., and Eric van Wincoop. 2003 “Gravity with Gravitas: A Solution to the Border Puzzle.” American Economic Review 93 (1): 170–92. Arcidiacono, Peter, and John Bailey. 2003. “Finite Mixture Distributions, Sequential Likelihood and the EM Algorithm.” Econometrica 71 (3): 933–46. Artuc, Erhan, Frederic Docquier, Çağlar Özden, and Christoper Parsons. 2015. “A Global Assessment of Human Capital Mobility: The Role of non-OECD Destinations.” World Development 65 (1): 6–26 Bamieh, Omar, Matteo Fiorini, and Bernard Hoekman. 2017. “Trends in Selected Connectivity-Related Policy Indicators for Europe and Central Asia.” Unpublished, European University Institute, Florence, Italy. Barro, R. J., and J. W. Lee. 2001. “International Data on Educational Attainment: Updates and Implications. Oxford Economic Papers 53: 541–63. Barsbai, Toman, Hillel Rapoport, Andreas Steinmayr, and Christoph Trebesch. 2017. “The Effect of Labor Migration on the Diffusion of Democracy: Evidence from a Former Soviet Republic.” American Economic Journal: Applied Economics 9 (3): 36–69. Batista, Catia, and Pedro C. Vicente. 2011. “Do Migrants Improve Governance at Home? Evidence from a Voting Experiment.” World Bank Economic Review 25 (1): 77–104. Beine, Michel, Simone Bertoli, and Jesús Fernández-Huertas Moraga. 2014. “A Practitioners’ Guide to Gravity Models of International Migration.” Working Papers 2014-03, FEDEA. Beine, Michel, Frédéric Docquier, and Çağlar Özden. 2011. “Diasporas.” Journal of Development Economics 95 (1): 30–41. Beine, Michel, and Khalid Sekkat. 2014. “Emigration and Origin Country’s Institutions: Does the Destination Country Matter?” Middle East Development Journal 6 (1): 20–44. 196  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Borjas, George J. 1995. “The Economic Benefits from Immigration.” Journal of Economic Perspectives 9 (2): 3–22. Bouton, Lawrence, Saumik Paul, and Erwin R. Tiongson. 2011. “The Impact of Emigration on Source Country Wages: Evidence from the Republic of Moldova.” Policy Research Working Paper 5764, World Bank, Washington, DC. Bussolo, Maurizio, Johannes Koettl, and Emily Sinnott. 2015. Golden Aging: Prospects for Healthy, Active, and Prosperous Aging in Europe and Central Asia. Washington, DC: World Bank. Castles, S., H. de Haas, and M. J. Miller. 2013. The Age of Migration: International Population Movements in the Modern World. London: Palgrave Macmillan. Cattaneo, Cristina, Carlo V. Fiorio, and Giovanni Peri. 2015. “What Happens to the Careers of European Workers When Immigrants ‘Take Their Jobs’?” Journal of Human Resources 50 (3): 655–93. Center for Systemic Peace. 2015. Polity IV Project, Political Regime Characteristics and Transitions, 1800–2015. Chauvet, L., and M. Mercier. 2013. “Migration and Elections in Mali. Does Migration Promote Democratization in Africa?” Unpublished, Paris School of Economics, Paris. Chellaraj, Gnanaraj, Keith E. Maskus, and Aaditya Mattoo. 2008. “The Contribution of International Graduate Students to US Innovation.” Review of International Economics 16 (3): 444–62. Colomer, J. M. 2000. “Exit, Voice, and Hostility in Cuba.” International Migration Review 34 (2): 423–42. Dempster, A. P., N. M. Laird, and D. B. Rubin. 1977. “Maximum Likelihood from Incomplete Data via the EM Algorithm.” Journal of the Royal Statistical Society, Series B 39 (1): 1–38. Docquier, Frédéric, Elisabetta Lodigiani, Hillel Rapoport, and Maurice Schiff. 2010. “Emigration and the Quality of Home Country Institutions.” Discussion Paper 2010-35, Université catholique de Louvain, Institut de Recherches Economiques et Sociales (IRES), Louvain, Belgium. Dustmann, Christian, Tommaso Frattini, and Anna Rosso. 2015. “The Effect of Emigration from Poland on Polish Wages.” Scandinavian Journal of Economics 117 (2): 522–64. Elsner, Benjamin. 2013. “Emigration and Wages: The EU Enlargement Experiment.” Journal of International Economics 91 (1): 154–63. Eurostat. 2017. Migrant Integration Statistical Book . http://ec.europa.eu/eurostat​ /­d ocuments/3217494/8081569/KS-01-17-539-EN-N.pdf/3eba7121-91fd-4512​ -aeb5-b820a55517e2. Felbermayr, Gabriel, and Isabella Reczkowski. 2012. “International Student Mobility and High-skilled Migration: The Evidence.” Working Paper 132, IFO Institute, Munich. https://www.econstor.eu/handle/10419/73806. Ferguson, J. 2003. “Migration in the Caribbean: Haiti, the Dominican Republic and Beyond.” Minority Rights Group International, London. Foged, Mette, and Giovanni Peri. 2015. “Immigrants’ Effect on Native Workers: New Analysis on Longitudinal Data.” IZA Discussion Paper 8961, Institute for the Study of Labor, Bonn. http://ftp.iza.org/dp8961.pdf. Galenson, David W. 2007. Old Masters and Young Geniuses: The Two Life Cycles of Artistic Creativity. Princeton, NJ: Princeton University Press. Hamilton, James D. 1990. “Analysis of Time Series Subject to Changes in Regime.” Journal of Econometrics 45 (1–2): 39–70. Hansen, L. O. 1988. “The Political and Socio-economic Context of Legal and Illegal Mexican Migration to the U.S. (1942-1984).” International Migration 26 (1): 95–107. Migration and Connectivity ●  197 Hausmann, R., and L. Nedelkoska. 2017. “Welcome Home in a Crisis: Effects of Return Migration on the Non-migrants’ Wages and Employment.” CID Faculty Working Paper 330, Center for International Development at Harvard University, Cambridge, MA. https://albania.growthlab.cid.harvard.edu/files/albaniagrowthlab/files/return​ _migration​_cidwp_330.pdf. Hoffman, B. 2005. “Emigration and Regime Stability: Explaining the Persistence of Cuban Socialism.” Journal of Communist Studies and Transition Politics 21 (4): 436–61. Horton, John, William R. Kerr, and Christopher Stanton. 2017. “Digital Labor Markets and Global Talent Flows.” NBER Working Paper 23398, National Bureau of Economic Research, Cambridge, MA. Huddleston, Thomas, Özge Bilgili, Anne-Linde Joki, and Zvezda Vankova. 2015. Migrant Integration Policy Index 2015. Barcelona/Brussels: CIDOB (Barcelona Center for International Affairs) and MPG (Migration Policy Group). Kerr, Sari Pekkala, William Kerr, Çağlar Özden, and Christopher Parsons. 2016. “Global Talent Flows.” Journal of Economic Perspectives 30 (4): 83–106. Kerr, William, and William Lincoln. 2010. “The Supply Side of Innovation: H-1B Visa Reforms and US Ethnic Invention.” Journal of Labor Economics 28: 473–508. Li, Xiaoyang, and John McHale. 2009. “Emigrants and Institutions.” Unpublished, University of Michigan, Ann Arbor, and National University of Ireland, Galway. Li, Xiaoyang, John McHale, and Xuan Zhou. 2013. “Does Brain Drain Lead to Institutional Gain? A Cross Country Empirical Investigation.” Unpublished, Sauder School of Business, University of British Columbia, and Queen’s School of Business, Queen’s University, Kingston, Ontario, Canada. Longhi, Simonetta, Peter Nijkamp, and Jacques Poot. 2005. “A Meta-analytic Assessment of the Effect of Immigration on Wages.” Journal of Economic Surveys 19 (3): 451–77. Mansoor, Ali, and Bryce Quillin. 2006. Migration and Remittances: Eastern Europe and the Former Soviet Union. Washington, DC: World Bank. http://documents.­ worldbank​ .org​/curated/en/183131468024337798/pdf/384260Migratio101OFFICIAL0USE0O NLY1.pdf. National Academies of Sciences, Engineering, and Medicine. 2017. The Economic and Fiscal Consequences of Immigration. Washington, DC: National Academies Press. Nguyen, Tu Chi. 2017. “Leveraging Emigration for Institutional Strengthening.” Unpublished, World Bank, Washington, DC. Niebuhr, Annekatrin. 2010. “Migration and Innovation: Does Cultural Diversity Matter for Regional R&D Activity?” Papers in Regional Science 89 (3): 563–85. https://doi​ .org/10.1111/j.1435-5957.2009.00271.x. OECD (Organisation for Economic Co-operation and Development). 2010. “Public Opinions and Immigration: Individual Attitudes, Interest Groups and the Media.” In International Migration Outlook 2010. Paris: OECD Publishing. http://dx.doi.org/10.1787/migr​ _outlook-2010-6-en. ———. 2016. Making Integration Work: Refugees and Others in Need of Protection. Paris: OECD Publishing. Özden, C., C. Parsons, M. Schiff, and T. L. Walmsley. 2011. “Where on Earth Is Everybody? The Evolution of Global Bilateral Migration, 1960-2000.” World Bank Economic Review 25 (1): 12–56. Pérez-Armendáriz, C., and D. Crow. 2010. “Do Migrants Remit Democracy? International Migration, Political Beliefs, and Behavior in Mexico.” Comparative Political Studies 43 (1): 119–48. Santos Silva, J. M. C., and Silvana Tenreyro. 2006. “The Log of Gravity.” Review of Economics and Statistics 88 (4): 641–58. 198  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Spilimbergo, Antonio. 2009. “Foreign Students and Democracy.” American Economic Review 99 (1): 528–43. Teodorescu, Daniel, and Tudorel Andrei. 2011. “The Growth of International Collaboration in East European Scholarly Communities: A Bibliometric Analysis of Journal Articles Published between 1989 and 2009.” Scientometrics 89 (2): 711–22. Tse, Emily. 2012. “Approaches to International Degree Recognition: A Comparative Study.” Working Paper, International Education Research Foundation, Culver City, CA. Van Mol, C. 2014. “The Reconstruction of a Social Network Abroad.” In Intra-European Student Mobility in International Higher Education Circuits: Europe on the Move (Palgrave Studies in Global Higher Education), 66–90. London: Palgrave Macmillan. https://doi.org/10.1057/9781137355447_4. World Bank. 2016a. Migration and Development: A Role for the World Bank Group. Washington, DC: World Bank. ———. 2016b. The Migration and Remittances Factbook. Washington, DC: World Bank. www.worldbank.org/prospects​/migrationandremittances. ———. 2018. Moving for Prosperity: Global Migration and Labor Markets. Policy Research Report. Washington, DC: World Bank. 5 Infrastructure Linkages: Cost, Time, and Networks The benefits of connectivity in transport infrastructure cannot be simply measured by kilometers of roads and rail, their condition, or density of connections. Evaluating transport connectivity in this way could lead to building too many roads or rail connections that do not pay back economic dividends in the long run. The Russian Federation might decide to build a six-lane road in Siberia, but unless it is connect- ing two important economic centers, or is providing access to natural resources that were previously unreachable, it may make little economic sense. A preferable approach is to measure the quality and importance of transport services in terms of availability, speed, cost, timeliness, and the economic value to what is being connected. Is it a road to nowhere, or a bridge between two important economic hubs? This chapter uses new transport service data and network analysis tools to measure the degree of transport service connectivity through roads, railroads, and to lesser extent, maritime transport of Europe and Central Asia (ECA) countries. A clear understanding of the benefits of within-country transport connectivity and how a country links to the broader regional transport network is critical to evaluating transport investment. Measuring the quality and value of transport service connectivity should allow for the design of strategies aimed at improving ­ the economic benefits of transport investment to countries or regions and help assess which interventions a country or region should focus on to achieve a strate- gic policy or economic objective (for example, attract traffic or increase the eco- nomic benefits to being in a network of countries). Taking such an approach can help us understand, for example, benefits from the proposed corridor develop- ment projects such as the Belt and Road Initiative (BRI)1 and the Trans-European Transport Network2 as well as sustainable financing schemes. 199 200  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Main Messages • Countries face various strategic choices in deciding on transport investments, including the target markets (e.g., domestic, neighbor countries, or strategic partners) and underlying goals (e.g., maximizing the economic activities reached by transport infrastructure, strengthening regional partnerships, or increasing resilience and improving the integration of networks). A country’s investment program may be judged a success in meeting some of these goals but not others; thus, the flexibility of the proposed methodology for evaluating transport investment is critical. • Transport investments may improve the probability that a transport flow will pass through a country or increase the importance of the country to the overall net- work. The former may generate gains from transit traffic and the opportunities from trade, foreign direct investment (FDI), knowledge sharing, and migration. By contrast, the latter implies that transport disruptions would have a more negative impact on other countries. Such network-based evaluations could facilitate regional agreements supporting mutually beneficial investment that reflects the impact of domestic transport on improving access in other countries. • With respect to reaching neighboring countries, the cost and time required for passenger travel varies greatly in ECA, while the cost and time for delivering a container varies little, except it is much higher for the Russian Federation, Turkey, and Central Asia. Countries with high costs for connecting to neighbors but low costs for connecting to ECA as a whole (e.g., Russia) could focus trans- port resources on improving connections with neighbors, while countries with low costs to connect to neighbors but high costs to connect to the ECA net- work (e.g., Central Asia) could focus on improving connections to the region as a whole, or to countries outside ECA. For freight transport, ECA is increasingly an integrated market. Thus, compared with passenger transport, the competi- tiveness of traders is more influenced by the logistics cost structure than by affordability. Connecting Cities and Neighbors: A Vicinity View of Transport Services in ECA To understand overall transport connectivity in ECA, we begin the analysis by evaluating connectivity within each country and between countries and their neighbors. ECA countries’ connectivity is based on direct measures of cost and time of travel or shipment for both passenger and freight transport for (a) domestic connectivity or connectivity among the five main cities within a country, or (b) neighbor connectivity or connectivity from the capital city of a given country to capital cities of each neighboring country.3 Connecting Cities within a Country Domestic connectivity is largely determined by geography and demographic factors. As an element of geography, the size of a country plays an important role in determining the cost and time of transport services. For example, it would take Infrastructure Linkages: Cost, Time, and Networks ●  201 an average Russian passenger 8–10 times longer to visit the country’s key cities than it would take an average passenger in the much smaller Balkan countries to visit the key cities within each Balkan country. Similarly, a Russian passenger would pay a staggering 50 euros on average to visit a main city from Moscow, while a person in the South Caucasus or in the Western Balkans would pay less than 10 euros to travel from the capital to another large city (figure 5.1, panel a). Distance also goes a long way toward explaining the long travel times in the big and sparsely populated countries of Central Asia. Natural elements, such as the location of forests, mountains, deserts, and raw materials (such as minerals), also affect the level and type of domestic connectivity. Considerations affected by indi- vidual choices, such as where people live, businesses locate, and services are pro- vided, as well as population density, also help define the type of connectivity that is optimal for a country. Normalized metrics of speed and unit costs adjusted by purchasing power provide a first proxy for the transport user experience with transport infrastructure and service quality.4 Income level is a good predictor of both the quality (speed) and affordability (unit cost adjusted for purchasing power) of domestic transport services for pas- sengers in ECA. For example, Austria, Denmark, France, Sweden, Switzerland, and Germany, all among the richest countries in ECA, fare very well in terms of service performance (speed) and also in terms how their domestic passengers can afford the services (figure 5.1, panel b, top-right quadrant). By contrast, poorer countries such as Moldova, Kosovo, Armenia, and Bosnia rank very low in both speed and affordability of their internal transport connections (figure 5.1, panel b, bottom-left quadrant). There are a few exceptions. Turkey, Ukraine, and Uzbekistan provide fast ser- vices that are not particularly affordable for the average domestic passenger. At the other extreme, passenger services in Azerbaijan, Norway, and Turkmenistan are affordable but the speed is low, likely because of their challenging geography and low population density. The market structure of the transport sector, taxes, fees, subsidies, and the level of government regulation can be sources of domestic cost differentials. Technology selection, geography, and network design (for instance, the number of transition points built into the system given by feeder road or track infrastructure and terminals) can be sources of domestic speed differentials.5 Cargo owners and freight forwarders have a different view of domestic con- nectivity. Except in Russia, where it takes close to two days to deliver a container from one main city to another, there is no significant difference in time to deliver a container within a country in all the subregions of ECA, which on average takes one day (figure 5.1, panel c). Some countries that serve as gateways to their neighbors perform poorly ­according to speed indicators. Belgium and the Netherlands have a surpris- ingly poor ranking for speed (figure 5.1, panel d) given their income levels. This is likely due to the high level of congestion in their highway systems; both countries feature as the top congestion hotspots in Europe according to the INRIX traffic scorecard.6 Another clear outlier on speed performance is Luxembourg, which suf- fers not so much from its own traffic but from transit flows through its territory. These countries face tension between the burden on internal infrastructure of increased transit traffic and the benefits of providing transportation services to the region. 202  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 5.1  Domestic connectivity Passengers a. Cost and time (average per region) b. Affordability and speed (ranking) Best NOR Low speed, high IRL KAZ South Caucasus Western Balkans affordability DNK AUT Western Balkans Advanced Europe SWE CHE TKM BEL DEU FRA Central Asia South Caucasus AZE NLD CZE Affordability ranking ITA FIN RUS Eastern Europe Central Europe GBR LVA EST LTU POL GRC ESP Central Europe Eastern Europe SVN HUN ROU PRT BGR TUR Advanced Europe Turkey MNE HRV GEO ARM BLR Turkey Central Asia MKD SRB UKR High speed, low KSV BIH affordability Russian Federation Russian Federation Worst MDA UZB 0 10 20 30 40 50 0 5 10 15 20 Worst Best Average passenger Average travel Speed ranking cost (euros) time (hours) Freight c. Cost and time (average per region) d. Unit costs and speed (ranking) Best DNK NOR Western Balkans Advanced Europe FIN SWE ISL EST NLD South Caucasus Central Asia CHE AUT DEU ESP FRA Central Europe BEL LVA IRL GRC Central Europe ITA Unit cost ranking SVN GBR Eastern Europe Eastern Europe CYP CZE SVK PRT ROU KAZ LTU South Caucasus HRV Advanced Europe HUN POL MNE RUS BLR Turkey Turkey AZE BGR ARM MKD SRB TKM Western Balkans GEO TUR Central Asia ALB BIH MLT UKR Russian Russian KSV UZB Worst TJK KGZ Federation Federation MDA 0 0.5 1.0 1.5 2.0 Worst Best 0 0 00 00 00 50 1,5 1,0 2,0 Speed ranking Average container Average container sending cost (euros) delivery time (days) Note: Passenger transport connectivity as measured here is multimodal, averaging across road, rail, and bus modes the price that must be paid to travel to a representative main city in the country. Only countries with complete data for time and cost for all modes (road, bus, and rail) are included. Affordability is estimated using the unit cost of services adjusted for purchasing power. Freight transport connectivity for a given city is measured as the average price to send a container from that city to the other main cities within a country. Luxembourg is not included in the analysis of affordability and speed because of data concerns. “Advanced Europe” includes countries in Western, Southern, and Northern Europe that signed the Maastricht Treaty or joined the European Union before 1995. Individuals in wealthier countries, such as the Scandinavian countries, France, the Netherlands, Austria, Italy, and Germany, may have a higher opportunity cost of time and thus are willing to accept higher absolute costs to achieve faster trans- port (figure 5.1, panel d, top-right quadrant). Countries with lower gross domestic product (GDP) per capita, such as Armenia, Moldova, and Azerbaijan, seem to prefer to sacrifice speed (quality) and make freight connectivity more affordable within the country (figure 5.1, panel d, bottom-left quadrant). Infrastructure Linkages: Cost, Time, and Networks ●  203 Connecting with Neighbors (Regional Angle) The cost and time required to connect with neighboring countries vary greatly within ECA. The average passenger in a Central Asian country pays close to 125 euros and travels close to two days (47 hours) to reach main cities in neighboring countries (figure 5.2, panel a), despite strong economic potential and the need for connections as a result of being landlocked. Connectivity to neighbors is low. The average cost faced by a Russian passenger is similar to that of an average Central Asian passenger, though, surprisingly, the average travel time to neighbors is lower (about 30 hours). By contrast, countries in the Western Balkans show the lowest average cost to connect with neighbors, primarily because of the proximity among the main cities in the region. Similarly, average travel times to connect with neighbors for passengers in the Western Balkans and Central Europe are less than 10 hours. Advanced Europe’s high-level road and rail infrastructure also delivers low travel times, but at somewhat higher cost. However, advanced Europe’s high incomes mean that affordability of passenger transport to neighboring countries is among the highest in ECA (figure 5.2, panel b, top-right quadrant) as transport is a small share of their total income. Poorer parts of ECA fare worse in terms of affordability (figure 5.2, panel b, bottom-left quadrant), with transport taking up a larger share of their relatively low income. Most Central European and Baltic coun- tries remain in the middle of the pack, which is reasonable given their infrastructure stock and their cost of living.7 For freight transport, ECA is increasingly an integrated market. Thus, compared with passenger transport, the competitiveness of traders is more influenced by the logistics cost structure than by affordability. Freight transport costs are determined by prices in both the origin and destination countries, and transport costs might not be symmetric because of load factors and market structures of the trucking and shipping industry. Relatively low unit transport costs might help countries like Albania, Armenia, Kazakhstan, and the Kyrgyz Republic, which export products with lower value added. The average cost and time required to ship a container from the capital city to the main city of neighboring countries varies little among most regions, except for Russia, Turkey, and Central Asia. Russia faces the highest shipment cost and time to its neighbors’ main cities (figure 5.2, panel c). This is not surprising, given that Russia has 11 neighbors including China in the Far East and Norway in Scandinavia.8 Turkey is the second most expensive place to send a container, followed by Central Asia. It takes twice as long to ship a container in some Central Asian countries and Russia compared with the best- performing regions. While not shown in the figures, the average cost to ship a container in land- locked ECA countries is lower than for coastal and island countries, although landlocked countries face slightly higher shipment times. Costal and island coun- tries would likely register better performance if the comparison were made based on the transport of bulk commodities as opposed to containers. This is due to the greater role of sea transport (a natural advantage of coastal and island countries) for such commodities.9 This implies that a geographic advantage does not necessarily translate into a cost advantage. Conceivably, landlocked 204  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 5.2  Neighbor connectivity Passengers a. Cost and time (average per region) b. Affordability and speed (ranking) Best NOR DNK LUX Western Balkans Western Balkans Low speed, high affordability AUT SWE CHE Central Europe Central Europe NLD ITA DEU EST SVN BEL Affordability ranking Eastern Europe Advanced Europe ESP LTU PRT South Caucasus Eastern Europe KAZ FRA AZE RUS SVK LVA POL Advanced Europe Turkey ROU GBR TUR Turkey South Caucasus GEO MNE BGR HRV HUN Russian ARM SRB BLR Central Asia BIH Federation MKD CZE High speed, Russian KSV MDA Central Asia Worst UKR low affordability Federation KGZ 0 50 100 150 0 10 20 30 40 50 Worst Best Average passenger Average travel Speed ranking cost (euros) time (hours) Freight c. Cost and time (average per region) d. Unit costs and speed (ranking) Western Balkans Central Europe CYP Best ARM KAZ ISL MKD AZE Central Europe Eastern Europe KGZ NOR TJK TKM UZB BLR Eastern Europe Western Balkans BIH EST Unit cost ranking LVA MDA ROU UKR KSV South Caucasus Advanced Europe ALB MNE SRB GEO POL HRV BGR GRC RUS Advanced Europe South Caucasus SVN SVK HUN ESP TUR Central Asia Turkey CZE PRT DEU NLD LTU FRA DNK Turkey Central Asia BEL SWE FIN ITA Russian Russian LUX Federation Federation Worst AUT CHE GBR MLT IRL 0 1,000 2,000 3,000 0 0.5 1.0 1.5 2.0 2.5 Worst Best Average container Average container Speed ranking sending cost (euros) delivery time (days) Note: Passenger transport connectivity for each country is measured as the average (over rail and road transport options) travel time and travel cost passengers incur to reach the main cities in neighboring countries, starting from the capital city. Only countries with complete data for travel time and cost for all modes (road and rail) are included. Freight transport connectivity for a given country is measured as the average price to send a container from its capital city to the main cities of neighboring countries. “Advanced Europe” includes countries in Western, Southern, and Northern Europe that signed the Maastricht Treaty or joined the European Union before 1995. countries can outperform coastal countries if they have good road and rail con- nections to their neighbors’ main cities, but it is costly. Landlocked countries have strong economic payoffs for good transit networks to connect with key trade outlets.10 Like what was observed for domestic freight connectivity, shippers and traders in Belgium and the Netherlands face the slowest speed to send containers to main cities in their neighboring countries. Again, the high level of congestion in the highway systems of these countries is likely the main the culprit. Infrastructure Linkages: Cost, Time, and Networks ●  205 From First Neighbors to Transport Networks: Connectivity as a Policy Objective An approach to connectivity that exploits the opportunities presented by the entire network, rather than just by neighboring countries, can increase the benefit from the movement of goods, services, capital, people, and ideas across ­ countries. Comparing the number of neighbors of the neighbors of a country, versus the number of direct neighbors, illustrates the importance of thinking in terms of trans- port networks. For individual countries of ECA, the aggregate number of countries that are neighbors or neighbors of neighbors varies from 2 to 22. Ukraine, Russia, Hungary, and Germany have the highest number of neighbors plus neighbor of neighbor countries, while Ireland, Portugal, and Iceland have the lowest number (­ figure 5.3, panel a). Similarly, the level of aggregate GDP next to a country varies significantly for each country, and it is not directly proportional to the number of neighbors (or neighbors of neighbors). Germany, France, Switzerland, Belgium, and Luxembourg have adjacent access to large economic centers (higher GDP), while Romania, Belarus, FYR Macedonia, Kazakhstan, and Georgia—with similar numbers of neighbors (or neighbors of neighbors)—have adjacent access to much lower aggregate GDP levels (figure 5.3, panel b). These basic examples underscore the importance for a country of incorporating a wider network (or the whole ECA net- work) in its decisions. A comparison of the costs faced by countries when only connecting with neighbors compared with when connecting with ECA as a whole reveals country-­ specific challenges or opportunities. For example, Russia has the highest aver- age cost to connect to its neighbors but the lowest average cost to connect to the whole of ECA (figure 5.4). Thus, improving access to the rest of the ECA countries may not represent a big cost-saving opportunity, while targeting next- door neighbors could bring more benefits. By contrast, Central Asian countries have low average costs to connect to neighbors but a high average cost to connect to the rest of the ECA network. South Caucasus and Western Balkan countries, which are small and centrally located, have very low costs to reach their neighbors and lower average costs to reach the whole ECA region com- pared with Central Asian or advanced European countries. Because of their centrality in ECA, countries in Eastern and Central Europe face relatively low costs to reach all ECA countries. As shown earlier, it also is important to assess connectivity in terms of time, particularly when products or the nature of the passenger trip require reliability and predictability. Perishable products might need cuts in time even at the expense of using more costly transport alternatives, thus shipping products by air rather than by land (e.g., as in the flower export business). Similarly, a business traveler may pay a premium for more rapid transportation, likely in proportion to the opportunity cost of the traveler’s time. Exporters optimally choose between modes and routes depending on their preference for a cheaper solution that might take longer or be less reliable, versus a more expensive one that is faster or more reli- able. Hummels and Schaur (2013) find that delays can impose high costs (each day in transit is equivalent to an ad valorem tariff of 0.6 to 2.1 percent), and that the 0 5 10 15 20 25 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 18,000 20,000 Ireland Cyprus Portugal Kyrgyz Republic Iceland Tajikistan Armenia 206  ●   Turkmenistan Cyprus Albania Kyrgyz Republic Bosnia and Herzegovina Tajikistan Kosovo Turkmenistan Montenegro Malta Greece Uzbekistan Macedonia, FYR Sweden Bulgaria Spain Uzbekistan United Kingdom Serbia Albania Armenia Greece Moldova Lithuania Estonia Moldova Kazakhstan Netherlands Turkey Bosnia and Herzegovina Croatia Estonia Latvia Kosovo Country itself Azerbaijan Azerbaijan Neighbors Norway Finland Finland Latvia Georgia Montenegro Portugal Norway Neighbors Romania Turkey Iceland Belgium Russian Federation a. Number of countries Denmark Ireland Luxembourg Malta Macedonia, FYR Neighbors of neighbors Lithuania Georgia Sweden Kazakhstan Belarus Croatia FIGURE 5.3  Nonlinear impact of connecting with neighbors of neighbors Ukraine Italy Neighbors of neighbors b. Aggregate level of adjacent GDP (billions of US dollars) Netherlands Bulgaria Slovak Republic Serbia Hungary Slovak Republic Slovenia Slovenia Denmark Czech Republic Poland France Czech Republic Switzerland Austria Belarus Spain United Kingdom Poland Italy Austria Luxembourg Romania Belgium Germany Switzerland Hungary France Russian Federation Germany Ukraine Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Infrastructure Linkages: Cost, Time, and Networks ●  207 FIGURE 5.4  Cost-based Russian Federation connectivity of the ECA Eastern Europe road transport network, by region Central Europe Western Balkans Euros South Caucasus Turkey Advanced Europe Central Asia 0 50 100 150 200 250 300 350 Neighbors All ECA Note: Regions are sorted with respect to the costs incurred to reach all Europe and Central Asia (ECA) countries. The figure shows average weighted road costs for each ECA region for two cases: (a) costs to connect to neighbor countries and (b) costs to connect to all ECA countries. Road costs are measured using the average speed on the road and the fuel costs in each country. Average costs are estimated as follows: m= Costc ∑ m distc d ∈∆(c ) m →d × costc →d , ∑ m distc d ∈∆(c ) →d in which Δ(c) is the set of countries of interest to be reached starting from the main city of country c, m distc m costc →d is the distance in kilometers, and →d is the cost (in euros) between the main cities of countries c and d via modality m. most time-sensitive trade flows involve parts and components trade, which may be one reason for the large and growing share of world trade shipped by air. Rankings of ECA countries by cost and time of transport (across transport modes) differ little. Central Asian countries have much higher costs (for both road and container transport) and much longer times than other ECA regions (Figure 5.5). Recent or expected infrastructure projects, gathered under the Belt and Road Initiative, might help integrate these countries and improve their connectivity. The islands Cyprus, Malta, and Ireland, as well as Spain and Portugal, ­ are also among the countries with the highest costs and time to reach the rest of the ECA network. The South Caucasus performs better in terms of costs com- pared to time, whereas Western Europe is the opposite. Eastern and Central European countries have relatively cheaper and faster connections to the rest of the network. The similarity in these rankings largely reflects road transport costs, which are determined in part by average speeds—reflecting the quality of ­ infrastructure—and are thus more correlated with time than are container prices, which reflect other parameters, such as logistics costs, the presence of rent-­ seekers, and the degree of competitiveness among service