102987 SACU in Global Value Chains: Measuring GVC integration, position, and performance of Botswana, Lesotho, Namibia, South Africa, and Swaziland Prepared by Jakob Engel1, with inputs from Deborah Winkler and Thomas Farole 1 Corresponding author: please contact on: Jakob.engel@gmail.com This is a Working Paper of the World Bank – it is being issued in an effort to share ongoing research. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent Table of Contents TABLE OF CONTENTS ......................................................................................................................... I LIST OF FIGURES ................................................................................................................................... II LIST OF TABLES ..................................................................................................................................... III ACKNOWLEDGEMENTS .................................................................................................................... IV ACRONYMS AND ABBREVIATIONS .................................................................................................... V 1. INTRODUCTION ......................................................................................................................... 1 1.1. WHY GVCS MATTER .................................................................................................................... 1 1.2. MEASURING GVC PARTICIPATION ................................................................................................... 1 1.3. THE RELEVANCE OF GVCS FOR SACU COUNTRIES ................................................................................ 2 1.4. OBJECTIVES AND STRUCTURE OF THIS NOTE ........................................................................................ 3 2. SCOPE AND METHODOLOGY ...................................................................................................... 4 2.1. DATA SOURCES AND BACKGROUND TO THE EORA DATABASE .................................................................. 4 2.2. SCOPE OF COVERAGE: GEOGRAPHICAL AND SECTORAL ........................................................................... 4 2.3. ASSESSMENT OF RELIABILITY OF EORA DATA AT THE NATIONAL AND SECTORAL LEVEL .................................... 5 3. SACU EXPORTS AND STRUCTURAL INTEGRATION IN GLOBAL TRADE NETWORKS ......................... 7 3.1. OVERALL TRADE INTEGRATION ........................................................................................................ 7 3.2. OVERVIEW OF MAIN TRADED SECTORS .............................................................................................. 7 3.3. PROXYING GVC INTEGRATION: TRADE IN INTERMEDIATES...................................................................... 9 3.4. POSITIONING IN GLOBAL TRADE NETWORKS ..................................................................................... 12 4. STEPPING INTO GVCS: MEASURING EXPORT VALUE-ADDED IN SACU ........................................ 16 4.1. WHY UNDERSTANDING (DOMESTIC) VALUE ADDED IS IMPORTANT IN STUDYING GVCS................................ 16 4.2. IS DOMESTIC VALUE ADDED INCREASING IN SACU COUNTRIES?............................................................. 17 4.3. WHERE IS DVA GROWTH COMING FROM? – SECTORAL ASSESSMENT ..................................................... 21 5. GLOBAL VALUE CHAIN PARTICIPATION AND POSITIONING........................................................ 27 5.1. GVC PARTICIPATION – INTRODUCTION AND OVERALL INDEX................................................................. 27 5.2. WHAT IS DRIVING GVC PARTICIPATION? – FORWARD VERSUS BACKWARD INTEGRATION ............................ 28 5.3. SECTORAL DRIVERS OF GVC PARTICIPATION ..................................................................................... 30 5.4. GEOGRAPHICAL DRIVERS OF GVC PARTICIPATION .............................................................................. 32 5.5. ASSESSING GVC POSITIONING ...................................................................................................... 33 6. SUMMARY CONCLUSIONS........................................................................................................ 36 REFERENCES ................................................................................................................................... 38 APPENDICES ................................................................................................................................... 40 APPENDIX 1: OVERVIEW OF EORA AND OTHER MRIOS ................................................................................ 40 APPENDIX 2: METHODOLOGY FOR VALUE-ADDED ANALYSIS .......................................................................... 43 APPENDIX 3: CORRESPONDENCE BETWEEN EORA SECTORS AND ISIC REV.3 ...................................................... 46 APPENDIX 4: EXPORTS AND IMPORTS FOR SACU COUNTRIES - COMPARING EORA AND COMTRADE ....................... 47 APPENDIX 5: GROWTH OF DVA EMBODIED IN GROSS EXPORTS, BY SECTOR....................................................... 58 APPENDIX 6: FVA IN EXPORTS AS A SHARE OF GROSS EXPORTS, BY SECTOR ....................................................... 60 APPENDIX 7: FORWARD AND BACKWARD INTEGRATION BY PARTNER COUNTRY .................................................. 63 i List of figures Figure 1: Global comparison of trade openness by national income level ................................................. 7 Figure 2: Seller and buyer functions .......................................................................................................... 10 Figure 3: Intermediates as a share of gross exports and imports (2000, 2012) ........................................ 11 Figure 4: Minimal spanning tree: trade in consumption goods (2010), SACU in red ................................ 14 Figure 5: Minimal spanning tree: trade in intermediate goods (2010), SACU in red ................................ 