providers. Thus, cost versus time performances of countries are more diverse when looking at con- tainer prices ­ (figure 5.5, panel b). Some countries, like Armenia, Kosovo, Turkey, Macedonia, FYR, and Greece, have relatively better connectivity in terms of con- tainer costs than for time. Others, like Montenegro, Slovenia, and Norway, have instead relatively better time connectivity than cost connectivity. Understanding 208  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia country specifics requires a deeper look into institutional factors, the quality of logistics, and the competitiveness of the transport sector. The rest of this section discusses three strategies that countries might choose in improving connectivity. Some countries could embrace a transport network development strategy based on targeted agreements, for instance, connecting to central hubs rather than engaging in bilateral connections, or emphasizing transport connectivity to key economic centers or political allies (strengthening partnerships). Other countries might aim to maximize the regional GDP that their transport networks unleash (maximizing potential). A final possible strategy can be rooted in strengthening existing transport corridors by increasing their resilience or just emphasizing the integration of ­ transport networks by shortening distances (fostering redundancy and integration). This is not an exhaustive list of options, and certainly they are not ­ mutually exclusive policy targets. Strengthening Partnerships: Connectivity through the Lens of Alliances There is a trade-off between slightly improving all connections and significantly improving a few well-chosen ones, which means going from a focus on the whole network to a few strategic partners. The choice of partnerships to strengthen through better connectivity may depend on many factors beyond distance and geographic constraints. Countries may wish to improve connections with large and sophisticated markets, which can offer large trade opportunities. The opportunity to participate in cross-border supply chains might push countries to focus on improving their connectivity to headquarters economies like Germany. A similar reasoning can be applied for the benefits from knowledge transfers. Learning from markets that are closer to the actual technological frontier might allow a country to leapfrog in productivity, or make incremental productivity gains, depending on its initial conditions. Comparing travel time and cost performance for the main three strategic poles in ECA (the European Union [EU], Russia, and Turkey) can help countries assess their geopolitical connectivity in ECA.11 Focusing on costs and time reveals com- plementary patterns of connectivity and provides a better picture of transport ser- vices than just looking at geographic connectivity. Disaggregating the previous results into connectivity toward the three main ECA poles helps countries position themselves with respect to the main poles. Western Balkans and Central Europe have high costs to reach Russia relative to the average cost of Focusing on reaching all EU countries, whereas the South Caucasus has higher costs costs and time reveals complementary pat- to reach all EU countries compared to reaching Russia. Turkey can be terns of connectivity reached at a relatively low cost from Western Balkans, South and provides a better Caucasus, and Central Asian countries. Being well connected to one picture of transport or several of these poles matters for ECA countries in terms of eco- services than just look- nomic and political opportunities. ing at geographic Comparing travel time and cost performances for different sets of connectivity. intended markets can help countries assess the probability of success of various improvements to transport connections. For simplicity, Infrastructure Linkages: Cost, Time, and Networks ●  209 FIGURE 5.5  Cost and time a. For roads 140 connectivity in the ECA network ISL 120 TKM Average weighted hours TJK 100 UZB KGZ KAZ 80 CYP AZE ARM IRL PRT MLT 60 ALBGEO KSV TURGRC ESP MKD BIH BGR MNE NOR MDA AUT ROU SRB HRV BELFRA ITA GBR SVN FIN CHE LUX NLD UKRSVK DEU CZE HUN EST DNK BLR LTULVA SWE POL 40 RUS 200 300 400 500 600 Average weighted costs b. For containers TKM 8 TJK ARM IRL PRT CYP KSV TUR GBR KGZ UZB Average weighted days MKD GRC ESP 6 BIH BEL NLD AZE LUX FRA ISL BGR HRV CHE SRB GEO KAZ CZE AUT DNK LTU LVA EST DEU ROU SWE HUN FIN MDA POL SVK RUS ALB NOR BLR 4 UKR MNE SVN MLT 2 ITA 4,000 6,000 8,000 10,000 Average weighted costs Note: Weighted cost indexes are computed by summing the costs for countries weighted and normalized by the distance between the two countries. Country-weighted costs are then averaged across regions using a simple average method. The green line in panel b shows the linear prediction. Average costs and times are estimated as follows: m= Costc ∑ m distc d ∈∆(c ) m →d × costc →d , ∑ m distc d ∈∆(c ) →d ∑ m m →d × distc timec →d m= d ∈∆(c ) Timec , ∑ m distc →d d ∈∆(c ) in which Δ(c) is the set of countries of interest to be reached starting from the main city of country c, m m costc m distc →d is the distance in kilometers, and →d and timec →d are the cost (in euros) and the time (in hours or days) between the main cities of countries c and d via modality m. 210  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia only five types of scenarios are compared. First is connecting only to neighbors. Trade or capital flows tend to increase with proximity, partly because of the importance of factors other than transport costs, such as migration diasporas and language or cultural proximity. Proximity is associated with knowledge trans- fers (Arrow 1969; Bahar, Hausmann, and Hidalgo 2014). Second is connectivity to the whole network of countries as a means to assess trade potential and knowledge transfers without any assumptions concerning the value of different connections. Third is connecting to just the largest economies in terms of GDP to target markets with the highest potential for trade opportunities and knowl- edge spillovers.12 Fourth is increasing connections to economies with the high- est levels of technological sophistication, which may help maximize trade and knowledge spillover opportunities if initial conditions are sufficiently strong (i.e., strong business and governance environment).13 And finally, increasing connec- tions to economies with similar levels of complexity captures opportunities with the highest potential for firms that would not be competitive in the largest or most complex economies. According to the proposed measure of cost connectivity (for simplicity and to capture the trade dimension, we focus on freight costs only), ECA countries can be grouped into three categories based on the strategies mentioned above. First, countries in the Western Balkans or Central Europe face lower costs and greater connectivity with the largest economies of Europe. Second, countries in Central Asia and the South Caucasus, together with Russia, face lower costs and higher connectivity toward countries with similar levels of production sophistication. Finally, countries in advanced Europe, Eastern Europe, and Turkey face similar costs to reach either the largest ECA economies or countries with higher produc- tion sophistication (figure 5.6). Countries with high costs to connect with ECA’s largest markets may decide whether to focus on improving the connectivity to those markets. For instance, they could alternatively decide to target markets that are currently accessible, improve connections to strategically selected partners, or develop and improve connectivity to different economic gravity centers. A comparison of Central Asia and Western Balkans illustrates the usefulness of this approach. Central Asian countries face an average cost to send a con- tainer to the largest ECA economies of 10,000 euros, more than double that of countries in Central Europe, Western Balkans, and even Turkey. Thus, large and expensive investments, and an unlikely leapfrog in technology (given initial conditions), would be required to enable Central Asian countries to gain eco- nomically in investing in greater connectivity to European markets. For Central Asia, aiming at strengthening ties with neighbors as a means to building cor- ridors to the larger ECA network seems a more cost-­ effective strategy.14 By contrast, Western Balkan countries are in a very different position than Central Asia, with very advantageous connectivity to the main economic gravity center of ECA. Clearly, variables other than transport interventions, such as characteristics of the business environment, level and type of technology available, amount of resources the country has available for investment, and other critical dimensions, also will determine which connections a country wants to prioritize. Infrastructure Linkages: Cost, Time, and Networks ●  211 FIGURE 5.6  Container cost a. Partnerships with alternative economic centers connectivity Central Europe Western Balkans Eastern Europe Advanced Europe Turkey Russian Federation South Caucasus Central Asia 0 2,000 4,000 6,000 8,000 10,000 Euros Turkey Russia EU countries b. Partnerships with strategic groupings of countries Russian Federation Eastern Europe Central Europe Turkey Western Balkans Advanced Europe South Caucasus Central Asia 0 2,000 4,000 6,000 8,000 10,000 Euros Most complex economies Economies with similar sophistication of production Only neighbors Largest economies All ECA Note: Weighted costs (in euros) are computed as follows: m= Costc ∑ d ∈∆(c ) m distc × m →d costc →d , ∑ m distc d ∈∆(c ) →d in which Δ(c) is the set of countries of interest to be reached starting from the main city of country c, m distc is the distance in kilometers, and cost m is the cost (in euros) between the main cities of →d c →d countries c and d via modality m. The resulting costs are averaged across regions. ECA = Europe and Central Asia; EU = European Union. 212  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Maximizing Potential: Connectivity through the Lens of Market Access An efficient transport network allows firms to reach foreign markets and serve both final producers buying intermediate inputs and consumers buying final products. One policy goal when designing a road or railway network is to maximize the GDP footprint of the ECA region made accessible by that transport system. This type of approach demands that the connectivity assessment incorporate the size of the reachable markets together with the cost of connectivity (box 5.1). Advanced Europe scores the highest realized potential for both road and rail networks in terms of out-of-pocket cost, and therefore is the best-performing region for market access connectivity (figure 5.7). That sets the yardstick for the other regions. Eastern and Central Europe also perform very well, accessing 85–90 percent of the market potential with their road networks. The Central Asia road network opens roughly 40 percent of the GDP attain- able by the road network of advanced Europe, which implies that Central Asia would need to more than double its road connectivity by reducing costs to have the same market access as the advanced Europe benchmark. Similarly, the South Caucasus and Turkey, which reach only 50 percent of the ECA market frontier, would need to cut their road transport costs in half to be at the con- nectivity performance of road networks in advanced Europe. Many countries’ ability to increase their market connectivity by improving efficiency is limited by BOX 5.1 Measuring Market Access The potential index using costs is estimated for Realized Potential is defined as the potential each country according to the following equation: attained by a country with respect to the maximum realized potential achieved by any country in the GDPd ∑ cost , m Potentialc = sample for that transport mode or network in the m d c →d geographic area considered: in which m m Potentialc = . ( ) c is the country analyzed and of origin Realized Potentialc d is each country of destination in the targeted max Potential m market including c Potentials are computed for each country and m is the transport modality m the unrealized potential is then estimated with cost  c →d is the cost from c to d using a specific respect to the best-performing country (or region) transport mode m  with cost m the cost of c →c of ECA. reaching main cities within a country The term costc m when c and d are not adja- →d Potential captures the amount of GDP a country cent countries is estimated by finding the optimal can reach with a unit of transport cost (one con- route between c and d in a virtual network that tainer, one private car, or one individual using the concatenates all collected data between neighbor- railway). It is unit-less. It also includes the domestic ing countries. For this, the Dijkstra algorithm con- potential for each country, defined as the domestic siders all possible paths between countries and GDP divided by the average cost of reaching the determines the shortest or cheapest route between main domestic cities. any pair of nonneighbor countries. Infrastructure Linkages: Cost, Time, and Networks ●  213 FIGURE 5.7  Realized Advanced Europe potential of connectivity to ECA markets relative to Eastern Europe advanced Europe Benchmark: Roads, railways, Central Europe and containers Western Balkans Russian Federation South Caucasus Turkey Central Asia 0 20 40 60 80 100 Road Containers Rail Note: The targeted market is defined as all countries in Europe and Central Asia (ECA). Potentials are averaged per region and then normalized by the highest potential. Albania, Cyprus, Iceland, Malta, and Tajikistan are removed from the measure of rail connectivity either because of their lack of rail connections with neighbor countries or because of limited data availability. long distances from markets and difficult terrain (figure 5.8, panel a, provides country data). Differences in realized potentials across ECA regions are greater for rail trans- port networks (figure 5.7). The second-best region after advanced Europe in accessing economic opportunities by railroads is Central Europe, achieving about half of the maximum potential attainable. Eastern Europe performs relatively well for roads but poorly for the rail network. Central Asia also remains well behind all regions in its railroad and container networks. Central Asia would have to increase its container connectivity (reduce costs) by a factor of three to have the same potential as advanced European coun- tries. Its rail connections barely capture 10 percent of the economic potential of advanced European countries, which is consistent with previous results and implies that Central Asian firms face huge obstacles to reaching markets in ECA. While Central Asian countries could improve their railroad connectivity by reducing prices, the long distances and costs inherent in reaching the largest ECA markets makes it unlikely that they could match the economic potential of other regions (figure 5.8, panel b).15 Fostering Linkages and Overall Integration: Connectivity through the Lens of Robustness Countries should be strategic in choosing which connections to focus on in improving their connectivity in the overall transport network. Some linkages to countries might become redundant, in the sense of not improving the overall integration of a country in the ECA network. Linkages and integration indexes characterize the relationship between having linkages to many neighboring 214  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia countries and being well integrated in the whole ECA network (box 5.2). The Czech Republic, the Slovak Republic, and Austria are the three most integrated countries in the whole ECA region. Central Asian countries remain poorly inte- grated in the whole network. Interestingly, some countries are well integrated despite having few linkages to neighbors (Lithuania, FYR Macedonia, the Czech Republic, Luxembourg, Estonia, FIGURE 5.8  Realized potential of connectivity to markets by country, 2016 Percent of benchmark (100 = best) a. Roads b. Railroads Luxembourg Germany Belgium Netherlands Switzerland United Kingdom France France Germany Belgium Slovak Republic Spain Netherlands Italy Czech Republic Switzerland Austria Austria United Kingdom Czech Republic Lithuania Slovak Republic Slovenia Western Europe Luxembourg Western Europe Ireland Sweden Latvia Slovenia Denmark Finland Italy Portugal Estonia Lithuania Spain Denmark Sweden Ireland Finland Norway Portugal Estonia Greece Latvia Norway Greece Malta Malta Cyprus Iceland Iceland Cyprus Belarus Belarus Eastern Eastern Europe Europe Ukraine Ukraine Moldova Moldova Hungary Poland Central Europe Central Europe Poland Hungary Croatia Croatia Bulgaria Romania Romania Bulgaria Serbia Serbia Western Balkans Western Balkans Montenegro Bosnia and Herzegovina Macedonia, FYR Montenegro Bosnia and Herzegovina Macedonia, FYR Kosovo Kosovo Albania Albania Russian Federation Russian Federation South Tur Rus South Tur Rus Caucasus key sia Caucasus key sia Turkey Turkey Georgia Azerbaijan Azerbaijan Georgia Armenia Armenia Kyrgyz Republic Kazakhstan Central Asia Central Asia Kazakhstan Uzbekistan Uzbekistan Kyrgyz Republic Tajikistan Turkmenistan Turkmenistan Tajikistan 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 continued Infrastructure Linkages: Cost, Time, and Networks ●  215 FIGURE 5.8  continued c. Containers Belgium Luxembourg Slovak Republic Netherlands Czech Republic Germany France Austria Switzerland Slovenia United Kingdom Western Europe Lithuania Latvia Spain Italy Estonia Finland Malta Norway Cyprus Ireland Greece Portugal Denmark Iceland Sweden Serbia Eastern Rus South Tur Western Balkans Macedonia, FYR Kosovo Montenegro Bosnia and Herzegovina Albania Turkey Europe sia Caucasus key Azerbaijan Georgia Armenia Russian Federation Belarus Ukraine Moldova Hungary Central Europe Poland Croatia Bulgaria Romania Kyrgyz Republic Central Asia Kazakhstan Uzbekistan Tajikistan Turkmenistan 0 10 20 30 40 50 60 70 80 90 100 Note: The targeted market is defined as all countries in Europe and Central Asia. the Netherlands (figure 5.9). Some countries are relatively poorly integrated despite having many linkages to neighboring countries (Uzbekistan, Greece, Italy, France, and Georgia). Exogenous constraints, such as the number of neighboring countries, are important in determining the degree of integration, but other fac- tors such as the state of the transportation network are more crucial in determining how well integrated a country is. 216  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia BOX 5.2 Linkages and Integration Being well integrated in the transport network measures help to understand the position of a depends on the number of connected neighbors country in a network: (a) linkages and (b) overall but also on the nature of the neighbors. After fac- integration. Linkages are given by the number of toring in the risks of being more vulnerable to connected neighbors (degree centrality) that a shocks, a country might benefit from having low country has. Overall integration is measured as the transport costs to a hub country instead of having average shortest paths between one country and several connections to nonhub countries. Two the rest of the network (closeness centrality). FIGURE 5.9  Linkages and overall integration 50 CZE SVK AUT HUN POL UKR BLR DEU LTU SRB 40 HRV RUS SVN MDA ROU BGR MKD LUX KSV CHE Ranking for integration 30 BIH LVA MNE BEL ALB FRA EST NLD TUR FIN ITA 20 DNK GRC GEO AZE SWE ARM GBR KAZ KGZ 10 NOR IRL ESP UZB MLT TJK TKM CYP PRT ISL 0 0 10 20 30 40 50 Ranking for linkages Note: Measures of linkages (unweighted degree centrality) and integration (closeness centrality) are used to rank countries. High ranks mean high centrality indexes. Countries with a similar rank for linkages have the same number of connected neighboring countries. Connectivity as a Collective Challenge: Centrality and Criticality Improving one segment of the whole transport network creates positive externali- ties for all other countries. Improving the quality of a road or reducing transport services costs between two countries not only improves the connectivity and the potential of the directly affected countries, but also many other countries and links Infrastructure Linkages: Cost, Time, and Networks ●  217 belonging to the same transport network or system. Such externalities are usually not fully reflected in the returns to transport investments, especially when they happen between countries. The EU allows for some cross-country coordination of interventions that, by design, internalizes many of these network externalities. The concrete instruments used are coordinated planning, budgets, and policies gov- erning infrastructure investments. The Trans-European Transport Network corri- dors investments, a flagship EU program, aim at improving transport infrastructure, mostly in segments located in emerging European markets. However, the improved connections will also greatly benefit advanced European economies, which have better access to Eastern and Asian markets and to important gateways. Considering the whole transport network, rather than the network of an indi- vidual country, helps to provide an understanding of the dynamic of transport systems. Some policies might be optimal from a country perspective but not opti- mal from the perspective of the whole network. Network measures characterizing the position of countries in the network provide stylized tools for understanding which countries might benefit from corridor paths between countries or which countries might get the most from their central or critical position. The network “positioning” indexes provide complementary information to understand the structure of the transport networks in ECA, but also which coun- tries benefit, or potentially could benefit, the most from corridors and trade routes (box 5.3). Solving the coordination problem among all ECA countries would facili- tate targeting of efficient investment that could benefit the region as a whole. It is also interesting to assess potential benefits and choose complementary policies that would help firms and workers fully reap the benefits of better connectivity with other countries. How important a country is in a transport network depends on its centrality and on its criticality. Higher transport centrality may bring benefits such as direct gains from transit traffic and increase potential opportunities from trade, FDI, knowledge sharing, and migration.16 Higher criticality implies that disruptions affecting the country’s transport infrastructure will have a more negative impact on the rest of the countries (or some countries) that connect to the country. Centrality and Criticality in the ECA Network Transportation infrastructure channels the movement of goods or people along major cross-country networks and, within networks, corridors. The comprehen- sive nature of the economic benefits for countries of being on a corridor or BOX 5.3 Centrality and Criticality Being in a central position to benefit from corridor given the transport network (PageRank centrality). routes depends on two criteria: (a) centrality and Criticality is given by the number of optimal routes (b) criticality. Centrality is given by the probability between all pairs of countries that pass through the for a transport flow to pass through this country country (betweenness centrality). 218  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia specific crossroads of a network remains an open question. For example, do roads or rail that pass though countries provide economic benefits if ancillary businesses associated with the corridor fail to materialize? However, transit flows may increase the export and import opportunities of firms along these routes or corridors, develop new sectors such as logistics services, and generate nonmate- rial benefits (flows of ideas and knowledge) to boost productivity if the economic and business environment is sufficiently attractive for investment. Firms located in transit countries may benefit from lower production costs and an improved ability to deliver on time. Higher transport centrality might be desirable for a country to increase participation in regional and global value chains or attract FDI or participate in development corridors (chapter 7 discusses the importance of participation in supply chains for developing countries). Centrality is a proxy for the ability to attract traffic and potential corridor spill- overs. However, the level of centrality of a country or city varies depending on the metrics used, which are defined by, among other things, the product, value chain, and services transported. The five most central countries in the ECA network differ, depending on whether cost or time is considered. In terms of transport cost, Austria, the Slovak Republic, and FYR Macedonia are in the top five of the most central countries for the network (figure 5.10, panel a) whereas in terms of time, Serbia, Ukraine and Russia are in the top five for the network (figure 5.10, panel b). France and Germany are among the top five measured in both time and cost ­ (figure 5.10). As discussed, whether traders focus on cost or time depends on several factors that are not considered here, including the nature of the goods or FIGURE 5.10  Centrality in the ECA network for container transport a. Cost-driven centrality b. Time-driven centrality CYP IRL PRT GBR ESP GRC ALB ARM MKD TUR MLT FRA KSV BGR GEO TJK LUXBEL MNE KGZ TKM ITA CHE NLD SRB AZE UZB BIH KAZ DEU ROU SVN AUT MDA HRV CZE ISL RUS DNK EST HRV HUN UKR BIH HUN SVK BLR LVA MNE POL NOR SRB SVN SVK LTU FIN KSV ROU UKR LTU SWE POL ALB MKD MDA BGR BLR AUT CZE FIN SWE GRC LVA ITA NOR MLT TUR RUS CHE DEU CYP EST DNK GEO AZE FRA LUX BEL NLD ISL ARM KAZ ESP KGZ GBR UZB PRT TJK IRL TKM Note: Circle size indicates level of centrality (larger diameter = greater centrality). For illustrative purposes, the circles representing the top five countries in centrality are colored in green. Lines between nodes indicate the presence in the physical network of an optimal corridor connecting countries. Locations of circles and countries are not linked to geography in any way. ECA = Europe and Central Asia. Infrastructure Linkages: Cost, Time, and Networks ●  219 services, whether the goods traded are part of a supply chain, and the presence of border costs.17 Identifying the most critical countries in transport networks reveals which coun- tries have more control over transportation network operability and if these coun- tries suffer a shock, what the implications for other connected countries would be (figure 5.11). This measure can help countries target investments to reduce their vulnerability to specific country shocks in accessing markets or other areas of the network. More generally, critical countries in the transport network are those where disruption would have a major impact on subnetworks or countries that can be, de facto, disconnected. Russia is the most critical country in the network of con- tainer costs in Eurasia (figure 5.11). Germany, Ukraine, Hungary, and Poland are among the five most critical countries. Islands or isolated countries have a very low criticality, as would be expected. As shown, while not a top-five country in terms of criticality, disruptions in the French transport network would affect the connectivity of Spain, Portugal, the United Kingdom, and Ireland to the rest of ECA. Portugal’s connection to the European network is contingent upon Spain, and so forth. FIGURE 5.11  Cost-driven CYP criticality in the ECA network ALB for container transport KSV MNE MKD BIH GRC SRB HRV BGR TUR MLT ROU SVN HUN MDA ITA AUT PRT ARM SVK ESP GEO UKR CHE CZE FRA POL LUX GBR AZE DEU BLR BEL IRL RUS LTU LVA NLD EST DNK NORFIN KAZ SWE KGZ ISL TJK UZB TKM Note: Circle size indicates level of criticality (larger diameter = greater criticality). For illustrative purposes, the circles representing the top five countries in criticality are colored in green. Lines between nodes indicate the presence in the physical network of an optimal corridor connecting countries. Locations of circles and countries are not linked to geography in any way. Results for time-driven criticality are not presented, as the results are very similar to those presented. 220  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Centrality as a Strategic Target Improving connectivity can change the centrality of a country in the transport net- work in many ways and will be accompanied by economic benefits. Increased centrality in a network, subnetwork, or corridor, when properly managed, is a desir- able objective for a country to benefit from a prospective or existing corridor.18 Looking at centrality indexes is a first attempt at understanding the broader ben- efits from transport connectivity at the country level, and can complement mea- surements of the direct user benefits provided in most cost-benefit analyses of projects and corridors. The centrality of a country (or node) in a network can be affected by the effi- ciency, and therefore, the transport costs of its network. A key question for policy makers is the distribution and level of benefits from alternative interventions, that is, the impact of reducing transport costs in a specific segment (as opposed to randomly chosen links). This information can help in making strategic decisions about which segment to invest in to increase the potential benefits from a network or corridor, and how to structure a project’s financing to make it sustainable and linked to the stream of benefits. This analysis can be illustrated by the case of Romanian investment choices. A one-third decrease in the costs of transporting a container traveling between Bulgaria and Romania would increase Romania’s centrality in the corridor by 14 percent, meaning that Romania’s probability of attracting cargo flows would increase by 14 percent. At the same time, Bulgaria would increase its centrality by 9 percent. Romania, however, could instead invest in reducing container trans- port costs in its trade with Ukraine. If the cost of moving a container in the Romania– Ukraine corridor were reduced by a third, Romania’s centrality would increase by 6 percent. Thus, investing in Romania’s connection with Bulgaria would have a larger impact on Romania’s centrality (table 5.1). Similarly, one can compare a decrease in the costs of shipping goods from Poland toward Germany, the Slovak Republic, or Ukraine. The resulting estimates suggest that gains in centrality are slightly higher when the segment Poland– Ukraine is improved. Kazakhstan would increase its centrality the most by reducing the cost of shipping goods to the Kyrgyz Republic. Interestingly, a decrease in costs between Russia and Kazakhstan does not increase the centrality of either country. This can be explained by Kazakhstan’s degree of isolation and the very high costs of such connections. Criticality and Market Access as Strategic Targets Alternatively, a policy maker might want to decrease transport costs in a specific segment to increase the criticality of its networks in a broader context or, in other words, the importance of its role as a country in supporting the reliability and sta- bility of the wider transport system. Increasing criticality can position a country as a transit country that “sells” or exports transport services to other countries. Similarly, when capturing network effects, decreasing container costs for a country would affect each country and link differently. In the case of Romania, transport costs for containers between Bucharest and the main cities of neighboring countries (Belgrade, Budapest, Chisinau, Kiev, Infrastructure Linkages: Cost, Time, and Networks ●  221 TABLE 5.1  Changes in Centrality Due to a Decrease in Container Transport Costs Percent Affected segment First beneficiary Second beneficiary Bulgaria–Romania Romania: 14 Bulgaria: 9 Poland–Germany Poland: 4 Germany: 2 Poland–Slovak Republic Poland: 5 Slovak Republic: 3 Kazakhstan–Russian Federation Russian Federation: 2 Kazakhstan: 0 Kazakhstan–Kyrgyz Republic Kazakhstan: 13 Kyrgyz Republic: 12 Kazakhstan–Uzbekistan Kazakhstan: 6 Uzbekistan: 4 Bosnia and Herzegovina–Serbia Bosnia and Herzegovina: 19 Serbia: 6 Bosnia and Herzegovina–Croatia Bosnia and Herzegovina: 15 Croatia: 4 Ukraine–Romania Romania: 6 Ukraine: 4 Ukraine–Slovak Republic Slovak Republic: 2 Ukraine: 1 Ukraine–Poland Poland: 7 Ukraine: 5 Armenia–Georgia Armenia: 28 Georgia: 8 Note: Changes in centrality are calculated as the normalized centrality PageRank index in which Centralityn,i is defined by   Centralityi − min Centrality j ∈ECA    , Centrality n, i =     max Centrality j ∈ECA  − min Centrality j ∈ECA      in which centralityi is the PageRank index for a given country i, min(Centralityj=ECA) is the minimum for all countries, and max (Centralityj=ECA) is the maximum for all ECA countries. Simulations consist in assuming a 30 percent decrease in transport costs in the listed links. Criticality is estimated before and after the cost shock. Percentage point changes are as reported. and Sofia) are assumed to be reduced by a third. The decrease in costs for the links out of Romania to reach its neighbors affects those direct costs but also, as expected, the costs of shipping goods from Romania to all ECA countries figure 5.12, panel a). Criticality increases for Romania and, simultaneously, for the (­ Czech Republic and Ukraine. However, it decreases for Bulgaria, France, Poland, Hungary and the Slovak Republic (figure 5.12, panel b). Cost reductions in the links going out of Romania would also increase its market access potential. However, the change in costs would also increase market poten- tial for all countries in Central Europe (figure 5.12, panel c). Admittedly, the largest increase is for Romania, but Croatia and Bulgaria would also accrue important increases in realized potential. If Kazakhstan’s transport costs for containers are reduced by one-half between Almaty and the main cities of neighboring countries (Bishkek, Moscow, and Tashkent) the centrality of neighboring countries and the network as a whole would increase significantly as a result of connections with Russia and Russia’s con- nections to the rest of ECA. The analysis shows that Kazakhstan, as a Central Asian country, is very isolated from the rest of ECA and would benefit highly, as would the other Central Asian countries, from lower-cost connections to other ECA coun- tries. Interestingly, the Kyrgyz Republic would see the largest increase in realized potential market access (figure 5.13). Collective Benefits of Improved Connectivity along a Corridor What happens if a broader, corridor-wide, improvement in connectivity is achieved? Consider the case of a one-third reduction in container costs in the West-East 222  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 5.12  Romania: Impact of a 30 percent decrease in container transport costs a. Costs to reach markets b. Relative index of criticality 50 DEU CZE RUS SVK ROU SRB HUN BLR UKR POL Neighbors 40 FRA KAZ UZB MKD BGR GEO HRV After cost reduction TUR DNK NOR SVN 30 GBR ITA FIN ESP SWE LVA GRC MNE AUT All ECA 20 BIH ALB LTU BEL KSV 10 0 2,000 4,000 6,000 Euros 10 20 30 40 50 Before cost reduction After cost reduction Before cost reduction c. Realized potential with respect to maximum potential attained in the sample Romania Bulgaria Croatia Poland Hungary 0 10 20 30 40 50 60 70 80 Percent of maximum potential attainable Before cost reduction After cost reduction Note: Realized potential is estimated based on Europe and Central Asia as a whole. Market access along the corridor is proxied by the potential indexes for markets located along the corridor: GDPp Potentialc , corridor = ∑ costc →p . p ∈Corridor corridor that starts in China (Shanghai) and goes to Germany through Kazakhstan, Russia, Belorussia, and Poland. This corridor contains segments that are central to the whole BRI. Market access potential (or GDP made accessible per unit of trans- port used) for each country along the corridor would increase on each of the seg- ments of interest defined, broadly speaking, by each pair of country/capital cities. Measuring the impact of cost reductions for containers on market access potential would enable us to better understand the location and size of the impact of some targeted investments of the BRI. Infrastructure Linkages: Cost, Time, and Networks ●  223 FIGURE 5.13  Kazakhstan: Costs and potential indexes before and after decrease in costs b. Realized potential with respect to maximum a. Costs to reach markets potential attained in the sample Uzbekistan Neighbors Turkmenistan Tajikistan Kyrgyz Republic All ECA Kazakhstan 0 2,000 4,000 6,000 8,000 10,000 0 5 10 15 20 25 30 35 Euros Percent of maximum potential attainable Before cost reduction After cost reduction Note: Realized potential is estimated based on Europe and Central Asia as a whole. Market access along the corridor is proxied by the potential indexes for markets located along the corridor: GDPp Potentialc , corridor = ∑ costc →p . p ∈Corridor TABLE 5.2  Segments Affected by 33 Percent Cost Reduction in Container Transport Percent Recipient of changes in market access potential Germany–Poland Poland–Belarus Belarus–Russia Russia–Kazakhstan Kazakhstan–China Germany 21 3 2 7 16 Poland 25 8 6 3 8 Belarus 7 3 26 7 8 Russian Federation 8 6 9 9 21 Kazakhstan 1 1 1 12 44 Note: Market access potential along the corridor is proxied by the potential indexes for markets located along the corridor: GDPp Potentialc , corridor = ∑ costc →p . p ∈Corridor A cost reduction of one-third for the Kazakhstan–China segment would have the largest impact on market potentials: Kazakhstan’s potential would increase by 44 percent because of cheaper access to China’s large GDP. Russia and Germany would also benefit substantially from this cost reduction. Improving the Belarus–Russia segment would mostly benefit Belarus. A cost reduction on the Germany–Poland segment would greatly improve the potential of both Germany and Poland, as well as increase Kazakhstan’s potential by 1 percent. The cost reduction on the Poland–Belarus segment has the smallest impact (table 5.2). 224  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Reducing container costs increases the market potential of each country along the corridor and illustrates the extent to which transport infrastructure investment is a collective problem. Improving a segment of a corridor generates positive externalities and boosts the economic opportunities of all countries along the corridor. However, these results do not consider costs or other poten- tial effects of transport improvements. Relocation of activities and people might offset some of the positive impacts of investing in these corridors, and the trade-offs between potential positive and negative effects should be further studied. Conclusion This chapter offers several insights into transport connectivity in ECA, from both methodological and strategic decision-making viewpoints. From a methodological standpoint, this chapter uses data on ECA transport costs to study transport services rather than just focusing on transport infrastruc- ture (e.g., kilometers of roads). It relies on a simplified collection of prices (or out- pocket costs) for different modes and purposes (passengers and cargo) for all ECA countries. The cost and time data collected focus on transport connectivity (a) within countries and (b) from a country and its immediate neighboring countries. As a means to generate estimates of cost and time connectivity for the whole network, network analysis is used to re-create paths from one country to any given country. This two-step approach enormously simplifies the data collection and facilitates connectivity analysis that can be adapted to the level of granularity desired and resources available. The introduction of a geographic attribute as part of the definition of connectivity is essential to the analysis. Connectivity and other measures of ­ accessibility to markets use diverse definitions of geographic spaces and mar- kets, including regional—for example, coverage of the ECA space—the largest economies or economic gravity centers, countries with similar levels of produc- tion sophistication, and so forth. This allows practitioners and policy makers to assess the trade-offs, opportunities, and limitations of various connectivity strategies. From a strategic standpoint, decisions to improve connectivity of specific transport network segments, corridors, or subnetworks are linked to the potential economic and trade strategies of a country. This approach nuances any possible performance evaluation of connectivity. Connectivity of a spe- cific country will be better or worse depending on what market the transport system aims to cover: domestic connectivity, connectivity to the ECA trans- port network, to neighbors only, to strategic partners, and so forth. Some countries might wish to develop connectivity to maximize their own GDP unleashed by their transport networks (maximizing potential). Other countries could embrace a transport network development strategy based on regional cooperation agreements, for instance, connecting to central hubs rather than Infrastructure Linkages: Cost, Time, and Networks ●  225 engaging in bilateral connections, or underscoring transport connectivity to key economic centers or political allies (strengthening partnerships). A third strategy can be rooted in strengthening existing transport corridors by increasing their resilience or just emphasizing the integration of transport networks by shortening distances (fostering linkages and integration). A coun- try can be highly successful in achieving one policy goal, and not that success- ful in achieving another. One policy goal could be feasible given the country’s geography and geopolitics, and another one might not be affordable given the fiscal space available. Thus, the flexibility of the methodology proposed is not trivial. Adopting a network perspective on the whole transport network, rather than focusing on an individual country, helps provide an understanding of the dynamics of transport systems. Network measures characterizing the position of countries in the network provide stylized tools for understanding which countries would ben- efit from corridor paths between countries or which countries would get the most from their central or critical position. Solving the Transport investments can generate positive externalities for coordination problem other countries in a corridor. Solving the coordination problem among all ECA coun- among all ECA countries would help policy makers target effi- tries would help policy cient investment that would maximize the benefits for all coun- makers target efficient tries in the region. The inability of the country making the investment that would investment to capture these benefits can lead to general under- maximize the benefits investment. More cooperation between countries, especially for all countries in the along corridors, could therefore increase the global benefits of region. transport investments. For ECA, the cost and time required for passenger travel, both within countries and to neighboring countries, varies greatly, with richer countries tend- ing to have both higher quality and affordability of domestic transport services for passengers. By contrast, the cost and time required to deliver a container within the domestic market or to the capital city of neighboring countries varies little across ECA regions, except for Russia, Turkey, and Central Asia. It takes twice as long to ship a container in some Central Asian countries and Russia as it does in the best-performing regions. Countries with high costs for connecting to neighbors but low costs for con- necting to ECA as a whole (e.g., Russia) could focus transport resources on improving connections with neighbors, while countries with low costs to con- nect to neighbors but high costs to connect to the ECA network (e.g., Central Asia) could focus on improving connections to the region as a whole, or to extraregional countries. It also is important to assess connectivity in terms of time, particularly when products (e.g., those traded through supply chains) or the nature of the passenger trip (e.g., for businesspeople with a high opportu- nity cost of time) require reliability and predictability. Cost and time rankings differ more for container transport compared to passenger travel, as container prices depend on various parameters, such as logistics costs and the degree of competitiveness among service providers. 226  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia A comparison of time and cost performance for different intended markets can help countries better assess the probability of success of transport inter- ventions. For example, the very high costs faced by Central Asia makes it dif- ficult to enter new markets with competitive prices, so a focus on an East-West corridor, building on access to neighbors and their access to global markets might be more productive. A second option is to maximize the regional GDP reached by transport networks. Advanced, Central, and Eastern European countries have the greatest market access, and Central Asia the lowest. Finally, countries should consider the value of targeted markets in the whole network. For example, strengthening connections with neighbors that are well con- nected may contribute most to the overall integration of the country within the region and globally. Annex 5A. Methodology and Data Overall Approach The first step is the data collection process, which was focused on a pre- defined network that covers direct links (a) among main cities within a country, (b) from capital cities in a specific country to capital cities of its neighboring countries, and (c) from a capital city to key air and maritime hubs. This analysis measures transport services as opposed to physical assets. Thus, the kilome- ters of roads, or number of airports or ports, is less relevant than observable transport service attributes such as cost, time, reliability, and frequency, among others. Several measures have been suggested to look at transport connectivity, with most of them focusing on physical infrastructure. Aggregate statistics such as the total road or rail track length in a country, or distance to high-quality infrastructure (such as airports, ports, highways), are examples of measures of connectivity based on the physical properties of a transport network. Using the presence of transport infrastructure, or its capacity, as a measure of connectiv- ity is a poor proxy for the quality of transport services or accessibility to a given location.19 In the case of airports, for example, capacity-based measures of transport connectivity “tend to underestimate the importance of small airports and overestimate it for large airports. Small airports may have high accessibility levels if they have few flights to well-connected hub airports” (Burghouwt and Redondi 2013, 36). Similarly, Briceno-Garmendia, Moroz, and Rozenberg (2015) illustrate the problem of traditional capacity-based measures in the context of road trans- port. They find that while the South Asia region is ranked as the best performer for road density, it is the worst performer in terms of perceived road quality (based on the World Economic Forum’s road quality perception ranking). These contradictions imply that a more nuanced measure of transport connectivity should cover not only the availability and capacity of transport infrastructure, but also the quality of transport services on the transport network. Furthermore, viewing transport in terms of proxies for user experiences—and not only in Infrastructure Linkages: Cost, Time, and Networks ●  227 terms of the stock of infrastructure capital—is a key part of any effort aimed at understanding connectivity. For metrics purposes, connectivity, among other transport services, is mea- sured as the cost and time for delivering a container to a destination and the cost and time for a passenger to reach a destination. The costs are the “out-of- pocket” payments made by a consumer for sending a container or by a passen- ger for using transportation, in other words, the “visible price” faced by end-users in the market. An exception is the cost of using private cars over the road net- work. Car travel out-of-pocket costs are represented by the cost of fuel, a func- tion of the fuel consumption that is linked to the vehicle speed and cost of fuel per liter.20 Data Collection Cost and time statistics were collected to capture both freight and passen- ger services along each country’s internal core transport network (domestic connectivity), the nodes connecting it to its neighbors (neighbor connectivity), and international transportation hubs (global connectivity). The modes of transport covered include road, rail, air, maritime, and multimodal. The data come from observable open sources, which should make results replicable and scalable. For data collection purposes, a country’s core network is defined as the road and rail network connecting the main city (capital or main commercial center) to the four most populated cities in each country, the main airport, the main port, key border crossings, and key urban roads (e.g., ring roads). In general, the cities considered were the five most populated cities in each country, with some exceptions for larger countries for which more than five cities were evaluated.21 For each country, a main city was defined, which in most cases corresponded to the capital. In some exceptions, a second or alternative city was chosen given its economic importance. That is the case, for example, of Frankfurt in Germany and Istanbul in Turkey. Primary data were collected on the time required to connect from node A to B (the main cities) and the out-of-pocket cost to connect from node A to B.22 Data were collected from open sources, hence, values and eventual results can be replicated and used in other countries and continents. Sources of primary data collected include open source platforms (i.e., Google Maps), logistics service pro- viders (i.e., UPS), travel sites (i.e., rome2rio, Skyscanner), online freight forwarders, price estimates, and bus schedules, among others. Geographic Scope and Coverage The countries covered in this study include those in the ECA space: 50 countries comprising the European Union, Eastern Europe and the Western Balkans, Turkey and the South Caucasus, and Russia and Central Asia.23 The country sample is diverse in terms of income, economic activity, trade ori- entation, geography, and population. To facilitate the analysis of the data and the characterization of patterns and messages, the chapter uses typologies 228  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia representing geopolitical, geographic, and socioeconomic angles. Three main typologies are used: region, location, and income (map 5A.1).24 Extending the Set of ECA Countries to China One extension would be the addition of China in the set of countries to reach through the transport network. It is important given the sizable trade flows between China and the EU countries and given the geographic location of Central Asian countries that are neighboring China. Focusing only on the ECA countries delivers a truncated perspective, especially for Central Asia. Adding China and access to ports south of Central Asia would reveal different opportuni- ties for those countries. Additional work would help in positioning Central Asia differently, especially within the context of the BRI. In this study, Central Asia is shown at the “far end” of the ECA region. Other work could add a different set of countries south and east of Central Asia. MAP 5A.1  Typologies used in the analysis a. Regions Region Advanced (Western, Northern, and Southern) Europe Kara Sea Barents Sea Central Europe Greenland Sea Eastern Europe Western Balkans Norwegian Sea Turkey Russian Federation South Caucasus Central Asia White Sea Iceland Finland Norway Sweden Gulf of Bothnia Russian Federation Gulf of Finland Estonia Estonia Denmark Latvia Inner Seas North Sea Denmark Denmark Baltic Sea Lithuania Denmark United Kingdom Irish Sea Belarus Ireland United Kingdom Netherlands Poland NORTH ATLANTIC Germany Bristol Channel OCEAN Belgium English Channel Luxembourg Czech Republic Ukraine Slovakia Mongolia Kazakhstan Liechtenstein Austria Hungary Moldova France Switzerland Slovenia Bay of Biscay Croatia Romania San Marino Serbia Bosnia and Herzegovina Golfe du Lion Italy Montenegro Black Sea Adriatic Sea Kosovo Bulgaria Georgia Caspian Sea Uzbekistan Andorra Macedonia Kyrgyz Republic Albania Armenia Azerbaijan Portugal Spain Balearic Sea Italy Tyrrhenian China Greece Turkey Azerbaijan Turkmenistan Sea Aegean Sea Tajikistan Italy Ionian Sea Malta Strait of Gibraltar Northern CyprusSyria Afghanistan Mediterranean Sea Greece Cyprus Iraq Iran continued Infrastructure Linkages: Cost, Time, and Networks ●  229 MAP 5A.1  continued b. Location Location Coastal Kara Sea Landlocked Barents Sea Greenland Sea Island Norwegian Sea White Sea Iceland Finland Norway Sweden Gulf of Bothnia Russian Federation Gulf of Finland Estonia Estonia Denmark Latvia Inner Seas North Sea Denmark Denmark Baltic Sea Lithuania Denmark United Kingdom Irish Sea Belarus Ireland United Kingdom Netherlands Poland NORTH ATLANTIC Germany Bristol Channel OCEAN Belgium English Channel Luxembourg Czech Republic Ukraine Slovakia Mongolia Kazakhstan Liechtenstein Austria Hungary Moldova France Switzerland Slovenia Bay of Biscay Croatia Romania San Marino Serbia Bosnia and Herzegovina Golfe du Lion Italy Montenegro Black Sea Adriatic Sea Kosovo Bulgaria Georgia Caspian Sea Uzbekistan Andorra Macedonia Kyrgyz Republic Albania Armenia Azerbaijan Portugal Spain Balearic Sea Italy Tyrrhenian China Greece Turkey Azerbaijan Turkmenistan Sea Aegean Sea Tajikistan Italy Ionian Sea Malta Strait of Gibraltar Northern CyprusSyria Afghanistan Mediterranean Sea Greece Cyprus Iraq Iran continued However, it remains challenging for Central Asia to access China nowadays. Central Asia is the region with the largest costs to send containers both to the main maritime hubs and to Shanghai compared with other ECA regions (figure 5A.1). Given that freight transport does not cover bulk trade, few containers are currently shipped from Central Asian countries toward China. Therefore, expanding this analysis to China does not improve the connectivity of Central Asian countries given the lack of affordable transport services toward the Chinese hub. The Way Forward The modes considered for this analysis include, for passengers, roads (personal cars and buses) and rail. For freight, the approach is slightly different and the breakdown is defined in terms of the packaging, that is, whether the freight goes in containers or parcel. The movement of a container is, by nature, multimodal. The data originally included data for parcel, but the analysis was inconclusive. The mode choice in each country depends on the availability of infrastructure as well 230  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia MAP 5A.1  continued c. Income Income quintile 1 lowest income per capita Kara Sea 2 Greenland Sea Barents Sea 3 4 5 highest income per capita Norwegian Sea White Sea Iceland Finland Norway Sweden Gulf of Bothnia Russian Federation Gulf of Finland Estonia Estonia Denmark Latvia Inner Seas North Sea Denmark Denmark Baltic Sea Lithuania Denmark United Kingdom Irish Sea Belarus Ireland United Kingdom Netherlands Poland NORTH ATLANTIC Germany Bristol Channel OCEAN Belgium English Channel Luxembourg Czech Republic Ukraine Slovakia Mongolia Kazakhstan Liechtenstein Austria Hungary Moldova France Switzerland Slovenia Bay of Biscay Croatia Romania San Marino Serbia Bosnia and Herzegovina Golfe du Lion Italy Montenegro Black Sea Adriatic Sea Kosovo Bulgaria Georgia Caspian Sea Uzbekistan Andorra Macedonia Kyrgyz Republic Albania Armenia Azerbaijan Portugal Spain Balearic Sea Italy Tyrrhenian China Greece Turkey Azerbaijan Turkmenistan Sea Aegean Sea Tajikistan Italy Ionian Sea Malta Strait of Gibraltar Northern CyprusSyria Afghanistan Mediterranean Sea Greece Cyprus Iraq Iran Note: In panel a, “Advanced Europe” refers to the group of countries that signed the 1993 Maastricht Treaty or joined the European Union before 1995. Panel c classifies countries according to their quintiles in the distribution of GDP per capita in the ECA region. as freight rates, frequency, and reliability. For example, a multimodal transport chain in countries such as the Netherlands, Belgium, and the Western Balkans is likely to involve an inland waterway or railway leg to transport a container domesti- cally or regionally, whereas in other countries it might only involve one mode, mostly trucking. While knowing each leg of the transport chain allows for a richer analysis, our data capture the most relevant information for shippers whose main interest is the market (equilibrium) price and shipment time rather than individual components of the mode used to move their cargo. Seen this way, the data are comprehensive enough to characterize the level of transport connectivity for con- tainerized cargo. An alternative approach would be to collect data on freight by mode but an economic model would be required to find the optimal choice of modes along the transport chain. The analysis of freight transport connectivity is based on the cost and ship- ment time of sending a container. Conspicuously missing are data for bulk, given that the share of containerized cargo movements versus bulk movements varies greatly in the ECA region. Eastern European and Central Asian economies mainly export minerals and agricultural products, which rely more heavily on Infrastructure Linkages: Cost, Time, and Networks ●  231 a. To main maritime hubs FIGURE 5A.1  Cost of and time required for freight Turkey Advanced Europe transport services Averages per region South Caucasus Turkey Eastern Europe Russian Federation Advanced Europe Central Europe Central Europe Western Balkans Western Balkans South Caucasus Russian Federation Central Asia Central Asia Eastern Europe 0 1,000 2,000 3,000 4,000 5,000 0 10 20 30 40 50 Average container sending Average container delivery cost (euros) time (days) b. To Shanghai Turkey Central Europe South Caucasus Advanced Europe Western Balkans South Caucasus Eastern Europe Russian Federation Central Europe Turkey Advanced Europe Central Asia Russian Federation Western Balkans Central Asia Eastern Europe 0 1,000 2,000 3,000 4,000 5,000 0 20 40 60 Average container sending Average container delivery cost (euros) time (days) Note: Freight connectivity to hubs for a given country is measured as the average price to send a container from its capital city to representative intercontinental global hubs: Los Angeles, Rotterdam, and Shanghai. “Advanced Europe” comprises countries in Northern, Southern, and Western Europe. modes that transport bulk freight. Kazakhstan, Russia, and Ukraine have some of the world’s most extensive freight railway systems. It is possible to get a dif- ferent ranking favoring these countries had one looked at the cost and time of sending bulk commodities. It is important to note that while containerized cargo is a small share of exports from these countries, it constitutes a large share of their imports of consumer and industrial products, which rely on intermodal 232  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia transport. The freight connectivity analysis is, therefore, relevant for all countries in the sample. When defining costs as out-of-pocket and using time from origin to destination, the study focuses on the variable observables by a user or decision maker. A deeper decomposition of the costs and time structure would be necessary to iden- tify the bottlenecks and key elements for policy recommendations. Notably, insti- tutional costs and red tape are embedded in the costs and times observed or faced by the user. It would also be interesting to consider the impact of border crossings and market structures like the presence of cartels on transport prices and time. A natural extension is to clarify the structure of the markups created by vari- ous institutional and market issues. For example, Atkin and Donaldson (2015) use the spatial distribution of prices to obtain an estimate of the whole transport costs. These costs include the distance factor as well as the markup due to the structure of the transport sector. The study does not explicitly consider border crossing times and costs sepa- rately. Conceivably, given the diversity of countries and presence of various eco- nomic blocs in the sample, these variables have different degrees of importance. While border-crossing delays and costs are major issues in the Western Balkans and countries such as Belarus, they are irrelevant for intra-EU movements. The shipment time and travel time data, which are gathered from open source web- sites should, in theory, reflect the difference in border crossing across countries. They are, however, likely to underestimate the time for inspections and customs clearance, especially for countries outside of the EU region. In addition to prolong- ing travel and shipment times, border-crossing problems lead to reliability issues, which have important implications for mode choice in particular and the level of realized connectivity in general. Our cost and time (speed)–based connectivity ranking could be biased upward for countries where there are well-known border- crossing problems such the Western Balkans. For others, for example, the Netherlands and Belgium, it could be biased downward where cargo moves reli- ably albeit at a slower speed. Open source data can be completed with survey- based data when available. The car travel out-of-pocket cost, which is estimated as a function of vehicle speed and fuel prices in each country, does not include tolls. Accounting for tolls would make the analysis richer but would require knowledge of specific routes and segments of the road system used to net out their effect, which is a tall order within the scope of the current study. While tolls are not applied universally in the ECA region, they are becoming common for new motorways in some coun- tries. To the extent that tolls are indicators of higher levels of service, the simple cost function should still reflect differences between countries through the speed variable. In terms of metrics, the study uses cost and time. It is known that reliability is a third key metric entering the decision making of transport users. For exam- ple, the Netherlands is poorly rated in terms of speed for freight and passenger connectivity. However, the country would rank very well when looking at reli- ability. Participants in global value chains have to send parts and components to the next stage on time. An accumulation of delays caused by unreliable Infrastructure Linkages: Cost, Time, and Networks ●  233 transport services will disturb the whole supply chain. Having reliable transport services is essential for countries and to decision makers. Further work should add reliability to time and costs as metrics for connectivity to get a more com- plete picture. The current study presents the material as separate connectivity indexes for each of the aspects analyzed, that is, domestic and regional indicators, time and cost connectivity, as well as passenger and freight. The team made the deci- sion not to create a global integrative connectivity index. Further thinking is necessary to determine whether an integrated index makes sense and if so, which methodology to use. Merging time and costs can be done using gener- alized cost methods, but difficulties emerge with properly defining the oppor- tunity costs for all the cases. Among others, the principal component approach could be used to aggregate elements into a unique connectivity measure per country. Having three indicators for domestic and regional connectivity is prob- ably the highest level of aggregation to hope for at this stage. Finally, the resulting indexes in this report can be used to answer key analytic questions about the role of transport services in boosting social and economic outcomes. A gender perspective would help assess to what extent men and women have the same opportunities in using transport services. Obstacles for women to efficiently use transport services can lower their economic opportuni- ties. Measuring the penetration of transport services in the whole territory can be used to gauge the extent to which less connected populations benefit less from the gains of border opening and cheaper products (Atkin and Donaldson 2015). In addition to trade opportunities, transport services also matter for peo- ple to access jobs and different services such as education and health. It could also help for cost-benefit and country-specific analyses. Thinking about strate- gic connectivity is particularly important for countries that are landlocked, like most countries in Central Asia. Measures of centrality and criticality help provide an understanding of the dynamics of the whole network instead of only focusing on links to neighboring countries or large hubs. In general, these indexes pro- vide quantified tools to help develop thinking about sectoral and country strategies. The natural extensions of this chapter include considering other groupings or adding countries outside ECA to the sample pool. As it is, the chapter mostly looks at three main economic poles or gravity centers in ECA (Russia, Turkey, and the EU) or to reach groupings defined by economic criteria. Further work could be done to analyze other key groupings defined by their historical or cultural links or prospective impact. For example, the countries of the Commonwealth of Independent States and their historic links or the involvement of China and its role in promoting projects within the BRI could be explored. China is important for most ECA countries as a major trading partner or a neighbor with large potential for trade and investment. Including other regions and considering the connectivity of ECA countries with these other regions would add more insights. The analysis would also benefit from includ- ing connections with the United States and other subsets of the ECA region. Spain and Portugal have intense connectivity with Central and South American 234  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia countries. France has historical links with Maghreb countries. Central Asian countries have historical links with Middle Eastern and South Asian countries. A broader set of countries would enrich the connectivity analysis and allow for more meaningful benchmarking. Notes 1. The BRI is an ambitious project that will reach 65 countries and 4.4 billion people, and leverage 40 percent of the world’s GDP. Billions of dollars’ worth of investments will be channeled toward infrastructure projects across Asia, Africa, and Europe. Six new land corridors will be rehabilitated, and the maritime connectivity, which consists of a net- work of planned ports and other coastal infrastructure projects, will be improved. This project will be transformative for all cities along the new corridors and will bring unprec- edented opportunities for their economies. 2. The Trans-European Transport Network is an initiative of the EU consisting of hundreds of projects (studies and civil works). The project’s main purpose is ensuring the cohe- sion, interconnection, and interoperability of the network, as well as access to it. Once completed, the Trans-European Transport Network projects will touch every EU mem- ber state and all modes of transport (European Commission 2017). 3. Data were also collected for connections to main air and maritime hubs but are not included in this analysis because of representativeness issues. 4. Data collected are that of transport prices or out-of-pocket costs, and may be influ- enced by subsidies and affected by the degree of competition in transport services markets (e.g., the potential for collusive practices by service providers). 5. While the results are not presented here, domestic connectivity can appear very dif- ferent from the perspective of a global passenger than a domestic passenger. For global passengers, higher speeds are strongly correlated with higher unitary prices. The few exceptions include the Czech Republic, Kazakhstan, Poland, Ukraine, and Uzbekistan, where passenger transport services are subsidized, a legacy effect from the Soviet era. Based on these metrics, countries in ECA with the highest GDP per capita—such as France, Spain, Denmark, Holland, Sweden, Switzerland, and Germany—that are known for their advanced highway and railway systems (using speed as proxy) are affordable for local passengers but not necessarily for an aver- age passenger from the rest of ECA. In sharp contrast, countries like Azerbaijan, Georgia, FYR Macedonia, Armenia, Tajikistan, and Kosovo provide very low-priced passenger transport services for global passengers but with an apparent diminished quality (using speed as proxy). 