15 Figure 6: From gross exports to domestic value added: decomposition of gross exports in the auto industry ............................................................................................................................................. 16 Figure 7: Compound annual growth rate of domestic value added embodied in gross exports, 2000-2011 .......................................................................................................................................................... 18 Figure 8: DVA embodied in gross exports as a share of GDP, 2000 and 2011 .......................................... 18 Figure 9: Compound annual growth rate of domestic value added embodied in gross exports, and of gross exports, 2000-2011 ........................................................................................................................... 19 Figure 10: DVA embodied in gross exports as share of gross exports, 2000 and 2011 ............................ 20 Figure 11: Compound annual growth rate of sectoral domestic value added embodied in gross exports for the Agriculture and Food & Beverages sectors, 2000-11............................................................ 23 Figure 12: DVA embodied in gross exports as share of gross exports for the agriculture and food & beverages sectors, 2000 and 2011 ................................................................................................... 24 Figure 13: Compound annual growth rate of sectoral domestic value added embodied in gross exports for the ............................................................................................................................................... 24 Figure 14: DVA embodied in gross exports as share of gross exports for hotels & restaurant sector, 2000 and 2011 ........................................................................................................................................... 25 Figure 15: Compound annual growth rate of sectoral domestic value added embodied in gross exports for the textiles & wearing apparel and transport equipment sectors, 2000-2011 .......................... 26 Figure 16: DVA embodied in gross exports as share of gross exports for the textiles & wearing apparel and transport equipment sectors, 2000 and 2011 .................................................................................. 26 Figure 17: GVC participation index 2000 and 2011 .................................................................................. 28 Figure 18: Foreign value added (left) and indirect value added (right) embodied in gross exports as share of gross exports, 2000 and 2011....................................................................................................... 29 Figure 19: Compound annual growth rate of foreign and indirect value added embodied in gross exports, 2000-2011 ......................................................................................................................................... 29 Figure 20: Compound annual growth of foreign value added and indirect value added by SACU country, by sector, 2000-2011 ........................................................................................................................ 30 Figure 21: Compound annual growth rate of sectoral domestic value added embodied in gross exports for the ............................................................................................................................................... 32 Figure 22: Import and Export Upstreamness and Domestic GVC length .................................................. 34 Figure 23: Main strands of empirical research on GVCs ........................................................................... 40 Figure 24: Compound annual growth rate of DVA embodied in gross exports by sector, 2000-2011 and 2006-2011 ......................................................................................................................................... 58 Figure 25: Foreign value added in exports by sector, 2000 and 2011 ...................................................... 60 Figure 26: Foreign and indirect value added in exports by source and destination, 2011 ....................... 63 ii List of tables Table 1: Structure of exports by SACU country (2013) – by SITC(2) section (1-digit classification) ............ 8 Table 2: Structure of imports by SACU country (2013) – by SITC(2) section (1-digit classification) ........... 9 Table 3: Intermediates as a share of gross exports (2012)- selected sectors ........................................... 11 Table 4: Intermediates as a share of gross imports (2012)- selected sectors ........................................... 12 Table 5: Centrality Ranking for SACU, World Ranking ............................................................................... 13 Table 6: Centrality Measures for South Africa and Peer Countries, World Ranking 2010, Total Intermediates.................................................................................................................................... 13 Table 7: DVA in exports for 2011, compound annual DVA growth, and DVA in exports as share of exports for 5 largest sectors by DVA, 2000-2011 .......................................................................................... 22 Table 8: Export upstreamness, selected sectors (2012) ............................................................................ 35 Table 9: Domestic GVC length, selected sectors (2012) ............................................................................ 35 Table 10: Overview of main MRIO databases ........................................................................................... 41 Table 11: Top 10 import and export sources (Eora) for Botswana, 2011 ................................................. 47 Table 12: Top 10 import and export sectors (Comtrade) for Botswana, 2011 ......................................... 48 Table 13: Top 10 import and export sectors (Eora) for Lesotho, 2011 ..................................................... 