6. INRIX traffic scorecard: http://www.dutchdailynews.com/netherlands-named-second​ -worst-country-for-traffic-congestion/. The high share of inland waterways in these countries can also be a factor for slower movement of cargo. 7. While results are not presented here, the story of neighbor connectivity can change significantly from the perspective of a global passenger rather a domestic passenger (if nominal unit costs are used instead of the unit costs adjusted by the purchasing power parity of the country of origin of the passenger). From the perspective of a global passenger, richer countries like France, Belgium, and the United Kingdom perform well in speed ranking but badly in cost ranking, and vice versa for poorer countries such as Armenia, the Kyrgyz Republic, Georgia, and Azerbaijan. The story shifts when unit costs are adjusted by purchasing power parity, as shown in the text. This has important impli- cations. Essentially in terms of affordability, passengers from advanced European econ- omies such as France, Belgium, Germany, and the Scandinavian countries—among others—are the most mobile in ECA, they can afford to travel to neighboring countries. Sadly, that is not the case for passengers from Central Asia, the South Caucasus, and the Western Balkans, who are largely trapped because of affordability issues. Infrastructure Linkages: Cost, Time, and Networks ●  235 8. The Democratic Republic of Korea and Mongolia are not included in the data. 9. Unfortunately freight costs for bulk cargo were not collected. 10. Earlier studies have shown that coastal countries may have poor accessibility if infra- structure (ports) is not adequate (i.e., only minor ports are located nearby or connec- tions are expensive). 11. Other important poles such as the United States and China are not considered here. More information is provided about connectivity to China in annex 5A. 12. The largest ECA economies that are considered here are France, Germany, Russia, the United Kingdom, and Italy. 13. A country’s level of production sophistication is defined using the index of complexity (Hausmann and Hidalgo 2014). The Economic Complexity Index measures the knowl- edge intensity of an economy by considering the knowledge intensity of the products it exports. This index is used to find, for each country, the eight economies with similar levels of production sophistication. Each country is compared with a different set of economies. The index only depends on the pattern of exports, which is used as a proxy for production factors, the stock of knowledge, the institutional and regulatory environ- ment, and the rest of production capacities. The most complex economies differ from the largest markets. According to the 2015 ranking of the MIT Atlas of Complexity, the five countries with the highest knowledge intensity are Switzerland, Austria, the Czech Republic, Germany, and Sweden. 14. How Central Asia connects with China, and how it can benefit from that economic grav- ity center, was, unfortunately, not included in this study, which limits the scope of analy- sis and data collection to ECA countries. This is a natural extension of this work and of the applicaiton of the proposed methodology to measure and assess connectivity. 15. Considering freight services toward China does not improve the connectivity of Central Asian countries. Prices to send containers toward Shanghai are much higher in Central Asian countries than in the rest of ECA (see “Extending the Set of ECA Countries to China” in annex 5A). 16. See chapter 1 for the analysis of the economic benefits of greater connectivity and the complementarity of different types of connections. 17. See “The Way Forward” in annex 5A. 18. “Properly managed” is an important nuance as a country cannot attract more traffic than it is able to handle domestically with its current infrastructure and services. Otherwise the infrastructure will collapse or start struggling with congestion. 19. Deichmann et al. (2004) note that summary statistics such as the total road length in a state or province or straight-line distance to ports or urban agglomerations are poor proxies for the complexity inherent in a national or regional transportation network. 20. If more time and resources were available, HDM-4 could be used to estimate more elaborate road and vehicle user costs considering aspects such as terrain, standard and class of each road link, and so on. For an exercise of this scope, the HDM-4 was not practical or affordable.   Out-of-pocket costs or prices faced by users reflect the quality of infrastructure and services. They also internalize issues pertaining to market structure of service providers and regulatory issues. Note that all these elements have an impact on the resulting “connectivity” faced by users. A natural follow-up analysis would be to assess the driv- ers of the proposed connectivity index, singling out as much as possible aspects per- taining to market structure, regulation, and institutions. Also, it is possible to use metrics in addition to time and costs, such as frequency and reliability, which could be the subject of further analysis. 21. In Russia ten cities were considered. In France, Germany, Italy, Kazakhstan, Poland, Spain, Turkey, Ukraine, and the United Kingdom, seven cities were included. In Greece, only cities on the mainland were considered. In Cyprus, Luxembourg, Malta, and Montenegro, only three cities were considered. For Germany, Kazakhstan, Switzerland, and Turkey, commercial centers (Frankfurt, Almaty, Zurich, Istanbul) were included instead of the capital cities. 236  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia 22. Other variables collected but not used in this specific analysis include connection fre- quency, and reliability (or variance) as from rush hour to non–rush hour time, or from daytime to nighttime.   In the original data collection, nodes were also defined as hubs. Two sets of hubs were defined. Frankfurt, Istanbul, and Moscow were selected as air hubs. Los Angeles, Rotterdam, and Shanghai were selected as the “illustrative” maritime hubs. This is per- haps the most controversial element of the proposed approach because the selection of hubs is exogenous to the data collection. Which cities are considered hubs, how many international hubs to include, and their regional representativeness are all ele- ments open to debate. For instance, Frankfurt, Istanbul, and Moscow are illustrative air hubs rather than representative ones. London, Paris, and Rome could have also been included. Given that the validaton of the pre-selected hubs was not widely discussed, the analysis of that data is excluded. 23. Albania, Armenia, Austria, Azerbaijan, Belarus, Belgium, Bosnia and Herzegovina, Bulgaria, Croatia, Cyprus, the Czech Republic, Denmark, Estonia, Finland, France, Georgia, Germany, Greece, Hungary, Iceland, Ireland, Italy, Kazakhstan, Kosovo, the Kyrgyz Republic, Latvia, Lithuania, Luxembourg, FYR Macedonia, Malta, Moldova, Montenegro, the Netherlands, Norway, Poland, Portugal, Romania, Russia, Serbia, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Tajikistan, Turkey, Turkmenistan, Ukraine, the United Kingdom, and Uzbekistan. 24. Other categories such as EU or European Economic Association membership, the size of the country, the density, and the level of inequality were considered. The final analy- sis only focuses on three dimensions. References Arrow, Kenneth J. 1969. “Classificatory Notes on the Production and Transmission of Technological Knowledge.” American Economic Review 59 (2): 29–35. Atkin, David, and Dave Donaldson. 2015. “Who’s Getting Globalized? The Size and Implications of Intranational Trade Costs.” Working Paper 21439, National Bureau of Economic Research, Cambridge, MA. http://www.nber.org/papers/w21439. Bahar, Dany, Ricardo Hausmann, and César Hidalgo. 2014. “Neighbors and the Evolution of the Comparative Advantage of Nations: Evidence of International Knowledge Diffusion?” Journal of International Economics 92 (1): 111–23. Briceno-Garmendia, C., H. Moroz, and J. Rozenberg. 2015. Road Networks, Accessibility, and Resilience: The Cases of Colombia, Ecuador, and Peru. An LCR Regional Study. Washington, DC: World Bank. Burghouwt, G., and R. Redondi. 2013. “Connectivity in Air Transport Networks: An Assessment of Models and Applications.” Journal of Transport Economics and Policy 47 (1): 35–53. Deichmann, U., M. Fay, J. Koo, and S. Lall. 2004. “Economic Structure, Productivity, and Infrastructure Quality in Southern Mexico.” Annals of Regional Science 38 (3): 361–85. European Commission. 2017. TEN-T Projects. https://ec.europa.eu/inea/en/ten-t​ /­ten-t-projects. Hausmann, Ricardo, and César Hidalgo. 2014. The Atlas of Economic Complexity: Mapping Paths to Prosperity. Cambridge, MA: MIT Press. Hummels, David L., and Georg Schaur. 2013. “Time as a Trade Barrier.” American Economic Review 103 (7): 2935–59. 6 Supply Chains in Europe and Central Asia: Connectivity through Cross- Border Production Fragmentation Cross-border production fragmentation, or specialization across stages of the ­ production process, has accelerated in recent decades. Many goods that were produced in single countries are now sliced in different bundles that are assigned to plants in different countries, as countries import intermediate inputs to combine them with domestic value added and reexport the whole product as a final good or as an input into the rest of the production process. This chapter is organized in three sections. The first addresses the rise in production fragmentation among European countries, creating the cluster “Factory Europe” that competes with “Factory Asia” and “Factory North America.” The second section investigates the different channels and implications of cross-country knowledge sharing when pro- duction is fragmented across countries, focusing on the existence of input-output linkages across sectors from different countries. The third discusses policies to increase the gains from participating in global value chains (GVCs). Main Messages • The expansion of supply chains has enabled more developing country firms to participate in the production of advanced products by specializing in one or several stages of the chain, and thus gain from foreign knowledge and learning by doing. Knowledge diffusion requires more direct forms of interaction than just trade and is heightened in supply chains that involve both movement of goods and know-how transfers, as well as movement of managers across stages 237 238  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia of production. Thus, despite large improvements in transport and information flows between countries, geographic proximity continues to be important to the coordination of input production as well as the transfer of “tacit knowl- edge.” Cultural proximity and migration networks have facilitated the develop- ment of successful regional supply chains in Europe. • Productivity growth in supply chains depends on the extent to which coun- tries are close to or well connected with the most productive countries to reap the benefits of cross-border knowledge sharing. Integration into the trade network and GVC participation are drivers of productivity growth. At the same time, the rise of trade in intermediate goods and services tends to increase the interdependence of countries and to make economies more volatile. • Different policies may be useful for increasing participation in supply chains through importing intermediate goods versus increasing exports of intermedi- ate goods. A large set of policies, from removing barriers to trade and foreign direct investment (FDI) to strengthening intellectual property protection and competitiveness reforms, are needed to participate and make the most of cross-border production fragmentation. Factory Europe Many supply chains can be described as “regional” production chains rather than global value chains, since they are composed of geographically proximate coun- tries. The production chains associated with auto parts trade in North America and the production and assembly of electronic components in Asia are the most famous ones. Other regional production chains are located in Europe and Central Asia (ECA). Siemens has divided its activities between engineering in Western Europe and assembly in Eastern Europe. French and German car makers have raised their productivity by offshoring part of their production in Eastern coun- tries (Gill and Raiser 2011; Marin 2010). Skoda in the Czech Republic makes high- tech components for Volkswagen, and Renault has opened assembly firms in Romania. This section focuses on “Factory Europe” and the importance of geographic proximity in the development of supply chains. Despite dramatic declines in trans- portation costs, proximity remains important in channeling knowledge and facili- tating the coordination of production stages across borders. In addition, geographic proximity is associated with similarities in culture and language, as well as migra- tion networks, that matter even more for production fragmentation than for tradi- tional trade. These diverse forms of proximity facilitate the movement of “tacit knowledge” along production stages across borders. Factory Europe in a Multipolar Economy The rise of production fragmentation as an important global phenomenon began in developed countries. For example, the 1965 Auto Pact between the United States and Canada increased trade and improved the efficiency of their auto-parts Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  239 supply chain. Between 1985 and 1995 trade in supply chains grew between high- tech and low-wage countries, referred to as “globalization’s second unbundling” (Amador and Cabral 2009; Baldwin and Lopez-Gonzalez 2015). Three regional clusters have emerged as zones of intense production fragmentation: “Factory Asia,” “Factory North America,” and “Factory Europe” (Baldwin and Lopez- Gonzalez 2015). These zones are composed of headquarters economies such as Germany, the United States, and China, which offshore part of their production to nearby “factory” economies with lower wages. Participation in supply chains is measured in terms of value-added flows rather than gross export flows across countries.1 Data on value-added flows from the Organisation for Economic Co-operation and Development (OECD) Trade in Value Added (TiVA) database2 can be used to construct visual network representations3 that capture the integration of countries into the global network of trade in value added (figure 6.1). The size of the country is given by a measure of its centrality in the network (figure 6.1 only shows the backbone of the network rather than all the flows across countries). In 2011, three groupings emerged around three central economies: Germany for the ECA region, China in Asia, and the United States for North America. The emergence of China as a main node for the Asian regional cluster is relatively recent when compared to the same network in 1995 in which the Asia cluster was FIGURE 6.1  Three clusters of countries emerge: “Factory Europe” around Germany, “Factory North America” around the United States, and “Factory Asia” around China Minimal spanning tree, value-added network, 2011 Source: Santoni and Taglioni 2015. Note: The minimal spanning tree is a network analysis technique that keeps only a subset of the links between all countries in the network of value-added flows. All countries are connected through a path that favors links that represent major flows of value added. Node size is given by the centrality measure in the network. 240  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 6.2  Smaller European countries, like the Czech Republic, are dominated by trade with Germany, while Germany is the headquarters of “Factory Europe,” which trades more globally a. Like that of several other countries, the Czech Republic’s trade is dominated by Germany b. Germany trades globally AUT AUT BEL BEL BGR BGR CHE CHE CHN CHN DEU CZE DNK DNK ESP ESP FIN FIN FRA FRA GBR GBR HUN HUN ITA ITA JPN JPN KOR KOR NLD NLD NOR NOR POL POL ROM ROM RUS RUS SVK SVK SWE SWE TUR TUR USA USA 0 10 20 30 40 0 2 4 6 8 10 Share of total flows (percent) Share of total flows (percent) Imports Exports Source: UN Comtrade data on the import and export structure of the Czech Republic and Germany, 2015. not apparent (Santoni and Taglioni 2015). Germany’s role as the headquarters economy of a large European cluster can be illustrated by comparing the export and import destinations for Germany and the Czech Republic (figure 6.2). Similar to other small European countries, the bulk of imports to and exports from the Czech Republic go to Germany, whereas Germany has a balanced portfolio of trade partners in the world. Asia and North America have similar patterns to those observed in Europe, although Europe has moved much more in terms of political and economic coop- eration, particularly under the European Union (EU). Germany, like the United States, does a great deal of supply-chain trade with its lower-wage neighbors. Several countries, such as the Czech Republic and Poland, import intermediate and final goods from Germany, and export intermediate products to Germany, where they undergo final processing for sale to third countries. However, Germany also engages in supply-chain trade with many other high-wage Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  241 neighboring nations (Austria, the Netherlands, and France). Unlike other regional supply chains, Factory Europe has three high-technology nations (other than the main node, Germany) with large manufacturing sectors: the United Kingdom, France and Italy. Production Fragmentation and Exports of Value Added as an Indicator of GVCs Production fragmentation has increased in Europe. Comparing data on trade mea- sured by gross and value-added flows shows the fragmentation of production through trade in intermediate goods. Traditionally, exports are simply measured by their gross value, which masks the value of imported component inputs.4 For example, a country that imports high-value-added inputs and assembles them into a final good may only provide a small domestic value-added contribution while exporting high-value goods. Trade statistics count the whole product as an export, whereas value-added trade flows only count the smaller domestic contribution. We use the ratio of gross exports to value-added exports to quantify the fragmen- tation of the production process due to the presence of supply chains.5 An increase in this ratio over time means that the importance of production fragmentation and supply chains are likely increasing. Overall, increased participation in supply chains (production fragmentation) is associated with more rapid growth in value added in exports. Production fragmen- tation (an increase in the ratio of gross exports to domestic value-added exports) increased between 2000 and 2011 for both Asia and Europe (but not in the North American Free Trade Agreement [NAFTA] region),6 with the largest increase in Asia (figure 6.3, panel a). At the same time, panel b of figure 6.3 shows that those regions with larger percentage point increases in production fragmentation also experienced greater average growth in overall value added in exports. Thus, higher growth in gross exports than in domestic value added is associated with more rapid growth of domestic value added in exports than in regions where the fragmentation of production occurred more slowly. EU advanced countries (EU15) and transition economies (EU13) are the second- and third-­ highest beneficiaries of this process, after Asia. Value added in exports is a better measure of the impor- tance of participation in international trade than gross value of exports, because it encompasses the potential benefits in terms of domestic employment and productivity growth. Greater production fragmentation in the EU13 countries and in Turkey was associated with more rapid growth of total value added in exports. Over the period 2000–11, Turkey and Poland experienced among the largest percentage increases in both production fragmentation (figure 6.4, panel a) and growth rate of exports of value added (figure 6.4, panel b). The Czech Republic, Bulgaria, and the Slovak Republic also experienced greater than average rates of increase in production fragmentation and exports of value added. In Romania, the increase in production fragmentation was moderate but exports of value added grew rapidly. Finally, Slovenia, the Russian Federation, and Hungary experi- enced decreases in production fragmentation and modest growth of exports of value added. 242  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 6.3  Higher a. Production fragmentation has increased in all regions except NAFTA production fragmentation due to supply chains is NAFTA associated with more rapid growth in value added in exports over 2000–11 EU15 EU13 Asia –5 0 5 10 15 Change in the ratio of exports to value added (percentage points) b. Growth in value added has increased most in Asia, but Europe is not far behind, and NAFTA has seen the slowest increases NAFTA EU15 EU13 Asia 0 5 10 15 20 Average annual growth of value added in exports (percent) Source: Organisation for Economic Co-operation and Development Trade in Value Added database. Note: NAFTA (the North American Free Trade Agreement) includes Canada, Mexico, and the United States. Asia includes China; Hong Kong SAR, China; Japan; Korea; and Taiwan, China. EU = European Union. Participation in Supply Chains: Backward and Forward Linkages The form of participation in supply chains differs in terms of the value added embodied in imports and exports (Hummels, Ishii, and Yi 2001; Koopman, Wang, and Wei 2012). Countries can participate in supply chains as importers of foreign value added produced in other countries (backward links—such as importing inter- mediate inputs) and exporters of domestic value added to other countries (forward links—such as exporting intermediate inputs). Backward links are measured as the percentage of foreign value added in domestic exports. Forward links are mea- sured as the percentage of domestic value added used as inputs in third-country exports. Countries that are specialized in sophisticated tasks that add more value to goods, like research and development, tend to have strong forward linkages. Countries that focus on low-value-added tasks, like assembly, tend to depend Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  243 FIGURE 6.4  Among the a. Production fragmentation, 2000–11 transition EU13 countries, HUN greater production RUS fragmentation is associated with a more rapid SVN increase in the flows of value ROU added in exports SVK DEU BGR POL CZE TUR –20 –10 0 10 20 30 Change in the ratio of exports to value added (percentage points) b. Exports of value added, 2000–11 SVN HUN DEU RUS CZE SVK BGR ROU POL TUR 0 5 10 15 20 Average growth per year (percent) Source: Calculations based on Organisation for Economic Co-operation and Development Trade in Value Added database. Note: Germany and Russia are included as reference countries. more on foreign value added and have higher backward linkages. For example, 32.4 percent of the gross exports of the EU28 countries are foreign value added (backward participation) whereas 20.8 percent of third countries’ exports are value added from the EU28 (forward participation) (table 6.1). The NAFTA region has the lowest share of f­oreign value added in gross exports (23.4 percent), or the lowest backward p ­ articipation, and the lowest share of domestic value added in third countries’ exports (19.7 percent), or the lowest forward participation. Such aggre- gate numbers can partly be explained by the size of countries in each zone. Bigger countries tend to trade less and to participate less in international supply chains 244  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE 6.1  The Importance of Imported Value Added in Exports (Backward Linkages) and Exported Value Added in Third Countries’ Exports (Forward Linkages) Differs by Region Backward and forward linkages, 2011 Backward linkages Forward linkages Foreign value added as a Domestic value added as a Region percentage of gross exports percentage of third countries’ exports NAFTA 23.4 19.7 EU28 32.4 20.8 Asia 34.5 20.9 Source: Organisation for Economic Co-operation and Development Trade in Value Added database. Note: EU = European Union; NAFTA = North American Free Trade Agreement. FIGURE 6.5  EU countries a. Growth of backward linkages, 2000–11 b. Growth of forward linkages, 2000–11 buy more foreign value added (backward linkages) NAFTA Asia than they sell to third countries (forward linkages) EU13 EU13 EU15 EU15 Asia NAFTA –5 0 5 10 15 20 0 10 20 30 Percent Percent Source: Organisation for Economic Co-operation and Development Trade in Value Added database. Simple averages across countries are calculated for each region. Note: EU = European Union; NAFTA = North American Free Trade Agreement. (OECD 2015a). Backward and forward linkages are the lowest for NAFTA, which gathers large ­countries with relatively lower participation in supply chains. These measures reflect different forms of involvement in value chains. A country that mostly assembles imported inputs (such as auto parts) into final goods and then exports them will have a high backward-participation index but a small forward-participation index. In contrast, a country that mostly produces intermedi- ate inputs (such as auto parts) and exports them to be assembled and reexported abroad will have a low backward-participation index but a high forward-­participation index (figure 6.5). Compared with other countries, EU13 countries tend to import relatively more foreign value added for their exports than they export domestic value added to third countries’ exports. On average, EU13 countries contribute less (than EU15 countries and NAFTA) to the production of the goods or services they export and rely more on foreign contributions (figure 6.5). They also contribute less to the production of exports of third countries. Between 2000 and 2011, all regions except NAFTA increased their participation in supply chains as importers of for- eign value added as well as exporters of domestic value added. The largest increase in backward linkages (left panel of figure 6.5) happened in Factory Asia Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  245 FIGURE 6.6  Participation a. Growth of backward linkages, 2000–11 b. Growth of forward linkages, 2000–11 in supply chains is MLT HRV heterogeneous among EST BGR ECA countries HUN LTU CYP LVA HRV CZE SVN DEU SVK SVK ROU ROU LVA POL LTU CYP CZE EST BGR SVN DEU HUN POL MLT –40 –20 0 20 40 –20 0 20 40 60 Percent Percent Source: Organisation for Economic Co-operation and Development Trade in Value Added database. and the largest increase in forward linkages (right panel) happened in Factory North America (NAFTA countries). Almost all ECA countries increased their supply-chain participation as exporters of domestic value added from 2000 to 2011, but only half have increased their participation as importers of foreign value added. Countries in Factory Europe, such as Poland, Bulgaria, and the Czech Republic, experienced the largest increase in participation as importers of foreign value added and an increase in participa- tion as exporters of domestic value added (figure 6.6). Other countries, such as Slovenia, Hungary, Malta, Cyprus, and Romania, mostly experienced large increases in participation as exporters of domestic value added. This may suggest that the recent integration of countries into the EU has increased their participation in ­ supply chains. The Role of Geographic Proximity Despite the fall in transport and information costs, geographic proximity continues to play an important role in production fragmentation across countries. Distance is a friction for both bilateral gross exports and value added in exports (Johnson and Noguera 2017). Overall, the fragmentation intensity elasticity with respect to dis- tance has remained stable since 1995, except for a dip in the first half of the 2000s for EU13 countries (figure 6.7).7 The effect of distance on production fragmentation remains important. While distance depresses the flows of both gross exports and value added in exports, gross exports fall more strongly with distance than do value-added exports— that is, the absolute value of the distance coefficient on gross exports is larger than the coefficient on value-added exports in all years (table within figure 6.7). The ratio of gross exports to value-added exports, that is, the index ­ 246  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 6.7  Supply chain 0 trade with close partners remains high –0.05 Elasticity of the ratio of gross –0.10 exports to value-added exports with –0.15 respect to distance –0.20 –0.25 –0.30 –0.35 1995 2000 2005 2008 2009 2010 2011 All countries EU13 EU15 a. Elasticity of flows of value added in exports World EU13 EU15 1995 –1.00 –1.35 –0.96 2005 –0.99 –1.24 –0.88 2011 –0.95 –1.17 –0.83 Percentage change from 1995 to 2011 4.80 13.47 13.55 b. Elasticity of flows of gross exports World EU13 EU15 1995 –1.27 –1.67 –1.17 2005 –1.29 –1.43 –1.03 2011 –1.28 –1.41 –0.99 Percentage change from 1995 to 2011 –0.31 15.53 15.22 Source: Data from the Organisation for Economic Co-operation and Development Trade in Value Added database. Note: The elasticities depicted in the figure result from a regression of each flow on the distance between origin and destination countries, with additional controls. Annex 6A provides additional details. “All countries” covers all countries included in the OECD database. of production fragmentation, decreases with distance (figure 6.7). Overall the tendency was an increase in the deterring effect of distance on the extent of production fragmentation in trade flows, suggesting an increase in regional sup- ply chains at the expense of global supply chains. Regarding ECA, the effect of distance is stronger for EU13 countries than for EU15 countries. Between 2000 and 2005, the patterns of participation in supply chains of EU13 and EU15 coun- tries converged as production fragmentation became less dependent on dis- tance for EU13 countries. One reason could be the entry of these countries into the EU and the deepening of supply chain links with richer EU countries like Germany. The role of regional trade agreements in increases in production frag- mentation among adopting partners has been shown for the EU and other agreements too (Johnson and Noguera 2017). Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  247 Why Does Distance Matter for Supply Chains? Despite the fall of transport and information and communication technology costs, geographic distance remains an important factor in shaping the information flows and connections required for supply chains. Two explanations are explored here: (a) the need for rapid transport, coordination, and agglomeration to support supply chains and (b) the association between distance and language or cultural similari- ties, as well as migrant diaspora proximity between the home and host countries. Transport, Coordination, and Agglomeration in Supply Chains Coordination costs, timeliness and agglomeration forces partly explain the continued importance of proximity. In order to understand firms’ location and trade decisions, three types of costs should be distinguished: transport costs, coordination costs, and wage costs. The fall of transport costs and lower foreign wages tends to increase the role of global supply chains, whereas higher coordination costs tend to reduce the importance of global supply chains and induce firms to locate production close to large markets. One difference between trade linked to production frag- mentation and traditional trade is the importance of coordina- Timeliness . . . is more tion costs in shaping the geographic patterns of trade. important across pro- Timeliness in the shipping and receipt of inputs is more impor- duction value chains tant across production value chains than it is for traditional final than it is for traditional goods trade. The absence of key intermediate inputs as a result final goods trade. of shipping delays or quality defects can have an important adverse impact on the whole supply chain (Hummels and Schaur 2013), and time costs are magnified by the number of stages in supply chains. Therefore, lead firms face trade-offs between reliability and cost effectiveness, particularly when choosing where to locate their activ- ities (Nicita, Ognivtsev, and Shirotori 2013; Taglioni and Winkler 2016). The benefits of agglomeration also play a role and may exceed the gains from using suppliers in more distant markets because they have lower wages and other natural endowments. Agglomeration externalities are due to better allocation of production factors across firms, easier transfer of knowledge across plants, and more efficient use of infrastructure and other public goods. Similar to results found in the “New Economic Geography” literature, lower- ing trade costs such as transport costs when they are high can produce a concentration of economic activity. The existence of tacit or noncodified knowledge transfers partly explains the importance of proximity for supply chains. Firms’ knowledge-based capital reflects their history of technology investments, their successes and failures, and the i­nteractions between their workers and other types of capital. Part of this knowledge, such as technological knowledge or knowledge that can be codified as standards or well-defined routines, can be replicated. Management know- how and techniques for reducing the cost of production can be transferred to suppliers. However, firm-specific knowledge can be difficult to replicate when it contains complex and abstract notions or when it is embodied in specific 248  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia employees or corporate systems (OECD 2013; Polanyi 1962). This “tacit knowl- edge” cannot be captured by blueprints or instruction manuals and requires more direct forms of human interaction than just trade to be diffused (Arrow 1969; Bahar, Hausmann, and Hidalgo 2014; Polanyi 1962). To avoid imitation by other firms, firms tend to increase the share of tacit knowledge and noncodified know-how (Thoenig and Verdier 2003). Thus, the exchange of complex tacit information accompanied by frequent face-to-face interactions is an important part of knowledge sharing across countries that participate in supply chains. Geographic Proximity Is Associated with Other Types of Proximity Other noneconomic and nongeographic issues, such as language, culture, social norms, or migration networks, may explain the continued importance of proxim- ity for participation in supply chains. Countries that are near each other often have other similarities, such as a common language, culture, or social norms. Migration networks and diaspora ties, which can be important Cultural and lan- guage proximity for trade and investment (Aleksynska and Peri 2014; Burchardi, makes it easier for Chaney, and Hassan 2016; Felbermayr and Toubal 2012; Gould managers to move 1994; Rauch and Trindade 2002), may also be stronger between and transfer knowl- countries located close to each other. Such ties may be more edge across stages important within supply chains than for traditional trade, given the of production and role of face-to-face interactions in transferring tacit knowledge for workers and firms across production stages. Cultural and language proximity makes it to absorb it. easier for managers to move and transfer knowledge across stages of production and for workers and firms to absorb it. In addition, the nature of the contracts between stages in supply chains (Antràs 2003; Helpman, Antràs, and Helpman 2008) and the often-differentiated nature of the products explain why such ties can be useful. When product specifications can- not be fully codified, so that cooperation relies on the transfer of tacit knowl- edge, it is difficult to write complete contracts to protect transactions. This increases the importance of reputation, social and spatial proximity, family and ethnic ties, and the like in managing transactions (Gereffi and Humphrey 2005). The Role of Regional Agreements The rise of regional agreements in the past two decades also has contributed to the development of regional supply chains. Integration agreements have been shown to be one of the forces behind the development of production linkages across coun- tries (Blyde, Graziano, and Volpe Martincus 2015; Hayakawa and Yamashita 2011; Johnson and Noguera 2017; Orefice and Rocha 2014). The 1965 US-Canada Auto Agreement, for example, was crucial to building supply chains involving US exports of engines and parts to Canada and Canadian assembly and exports of finished cars and trucks to the United States (Hummels, Rapoport, and Yi 1998). Deep regional agreements (those, like the EU, that cover many economic dimensions beyond trade policies—see chapter 7) have played an important role in the development of sup- ply chains. In addition to reducing customs barriers and the likelihood of future protectionist measures, deep regional agreements can promote production frag- mentation by encouraging FDI flows (see the spotlight to chapter 2). Figure 6.8 shows the rise in FDI flows to new EU entrants. Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  249 FIGURE 6.8  European FDI 80 inflows into the EU13 countries increased FDI inflow (billions of dollars) 60 dramatically after the 2003 entry of those countries into the European Union 40 Inflows of FDI into the EU13 20 0 1990 1995 2000 2005 2010 Year East Asia and Pacific Latin America and the Caribbean North America Europe and Central Asia Middle East and North Africa Sub-Saharan Africa Source: Calculations based on data from Peterson Institute for International Economics, Transactional FDI Dataset (created by Jacob F. Kirkegaard; see Kirkegaard 2013). Note: FDI = foreign direct investment. Are There Only Benefits from Increased Interdependence of Countries? Globalization’s second unbundling has not just involved more goods crossing bor- ders; it also has heightened international mobility of managerial and manufactur- ing know-how. When Volkswagen makes car parts in Poland, they do not only rely on local know-how; they bring Volkswagen technology, Volkswagen management, Volkswagen logistics, and other types of know-how needed. Polish-made parts must fit with precision into the German company’s production network. A country’s capacity to benefit from this knowledge sharing depends on its integration in trade, FDI, capital, and migration networks and its links with countries or sectors that are large producers of ideas or innovations. The extent to which countries benefit from these knowledge transfers also depends on their capacity to receive and absorb this knowledge. This section investigates the different channels of cross-country knowledge shar- ing linked to production fragmentation. It discusses whether the most central sectors or countries given their international input-output linkages are sectors or countries that are likely to produce new ideas. It also considers whether the increasing inter- dependence of countries through trade is conducive to more positive or negative shocks and increases the vulnerability of economies. Productivity Growth, Value-Added Growth, and Participation in Global Production Networks Firms exchange know-how through supply chains, which increases productivity and spurs innovation. Along the stages of production in different countries, the lead firms and the key suppliers exchange blueprints, technicians, managerial 250  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia practices, and productivity-enhancing techniques. Compared to trade in final goods, supply chains can facilitate the transfer of “tacit knowledge” and learning at a more rapid pace (Taglioni and Winkler 2016). Traditional trade of goods allows countries to learn new things through the technology embodied in the goods. In a supply-chain framework, slicing the production requires frequent interaction between the staff of the suppliers at one stage and the managers of the lead firms or those of firms at other stages of the production chain. Production fragmentation across countries can increase output, productivity, value added, and job creation through several channels. First, increasing supply chain participation can boost productivity of firms that have access to cheaper and better inputs (backward links) and boost the sales of domestic intermediates used in third countries’ exports (forward links). Productivity gains may also come from a finer division of tasks (similar to factor-augmenting technical change; Grossman and Rossi-Hansberg 2008), increased competition, and a greater diversity of input variet- ies. The entry of foreign firms or new opportunities of importing better inputs may have a procompetitive effect on domestic firms. Second, participation in supply chains leads to knowledge transfers through the movement of goods, capital, and people and skill upgrading through increased demand for skilled labor (Baldwin and Robert-Nicoud 2014; Li and Liu 2005). FDI spillovers from foreign affiliates to local firms are discussed in chapter 3. Finally, supply chain participation may increase investments in infrastructure that benefit the whole economy (see box 6.1). BOX 6.1 Global Value Chain Spillovers in Romania H.Essers and Oracle are two examples of foreign Oracle is a major multinational company head- companies investing in Romania that illustrate the quartered in the United States, specialized in benefits from foreign investments. developing and marketing database software and H.Essers is a leading European logistics firm technology, cloud-engineered systems, and enter- focusing on chemicals, pharmaceuticals, health care, prise software products. In the mid-2000s, it cre- and high-quality products, with headquarters in ated a branch in Bucharest and began to hire local Belgium. After its integration with a Dutch company software engineers for its routine software devel- located in Romania, the Belgian firm increased its opment. In addition to short-term spillovers, its presence in Romania, increasingly looking toward entry has spurred a new generation of entrepre- Eastern Europe and Central Asia. Knowledge and neurs, who got their start at Oracle in Bucharest, to know-how coming from traditional logistics hubs like create their own businesses. One of them is the Netherlands and Belgium benefited Romania by Softelligence, a Romanian software company improving its logistics performance. Logistics is the designing tailored mobile applications for financial backbone of supply chains, as it makes production institutions. The low cost of entry for new entrepre- fragmentation and the smooth coordination of its neurs in this industry, coupled with low wages, a stages possible. Knowledge spillovers happen qualified workforce, and an excellent internet net- through clients being educated about good practices work, has boosted this sector and diversified the on norms, information technology, and cold chains. economy. Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  251 Participation in value chains also is associated with greater labor productivity.8 Participation in supply chains is broader than just trade integration and contributes to labor productivity through multiple channels: FDI spillovers, exchange of mana- gerial know-how, or increasing competition, as discussed in earlier chapters. In OECD countries, growth in participation in supply chains is positively correlated with growth in real labor productivity by country and year (figure 6.9).9 An increase of 1 point of growth in GVC participation is associated with a significant increase of 0.27 point of growth in labor productivity.10 This association does not prove that participation in supply chains causes increased productivity, as more productive countries may also be more likely to participate in supply chains. Addressing endog­ eneity is difficult, although Constantinescu, Mattoo, and Ruta (2017) find that participation in GVCs boosts productivity. Based on a panel estimation cover- ing 13 sectors in 40 countries over 15 years, they find that participation in GVCs is a significant driver of labor productivity. Backward participation emerges as par- ticularly important. An increase of 10 percent in the level of GVC participation increased average productivity by 1.7 percent. They also suggest that trade that is not GVC related also has a positive impact on productivity, but the relationship is less robust. Which channels are the most important remains to be determined. Participation in value chains is also associated with higher domestic value added at the sector level (figure 6.10). Causality issues are addressed by Kummritz (2016), who confirms a positive effect of GVC participation for sectors. In addition, GVC participation measured by both forward- and backward-linkage indicators generates robust gains for participating countries with a larger effect coming from forward linkages. Such gains are also independent of country’s per capita income level (Kummritz 2016). This sheds light on current debates on the transmission FIGURE 6.9  Participation in 0.10 global value chains is correlated with higher labor productivity Growth in labor productivity (percent) 0.05 Growth in labor productivity versus growth in global value chain participation, 2009–11 0 –0.05 –0.10 –0.20 –0.10 0 0.10 Growth in GVC participation (percent) Source: World Bank labor productivity data and country global value chain participation index for member countries of the Organisation for Economic Co-operation and Development over the period 2009–11. Note: Each dot in the figure represents one country for one year. GVC = global value chain. 252  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 6.10  Participation 150 in global value chains is correlated with higher domestic value added at the Predicted growth in value added (percent) sector level 100 50 0 –50 –50 0 50 100 150 Lagged growth in GVC participation (percent) Source: Sectoral global value chain and domestic value-added data for Organisation for Economic Co- operation and Development (OECD) countries for 1995, 2000, 2005, 2008, 2009, and 2011 from the OECD’s Trade in Value Added database (ISIC 3). Note: Each dot represents a sector-country-year pair of global value chain (GVC) participation and value added. GVC participation is defined as the sum of the backward and forward linkages. The predicted growth of value added results from a simple ordinary least squares regression with industry and country fixed effects. The figure also reports the confidence interval surrounding the linear fit. Growth in GVC participation is lagged to minimize the reverse-causality problem. channels between GVC participation and development as well as some current debates on the nature of the gains or losses from GVC participation. Input and Output Linkages: An Increasing Interdependence of Countries Greater participation in supply chains can increase firms’ dependence on other sec- tors and economies. The acceleration of production fragmentation has created a global economy in which some sectors increasingly rely on other sectors through the supply of inputs for their own production and exports. Recent studies have focused on the fact that significant aggregate fluctuations may originate from sector or firm shocks (Carvalho et al. 2012). Economic transmission can happen through demand- side shocks (to input-supplying industries) or through supply-side shocks (to cus- tomer industries). The structure of economies in which sectors are becoming more interdependent might create “cascade effects” where shocks can propagate to a large part of the economy and across borders. Recently, “the Dieselgate” scandal not only directly affected the activities of Volkswagen in Germany, but also indirectly affected the multiple suppliers of its production chain in Germany and in other European countries. Large multinationals whose external suppliers and affiliates are spread all over European countries are a major channel for propagation of shocks. This section examines the interdependence of sectors through their participa- tion in production networks. If the production network is dominated by a few sec- tors, fluctuations in these sectors are likely to propagate and affect economic Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  253 FIGURE 6.11  Sectors from Food and Retail in France advanced EU15, transition beverages in Retail in Italy France EU13, and non-EU countries Retail in are interdependent, but the Germany most central sectors are from EU15 countries Machinery in Germany Retail in Russian Federation Chemistry in Transport and Germany storage in Germany Motor vehicles, trailers, and semitrailers in Germany Construction in Germany Services in France Note: The figure shows a network visualization of the 2011 input-output network for Europe and Central Asia (ECA) member countries of the Organisation for Economic Co-operation and Development using the minimum spanning tree method. EU15 countries are shown in yellow, EU13 countries in orange, and ECA non-EU countries in red. Node size is determined by the PageRank measure in the ECA production network. The most central sectors in the ECA production network are motor vehicles, trailers, and semitrailers in Germany; wholesale and retail trade in Italy; machinery and equipment in Germany; wholesale and retail trade in Germany; and wholesale and retail trade in France. aggregates.11 In addition to the benefits from knowledge transfers and increased trade opportunities, increased production fragmentation could increase the vola- tility of sectors or large economies, driven by shocks coming from different sectors or from different countries. It is therefore important for countries to adapt their economic and social structures to a potential for increasing volatility. Finding the central sectors and the major cross-border links is impor- tant to gaining an understanding of how positive or adverse shocks spread through production networks in the ECA region. We can use the Finding the central sectors and the major tools of network analysis to determine the extent to which sectors and cross-border links is countries are central and influence this network. We compute a mea- important to gaining an sure of influence in the network similar to that used in chapter 1, the understanding of how PageRank centrality, based on the flows of inputs going from one positive or adverse 12 shocks spread country-sector pair to another. A country or a sector that is central through production might be able to spread ideas to the rest of the network, but might also networks. more frequently receive shocks from the rest of the network. The ECA production network is organized around several clusters that include sectors from different parts of the region (figure 6.11). Each color 254  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE 6.2  What Sectors or Countries Are Expected to Have the Largest Impact on the Rest of the ECA Economies When They Face Either a Positive or a Negative Shock? Country outside Ranking Country/sector Country/sector outside EU15 Country (average) EU15 (average) Sectors (average) 1 Germany/vehicles Russian Federation/retail Germany Russia Retail 2 Italy/retail Poland/retail Italy Turkey Construction 3 Germany/machinery Russian Federation/ France Poland Transport construction 4 Germany/retail Turkey/retail Russian Czech Food and Federation Republic beverages 5 France/retail Russian Federation/ Turkey Croatia R&D transport 6 Russian Federation/retail Turkey/transport Spain Lithuania Machinery 7 Germany /construction Turkey/food and beverages Sweden Latvia Government 8 Germany/food Russian Federation/ Poland Cyprus Hotels/ government restaurants 9 France/R&D Poland /construction Finland Bulgaria Real estate 10 France/ construction Poland/food and beverages Belgium Hungary Others Source: Organisation for Economic Co-operation and Development Structural Analysis database, 2011. Note: EU = European Union; R&D = research and development. represents one of the three regions of ECA (EU15, EU13, and the non-EU area). Having sectors from different regions in the same production cluster illustrates the interdependence of country-sectors across most countries of ECA through input- output linkages. Table 6.2 ranks the most important pairs of country-sectors in the ECA region and in the ECA region excluding EU15 countries. It also indicates the most central countries and sectors in the ECA production network. Interestingly, the motor vehicle sector in Germany is the most central in the ECA production network. This sector largely relies on regional value chains to organize its produc- tion. The retail sectors in Italy, Germany, France and Russia are all very central. The machinery and equipment sector in Germany is also among the most central sec- tors. Outside of the EU15 countries, sectors in Russia, Turkey, and Poland appear the most central in the production network. Germany, Italy, and France are the most central countries in the ECA trade production network, followed by Russia and Turkey (table 6.2, map 6.1). The least central countries are Portugal, the Baltic countries, and Eastern European countries. Interdependence in the Global Network and Volatility The intensification of trade links between sectors across the ECA region has increased the interdependence of sectors. Countries or sectors that are more inte- grated into a trade network (measured by using network analysis tools to compute an integration index for sectors in the input-output trade network) tend to have output growth rates that are more correlated (annex 6B and figure 6.12).13 Thus, trade integration seems to increase output interdependence across sectors. Indeed, a sector that produces goods or services that are increasingly demanded will import more inputs from other sectors. A sector producing intermediate goods that experiences a positive productivity shock will export cheaper or better inputs. By contrast, a negative shock in a final-product sector or an intermediate-product sector will negatively affect the sectors that are using these products or selling products to them. Production fragmentation by dividing stages of production Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  255 MAP 6.1  Which countries are the most central in the ECA production network? Average centrality per country, 2011 High Medium Modest Low Note: The map shows average centrality measured across sectors at the country level: the darker the color (and thus the higher the number for centrality), the higher the importance of the country in the Europe and Central Asia (ECA) network, given sector linkages. Some ECA countries are not included in the analysis for data availability reasons. FIGURE 6.12  Sectors that 0.5 are more integrated in a production trade network move more together ECA region Predicted growth correlation 0.4 0.3 0.2 0 100 200 300 400 500 Distance in the input-output network Source: Calculations using data from the Organisation for Economic Co-operation and Development. Note: The red line in the figure represents the best linear fit. A shorter distance in the input-output network means that sectors are more integrated. 256  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 6.13  Imports of 150 intermediate goods are more volatile than imports Imports (billions of constant 2010 US dollars) of final goods 100 50 0 1970 1980 1990 2000 2010 Intermediate goods Final goods Source: United Nations Commodity Trade Statistics Database data using Broad Economic Categories classifications. across countries tends to increase the interdependence of sectors across borders. Economies might therefore become increasingly vulnerable to external shocks. Finally, trade in intermediate goods tends to be more volatile than trade in final (capital and consumption) goods (Sturgeon and Memedovic 2011). From 1967 to 2014, imports of intermediate goods appear to have been more volatile than trade in final goods, given the higher variation observed during major crises (the oil shock of 1979, the Asian crisis, and the global financial crisis in 2008) or sectoral bubbles (for example, the 2001 internet bubble) (figure 6.13). This supports the notion of “bullwhip” effects of recessions and business cycles. Parts and compo- nents shipments are more affected than final goods shipments, because final goods producers tend to draw down parts inventories and delay reordering during periods of uncertainty (Escaith, Lindenberg, and Miroudot 2010). Different Policies for GVC Upgrading Participating in value chains can be growth enhancing, especially by creating channels for knowledge transfers, but not all countries have fully benefited from the rise of cross-border production fragmentation. Different types of economic upgrading and participation in value chains call for different policies and depend on each country’s stage of development. Policies that play a role in economic upgrading should reinforce physical connectivity (infrastructure, trade, and invest- ment policies), improve the productivity of labor (education and skills policies), and improve the overall domestic economic environment (business climate, labor market flexibility, financial institutions, and so on). Diverse policies support increasing supply chain integration by strengthening backward and forward linkages (Kummritz and Quast 2016; Kummritz, Taglioni, Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  257 and Winkler 2017). A wide spectrum of policies can play a role for increasing GVC participation. Importing more foreign inputs requires increasing connectivity by improving infrastructure as well as trade and investment policies (backward partici- pation). Exporting domestic value added that is integrated into third countries’ exports (forward participation) requires countries to increase productivity to be more competitive in the global marketplace (Kummritz, Taglioni, and Winkler 2017). Higher GVC participation is associated with a higher share of manufactur- ing, better logistics, and lower trade barriers (see annex 6c for empirical results). Large countries tend to have lower participation in GVCs. Beyond trade or invest- ment policies, broader policies are necessary to upgrade GVC participation, in order to increase productivity or domestic value added. Upgrading forward link- ages, rather than backward linkages, contributes more to increasing the broad economic gains (Kummritz, Taglioni, and Winkler 2017). Higher forward linkages are strongly associated with better Doing Business indicators. Other nontraditional policies, such as business climate and institutions, financial development, labor market policy, education and skills, product standards and innovation, as well as labor, social, and environmental standards, have been shown to play a role in upgrading value chain participation (Kummritz, Taglioni, and Winkler 2017). Policies for Countries in Factory Europe Different policies are appropriate for countries at different stages of GVC upgrad- ing. Figure 6.14 shows that most countries are more integrated as buyers of foreign value added (backward linkages) than as sellers of domestic value added (forward linkages). Some countries, such as Croatia and Romania, have low backward partici- pation. Improving their connectivity and their trade and investment policies could lead to higher backward linkages. Some countries, like Hungary, the Slovak Republic, the Czech Republic, Bulgaria, and Slovenia, have high backward partici- pation but lower forward participation. If they target increasing their forward FIGURE 6.14  Many EU13 Croatia countries have high Cyprus backward participation but Lithuania low forward participation Romania Backward- and forward- Germany participation indexes across Latvia countries, 2011 Poland Estonia Slovenia Malta Bulgaria Czech Republic Slovak Republic Hungary 0 10 20 30 40 50 60 Forward Backward Source: Organisation for Economic Co-operation and Development Trade in Value Added database. 258  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia linkages, the focus should be on improving the productivity of their firms to increase the exported value added in third countries’ exports. For the rest of the ECA countries, participation in supply chains remains limited. While EU13 countries and Turkey have increased their imports of intermediates as well as their exports, other countries tend to import many final goods and export relatively little outside of raw materials. Figure 6.15 looks at three key GVCs that are important for the ECA region: apparel and footwear, electronics, and automotive goods. The electronics and automotive industries have been extremely important drivers of supply chain development (Sturgeon and Memedovic 2011). Different patterns of integration emerge across the different regions in ECA (excluding the high-income countries). EU13 countries and Turkey have increased their participa- tion in the automotive and electronics production chains through a rise in both imports and exports and in both intermediate and final goods. The largest exports are for final electronics, final vehicles, and intermediate vehicles. In contrast, ECA non-EU countries have substantially increased their imports, mostly in final electronics and final vehicles, but have only slightly increased their exports in these manufactured goods. They also import few intermediate products. This pattern differs greatly from the integration pattern observed in the EU13 coun- tries and Turkey, which have successfully integrated into regional and global supply chain trade. Despite large heterogeneity of firms’ supply chain participation across the ECA region, both domestic and foreign firms have stronger backward linkages as buy- ers of value added than forward linkages as sellers of value added. At the firm level, figure 6.16 shows the intensive and extensive margins for both imports (left panels) and exports (right panels) for domestic versus foreign firms. In most regions FIGURE 6.15  ECA Non-EU countries mainly import final goods and export little Imports and exports for ECA non-EU countries a. Imports b. Exports 80 80 60 60 Imports (billions of dollars) Exports (billions of dollars) 40 40 20 20 0 0 1990 1995 2000 2005 2010 2015 1990 1995 2000 2005 2010 2015 Final apparel and footwear Intermediate apparel and footwear Final electronics Intermediate electronics Final textiles Intermediate vehicles Final vehicles Source: World Bank, Global Value Chain database. Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  259 FIGURE 6.16  Heterogeneous participation in trade of domestic and foreign firms a. Share of importers and share of inputs imported Share of firms using foreign inputs Share of foreign inputs Balkans Balkans Central Asia Central Asia Central Europe Central Europe Other Eastern Europe Other Eastern Europe Russian Federation Russian Federation South Caucasus South Caucasus Turkey Turkey 0 20 40 60 80 100 0 20 40 60 b. Share of exporters and share of exported production Share of firms exporting Share of sales exported Balkans Balkans Central Asia Central Asia Central Europe Central Europe Other Eastern Europe Other Eastern Europe Russian Federation Russian Federation South Caucasus South Caucasus Turkey Turkey 0 20 40 60 80 100 0 20 40 60 80 Domestic firms Foreign firms Source: World Bank, Enterprise Survey database. (excluding the EU15 countries), a large share of foreign-owned firms’ inputs are foreign inputs. However, few of them export, and they export a small part of their production. Overall, foreign firms tend to first target the domestic market rather than looking to reexport their production. This shows that the primary purpose of these foreign firms is not to be involved in global production chains. One excep- tion is that most foreign firms in Turkey export, and they export a large share of 260  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia their production (figure 6.16, panel b). They seem to be well integrated in supply chains, especially in the automobile and textile industries. Except for Turkey, domestic firms in ECA rely more on importing foreign inputs than on exporting their production (they have stronger backward linkages than forward linkages). This reflects a lack of competitiveness of domestic firms in many ECA countries that do not or cannot compete with other firms in foreign markets. Most ECA countries outside the EU need to improve their connectivity, their trade and investment policies, and their business climates. Figure 6.17 shows the current levels of policies regarding trade, logistics, and the business climate. Central Asian countries (Tajikistan, Kazakhstan, and the Kyrgyz Republic) as well as Russia and Ukraine perform poorly compared with the rest of ECA. In this group, the Kyrgyz Republic performs better in terms of trade policies, while Russia and Kazakhstan have better business climates. Turkey is an outsider that should improve its business climate and its trade policies. A second cluster is formed by the Balkans and South Caucasus countries (FYR Macedonia, Serbia, Montenegro, Albania, Georgia, and Armenia). They overall have good trade policies but poor-quality logistics. The busi- ness climate varies a lot in this group, with Albania and Serbia having the poorest regulatory environment. Eastern European EU members (the Czech Republic, Hungary, the Slovak Republic, and Slovenia) could still improve their business cli- mates, with Doing Business indicators between 70 and 80 on a 100-point scale. Compared with the others, Bulgaria performs poorly in terms of logistics perfor- mance. Such various policies are complementary to GVC participation to increase the broader economic gains for each country (Kummritz, Taglioni, and Winkler 2017). FIGURE 6.17  Many ECA CZE countries can still reduce LTU trade barriers and improve 3.5 TUR HUN logistics and the business POL LVA EST environment to increase SVK Logistics performance index their supply chain SVN HRV participation 3.0 Trading across Borders, BGR Logistics Performance, and KAZ Doing Business Indexes UKR RUS MDA BIH 2.5 MKD GEO BLR ALB ARM KGZ TJK 2.0 60 70 80 90 100 Trade performance (distance to the frontier) Doing Business Index DB−DTF<=50 6090 Source: World Bank, Trading across Borders, Logistics Performance, and Doing Business Indexes (most recent data available). Note: The Doing Business (DB − DTF) and Trading across Borders Indexes are given by the distance to the frontier index for each country. A higher number for a country means that it is closer to best practices. Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  261 Conclusion Cross-border production fragmentation has increased connectivity across countries in Europe and Central Asia. Countries can benefit from being better connected by improving their access to ideas and innovations that support economic growth. Supply chains not only increase trade in goods or services, but also enhance move- ment of capital, people, and ideas. In particular, they promote the transfer of “tacit knowledge” across production stages as well as more traditional forms of knowl- edge sharing through increasing participation in the trade and FDI global networks. Policies to increase supply chain participation should be tailored to the needs and particularities of each country. However, an increasing interdependence of countries might also tend to increase the volatility of their economies. Complementary poli- cies should be adopted to minimize the risks from increased global interdepen- dence to fully reap the benefits of participation in supply chains. Annex 6A. Elasticities of Value Added in Exports, Gross Exports, and Fragmentation Intensity Following Johnson and Noguera (2017), this annex describes how changes in bilat- eral value added exports versus gross exports are shaped by bilateral trade frictions. The analysis here focuses on one common proxy for bilateral frictions: distance. FIGURE 6A.1  Elasticities of exports of value added and gross exports, and of the ratio of gross exports to value added, for the EU13, EU15, and all countries a. Elasticities of exports of value added and gross exports 0 –0.00005 –0.00010 –0.00015 –0.00020 –0.00025 –0.00030 –0.00035 –0.00040 –0.00045 –0.00050 1995 2000 2005 2008 2009 2010 2011 Exports of value added: all countries Exports of value added: ECA region Gross exports: all countries Gross exports: ECA region continued 262  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 6A.1  continued b. Elasticity of the ratio of gross exports to value added 0 –0.05 –0.10 –0.15 –0.20 –0.25 –0.30 –0.35 1995 2000 2005 2008 2009 2010 2011 All countries EU13 EU15 Source: Organisation for Economic Co-operation and Development Trade in Value Added database. Note: Elasticities of value added and exports with respect to distance (dist) are given by ∂(VA ) / VA εVAX = = βVA , ∂ dist / dist ∂(EXP ) / EXP εX = = βX . ∂ dist / dist The elasticity of the production fragmentation intensity with respect to distance (dist) is defined by the following formula:  EXP  EXP ∂   VA  VA εVAX = = βVAX = β X − βVA . ∂ dist dist Elasticities are measured for all countries and for ECA countries only. To measure the elasticity of production fragmentation with respect to distance, we look at how gross exports (xijt), value-added exports (vaijt), and gross-exports- to-value-added-in-exports ratios from country i to country j at time t respond to bilateral distance. To answer these questions, we estimate gravity-style regressions for each of the three variables of interest: ( ) y log y ijt = φit y + φjt ( ) + βty log distij + εijt . { y y The terms y ijt ∈ xijt , vaijt ,VAXijt , φit , φjt }{ } are importer-year and exporter- year fixed effects and βty the time-varying coefficient on bilateral distance (distij) for outcome yijt. Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  263 Annex 6B. Interdependence of Countries Using the OECD Input-Output database, distance measures are given by the closeness network measures in the input-output network in 2005. Correlations across sectoral output growth come from the OECD Structural Analysis database and cover the years from 1995 to 2011. The regression table includes the measure of supply chain integration from the 2005 input-output network and shows its correlation over the whole period with sectoral value-added growth. Country, sec- ­ tor, and interaction sector dummies are added as controls. TABLE 6B.1  Sectors That Are More Integrated in the Production Network Are More Correlated Sector comovement Variable (1) (2) Integration index 1.58e−05* 1.63e−05* (9.50e−06) (9.47e−06) Constant 0.275*** 0.379*** (0.00881) (0.0139) Number of observations 76,452 76,452 R2 0.047 0.069 Country dummies Yes Yes Sector dummies Yes Yes Interaction dummies No Yes Note: Robust standard errors are in parentheses. Comovement is measured by the correlation of sectoral outputs over the chosen period. The integration index measures the distance in the input-output network based on observed contributions of one sector in another. Significance level: * = 10 percent, *** = 1 percent. Annex 6C. Regression of Backward- and Forward- Participation Indexes over a Set of Policy Variables The variables to explain are the backward- and forward-participation indexes using the OECD Trade in Value Added database. • Country variables: gross domestic product (GDP), the size of the manufacturing sector, the population (POP), the total tax rate as a percentage of commercial profits, the minimum distance to a headquarters economy (Germany, China, or the United States). • Connectivity variables: the quality of logistics (the Logistics Performance Index), trade policies measured by Trading across Borders indexes, investment policies measured by FDI restrictions. • Business Climate variables: the Doing Business index. • Year fixed effects: 2008–09–10–11. • Countries covered: EU countries. 264  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE 6C.1  Variables for Global Value Chain Participation and Forward Linkages Global value chain Variable participation Forward linkages Log of GDP −3.604*** 0.724 Share of manufacturing in GDP 0.415*** −0.007 Population −2.11e−08 7.03e−09 Simple tax policies −0.025 0.071* Geographic distance −0.003 −0.001 FDI restrictions −23.623 2.497 LPI 10.53*** −4.027* Trading across Barriers −0.444*** −0.165 Doing Business −0.037 0.447*** Contiguity to Germany 1.647 1.723 Constant 155.5** −5.493 Number of observations 95 95 R2 0.633 0.415 Time fixed effects Yes Yes Note: FDI = foreign direct investment; LPI = Logistics Performance Index. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. Notes 1. Final goods and services are composed of inputs from several countries. The flows of goods and services within supply chains are not reflected in conventional measures of international trade. 2. Recent initiatives to measure supply chain activity using harmonized intercountry input- output tables (OECD 2015b; Timmer et al. 2014) led to the release of the OECD TiVA database in 2013. It provides a decomposition of gross trade flows into domestic and foreign value added. 3. Santoni and Taglioni (2015) use the Katz-Bonacich metrics as a measure of integration and show a simplified version of the whole network using the minimal spanning tree method. 4. To measure the flows of value added, national input-output tables are linked together using bilateral trade data to form a global input-output table that shows both final and intermediate good shipments between countries. All domestic contributions are tracked to determine the value-added content of exports until the final good reaches the final demand. 5. The inverse ratio of value-added exports to gross exports can also be found in the lit- erature on supply chains. 6. Production fragmentation started earlier in NAFTA than in the other regions. In addi- tion, trade flow measures are biased by the fact that the United States is only one country and interstate trade is not considered. 7. Annex 6A shows the details of the computation. 8. We consider total factor productivity (TFP) and labor productivity separately because of the difficulties in measuring productivity (for example, indexes of TFP suffer from mea- surement errors). Since TFP and labor productivity are measured differently, showing that both are related to supply chain participation provides greater confidence in empirical findings. 9. Participation in supply chains is measured as the sum of the foreign value added embodied in exports (backward linkages) and the domestic value added in exports that the direct importer exports further or that returns home as imports (forward linkages). Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  265 10. This results from regressing growth in labor productivity on growth in GVC participation with country fixed effects over the given period. The coefficient of the regression is significant and equal to 0.27. 11. A recent strand of the economic literature has studied how the structure of domestic production networks can affect aggregate performance (Acemoglu and Jensen 2015; Carvalho 2014). A few recent studies have focused on the importance of interconnec- tions between firms to understand how micro disturbances can affect macro perfor- mance (Carvalho and Grassi 2015). For example, the 2007–09 global financial crisis showed how the linkages between financial institutions contributed to the impact on economic growth and unemployment in most ECA countries. Other references on global supply chains include Antràs and Chor 2013; Chaney 2014; Costinot, Vogel, and Wang 2013; and di Giovanni and Levchenko 2012. 12. The contribution of the sector to another sector determines the strength of a link in the global network. We use the OECD input-output network in 2011 to highlight the most influential sectors, and then the most influential countries in this network. 13. Distance measures are given by the network closeness measures in 2005. Correlations across sectoral outputs come from the OECD Structural Analysis database and cover the years from 1995 to 2011. For more details, see annexes 6B and 6C. References Aleksynska, M., and G. Peri. 2014. “Isolating the Network Effect of Immigrants on Trade.” World Economy 37 (3): 434–55. https://doi.org/10.1111/twec.12079. Amador, J., and S. Cabral. 2009. “Vertical Specialization across the World: A Relative Measure.” North American Journal of Economics and Finance 20 (3): 267–80. https:// doi.org/10.1016/j.najef.2009.05.003. Antràs, P. 2003. “Firms, Contracts, and Trade Structure.” Quarterly Journal of Economics 118 (4): 1375–1418. Antràs, P., and D. Chor. 2013. “Organizing the Global Value Chain.” Econometrica 81 (6): 2127–204. https://doi.org/10.3982/ECTA10813. Arrow, K. J. 1969. “Classificatory Notes on the Production and Transmission of Technological Knowledge.” American Economic Review 59 (2): 29–35. Bahar, D., R. Hausmann, and C. Hidalgo. 2014. “Neighbors and the Evolution of the Comparative Advantage of Nations: Evidence of International Knowledge Diffusion?” Journal of International Economics 92 (1): 111–23. Baldwin, R., and J. Lopez-Gonzalez. 2015. “Supply-Chain Trade: A Portrait of Global Patterns and Several Testable Hypotheses.” World Economy 38 (11): 1682–721. https:// doi.org/10.1111/twec.12189. Baldwin, R., and F. Robert-Nicoud. 2014. “Trade-in-Goods and Trade-in-Tasks: An Integrating Framework.” Journal of International Economics 92 (1): 51–62. https://doi​ .org/10.1016/j.jinteco.2013.10.002. Blyde, J., A. Graziano, and C. Volpe Martincus. 2015. “Economic Integration Agreements and Production Fragmentation: Evidence on the Extensive Margin.” Applied Economics Letters 22 (10): 835–42. https://doi.org/10.1080/13504851.2014.980569. Burchardi, K. B., T. Chaney, and T. A. Hassan. 2016. “Migrants, Ancestors, and Investments.” Working Paper 21847, National Bureau of Economic Research, Cambridge, MA. Carvalho, V. 2014. “From Micro to Macro via Production Networks.” Journal of Economic Perspectives 28 (4): 23–48. https://doi.org/10.1257/jep.28.4.23. 266  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Carvalho, V., D. Acemoglu, A. Ozdaglar, and A. Tahbaz-Salehi. 2012. “The Network Origins of Aggregate Fluctuations.” Econometrica 80 (5): 1977–2016. Carvalho, V., and B. Grassi. 2015. “Large Firm Dynamics and the Business Cycle.” Working Papers in Economics 1556, Faculty of Economics, Cambridge University, Cambridge, UK. Chaney, T. 2014. “The Network Structure of International Trade.” American Economic Review 104 (11): 3600–34. https://doi.org/10.1257/aer.104.11.3600. Constantinescu, C., A. Mattoo, and M. Ruta. 2017. “Does Vertical Specialization Increase Productivity ?” Policy Research Working Paper 7978, World Bank, Washington, DC. Costinot, A., J. Vogel, and S. Wang. 2013. “An Elementary Theory of Global Supply .org/10.1093/restud​ Chains.” Review of Economic Studies 13 (80): 109–44. https://doi​ /­rds023. di Giovanni, J., and A. A. Levchenko. 2012. “Country Size, International Trade, and Aggregate Fluctuations in Granular Economies.” Journal of Political Economy 120 (6): 1083–132. https://doi.org/10.1086/669161. Escaith, H., N. Lindenberg, and S. Miroudot. 2010. “International Supply Chains and Trade Elasticity in Times of Global Crisis.” MPRA Paper 20478, University Library of Munich, Munich, Germany. Felbermayr, G. J., and F. Toubal. 2012. “Revisiting the Trade-Migration Nexus: Evidence from New OECD Data.” World Development 40 (5): 928–37. https://doi.org/10.1016/j​ .worlddev.2011.11.016. Gereffi, G., and J. Humphrey. 2005. “The Governance of Global Value Chains.” Review of International Political Economy 12: 78–104. https://doi.org/10.1080/09692290500049805. Gill, I. S., and M. Raiser. 2011. Golden Growth. Washington, DC: World Bank. https://doi​ .org/10.1596/978-0-8213-8965-2. Gould, D. 1994. “Immigrant Links to the Home Country: Empirical Implications for U.S. Bilateral Trade Flows.” Review of Economics and Statistics 76 (2): 302–16. Grossman, G. M., and E. Rossi-Hansberg. 2008. “Trading Tasks: A Simple Theory of Offshoring.” American Economic Review 98 (5): 1978–97. https://doi.org/10.1257/aer.98.5.1978. Hayakawa, K., and N. Yamashita. 2011. “The Role of Preferential Trade Agreements (PTAs) in Facilitating Global Production Networks.” Discussion Paper 280, Institute of Developing Economies, Japan External Trade Organization (JETRO), Tokyo. Helpman, E., P. Antràs, and E. Helpman. 2008. The Organization of Firms in a Global Economy. Cambridge, MA: Harvard University Press. Hummels, D. L., J. Ishii, and K.-M. Yi. 2001. “The Nature and Growth of Vertical Specialization in World Trade.” Journal of International Economics 54 (1): 75–96. https://doi​ .org/10.1016/S0022-1996(00)00093-3. Hummels, D. L., D. Rapoport, and K.-M. Yi. 1998. “Vertical Specialization and the Changing Nature of World Trade.” Economic Policy Review 4 (2): 79–99. Hummels, D. L., and G. Schaur. 2013. “Time as a Trade Barrier.” American Economic Review 103 (7): 2935–59. https://doi.org/10.1257/aer.103.7.2935. Johnson, R., and G. Noguera. 2017. “A Portrait of Trade in Value-Added over Four Decades.” Review of Economics and Statistics 99 (5): 896–911. Kirkegaard, J. F. 2013. “New Avenues for Empirical Analysis of Cross-Border Investments: An Application for the ASEAN Members and Middle and Low Income Country Outward Investments.“ PhD dissertation, Johns Hopkins University, Baltimore, MD. Koopman, R., Z. Wang, and S.-J. Wei 2012. “Estimating Domestic Content in Exports When Processing Trade Is Pervasive.” Journal of Development Economics 99: 178–89. Kummritz, V. 2016. “Do Global Value Chains Cause Industrial Development?” Working Paper 2016-01, Centre for Trade and Economic Integration, The Graduate Institute, Geneva. Supply Chains in Europe and Central Asia: Connectivity through Cross-Border Production Fragmentation ●  267 Kummritz, V., and B. Quast. 2016. “Global Value Chains in Low and Middle Income Countries.” Working Paper 2016-10, Centre for Trade and Economic Integration, The Graduate Institute, Geneva. Kummritz, V., D. Taglioni, and D. Winkler. 2017. “Economic Upgrading through Global Value Chain Participation: Which Policies Increase the Value Added Gains?” Policy Research Working Paper 8007, World Bank, Washington, DC. Li, X., and X. Liu. 2005. “Foreign Direct Investment and Economic Growth: An Increasingly Endogenous Relationship.” World Development 33 (3): 393–407. https://doi​ .org/10.1016/j.worlddev.2004.11.001. Marin, Dalia. 2010. “The Opening Up of Eastern Europe at 20—Jobs, Skills, and ‘Reverse Maquiladoras’ in Austria and Germany.” Working Paper 421, Bruegel, Brussels. Nicita, A., V. Ognivtsev, and M. Shirotori. 2013. “Global Supply Chains: Trade and Economic Policies for Developing Countries.” Blue Papers Series 55, United Nations Conference on Trade and Development, Geneva. OECD (Organisation for Economic Co-operation and Development). 2013. Interconnected 264189560-en. Economies. Paris: OECD. https://doi.org/10.1787/9789​ ———. 2015a. “The Participation of Developing Countries in Global Value Chains: Implications for Trade and Trade Policy.” Trade Policy Note, OECD, Paris. ———. 2015b. Trade in Value Added (TIVA) Indicators Database (October). OECD, Paris. https://doi​.org/10.1787/tiva-data-en. Orefice, G., and N. Rocha. 2014. “Deep Integration and Production Networks: An Empirical Analysis.” World Economy 37 (1): 106–36. https://doi.org/10.1111/twec.12076. Polanyi, M. 1962. “Tacit Knowing: Its Bearing on Some Problems of Philosophy.” Reviews of Modern Physics 34 (4): 601–16. https://doi.org/10.1103/RevModPhys.34.601. Rauch, J. E., and V. Trindade. 2002. “Ethnic Chinese Networks in International Trade.” Review of Economics and Statistics 84 (1): 116–30. https://doi.org/10.1162/003​ 465302317331955. Santoni, Gianluca, and Daria Taglioni. 2015. “Networks and Structural Integration in Global Value Chains.” In The Age of Global Value Chains, ed. João Amador and Filippo di Mauro. London: Centre for Economic Policy Research. Sturgeon, T. J., and O. Memedovic. 2011. “Mapping Global Value Chains: Intermediate Goods Trade and Structual Change in the World Economy.” UNIDO Working Paper 5: 52. Taglioni, D., and D. Winkler. 2016. Making Global Value Chains Work for Development. Washington, DC: World Bank. https://doi.org/10.1596/978-1-4648-0157-0. Thoenig, M., and T. Verdier. 2003. “A Theory of Defensive Skill-Biased Innovation and Globalization.” American Economic Review 93 (3): 709–28. https://doi.org/10.1257​ /­​000282803322157052. Timmer, M. P., A. A. Erumban, B. Los, R. Stehrer, and G. J. de Vries. 2014. “Slicing Up Global Value Chains.” Journal of Economic Perspectives 28 (2): 99–118. https://doi​ .org/10.1257/jep.28.2.99. 7 ECA Policies for Improving Connectivity Countries in Europe and Central Asia (ECA) have made important progress in furthering regional and global connectivity along the several policy dimensions discussed in this report, including trade, foreign direct investment (FDI), supply chains, migration, internet and telecommunications, and transport. ECA countries have taken critical steps to increase integration and connectivity along many of these dimensions, yet important challenges remain. This chapter considers the historical, political, and economic developments that have led to greater connec- tivity in many parts of ECA and how policies influenced this connectivity. We con- sider data on selected connectivity-related policies in the ECA region and comparator countries and regions including tariffs, FDI policies, preferential trade agreements (PTAs), bilateral investment treaties (BITs), product market regulations (PMRs), and domestic regulatory reforms in transition countries (transition indica- tors). In addition, comovements across the different policy areas for ECA as a whole and ECA subregions are analyzed as a means to understand whether con- nectivity policies pursued by ECA countries are moving in the same direction or are at odds with each other. Main Messages • ECA countries generally have supported greater international connectivity through reductions in most-favored-nation (MFN) tariffs, increased numbers of 269 270  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia PTAs and BITs, reductions in regulatory restrictions on FDI, improved domestic economic governance in general and the regulation of key network sectors in particular, and a process of policy transition toward Organisation for Economic Co-operation and Development (OECD) standards. Regional integration through PTAs has been faster in ECA than elsewhere, although less so with BITs. However, ECA is less successful than other regions in domestic product market governance. • The trend in policies that improve connectivity in ECA slowed significantly after the early 2000s. Little change is observed in tariff liberalization (as of the begin- ning of the 2000s), the use of BITs (as of the end of the 2000s), and reductions of FDI regulatory restrictions and product market liberalization (as of 2010). • Policies toward regional integration have varied greatly across ECA countries. Northern, Southern, and Western European high-income countries tend to have lower tariffs, higher global and extraregional integration through PTAs, and lower regulatory restrictions on FDI. Former centrally planned economies in Central and Eastern Europe still rank lower on the quality of domestic gover- nance in infrastructure sectors than countries in other ECA subregions. Countries tend to be consistent in their policies aimed at improving connectivity; decreases in MFN tariffs and increases in the number of BITs go hand in hand. However, some country groups, particularly the non-high-income ECA coun- tries in Eastern Europe and Central Asia, tend not to consistently implement connectivity-friendly ­policies across different dimensions. Introduction The set of policies that are relevant for international connectivity is multidimen- sional, encompassing measures that affect trade, FDI, supply chains, migration, and transport infrastructure. While policies affecting connectivity are determined on an autonomous, independent basis by governments, they have implications for foreign countries, and thus are often the focus of international agreements and cooperation. The ECA region has a rich history of progress on enhancing international connectivity. The region is unique both in the distinct character of ­ connectivity-related initiatives that have been pursued over time and more gener- ­ ally the integration of countries in this region into the broader world economy. An important dimension of this uniqueness is the role that has been played— and continues to be played—by formal regional integration arrangements between subsets of ECA countries. The most prominent feature of international economic policy cooperation in ECA is undoubtedly the gradual expansion of what is now the European Union (EU). Starting with a sectoral integration initiative among six European states—the 1951 European Coal and Steel Community—and a much more ambitious agreement to form a European Economic Community in 1957, over time the European Economic Community grew incrementally both in terms of issue coverage and the depth of policy cooperation. It is now an eco- nomic union spanning the free movement of goods, services, capital, and people with associated supra-national common institutions and a common currency that has been adopted by 19 EU member states. ECA Policies for Improving Connectivity ●  271 Concurrently with the gradual process of deepening economic cooperation between EU member states there has been a process of widening the EU to encompass additional countries. Currently membership stands at 28, with 7 ­countries formally accepted as accession candidates.1 A major feature of European integration in the past 20 years has been the process of accession— most notably by 10 Baltic and Central European countries (Estonia, Latvia, Lithuania, Poland, Hungary, the Czech Republic, the Slovak Republic, and Slovenia in 2004, followed by Bulgaria and Romania in 2007). Until the dissolution of the former Soviet Union, these countries had been part of the second major regional bloc that dominated the ECA region: the Council for Mutual Economic Assistance (CMEA or COMECON), led by the former Soviet Union. The 10 nations that acceded to the EU in 2004–07 had all been CMEA members in one form or another until it ceased to operate in 1991 following the breakup of the Soviet Union. The demise of the Soviet Union was followed by a looser form of economic inte- gration and cooperation between the Russian Federation and the former Soviet Republics—the Commonwealth of Independent States (CIS). Starting late in the first decade of the 2000s, Russia sought to deepen the CIS into a common market and economic union and pursued a process of deepening economic integration with a subset of its neighbors, through the creation of a Eurasian Economic Union. This currently comprises Armenia, Belarus, Kazakhstan, and Russia. Trade agreements have been a central feature of the EU’s engagement with countries in the “European neighborhood,” both those that were (are) eli- gible for EU membership and those that are not. The EU currently has Trade agreements have more than 50 PTAs in place, with another 80 or so in the pipeline— been a central feature both agreements that have been negotiated and are waiting ratifica- of the EU’s engage- tion and agreements that are in the process of negotiation.2 The EU’s ment with countries in approach toward reciprocal trade agreements has shifted over time the “European neigh- from “shallow” trade agreements that centered mostly on the liber- borhood,” both those alization of merchandise trade toward deeper agreements that also that were (are) eligible liberalize trade in services, public procurement markets, and cross- for EU membership border investment and include disciplines on the implementation of and those that are not. national regulatory regimes. EU trade agreements vary across partners in depth and design. The EU has customs union agreements with a small number of neighboring states, such as Turkey, and deeper arrangements with European countries that have elected not to join the EU—for example, Norway and Switzerland—that provide these countries with full access to the European Single Market through the European Economic Association agreement. The EU has developed so-called deep and comprehensive free trade agreements (DCFTAs) that include various elements of EU law (the acquis communautaire). These are on offer to neighboring countries and are intended to be instruments to support convergence in the partner with specific areas of EU legislation and regula- tion that pertain to the operation of the Single Market (Hoekman 2016). DCFTAs differ from earlier -vintage EU trade agreements with neighboring countries in having less “soft law” language and establishing specific, binding (enforceable) disciplines that aim at the (gradual) convergence of policies in covered areas with those of the EU (Langbein and Wolczuk 2012). An implication of DCFTAs anchored on adoption 272  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia of the acquis is that partner countries would move away from Russian regulatory standards, raising worries by Russian enterprises that they would be negatively affected by the adoption of EU norms and standards by European neighborhood countries (Hoekman, Jensen, and Tarr 2013). A recent development has been a shift toward a less EU-norm centric, more pragmatic strategy when pursuing DCFTAs, reflected in less emphasis on making EU law the focal point for deep agreements (Hoekman 2016). There is increasing recognition among European policy makers that the approaches pursued by the EU since the collapse of the Soviet Union that were centered on the concept of a “normative power Europe” and a focus on exporting European values and regula- tory norms to partner countries has not delivered the desired results (Langbein 2014; European Commission 2015). The EU itself is becoming more contested by European polities. Proposals by the European Commission to revamp long-­ standing approaches toward investor-state disputes under BITs are another indica- tion of a recognition that new approaches are needed to govern international economic cooperation. The decision by the UK government in 2016 to leave the EU will require the remaining 27 member states to determine how to structure a deep economic integration arrangement with the United Kingdom. This may build on recent agreements that have been concluded with Canada and Japan or be substantially more ambitious—the eventual outcome will depend on the objec- tives of the UK government, which have yet to be fully articulated. While the EU is in a state of flux, confronting major challenges and questions regarding the future of further deepening of cooperation as opposed to reducing the extent of integration both among the membership and with non-EU countries, it has played a major role in providing a focal point for efforts to enhance regional connectivity. The primary purpose of this chapter is to provide a descriptive analy- sis of a set of policy indicators that are salient from a connectivity perspective, focusing on both domestic policies and the extent to which countries have engaged in international agreements with partners that entail disciplines and liberalization of the relevant policy. ­ Trade Costs as a Focal Point for Connectivity Extensive research has shown that from a development and growth perspective, lowering trade and transactions costs for firms is a key dimension of enhancing connectivity. High trade costs reduce competitiveness of firms and the ability of an economy as a whole to exploit its comparative advantages. Trade costs are a func- tion of a mix of exogenous variables (e.g., location) and policy (Moïse and Le Bris 2015). Restrictive trade policies, markets that are difficult for new entrants to contest because of restrictive business practices of a dominant supplier or state- owned enterprise, PMR that impedes entry as opposed to addressing market fail- ures, barriers to FDI, and restrictive visa regimes that make it difficult for employees and professionals to cross borders to supply services or establish contacts with potential suppliers or customers (see chapter 4) are all examples of policies that raise trade-related operating costs for firms, which in turn may increase the prices of goods and services for consumers and reduce the demand for workers and thus negatively affect household incomes. ECA Policies for Improving Connectivity ●  273 A challenge for analysts (and policymakers) is to differentiate between trade cost–creating measures that generate social waste and those that do not. Abstracting from tariffs, which remain a burden on international exchange even though the average level of tariffs has dropped substantially in the past 30 years, most trade policy instruments used by countries comprise nontariff measures (NTMs): regulatory policies pertaining to product quality, health, and safety stan- dards for goods and services (e.g., transport, logistics, finance, and professionals).3 Taking action to reduce trade costs by facilitating the movement of goods, ser- vices, investment, and people by necessity implies focusing on the substance and implementation (enforcement) of NTMs. Many NTMs have been put in place for good reasons, that is, to address market failures or to pursue specific noneconomic social objectives. Policy consistency requires that efforts to reduce trade costs not undercut the realization of the legiti- mate objectives that motivate the regulatory policies (NTMs) a country has put in place. International cooperation is one mechanism governments can use to bal- ance a process aimed at reducing the trade costs generated by differences in regulatory regimes that affect connectivity. The demand for such balancing has increased as a result of the growth in international value chain–based production networks in recent decades. This has led to an increasing number of firms support- ing trade facilitation initiatives broadly defined as opposed to lobbying for policies to restrict trade and factor movement (Gawande, Hoekman, and Cui 2014; Baldwin 2016). The policy agenda has shifted toward efforts to facilitate trade and to reduce the trade costs created by regulatory heterogeneity while ensuring that regulatory objec- tives (such as health and safety) are met. This is a more complex agenda than one that centers on removing welfare-reducing border barriers such as tariffs. Trade policy today increasingly involves the use of NTMs that are not necessarily designed to restrict or to encourage trade but that address nontrade regulatory objectives such as product safety, environmental protection, national security, or intellectual property protection. Trade agreements, both those at the multilateral level of the World Trade Organization (WTO) and The policy agenda has bilateral and regional PTAs, are the instrument of choice for gov- shifted toward efforts ernments to pursue reductions in NTM-related trade costs. One to facilitate trade and function of trade agreements is to establish what types of NTMs to reduce the trade costs created by regu- should be banned because they are simply protectionist. An latory heterogeneity example is quantitative restrictions. These are prohibited in prin- while ensuring that ciple by the WTO and EU PTAs outside of agriculture where tariff regulatory objectives rate quotas continue to prevail for some products. More generally, are met. trade agreements provide frameworks regulating the use of NTMs. For example, a common form of NTMs is product standards, and, more generally, PMR. These are generally aimed at ensuring the health and safety of con- sumers. The WTO imposes rules on countries regarding how they may pursue such regulation, for example, by encouraging the use of international standards where they exist and requiring countries to notify the WTO regarding new product stan- dards if they are not compliant with or based on internationally agreed standards that have been developed by specialized international bodies. The extent to which countries notify regarding noncompliant standards is one indicator of integration 274  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia (connectivity) of their economies as it reveals implicitly to what extent a WTO mem- ber has chosen to adopt national norms that diverge from international standards. Agreeing to a common set of rules on the use and implementation of NTMs without undercutting the ability to pursue legitimate regulatory objectives enhances joint welfare. International cooperation and rules on NTMs generate not just benefits in terms of economic gains associated with lower trade costs, but also in terms of connectivity, interconnection, and the reduction of coordination exter- nalities related to public goods such as environmental and labor standards. No matter what a country’s strategy is with respect to industrial policy and trade or the extent to which it makes use of NTMs, minimizing the transactions costs and uncertainty associated with their implementation is important in reducing the real resource (welfare) costs of NTMs. There is therefore a strong connection between efforts to streamline and rationalize the use of NTMs (e.g., Cadot, Malouche, and Sáez 2012) and enhancing connectivity. Connectivity and trade cost concerns often are reflected in a focus on trade in goods and related FDI flows. The need to also consider services trade costs is often neglected. Services directly matter for connectivity, as many of the networks that define connectivity levels comprise services sectors. But they also matter more gener- ally because all firms use services as inputs into production. Input costs that are higher than they would be in an environment in which services trade costs were lower act as a tax on domestic industries and reduce their competitiveness. The stylized fact here is that trade costs for services are much higher than trade costs for goods (Miroudot and Shepherd 2016). The result is to reduce the volume of trade in services, and thus to reduce the access firms and households have to low-cost services. Services trade costs are high in part because of the characteristics of services: trade often requires movement of people or establishment of a commercial pres- ence (FDI). This implies that many policies and their administration may affect trade costs. Two dimensions are important in this regard: (a) regulatory policies that apply to all firms, both national and foreign; and (b) policies that are designed to discriminate against foreign providers or consumption abroad. Regulatory poli- cies vary across countries for any given sector and the resulting heterogeneity is an important source of international trade costs. High services trade costs also reflect in part that regulatory policies may discriminate against foreign providers. Examples include nationality requirements or banning access to markets as is the case in many countries for segments of the transport, communications, or profes- sional services sectors. Research has shown that barriers to trade and investment in services are often much higher than for goods. Although information on services trade policy is limited, recent compilations of prevailing policies across countries by the OECD and the World Bank have shown that barriers to trade in services are often high, with significant variation across countries and sectors, translating into estimates of ad valorem tariff equivalents that are greater than trade barriers for goods (Jafari and Tarr 2017).4 The effect of trade and FDI policy instruments is in part determined by institu- tional variables (e.g., Rodriguez and Rodrik 2001; Freund and Bolaky 2008; Fiorini and Hoekman 2017a, 2017b). Beverelli, Fiorini, and Hoekman (2017) show that the economic effects of services trade policies on manufacturing industries are moder- ated by the quality of economic governance institutions in the importing country. ECA Policies for Improving Connectivity ●  275 Lower services trade restrictiveness is found to increase downstream manufactur- ing productivity only in countries with good economic governance (as proxied by indicators of control of corruption, rule of law, and the quality of regulatory institu- tions). This moderating effect prevails with respect to trade policies that target services provision through foreign establishment (FDI) more than cross-border trade in services. This result reflects the intangibility and nonstorability of services, which mean that foreign providers must invest in local production facilities (estab- lish a commercial presence) to be able to contest the relevant market. The bottom line is that regulatory regimes for products matter—they help determine the ability of firms to benefit from actions that aim to integrate markets. Thus, PMR is one policy area that should be considered a determinant of the level of effective con- nectivity that prevails in a market. Indicators of Trade Cost–Related Policies What follows focuses on six policy areas that affect the cost of interaction between pairs of countries and between countries and the rest of the world: import tariffs, engagement in trade agreements that reduce tariffs on a preferential basis, poli- cies toward inward FDI, the use of BITs to provide protection to investors from expropriation and adverse changes in investment policies, PMRs, and European Bank for Reconstruction and Development (EBRD) transition indicators (on sectoral domestic regulation). Policies regulating international mobility of people (visa regimes) and policies toward integration of migrants are discussed in chapter 4. The time periods and country coverage for these variables are listed in table 7.1. The choice of these specific variables reflects in part data availability but more important is that they relate closely to the trade cost discussion in the previous section and the different dimensions of connectivity that are the focus of previ- ous chapters of this report.5 Some of the variables measure policies that apply at the border and increase costs—for example, tariffs. Tariffs are of course less important today than a few decades ago, but differences in average tariff levels provide information on the extent to which countries have opened their markets to foreign competition. Unfortunately, comprehensive comparable time series data on NTMs do not exist. However, as noted above, NTMs are highly corre- lated with PMR, and the intensity with which a country has pursued PTAs is a good indicator of the degree to which countries are willing to agree to disciplines on the use of NTMs. The same is true for the extent of BITs negotiated—BITs are an instrument that affects FDI policies, and Tariffs are of course BITs are concluded to improve the investment climate confronting less important today foreign firms. Measures of the degree to which FDI is restricted is than a few decades particularly salient as a proxy for barriers to trade in services, as FDI ago, but differences in is a major “mode of supply” for services firms. Measures of PMR average tariff levels are similarly a good proxy for connectivity because they capture provide information on the extent to which the extent of barriers to entry and the extent to which a nation has countries have opened put in place good regulatory practices. For many ECA countries their markets to for- there is a unique time series measure of convergence toward good eign competition. regulatory ­practices compiled by the EBRD—the so-called transition indicators. 