49 Table 14: Top 10 import and export sectors (Comtrade) for Lesotho, 2011 ............................................ 49 Table 15: Top 10 import and export sectors (Eora) for Namibia, 2011 ..................................................... 50 Table 16: Top 10 import and export sectors (Comtrade) for Namibia, 2011 ............................................ 50 Table 17: Top 10 import and export sectors (Eora) for South Africa, 2011 .............................................. 51 Table 18: Top 10 import and export sectors (Comtrade) for South Africa, 2011...................................... 52 Table 19: Top 10 import and export sectors (Eora) for Swaziland, 2011 .................................................. 53 Table 20: Top 10 import and export sectors (Comtrade) for Swaziland, 2011 ......................................... 53 Table 21: Non-exported output and standard deviation of 15 key sectors for Botswana, 2000 and 2011 .......................................................................................................................................................... 54 Table 22: Non-exported output and standard deviation of 15 key sectors for Lesotho, 2000 and 2011 . 55 Table 23: Non-exported output and standard deviation of 15 key sectors for Namibia, 2000 and 2011 56 Table 24: Non-exported output and standard deviation of 23 key sectors for South Africa, 2000 and 2011 .......................................................................................................................................................... 56 Table 25: Non-exported output and standard deviation of 15 key sectors for Swaziland, 2000 and 2011 .......................................................................................................................................................... 57 iii ACKNOWLEDGEMENTS This report was commissioned by the World Bank’s Trade and Competitiveness Global Practice and carried out between June and December 2014. Valuable comments and advice were received by Daria Taglioni and Deborah Winkler (World Bank) as well as by Marie-Agnes Jouanjean (Research Fellow, Overseas Development Institute). Permission to use the Eora database, as well as helpful advice and information on the data by the Eora team, and particularly Dan Moran, is gratefully acknowledged. iv ACRONYMS AND ABBREVIATIONS AfDB African Development Bank BLNS Botswana, Lesotho, Namibia, and Swaziland CAGR compound annual growth DVA domestic value added fob free on board FVA foreign value added GDP gross domestic product GVC global value chain I2E importing to export IOT input output table ISIC International Standard Industrial Classification IVA indirect value added MRIO multi-region input output table ODI Overseas Development Institute OECD Organisation for Economic Cooperation and Development SACU Southern African Customs Union SADC Southern African Development Community SUT supply and use table TiVA Trade in Value Added (database) WIOD World Input Output Database WTO World Trade Organization v 1. INTRODUCTION 1.1. Why GVCs matter Once concentrated among a few large economies, global flows of goods, services, and capital now reach an ever larger number of economies worldwide. Global trade in goods and services increased 10 times between 1980 and 2011, while FDI flows increased almost 30-fold. The sales from foreign-owned firms amount to $26 trillion. As many as 3,000 bilateral investment treaties have been signed to create the framework of deep agreements needed not only to facilitate the global movement of final goods and services but also to internationalize entire processes of production. All these flows have grown over time, creating increasingly dense and complex networks. One of the most significant reasons behind this transformation in global trade and investment has been the rise of global value chains (GVCs). Falling transport costs, greater global openness and cooperation on trade policy, and the ICT revolution have allowed production processes to be increasingly unbundled and shared across countries. Developing countries now join GVCs to further increase their economic competitiveness and they industrialize by densifying their participation. This is a huge change from the 20th century when countries had to build entire supply chains domestically to become competitive internationally (Baldwin 2006; see Antràs and Rossi-Hansberg 2009 and Ahmad 2013 for an overview of the literature). An implication of this is that GVCs denationalize comparative advantage, and this changes the options facing developing and developed nations, participants and non-participants. Globally competitive ‘lead firms’ knit together national comparative advantages to build components in the most cost‐effective locations. Factories in developing nations have become full-fledged participants in international manufacturing networks. These factories are no longer are merely importing parts for assembly, in order to service domestic markets; rather, they are exporting parts and components used in some of the most sophisticated products on the planet. In short, 20th century globalization was about made ‐ here ‐ sold‐there goods crossing borders: the trade system helped those nations able to produce finished goods domestically, sell products abroad. But 21st century globalization is also about factories crossing borders, so intra-factory flows of goods, know-how, investment, training, ideas, and people are now international commerce. The trade system now can help nations make things, not just sell things. 