276  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia TABLE 7.1  Policy Measures and Indicators Variable Starting year Ending year Countries covered Tariffs 1988 2015 186 FDI restrictiveness 1997 2015 59 PTAs 1988 2015 189 BITs 1988 2015 177 Horizontal PMR 1998 2013 47 Sectoral PMR 1975 2013 47 Transition EBRD 1989 2012 20 Sources: Tariff data are from World Bank, World Integrated Trade Solution. Foreign direct investment (FDI) policy measures are sourced from the Organisation for Economic Co-operation and Development (OECD) FDI Regulatory Restrictiveness Indicators (FDIRRI) database. Preferential trade agreement (PTA) data come from the World Bank database on Content of Deep Trade Agreements (http://data​ worldbank.org/data-catalog/deep-trade-agreements). Data on bilateral investment treaties (BITs) are .­ from the United Nations Conference on Trade and Development investment hub database. Horizontal product market regulation (PMR) is sourced from the OECD’s PMR Economy Wide Database. Sectoral PMR indicators are from the OECD’s PMR Energy, Transport, and Communications Database. Transition EBRD data are from the Transition Indicators Database of the European Bank for Reconstruction and Development (EBRD). Note: The simple average most-favored-nation tariff is taken as the measure for “Tariffs.” The “FDI restrictiveness” indicator measures statutory restrictions on inward FDI; a higher value means more restrictions. To capture integration through PTAs (BITs), the number of trade agreements (investment treaties) to which each country belongs is used. “Horizontal PMR” refers to indicators of the restrictiveness of product market regulation that applies to the economy as a whole, that is, measures that pertain to all types of economic activity, independent of sector; a higher value indicates more restrictive regulation. “Sectoral PMR” is the aggregate restrictiveness of product market regulation that is specific to sectors that matter for connectivity: energy, telecommunications, transport, and distribution; a higher value reflects more restrictive regulation. “Transition EBRD” refers to the transition indicators compiled by the EBRD, which measure the degree to which policies are equivalent to the standards prevailing in industrial market economies; higher values imply greater convergence toward best practice. The data reveal several patterns. First, there is a clear trend toward a policy environment that is supportive of greater international connectivity. This is apparent across all policy instruments for which data are reported before 2000. Across the different time spans for which data on respective policies are reported, ECA countries have on average decreased MFN tariffs, increased the number of PTAs and BITs, unilaterally reduced regulatory restrictions to FDI, improved domestic economic governance in general and the regulation of key network sectors in particular, and undergone a process of policy transition toward OECD standards. Comparing the evolution of ECA countries’ policy stances with those in the rest of the world, on average ECA as a whole (including the EU) is a leader in cooperat- ing with partner countries through PTAs and BITs, reflecting a relatively faster pace of intraregional integration than elsewhere. Conversely, the average ECA country is more of a follower and more restrictive than non-ECA regions when it comes to domestic economic governance. For many policy instruments, the observed positive trend toward a more connectivity-friendly policy environment in ECA slowed significantly after the ­ early 2000s. In some cases, the data reveal convergence toward a “steady state” with little change observed. This is the case for tariff liberalization (as of the begin- ning of the 2000s), the use of BITs (as of the end of the 2000s), and reductions of FDI regulatory restrictions and product market liberalization (as of 2010). Disaggregating these trends and patterns across subsets of ECA coun- tries, whether on the basis of geography or per capita incomes, often reveals ECA Policies for Improving Connectivity ●  277 very heterogeneous policy stances across country groups. Northern, Southern, and Western European high-income countries (HICs) converge on lower tar- iffs and higher global and extraregional integration through PTAs and have lower regulatory restrictions on FDI. Transition to higher quality of domestic governance in infrastructure sectors is observed for Central and Northern Europe compared with former Soviet Union countries in other ECA subregions. Analysis of comovements across policy instruments reveals the extent to which there is balance in connectivity-related policies—that is, whether for a given country, subregion or subgroup policies tend to move in the same (com- plementary) direction over time. Insofar as divergent policy trajectories are observed across instruments, there is a lack of policy consistency in terms of supporting greater connectivity. On average at the global level, countries would pursue balanced connectivity if they implement reforms that are consis- tent with each other and this pattern increases over time. The data reveal that different policies often move in the same direction, suggesting a balanced connectivity objective—for example, countries decrease MFN tariffs and at the same time increase the number of BITs. Some country groups, however, seem to move in the opposite direction and fail to consistently implement connectivity-friendly policies across different dimen- sions. This is the case especially among the non-HIC ECA countries. The data show substantial heterogeneity in the comovement of MFN tariffs and BITs between the two large ECA subregions defined by the World Bank: (a) Europe + Western Balkans; and (b) Eastern Europe + Central Asia. Only the first of these subregions demonstrates policy consistency. MFN Tariffs Figure 7.1 reports data on simple average MFN import tariffs, using the World Integrated Trade Solution database.6 The data are aggregated to simple aver- ages for seven global regions, including ECA, and for three time periods. In comparison with other regions, the aggregate pattern in ECA tariff evolution is very stable, characterized by a smooth trend toward higher integration. All other regions but North America (Canada and the United States) display higher aver- age levels of tariff protection. Within the ECA region there are heterogeneous patterns. While convergence to lower tariffs is almost ubiquitous, significantly higher integration is reached by the EU and Western Balkans region (figure 7.2) driven by Western, Southern, and Northern Europe. As shown in figure 7.3, these three subregions converged on a policy steady state of less than 5 percent as of the beginning of the 2000s. This is the case in particular for ECA HICs (figure 7.4). Although tariffs are just one type of trade policy, focusing on trends in average MFN tariffs is relevant because tariffs are particularly important as a barrier to value chain participation. The three world regions with the lowest tariffs (North America, ECA, and East Asia and Pacific) are the most regionally integrated and the domi- nant users of regional supply chains (Baldwin 2016). Low MFN tariffs are often 278  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 7.1  Tariffs across global regions 20 MFN tariff (simple average) 15 10 5 0 1988–96 1997–2005 2006–15 East Asia and Pacific Latin America and the Caribbean North America Sub-Saharan Africa Europe and Central Asia Middle East and North Africa South Asia Note: MFN = most favored nation. FIGURE 7.2  Tariffs in the 12 main ECA subregions 10 MFN tariff (simple average) 8 6 4 19 8 19 9 19 0 19 1 19 2 19 3 19 4 19 5 19 6 19 7 19 8 20 9 00 20 1 02 20 3 04 20 5 06 20 7 20 8 20 9 20 0 20 1 20 2 20 3 20 4 15 8 8 9 9 9 9 9 9 9 9 9 9 0 0 0 0 0 0 1 1 1 1 1 19 20 20 20 20 European Union and Western Balkans Eastern Europe and Central Asia Note: MFN = most favored nation. complemented by zero bilateral tariffs because of PTAs. They may be offset by NTMs—some countries have been shown to replace tariffs with various NTMs— but as noted above comprehensive data on NTMs are not available. The PTA, FDIRRI, PMR, and EBRD indicators are all proxies for the level of and trends in applied NTMs across countries. ECA Policies for Improving Connectivity ●  279 FIGURE 7.3  Tariffs in ECA 15 regions MFN tariff (simple average) 10 5 0 1988−96 1997−2005 2006−15 Western Europe Central Europe Western Balkans Central Asia Turkey Southern Europe Northern Europe South Caucasus Russian Federation Other Eastern Europe Note: MFN = most favored nation. FIGURE 7.4  Tariffs in ECA 20 countries by income group MFN tariff (simple average) 15 10 5 19 8 89 19 0 91 19 2 93 19 4 19 5 96 19 7 19 8 99 20 0 20 1 20 2 20 3 04 20 5 06 20 7 08 20 9 20 0 20 1 12 20 3 20 4 15 8 9 9 9 9 9 9 0 0 0 0 0 0 0 1 1 1 1 19 19 19 19 19 20 20 20 20 20 High-income ECA countries Lower-middle-income ECA countries Upper-middle-income ECA countries Note: ECA = Europe and Central Asia; MFN = most favored nation. Foreign Direct Investment Policies The OECD has developed an aggregate indicator of a variety of regulatory poli- cies that restrict inward foreign investment, the FDI Regulatory Restrictiveness 280  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia Index (FDIRRI). What follows uses the most aggregate version of the index, which encompasses equity restrictions, restrictions in the form of screening and approval requirements, and restrictions on the nationality of key personnel. The indicator takes values between 0 and 1, with 0 denoting no restrictions and 1 maximum restrictiveness. An important limitation for the scope of the analysis is that the country coverage of the database is not complete when it comes to the ECA region. In particular, there exists no information for countries in the Western Balkans and the South Caucasus subregions. The average ECA country (among those covered in the database) shows a very high degree of integration, with restrictiveness scores among the lowest in the database (figure 7.5). Since the mid-2000s, ECA has maintained less restrictive policies than the United States, Canada, Brazil, India, and China. Figures 7.6–7.8 unpack the average ECA scores across subregions and income groups. All four European regions show a pattern of openness starting in 1997, converging to stable and relatively similar scores, all below 0.05, in 2010. Turkey and Russia show a similar pattern, with the former converging to a degree of restrictiveness slightly above that of Western Europe and the latter to a relatively much more restrictive regulatory framework (score slightly below 0.2). After 2010, progress toward further integration (liberalization) seems to have stopped in both HICs and upper-middle-income countries (UMICs). The average ECA lower-middle-income country (LMIC) instead reveals some policy progress toward removing FDI regulatory barriers between 2010 and 2015. Figures 7.9–7.12 replicate the descriptive analysis of figures 7.6–7.8 for the FDIRRI scores for two specific services sectors that are particularly relevant for con- nectivity: transport and telecommunications. The main patterns hold for these two services sectors with the main difference being that restrictions for FDI in transport FIGURE 7.5  FDIRRI in ECA 0.6 and selected countries 0.4 Total FDI Index 0.2 0 1997 2006 2015 ECA United States Canada Brazil China India Note: ECA = Europe and Central Asia; FDIRRI = Foreign Direct Investment Regulatory Restrictiveness Index. ECA Policies for Improving Connectivity ●  281 FIGURE 7.6  FDIRRI in ECA subregion I 0.10 0.08 Total FDI Index 0.06 0.04 0.02 03 06 10 11 12 13 14 15 97 20 20 20 20 20 20 20 20 19 Western Europe Southern Europe Central Europe Northern Europe Note: ECA = Europe and Central Asia; FDIRRI = Foreign Direct Investment Regulatory Restrictiveness Index. FIGURE 7.7  FDIRRI in ECA 0.4 subregion II 0.3 Total FDI Index 0.2 0.1 0 06 10 11 12 13 15 97 03 14 20 20 20 20 20 20 19 20 20 Central Asia Russian Federation Turkey Other Eastern Europe Note: ECA = Europe and Central Asia; FDIRRI = Foreign Direct Investment Regulatory Restrictiveness Index. services appear to be of a significantly greater magnitude. It is also worth noticing how the two subregions Central Asia and Other Eastern Europe tend to out­ perform Russia and Turkey in terms of openness toward FDI in transport services (see 2015 data in figure 7.10). 282  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 7.8  FDIRRI in ECA countries by income group 0.30 0.25 Total FDI Index 0.20 0.15 0.10 0.05 97 03 06 10 11 12 13 14 15 19 20 20 20 20 20 20 20 20 High-income ECA countries Lower-middle-income ECA countries Upper-middle-income ECA countries Note: ECA = Europe and Central Asia; FDIRRI = Foreign Direct Investment Regulatory Restrictiveness Index. FIGURE 7.9  FDIRRI for 1.0 communications and transport in ECA and selected countries 0.8 0.6 0.4 0.2 0 1997 2006 2015 ECA telecommunications transport Brazil telecommunications transport United States telecommunications transport China telecommunications transport Canada telecommunications transport India telecommunications transport Note: ECA = Europe and Central Asia; FDIRRI = Foreign Direct Investment Regulatory Restrictiveness Index. ECA Policies for Improving Connectivity ●  283 FIGURE 7.10  FDIRRI for 0.6 communications and transport in ECA subregions 0.4 0.2 0 1997 2006 2015 Western Europe telecommunications transport Central Asia telecommunications transport Southern Europe telecommunications transport Russian Federation telecommunications transport Central Europe telecommunications transport Turkey telecommunications transport Northern Europe telecommunications transport Other Eastern Europe telecommunications transport Note: ECA = Europe and Central Asia; FDIRRI = Foreign Direct Investment Regulatory Restrictiveness Index. FIGURE 7.11  FDIRRI for 0.5 transport in ECA countries by income group 0.4 0.3 0.2 0.1 03 06 0 1 14 15 97 12 13 1 1 20 20 20 20 20 20 19 20 20 High-income ECA countries Lower-middle-income ECA countries Upper-middle-income ECA countries Note: ECA = Europe and Central Asia; FDIRRI = Foreign Direct Investment Regulatory Restrictiveness Index. 284  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 7.12  FDIRRI for communications in ECA 0.25 countries by income group 0.20 0.15 0.10 0.05 0 11 06 10 12 13 97 03 14 15 20 20 20 20 20 19 20 20 20 High-income ECA countries Lower-middle-income ECA countries Upper-middle-income ECA countries Note: ECA = Europe and Central Asia; FDIRRI = Foreign Direct Investment Regulatory Restrictiveness Index. Preferential Trade Agreements The number of PTAs a country has concluded is an important country-specific measure of policy toward integration and (regional) connectivity. As discussed in the introduction, PTAs are instruments to lower trade costs of a regulatory nature, as well as mechanisms to remove MFN tariffs for trade among the part- ners. The data that follow simply show the existence of PTAs and do not con- sider the coverage or depth of the PTAs. This is obviously a very important factor from a connectivity and integration perspective. Ideally, we would like to weight PTAs according to how comprehensive they are and the degree to which they are binding (enforceable). We use a simple count measure here to avoid subjective assessments of which PTAs are more “serious” than others. Analysis of the depth of PTAs is addressed in a World Bank project that comprehen- sively codes the content of PTAs (Hofmann, Osnago, and Ruta 2017). This permits deeper analysis of the differences across countries in this regard and their effects—a task that is not undertaken here. As a rule of thumb, PTAs between HICs tend to be more comprehensive than PTAs between developing countries. Agreements with the EU as a partner will always cover NTMs as well as tariffs, but they vary substantially in terms of coverage of services trade and investment policies and public procurement. Chapter 3 considers the impact of deep PTAs on attracting FDI. We construct a measure of the intensity of the use of PTAs from the database on PTAs compiled by the World Bank (see Hofmann, Osnago, and Ruta 2017). ECA stands out as the region with the highest use of PTAs. In the period 2000–15, ECA countries on average were members of almost 20 PTAs that were in force ECA Policies for Improving Connectivity ●  285 FIGURE 7.13  Preferential 20 trade agreements across global regions Number of enforced 15 preferential trade agreements 10 5 0 1958−79 1980−99 2000−15 East Asia and Pacific Middle East and North Africa Sub-Saharan Africa Europe and Central Asia North America Latin America and the Caribbean South Asia (implemented) (see figure 7.13 plotting the number of enforced PTAs averaged across countries within each region and across years within each period). This com- parative pattern with respect to other regions holds across intra- and extraregional integration as shown respectively in figure 7.14 and figure 7.15. European coun- tries are market leaders in their pursuit of regional integration: the average score of ECA countries dwarfs those of all other regions. The aggregate policy performance of ECA hides important heterogeneity across subregions and income groups. The rapid pace toward greater regional integration has been driven by EU member states plus—to a lesser extent— Turkey and Russia (see figure 7.16). The growth in Russian PTAs during the last period is a reflection of the demise of the former Soviet Union and the creation of the CIS and associated new PTAs. From an income group perspective PTAs are dominated by HICs ( ­figure 7.17), but the average number of PTAs signed by upper- and lower-middle-income ECA countries is always higher than for other regions, with the exception of North America. One reason for the spike in HIC PTAs is the decision by the EU to reengage in PTA negotiations in 2006, following the adoption of the 2006 Global Europe ­ communication, which removed a de facto moratorium on new PTAs in favor of cooperation through the WTO. As shown in figures 7.18 and 7.19, all non-EU ECA subregions except for Turkey and—from an income perspective—LMICs, have signed PTAs only with other ECA countries. EU countries were the leading actors in extraregional PTA integration at the end of the 1970s, but their pursuit of extraregional integration accelerated substantially since the mid-1990s. The result is an equal distribution of PTAs within and beyond the ECA region in 2015. 286  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 7.14  Preferential trade agreements across 10 global regions: Intraregional integration Number of enforced preferential trade agreements within the region 5 0 1958−79 1980−99 2000−15 East Asia and Pacific Middle East and North Africa Sub-Saharan Africa Europe and Central Asia North America Latin America and the Caribbean South Asia FIGURE 7.15  Preferential 8 trade agreements across global regions: Extraregional integration Number of enforced 6 preferential trade agreements beyond the region 4 2 0 1958−79 1980−99 2000−15 East Asia and Pacific Middle East and North Africa Sub-Saharan Africa Europe and Central Asia North America Latin America and the Caribbean South Asia ECA Policies for Improving Connectivity ●  287 FIGURE 7.16  Preferential trade agreements in ECA 30 subregions Number of enforced preferential trade agreements 20 10 0 1958−79 1980−99 2000−15 Western Europe Northern Europe Central Asia Other Eastern Europe Southern Europe Western Balkans Russian Federation Central Europe South Caucasus Turkey Note: ECA = Europe and Central Asia. FIGURE 7.17  Preferential 40 trade agreements in ECA countries by income group Number of enforced 30 preferential trade agreements 20 10 0 58 70 80 00 10 90 15 19 19 19 20 20 19 20 High-income ECA countries Lower-middle-income ECA countries Upper-middle-income ECA countries Note: ECA = Europe and Central Asia. 288  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 7.18  Preferential 100 trade agreements in ECA subregions: Intra- versus extraregional integration 80 Percentage of enforced preferential trade agrements 60 with ECA countries 40 20 0 1958−79 1980−99 2000−15 Western Europe Northern Europe Central Asia Other Eastern Europe Southern Europe Western Balkans Russian Federation Central Europe South Caucasus Turkey Note: ECA = Europe and Central Asia. FIGURE 7.19  Preferential 100 trade agreements in ECA countries by income group: Intra- versus extraregional 90 integration Percentage of enforced preferential trade 80 agreements with ECA countries 70 60 50 58 70 80 90 00 10 15 19 19 19 19 20 20 20 High-income ECA countries Lower-middle-income ECA countries Upper-middle-income ECA countries Note: ECA = Europe and Central Asia. Bilateral Investment Agreements Similar patterns to those observed for PTAs emerge when looking at BITs. The descriptive analysis in this section relies on the UNCTAD investment hub data- base, which includes some 3,000 BITs.7 Investment agreements are instruments to reduce uncertainty for foreign investors regarding the policy environment they will confront after investing and that provide investors with security that they will be ECA Policies for Improving Connectivity ●  289 given fair and equitable treatment and not be expropriated without obtaining adequate compensation. In general, there is a great deal of commonality across most BITs in terms of substantive obligations. This has been changing in recent years following public concern regarding the use (perceived abuse) of arbitration to address disputes between investors and host governments regarding actions by governments that are deemed by inves- tors to violate the provisions of the BIT. The major developments in this area have been quite recent and center on the allocation of responsibility for BITs to the European Commission (as opposed to the member states) and the EU decision in 2016 to shift away from providing for arbitration to settle disputes toward the use of an investment court system. Figure 7.20 plots regional averages for the number of enforced BITs a country in the region is part of (averaged across years in three periods). ECA emerges as the first region in terms of integration though the BIT as a policy tool. Looking at the average number of enforced BITs with other countries in the same region and in other regions, respectively, it is apparent how ECA’s leading position is driven by intraregional integration (figures 7.21 and 7.22). Indeed, almost 60 percent of all BITs signed by the average ECA country involve a partner in ECA. Figure 7.23 reveals a shift from almost zero use of BITs in the first period to a policy stance in which the average ECA country is a partner in almost 50 BITs on average between 2000 and 2016. With the exception of Western Europe clearly anticipating these patterns, the transition phase started at the beginning of the 1990s and finished at the end of the first decade of the 2000s for all ECA subre- gions. After that, the average number of BITs is rather constant, with the resulting policy steady states spanning a range from a minimum of 30 BITs signed on aver- age in Central Asia to a maximum of almost 80 in Western Europe. Figure 7.24 FIGURE 7.20  Bilateral 50 investment treaties across global regions 40 Number of enforced bilateral investment treaties 30 20 10 0 1959–79 1980–99 2000–16 East Asia and Pacific Latin America and the Caribbean North America Sub-Saharan Africa Europe and Central Asia Middle East and North Africa South Asia 290  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 7.21  Bilateral 25 investment treaties across global regions: Intraregional integration 20 Number of enforced bilateral investment treaties within the region 15 10 5 0 1959–79 1980–99 2000–16 East Asia and Pacific Latin America and the Caribbean North America Sub-Saharan Africa Europe and Central Asia Middle East and North Africa South Asia FIGURE 7.22  Bilateral investment treaties across 30 global regions: Extraregional integration Number of bilateral investment treaties beyond the region 20 10 0 1959–79 1980–99 2000–16 East Asia and Pacific Latin America and the Caribbean North America Sub-Saharan Africa Europe and Central Asia Middle East and North Africa South Asia summarizes these patterns showing that an anticipated transition and a higher steady-state level of integration characterizes the average ECA HIC with respect to the average UMIC and LMIC. As for intra- versus extra-ECA integration, Western and Northern Europe and HICs show a clear pattern from disproportionate extraregional use of BIT toward a rather balanced mix of intra- and extraregional integration (see figure 7.25 and figure 7.26). When considering the average HIC in ECA, half of its BITs are signed ECA Policies for Improving Connectivity ●  291 FIGURE 7.23  Bilateral 80 investment treaties in ECA subregions Number of enforced bilateral 60 investment treaties 40 20 0 1959–79 1980–99 2000–16 Western Europe Northern Europe Central Asia Other Eastern Europe Southern Europe Western Balkans Russian Federation Central Europe South Caucasus Turkey Note: ECA = Europe and Central Asia. FIGURE 7.24  Bilateral 60 investment treaties in ECA countries by income group Number of enforced bilateral investment treaties 40 20 0 59 70 80 90 00 10 16 19 19 19 19 20 20 20 High-income ECA countries Lower-middle-income ECA countries Upper-middle-income ECA countries Note: ECA = Europe and Central Asia. with other ECA countries and half with non-ECA ones. Similar balanced stances are reached by Southern and Central Europe as well as by Turkey and Russia with the difference that they were starting from a disproportionate intraregional use of the BIT policy tool. A rather strong average bias toward intraregional BITs is appar- ent for the Western Balkans, the South Caucasus, Central Asia, Other Eastern Europe, and, from the perspective of income categories, for both UMICs and LMICs. 292  ●   Critical Connections: Why Europe and Central Asia’s Connections Matter for Growth and Stability FIGURE 7.25  Bilateral 100 investment treaties in ECA subregions: Intra- versus extraregional integration 80 Percentage of enforced bilateral investment treaties 60 with ECA countries 40 20 0 1959–79 1980–99 2000–16 Western Europe Northern Europe Central Asia Other Eastern Europe Southern Europe Western Balkans Russian Federation Central Europe South Caucasus Turkey Note: ECA = Europe and Central Asia. FIGURE 7.26  Bilateral trade 100 agreements in ECA countries by income group: Intra- versus extraregional 80 integration Percentage of enforced bilateral investment treaties 60 with ECA countries 40 20 0 59 70 80 90 00 10 16 19 19 19 19 20 20 20 High-income ECA countries Lower-middle-income ECA countries Upper-middle-income ECA countries Note: ECA = Europe and Central Asia. Product Market Regulation This section provides some descriptive evidence on the patterns characterizing domestic PMR and regulatory reforms over time that are potentially relevant for various dimensions of connectivity. It discusses indicators of PMR developed by the OECD, focusing both on horizontal and sectoral regulation. The broader ECA Policies for Improving Connectivity ●  293 economy-wide regulation of product markets is relevant from a connectivity dimension—for example, the role of the state in the economy and the extent to which regulatory regimes impede entry into sectors by new firms. If a country has relatively closed markets internally, this will have an impact on external connectiv- ity by affecting international competitiveness. Moreover, as discussed in the intro- duction, horizontal regulation and the quality of economic governance in a country have been found to determine the effect of efforts to integrate economies with the rest of the world. More directly relevant to connectivity is sectoral regulation of activities that directly affect connectivity and trade costs: regulation of entry into and the operation of energy, transport, and telecommunications network indus- tries. We also present data on the use of product standards by ECA and compara- tor countries, focusing on the extent to which countries notify the WTO regarding standards that diverge from international norms. These data provide an indication of the degree to which countries adopt international standards and participate in the WTO. Product Market Regulation: Aggregate “Horizontal” Indicators The data used in this subsection come from the PMR Economy Wide database, which contains measures of the degree of policy restrictiveness implied by domes- tic regulatory regimes. More precisely, it captures horizontal barriers to entrepre- neurship, barriers to trade and investment, and barriers embedded in the scope and nature of state control of the economy. All indicators range from 0 (no restric- tions) to 6 (maximum restrictiveness). As with other OECD databases, the country coverage of the ECA region is not complete—there is no consistent information on the Western Balkans, the South Caucasus, Central Asia, Other Eastern Europe, and Russia. The time dimension of the data consists of four observations: 1998, 2003, 2008, and 2013. Figure 7.27 plots the values of PMR Economy Wide overall score (horizontal PMR) for the average covered ECA country and compares it with the same score in a number of selected non-ECA countries (China and India are not observed at the beginning of the sample while the United States is not observed for 2013). While relatively less restrictive than Brazil, China, and India, the average ECA coun- try imposes higher restrictions than either Canada or the United States. Moreover, figure 7.27 reveals a very slow pace of policy progress for the average ECA country (especially between 2008 and 2013). A low degree of policy progress appears also across available subregions in ECA, in particular for Western and Northern Europe since 2008 (figure 7.28). This pattern is reflected in the almost negligible change between the 2008 score for HICs and the 2013 one (figure 7.29). Product Market Regulation: Sectoral Indicators This subsection uses a different PMR database denoted as the PMR ETCR data- base. PMR ETCR contains annual measures of the degree of policy restrictiveness implied by domestic regulatory regimes for specific sectors (energy, transport, 294  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 7.27  Horizontal 4 product market regulation in ECA and selected countries Index, 0 (least restrictive) to 6 (most restrictive) 3 2 1 0 1998 2003 2008 2013 ECA United States Canada Brazil China India Note: ECA = Europe and Central Asia. FIGURE 7.28  Horizontal 4 product market regulation in ECA subregions Index, 0 (least restrictive) to 6 (most restrictive) 3 2 1 0 1998 2003 2008 2013 Western Europe Southern Europe Central Europe Northern Europe Turkey Note: ECA = Europe and Central Asia. and communications) from 1975 to 2013. As was true for the economy-wide PMR, indicators range from 0 (no restrictions) to 6 (maximum restrictiveness). The coun- try coverage of the ECA region is again limited. Figure 7.30 plots the PMR ETCR overall score for the average covered ECA country as well as for a number of selected non-ECA countries (China, India, and the United States are not observed at the beginning of the sample). ECA Policies for Improving Connectivity ●  295 FIGURE 7.29  Horizontal 4 product market regulation in ECA countries by income group Index, 0 (least restrictive) to 6 3 (most restrictive) 2 1 0 1998 2003 2008 2013 High-income ECA countries Non-high-income ECA countries Note: ECA = Europe and Central Asia. FIGURE 7.30  Product 6 market regulation: Aggregate energy, transport, and communications regulations 4 in ECA and selected countries Index, 0 (least restrictive) to 6 (most restrictive) 2 0 1975–87 1988–2000 2001–13 ECA United States Canada Brazil China India Note: ECA = Europe and Central Asia. Figure 7.30 reveals that in terms of PMRs the average ECA country is always more restrictive than either Canada or the United States. Moreover, a halt in policy prog- ress is observed at the end of the sample for the average ECA country. Figures 7.31–7.34 plot the average score for PMR ETCR aggregate (time series), airlines, rail, and telecoms (averages across periods) in a number of ECA subre- gions. Available data confirm a number of general policy patterns. The data reveal a phase of policy progress across all sectors and all covered country groups 296  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 7.31  Product 6 market regulation: Aggregate energy, transport, and communications regulations 5 in ECA subregions Index, 0 (least restrictive) to 6 (most restrictive) 4 3 2 75 80 90 00 10 13 19 19 19 20 20 20 Western Europe Southern Europe Central Europe Northern Europe Turkey Note: ECA = Europe and Central Asia. FIGURE 7.32  Product 6 market regulation: Regulations regarding airlines in ECA subregions Index, 0 (least restrictive) to 6 (most restrictive) 4 2 0 1975–87 1988–2000 2001–13 Western Europe Southern Europe Central Europe Northern Europe Turkey Note: ECA = Europe and Central Asia. (with the notable exception of Turkey in the rail sector). The removal of regulatory restrictions began in the early 1990s and in many cases stopped toward the end of the first decade of the 2000s. Looking at the aggregate ETCR scores in figure 7.31, a small increase in restrictiveness can be observed for all subregions except Western Europe. As shown in figure 7.35, this pattern seems to be driven by poli- cies in HICs. ECA Policies for Improving Connectivity ●  297 FIGURE 7.33  Product 6 market regulation: Regulations involving railways in ECA subregions Index, 0 (least restrictive) to 6 (most restrictive) 4 2 0 1975–87 1988–2000 2001–13 Western Europe Southern Europe Central Europe Northern Europe Turkey Note: ECA = Europe and Central Asia. FIGURE 7.34  Product 6 market regulation: Regulations regarding telecommunications in ECA subregions Index, 0 (least restrictive) to 6 4 (most restrictive) 2 0 1975–87 1988–2000 2001–13 Western Europe Southern Europe Central Europe Northern Europe Turkey Note: ECA = Europe and Central Asia. Product Market Regulation: SPS and TBT Notifications to the WTO A final dimension of PMR that is relevant to connectivity and trade costs is con- vergence of national standards regimes with those prevailing internationally. There are two major types of national product standards that affect international trade and investment: health and safety norms for plants, animals, and humans— so-called sanitary and phytosanitary (SPS) measures and safety standards for 298  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 7.35  Product 6 market regulation: Aggregate energy, transport, and communications regulations 5 in ECA countries by income group Index, 0 (least restrictive) to 6 (most restrictive) 4 3 2 75 80 90 00 10 13 19 19 19 20 20 20 High-income ECA countries Non-high-income ECA countries Note: ECA = Europe and Central Asia. nonfood products—so-called technical product regulations, called technical bar- riers to trade (TBT) in the WTO. The WTO has specific disciplines that apply to the use of both SPS and TBT measures, including provisions that call on WTO mem- bers to adopt international standards if they exist and to notify regarding new proposed standards that may have an impact on trade and that are not based on internationally agreed-upon norms. Figures 7.36 and 7.37 show descriptive evidence on the number of notifica- tions of SPS and TBT measures by countries (data on notifications are taken from the WTO). More precisely, the two figures report averages across WTO member countries within groups averaged across years within three seven-year periods starting in 1995 (the year the WTO was established). Observe that for all WTO members as a whole (the “world”) the number of notifications of SPS and TBT measures has been increasing steadily since 1995, especially in the most recent period (2010–17). This increase over time is an indication of the increasing prevalence of this type of NTM, although, as noted in the introduc- tion, it does not necessarily reflect a desire to discriminate against foreign suppliers. What the data do reveal is that the EU and other WTO members make relatively more frequent use of standards that are not based on interna- tional norms—or adopt standards for which no internationally agreed-upon norms exist. The EU is the only part of ECA that notifies the WTO extensively— other ECA countries notify much less than the world (WTO-member) average. That said, Russia and Turkey have notified the WTO regarding more standards in recent years, whereas ECA countries that acceded to the EU in 2004 stopped notifying the WTO after accession, reflecting the fact that this is an EU compe- tence (part of the common commercial policy). ECA Policies for Improving Connectivity ●  299 FIGURE 7.36  Technical 80 barriers to trade and sanitary and phytosanitary 60 notifications in ECA Average annual number of notifications 40 20 0 1995−2001 2002−09 2010−17 European Union TBT SPS Russian Federation TBT SPS Central Europe, Accession 2004 TBT SPS Turkey TBT SPS South Caucasus TBT SPS Other Eastern Europe TBT SPS Central Asia TBT SPS World TBT SPS Note: ECA = Europe and Central Asia; SPS = sanitary and phytosanitary; TBT = technical barriers to trade. FIGURE 7.37  Technical 80 barriers to trade and sanitary and phytosanitary notifications in ECA across 60 income regions Average annual number of notifications 40 20 0 1995–2001 2002–09 2010–17 High-income ECA countries TBT SPS Upper-middle-income ECA countries TBT SPS Lower-middle-income ECA countries TBT SPS Note: ECA = Europe and Central Asia; SPS = sanitary and phytosanitary; TBT = technical barriers to trade. EBRD Transition Indicators Domestic policy reforms in the former Soviet Union countries across both Europe and the Western Balkans and the Eastern ECA regions (often missing from the PMR coverage) can be analyzed empirically using the Transition Indicators Database developed and managed by the EBRD. This database contains a 300  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia number of horizontal as well as sector-specific policy variables capturing the degree of transition from the policy stance in 1989. Each indicator takes the refer- ence value of 1 in 1989. In subsequent years, indicators vary between 1 (no prog- ress) and 4 (OECD policy standard). Among the various indicators in the database, we report a simple average of the scores for five infrastructural sectors: electric power, railways, roads, telecom- munications, and water and wastewater. Figure 7.38 plots the average scores for this aggregate indicator for different ECA subregions; figure 7.39 reports the aver- age scores for each income group. Former Soviet Union countries in Central and Northern Europe appear to be pioneers of transition with a steep increase in their scores from the beginning of the sample until the end of the 1990s. Since the early 2000s the pace of domestic reforms for these two groups slowed significantly, entering a slower but still posi- tive trend of policy effort. The trend becomes flat in the second half of the 2000s for the average member of the Central Europe subregion, suggesting a stop in policy progress at levels of domestic governance still below the OECD standard. An overall trend of policy progress emerges across all other subregions, but the usual pattern of heterogeneous policy stances holds for transition indicators as well. The distance from the OECD policy standard for the average country in Central Asia is four times the distance of the average country in Central or Northern Europe. The pattern of transition toward heterogeneous policy frameworks is con- firmed when adopting an income perspective in defining subsets of countries: HICs reach higher standards in terms of domestic governance of infrastructural sectors, followed in order by UMICs and LMICs, which are relatively close to each other. FIGURE 7.38  Aggregate 3.5 EBRD Transition Indicators for infrastructure in ECA subregions 3.0 2.5 2.0 1.5 1.0 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 19 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 20 Central Europe Northern Europe (Baltic) Western Balkans South Caucasus Central Asia Russian Federation Turkey Other Eastern Europe Note: EBRD = European Bank for Reconstruction and Development; ECA = Europe and Central Asia. ECA Policies for Improving Connectivity ●  301 FIGURE 7.39  Aggregate 3.5 EBRD Transition Indicators for infrastructure in ECA 3.0 countries by income group 2.5 2.0 1.5 1.0 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 19 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 20 High-income ECA countries Lower-middle-income ECA countries Upper-middle-income ECA countries Note: EBRD = European Bank for Reconstruction and Development; ECA = Europe and Central Asia. Policy Comovements—Are Policies Consistent? This ECA flagship report establishes that balanced connectivity is an important driver of growth. Being connected along one dimension, for example, trade, is not enough to enhance growth; countries need to be connected in many dimen- sions to exploit complementarities between different types of connections that can enhance economic growth. This section explores whether connectivity- related policies move in the same direction within countries across time. The exercise can be interpreted as an evaluation of the extent to which countries are coherent in their policies, for example, if a country decreases tariffs over time, does it tend also to reduce barriers to mobility or lessen the restrictiveness of policies toward FDI? We consider the connectivity-related policies discussed in the foregoing sections, plus policies on mobility restrictions and migrant integration (from ­ chapter 4) for up to 200 countries, depending on data availability. The indicators span the simple average MFN import tariff (denoted as Tariff); the number of PTAs to which each country belongs; the average value of the horizontal and sectoral OECD PMR indicators (PMR_H and PMR_S); a measure of policies of immigrant integration—the Migrant Integration Policy Index (MIPEX); a measure of statutory restrictions on FDI; the average of several EBRD transition indicators that measure reform progress toward best practices observed in industrial mar- ket economies; the number of BITs signed by each country; and a measure of restrictiveness of visa requirements imposed by each country, the Mobility Barriers Index (MBI). Each policy variable covers a different set of countries 302  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia and time period. In most cases data start in the late 1980s or early 1990s. Therefore, the sample of countries and years changes somewhat for each esti- mated correlation. Table 7.1 provides a description of the countries and the time coverage for each policy variable. Table 7.2 reports overall correlations between all policies in all countries and years covered.8 As expected, trade policies move in the same direction. Tariffs are negatively correlated with PTA and BIT, and positively correlated with FDI. Thus, lower tariffs are associated with more PTAs and BITs, and lower FDI restrictions. Moreover, FDI is negatively correlated with PTA and BIT, and posi- tively correlated with Tariff: lower FDI restrictiveness is associated with more BITs and PTAs and lower tariffs. These correlations suggest that countries simultaneously reduce barriers to trade, both unilaterally (MFN tariffs) and through PTAs (which generally include a focus on NTMs), and increase their openness to foreign investment. Moreover, countries more open to trade perform better in terms of reducing the restrictiveness of regulation (as measured by the PMR) and convergence toward what the EBRD defines as good market economy regulatory practice. This is reflected in the positive correlation between Tariff and PMR_H and PMR_S, respectively (more restrictive PMRs are associated with higher tariffs), and the neg- ative correlation between Tariff and EBRD—higher tariffs are associated with lower EBRD indicators (less convergence toward good practices). The mobility of ­ people tends to be more restricted if trade barriers are lowered, as shown by the negative correlation between openness to trade and investment and MBI/MIPEX, suggest- ing a political trade-off between trade and investment openness on the one hand and immigration policies on the other. TABLE 7.2  Comovements of Connectivity Policies Tariff PTA PMR PMR_S MIPEX FDIRRI EBRD BIT MBI Tariff 1.000 −0.462 0.672 0.599 −0.284 0.309 −0.503 −0.341 −0.352 PTA −0.462 1.000 −0.447 −0.580 −0.154 −0.645 0.484 0.624 −0.581 PMR_H 0.672 −0.447 1.000 0.793 −0.458 0.283 −0.500 −0.033 −0.227 PMR_S 0.599 −0.580 0.793 1.000 −0.263 0.178 −0.876 −0.672 0.237 MIPEX −0.284 −0.154 −0.458 −0.263 1.000 0.118 0.219 0.025 −0.176 FDIRRI 0.309 −0.645 0.283 0.178 0.118 1.000 −0.523 −0.267 0.065 EBRD −0.503 0.484 −0.500 −0.876 0.219 −0.523 1.000 0.801 0.095 BIT −0.341 0.624 −0.033 −0.672 0.025 −0.267 0.801 1.000 −0.156 MBI −0.352 −0.581 −0.227 0.237 −0.176 0.065 0.095 −0.156 1.000 Note: The table reports estimates of the Spearman correlation coefficient between different connectivity policies across countries and time. Tariff is the most-favored-nation tariff. PTA is the total number of preferential trade agreements to which each country belongs. PMR_H is the average value of different Organisation for Economic Co-operation and Development (OECD) indicators of product market regulation; a higher value means more regulation. PMR_S is the average value of different OECD indicators of product market service regulation; a higher value means more regulation. MIPEX, the Migrant Integration Policy Index, measures policies regarding immigrant integration; a higher value means more favorable integration. FDIRRI is the FDI Regulatory Restrictiveness Index, which measures statutory restrictions on foreign direct investment; a higher value means more restrictions. EBRD is the average of several transition indicators that track convergence over time toward best practices; a higher value means greater convergence toward a market economy. BIT is the total number of bilateral investment treaties signed by each country. MBI, the Mobility Barriers Index, measures the strictness of visa requirements imposed by each country; a higher value means greater strictness. Each policy variable is measured for different sets of countries and for different time periods. Therefore, the sample of countries and the years used change somewhat for each correlation estimated in the table. Refer to Table 7.1 for a description of the countries and time coverage of each policy variable. ECA Policies for Improving Connectivity ●  303 Countries with more restrictive regulatory regimes do less well in converging toward the standards of industrial market economies, reflected in a negative cor- relation between the PMR and the EBRD indexes. Better migration integration policies are associated with less restrictive PMR and a freer market economy, as suggested by the negative correlation between MIPEX and PMR, and a positive correlation between MIPEX and EBRD. Better migration integration policies, con- versely, are associated with more restrictions on inward FDI, as indicated by the positive correlation between MIPEX and FDI restrictions. Finally, lower tariffs are associated with greater restrictions on mobility of people, given the negative cor- relation between Tariff and MBI. The correlations between Tariff and PTA, BIT, and FDIRRI are a good measure of consistent international policy to the extent that it measures how countries choose to enhance connectivity through two alternative forms of market integration—trade versus investment. The correlation between Tariff and PMR, ­ instead, is a relevant measure of consistent domestic policy to the extent that it captures how countries choose to be connected to other countries and at the same time enhance connectivity by improving internal competition. Table 7.2 suggests that in most cases countries are relatively coherent in their policies toward connectivity. Countries that impose high tariffs tend to also sign fewer BITs. Higher tariffs are associated with lower product market liberalization, suggesting that international integration, measured by the cor- relation between Tariff and BIT, goes hand in hand with domestic integration, measured by the c ­ orrelation between Tariff and PMR. Figure 7.40 illustrates that countries have become more policy coherent over time, with the positive correlation between Tariff and PMR increasing (lower tariffs and less restrictive PMR). Similarly, the negative ­ correlation between Tariff and BIT has decreased over time (lower tariffs and more BITs). Figure 7.41 shows that the negative FIGURE 7.40  Global 0.6 evolution of selected policy comovements 0.4 Spearman correlation coefficients 0.2 0 –0.2 –0.4 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 8 0 09 10 11 12 13 14 15 20 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Tariff–product market regulation Tariff–bilateral investment treaty Tariff–Mobility Barriers Index 304  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 7.41  Evolution of selected policy 0.5 comovements across income groups Spearman correlation coefficients 0 –0.5 –1.0 Low Lower middle Upper middle High income income income income Tariff–product market regulation Tariff–bilateral investment treaty correlation between Tariff and BIT is driven by upper-­ - and high-income middle​ countries. There are substantial differences in domestic policy consistency between lower-middle-income and high-income countries. While the former show a negative correlation between Tariff and PMR, suggesting policy incon- sistency, the latter show a positive correlation between the two variables, sug- gesting policy consistency. Another measure of policy consistency is the correlation between the MBI and Tariff. Countries may substitute stricter movement of people for stricter movement of goods, as documented by the negative correlations between Tariff and MBI in table 7.2. Figure 7.42 shows that this applies to most HICs except Iceland, Austria, the United States, and Great Britain. Unlike other policy variables, Tariff and BIT are measured for almost all coun- tries and hence permit comparisons across geographic areas. Figure 7.43 shows the heterogeneity in the Tariff-BIT correlation between different geo- graphic areas within the ECA region. All country groups within ECA are policy coherent except Central Asia, which has a strong positive correlation between Tariff and BIT (that is, higher tariffs are associated with more BITs). Figure 7.44 compares different global regions. North America is more consis- tent in terms of policies, whereas ECA, Sub-Saharan Africa, and Latin America and the Caribbean show far less policy consistency. East Asia and Pacific, the Middle East and North Africa and South Asia, instead, are policy inconsistent with a posi- tive correlation between Tariff and BIT. ECA Policies for Improving Connectivity ●  305 FIGURE 7.42  Tariff–Mobility 1.0 Barriers Index comovements ISL across countries imposing mobility restrictions Spearman correlation 0.5 coefficients GBR 0 NOR AUT USA DNK SWE DEU GRC LUX NLD PRT ESP BEL ITA SVN –0.5 MLT FRA HUN ROU CZE SVK FIN BGR EST LVA POL –1.0 0.5 FIGURE 7.43  Tariff–bilateral investment treaty comovements across ECA subregions 0 Spearman correlation coefficients –0.5 –1.0 pe pe e e ns us sia n y pe rke op op tio lka s ro ro ro lA ca ur ur ra Tu Eu Eu Eu Ba lE nE au ra de rn rn rn nt ra hC Fe rn er te he te Ce nt te rth ian ut es as ut Ce es So No W rE So ss W Ru he Ot Finally, figure 7.45 compares two macroregions within ECA: the EU and Western Balkans, and Eastern ECA. The correlation between Tariff and BIT is negative in the first group (policy consistency for EU and Western Balkans) and positive in the lat- ter (policy inconsistency for Eastern ECA). This suggests substantial differences in the extent of policy consistency within the ECA region. 306  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia FIGURE 7.44  Tariff–bilateral 0.5 investment treaty comovements across global regions 0 Spearman correlation coefficients –0.5 –1.0 ific ia an a ca ia a fric ric As As eri be ac Af hA rib Am al uth dP ran ntr Ca ort an So rth Ce ha dN he ia No Sa nd As dt an b− ea an st st Su Ea rop ca Ea eri Eu le Am dd Mi tin La FIGURE 7.45  Tariff–bilateral investment treaty 0.2 comovements across ECA macrosubregions Spearman correlation coefficients 0 −0.2 −0.4 European Union and Eastern Europe Western Balkans and Central Asia Note: ECA = Europe and Central Asia. ECA Policies for Improving Connectivity ●  307 Conclusion ECA countries have been global leaders in cooperation through PTAs and BITs, and among HICs in facilitating immigration. However, the ­ average ECA country is more restrictive than non-ECA regions in domestic regulations and migrant inte- gration policies. The trend toward more open policies slowed significantly, particu- larly after the first decade of this century. Little progress was made in tariff liberalization (as of the beginning of the 2000s), the use of BITs (as of the end of the 2000s), or reductions of FDI regulatory restrictions and product market liberal- ization (as of 2010). It appears that ECA countries mostly pursued complementary policies across many policy dimensions of connectivity, particularly in tariff reductions, investment treaties, and lower FDI restrictions. Countries that are more open to trade also tend to have less restrictive domestic regulatory regimes. Nonetheless, lower trade barriers are not always associated with lower restrictions on immigration or product market restrictiveness, and some countries rely heavily on other ECA part- ners for global connectivity. Most higher-income countries have pursued comple- mentary policies in most areas of connectivity, but LMICs less so (e.g., lower tariffs are not uniform across partner countries and are associated with higher regulatory restrictions). For the average country in Central Asia, challenges remain in improv- ing the attractiveness of the business environment, including product market restrictions and infrastructure gaps. Notes 1. In 2019 the number of EU member states is expected to drop to 27, following the exit of the United Kingdom. 2. See http://trade.ec.europa.eu/doclib/docs/2006/december/tradoc_118238.pdf. 3. See UNCTAD (2015) for an international classification of different forms of NTMs. 4. See Services Trade Restrictions Database (http://iresearch.worldbank.org/servicetrade​ /aboutData.htm) and OECD Services Trade Restrictiveness Index (http://www.oecd​ .org/tad/services-trade/services-trade-restrictiveness-index.htm). The negative effects of policies restricting access of foreign producers to services markets on downstream productivity performance have been estimated in country studies (e.g., Arnold et al. 2011 for the Czech Republic; Arnold et al. 2016 for India) and across countries using both firm- and industry-level data (e.g., Barone and Cingano 2011; Bourlès et al. 2013; Hoekman and Shepherd 2017). 5. The focus of this chapter is descriptive. It provides an overview of the levels and trends in applied policy that directly affect many of the dimensions of connectivity considered in this report. The aim is to provide information to help understand what has been done by ECA in different policy domains that affect connectivity of countries and to place ECA policies in a comparative context. 6. To account for measurement error, we recode as missing the values of simple average MFN import tariffs when reported as equal to 0. Within the ECA region this is the case for Estonia (1998 and 1999), the Kyrgyz Republic (1995), Turkmenistan (1998), and Switzerland (1990; 1993–2015). 308  ●   Critical Connections: Promoting Economic Growth and Resilience in Europe and Central Asia 7. See http://investmentpolicyhub.unctad.org/IIA. 8. It is not possible to transform all of our indicators so that they point in the same direc- tion for ease of interpretation, that is, a higher positive (negative) number denotes greater (lower) policy restrictiveness and connectivity. Doing so would generate incon- sistencies in the definition and interpretation of the variables in the preceding sections. References Arnold, J., B. Javorcik, M. Lipscomb, and A. Mattoo. 2016. “Services Reform and Manufacturing Performance. Evidence from India.” Economic Journal 126 (590): 1–39. Arnold, J., B. Javorcik, and A. Mattoo. 2011. “Does Services Liberalization Benefit Manufacturing Firms? Evidence from the Czech Republic.” Journal of International Economics 85 (1): 136–46. Baldwin, R. 2016. The Great Convergence: Information Technology and the New Globalization. Cambridge, MA: Harvard University Press. Barone, G., and F. Cingano. 2011. “Service Regulation and Growth: Evidence from OECD Countries.” Economic Journal 121 (555): 931–57. Beverelli, C., M. Fiorini, and B. Hoekman. 2017. “Services Trade Policy and Manufacturing Productivity: The Role of Institutions.” Journal of International Economics 104: 166–82. Bourlès, Renaud, Gilbert Cette, Jimmy Lopez, Jacques Mairesse, and Giuseppe Nicoletti. 2013. “Do Product Market Regulations in Upstream Sectors Curb Productivity Growth? Panel Data Evidence for OECD Countries.” Review of Economics and Statistics 95 (5): 1750–68. Cadot, O., M. Malouche, and S. Sáez, 2012. Streamlining Non-tariff Measures: A Toolkit for Policy Makers. Washington, DC: World Bank. European Commission. 2015. “Joint Communication to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions: Review of the European Neighbourhood Policy.” JOIN(2015) 50 final, November 11. European Commission, Brussels. http://eeas.europa.eu/archives/docs​ /­enp/documents/2015/151118_joint-communication_review-of-the-enp_en.pdf. Fiorini, M., and B. Hoekman. 2017a. “Economic Governance, Regulation and Services Trade Liberalization.” Working Paper 2017/47, Robert Schuman Centre for Advanced Studies, European University Institute, Fiesole, Italy. http://cadmus.eui.eu/bitstream​ /­handle/1814/48006/RSCAS_2017_47.pdf?sequence=1&isAllowed=y. ———. 2017b. “Services Trade Policy, Domestic Regulation and Economic Governance.” European Economy Discussion Paper 2017/058, Publications Office of the European Union, Luxembourg. https://ec.europa.eu/info/sites/info/files/dp058_en.pdf. Freund, C., and B. Bolaky. 2008. “Trade, Regulations, and Income.” Journal of Development Economics 87 (2): 309–21. Gawande, K., B. Hoekman, and Y. Cui. 2015. “Global Supply Chains and Trade Policy Responses to the 2008 Financial Crisis.” World Bank Economic Review 29 (1): 102–28. Hoekman, B. 2016. “Deep and Comprehensive Free Trade Agreements.” Working Paper 2016/29, Robert Schuman Centre for Advanced Studies, European University Institute, Fiesole, Italy. Hoekman, B., J. Jensen, and D. Tarr. 2014. “A Vision for Ukraine in the World Economy: Defining a Trade Policy Strategy that Leverages Global Opportunities.” Journal of World Trade 48 (4): 795–814. ECA Policies for Improving Connectivity ●  309 Hoekman, B. and B. Shepherd. 2017. “Services Productivity, Trade Policy, and Manufacturing Exports.” World Economy 40 (3): 499–516. Hofmann, C., A. Osnago, and M. Ruta. 2017. “Horizontal Depth: A New Database on the Content of Preferential Trade Agreements.” Policy Research Working Paper 7981, World Bank, Washington, DC. Jafari, Y., and D. Tarr. 2017. “Estimates of Ad Valorem Equivalents of Barriers against Foreign Suppliers of Services in Eleven Services Sectors and 103 Countries.” World Economy 40 (3): 544–73. Langbein, J. 2014. “European Union Governance towards the Eastern Neighbourhood: Transcending or Redrawing Europe’s East-West Divide?” Journal of Common Market Studies 52 (1): 157–74. Langbein, J., and K. Wolczuk. 2012. “Convergence without Membership? The Impact of the European Union in the Neighbourhood: Evidence from Ukraine.” Journal of European Public Policy 19 (6): 863–81. Miroudot, S., and B. Shepherd. 2016. “Trade Costs and Global Value Chains in Services.” In Research Handbook on Trade in Services, edited by M. Roy and P. Sauvé, 66–84. Cheltenham, UK: Elgar. Moïse, E., and F. Le Bris. 2015. “Trade Costs: What Have We Learned? A Synthesis Report.” OECD Trade Policy Paper 150, Paris. Rodriguez, F., and D. Rodrik. 2001. “Trade Policy and Economic Growth: A Skeptic’s Guide to the Cross-National Evidence.” In NBER Macroeconomics Annual, edited by Ben S. Bernanke and Kenneth Rogoff, 261–325. Cambridge, MA: MIT Press. UNCTAD (United Nations Conference on Trade and Development). 2015. “International /­ Classification of Non-tariff Measures.” UNCTAD/DITC​ TAB/2012/2/Rev.1, United Nations, Geneva. ECO-AUDIT Environmental Benefits Statement The World Bank Group is committed to reducing its environmental footprint. In support of this commitment, we leverage electronic publishing options and print-on-demand technology, which is located in regional hubs worldwide. Together, these initiatives enable print runs to be lowered and shipping distances decreased, resulting in reduced paper ­ consumption, chemical use, greenhouse gas emissions, and waste. We follow the recommended standards for paper use set by the Green Press Initiative. The majority of our books are printed on Forest Stewardship Council (FSC)–certified paper, with nearly all containing 50–100 percent r­ ecycled content. The recycled fiber in our book paper is either unbleached or bleached ­ using totally chlorine-free (TCF), p ­ rocessed chlorine-free (PCF), or enhanced elemental chlorine-free (EECF) processes. More information about the Bank’s environmental philosophy can be found at http://www.worldbank.org/corporateresponsibility. T he countries of the Europe and Central Asia region, along with much of the rest of the world, find themselves engaged in a revival of one of the fundamental questions of economic policy: how much to open to the rest of the world. This question now dominates the political economy of the region, not just within the advanced economies of the European Union but also among the region’s emerging market economies. In Critical Connections, the World Bank offers new research on the process of economic integration, showing its potential benefits without ignoring the downsides. The report examines how trade, investment, migration, and other linkages among countries drive economic growth in the Europe and Central Asia region. It breaks new ground by using a multidimensional approach that recognizes how each connectivity channel is likely to be affected by the strength of other channels. The multidimensional view offered by this approach makes it clear that diversity in country connections and balance in all channels of connectivity are critical for achieving the greatest impact on growth and economic resilience. Europe and Central Asia provides a great laboratory for observing the role of multidimensional connectivity in action. The region’s 47 countries vary widely in the degree of openness of their economies. Its collective experience shows how the various elements of cross-border connectivity work together to accelerate progrowth knowledge transfers, which in turn boost productivity through participation in today’s global value chains. Which countries a country has as its economic partners might be just as important as the type of connection it has with them, because being well connected to highly connected countries can provide benefits beyond being well connected to comparatively isolated countries. Although greater connectivity can expose countries to external shocks, the report presents a fact-based argument for policies that seek to build deeper and more diverse connections within the Europe and Central Asia region and globally. The message is timely. Europe’s once-confident march toward economic integration has slowed over the past decade, with voices in many countries questioning the wisdom of opening to the global economy. Critical Connections serves as a reminder to citizens and policy makers that greater regional and global connectivity has been a tremendous “convergence machine,” raising living standards in lower-income countries toward those of wealthier middle- to high-income countries. By exploring multidimensional connectivity and its impact, the report provides a framework for understanding the many benefits and challenges of globalization and helps provide information for policy discussions and actions that recognize how the various aspects of connectivity might work together to deliver resilient and faster growth. Europe and Central Asia Studies Europe and Central Asia Studies feature analytical reports on the main challenges and opportuni- ties faced by countries in the region, with the aim of informing a broad policy debate. Titles in this regional flagship series undergo extensive internal and external review prior to publication. ISBN 978-1-4648-1157-9 SKU 211157