1.2. Measuring GVC participation In this context, understanding a country’s current participation in value chains is beneficial to ensuring that its industrial and trade policies can facilitate sustainable productivity gains and increased quality employment within higher value-added sectors. But examining trade participation and performance through a GVC lens requires a revised way of measuring and analyzing cross-border and cross-industry flows in goods and services. In particular competitiveness in specific components and tasks (rather than comparative advantage in end products) is paramount, enabling participation within larger production networks and, in turn, increased value addition generated domestically over time. Indeed, from a country perspective what matters ultimately is the value addition generated in the country from its export activity, and whether it increases (nominally) over time. This is not a new question for economics. Value addition is a function of productivity, but it is associated with the breadth, variety, and sophistication of tasks and activities in which a country specializes. The concept of domestic value added in exports is, therefore, an essential concept to understand the importance of GVC trade for a country. This concept allows us to distinguish the foreign and domestic content of a country’s exports, at it also 1 accounts for the fact that some of the imported inputs may contain domestic value added that is processed in a foreign location and reimported. This concept is explained in more detail in Appendix 1. Analysis of value added trade is based on the use of input-output tables which, while sacrificing the specificity of using customs classifications on parts or components, allow tracking usage explicitly at a sectorally disaggregated level and differentiating between transactions that are intermediates and those consumed as final demand by firms, governments, or consumers. This has been central to a recent proliferation of studies examining the development of value added in production and trade, starting with Hummels et al. (2001) and more recently elaborated by Koopman et al. (2010), Foster et al. (2011), and Johnson and Noguera (2012), among others.2 1.3. The relevance of GVCs for SACU countries GVCs offer potential new opportunities for the five countries of the Southern African Customs Union (SACU): Botswana, Lesotho, Namibia, South Africa, and Swaziland. With global offshoring continuing to grow and wages rising in China and elsewhere in East Asia, substantial migration of value chains to Sub- Saharan Africa is expected (Lin, 2011). The SACU region – with its abundance of natural capital and surplus labor, along with a relatively high quality infrastructure and institutional environment – may be in a good position to attract GVC-oriented investment. Beyond assembly manufacturing that is typical of GVCs (e.g. apparel, electronics, automotive), the region should also be well-placed to compete as a location for value-addition to agricultural and mineral commodities (“beneficiation”). Both types of investment would not only drive exports and have the potential to create significant employment, but also support productivity upgrading by accessing global technologies and knowledge. While across the region there is significant interest in facilitating integration into GVCs as well deepening integration of regional value chains (RVCs), there remains limited evidence of the extent of current integration, either globally or regionally. The most systematic assessment to date comes in the African Economic Outlook (AfDB 2014), which notes that the wider Southern Africa is currently leading the continent in terms of GVC participation, accounting for 40% of Africa’s combined backward and forward integration, driven primarily by South Africa. In discussing the relevance of GVCs for the five SACU countries individually, the aforementioned African Economic Outlook summarizes the main GVC-related constraints and opportunities for each African country, including short profiles of the SACU countries (see Box 1). Substantial analysis has also been carried out within the more qualitative GVC literature of particular value chains in the SACU countries, including South Africa’s automobile sector (c.f. Barnes and Kaplinsky 2000; Black 2001; Barnes and Morris 2008) and horticulture industry (c.f. Barrientos and Visser 2012), and Lesotho’s and Swaziland’s apparel industries (c.f. Morris, Staritz and Banes 2011; Staritz and Morris 2013). However, with the exception of South Africa, which is featured in the recently released WIOD and OECD/WTO TiVA multi-region input-output databases, little empirical data on GVC participation, positioning, and performance is available for the other SACU countries. Box 1: African Economic Outlook’s assessment of current and potential value chains for SACU countries Drawing on the African Economic Outlook’s country notes (AfDB 2014), the region’s most recent trade policy review (WTO Secretariat 2009), as well as a cursory review of country strategy documents, the SACU countries have, when compared to other African countries, made greater inroads in terms of value chain integration. Botswana has a relatively open economy, and has benefited substantially from its natural resource boon. The sectors in which Botswana is most engaged in global trade are mining, vehicles, textiles, beef and tourism, with 2 A third approach lies in using customs data on processing trade (Gorg 2000, Swenson 2013, Baldwin and Lopez-Gonzalez 2013). 2 diamonds to the US, Europe, and Japan as the largest export commodity. Currently its manufacturing exports are only growing slowly and have low domestic value added and high import content. Increasing value addition from mining (by, for example, carrying out more processing and manufacturing domestically) as well as the growing tourism sector are seen to have substantial potential. The clothing, textile, and livestock value chains are of central importance for Lesotho. Clothing accounts for 60% of total exports and employs 80% of the country’s manufacturing labor force, functioning as the largest employer outside of government. Currently, Lesotho’s clothing sector participates primarily within US -based buyer-driven GVCs but there is scope to expand further the recently growing market share in South Africa, given Lesotho’s geographical advantages over Chinese imports. Livestock is currently the largest contributor to agricultural value added and could be further developed as an export sector. This will, however, require overcoming significant challenges including poor nutrition and low quality products, weak market links, and limited access to financial services. For textiles, there are concerns about inadequate skills. Despite recent efforts at diversification – particularly in the fish and meat processing and mineral industries – Namibia currently remains relatively minimally integrated into value chains. The country’s extraction and processing of minerals is the main growth driver but given its relative capital-intensity, it has only limited employment impacts. While constrained by skills shortages (especially at the mid-level) and regulatory obstacles, Namibia’s proximity to South Africa and its well-developed infrastructure offers potential to connect to regional and global supply chains. This has motivated the recent development of export-processing zones and the granting of special manufacturing incentives. South Africa is unique on the continent for the scale of its participation (and in some cases leadership) within GVCs, including the automobile, mining, finance, and agriculture sectors. In manufacturing (and particularly automobiles) it serves as an assembly hub for Africa, and this industry accounts for more than 6% of GDP. Mining, which is predominantly locally owned, is even more significant, accounting for 19% of GDP while the sector has substantial spillovers into financial services and housing. The finance and retail industries also have substantial presences in other African countries. According to the AEO, South Africa’s advantages pertain particularly to skills, research capacities, as well as well-developed and dense networks of local supply industries and services. Despite Swaziland’s declining attractiveness to foreign investors (following the investment boom of the 1990s), the role of GVCs remains significant and investment stock remains high considering the size of the economy. Its main exports include sugar and sugar products, forestry, processed fruit, textiles, soft drinks, and some diverse manufactures. Value addition is hampered by the only limited stages of production that the country engages in and, in turn, a heavy reliance on South Africa for goods and services inputs. Source: http://www.africaneconomicoutlook.org/en/countries/ 1.4. Objectives and structure of this note This note is intended provide an overview of SACU countries’ participation and performance in GVCs, drawing on several data sources and indicators, and most importantly the recently released 189-country Eora multi-region-input-output (MRIO) database (Lenzen et al. 2012, 2013). Following this introduction, the note is structured in five additional sections. Section 2 discusses in greater detail the scope of the report, including the data sources and methodological approaches, as well as their respective limitations. Section 3 looks at structural integration in trade, including the degree to which SACU countries import and export intermediates. Section 4 analyzes trends in value-added exports as a first step in exploring GVC participation. Section 5 hones in on the core measures of GVC participation and a brief analysis of SACU countries’ position in GVCs. Finally, Section 6 concludes by bringing together the main findings from the analysis. 3 2. SCOPE AND METHODOLOGY 2.1. Data sources and background to the Eora database This note makes use of several data sources to carry out the analysis. Aggregate trade data and data on trade in intermediates (Section 3) comes from UN Comtrade, as does the analysis of upstreamness (Section 5). In addition, the note draws on indicators of domestic value added embodied in gross exports based on the Francois et al. (2013) database3. The most important data source for the analysis of GVCs, however, is the Eora database – this is discussed in more detail below. The simplified Eora database is disaggregated into 189 countries and 26 sectors per country (including a ‘rest of world’ sector that captures statistical discrepancies)4. It is thus the only MRIO database that has relatively comprehensive coverage for sub-Saharan Africa. This makes it well suited for longitudinal analysis of value chain integration of developing countries not included in other datasets. 5 The Eora database, much like the OECD/WTO’s TiVA database, uses available information to produce measures of trade in value-added for all countries. In order to produce a contiguous and continuous dataset, values has been interpolated for countries lacking necessary data. Eora has a historical time series spanning 1990-2011 based on an iterative process using an initial year estimate for 2000, overlaying estimates for 1999 and 2001, respectively with new data, and then re-balancing. In the past year, some analysis derived from the Eora database has been published in the African Economic Outlook (AfDB 2014) and the World Investment Report (UNCTAD 2013). As such, the Eora dataset allows for an approximate replication, albeit at a somewhat lower level of precision, of the kind of analysis undertaken for other countries using the WIOD or TiVA databases. Appendix 1 provides a more detailed discussion of the development of different MRIOs, and Eora specifically. In terms of key indicators, this report draws on the methodologies first developed by Hummels, Ishii, and Yi (2001) in measuring vertical specialization and in turn formalized by Koopman, Powers, Wang, and Wei (2011) (and later Koopman, Wang, and Wei 2014) to derive some of the most commonly used trade-in-value-added indicators, including domestic and foreign value-added, as well as value added embodied in other country’s intermediate inputs – i.e. forward and backward integration (see Appendix 2). Thus far most of these indicators have only been available publically for developed and other emerging economies through the WTO-OECD TiVA database and the WIOD database (both released in 2013). 2.2. Scope of coverage: geographical and sectoral In order to provide a meaningful context for comparative analysis with the five SACU countries, 14 ‘peer countries’ in sub-Saharan Africa, South Asia, and Latin America have been selected. For each of the five SACU countries, decomposed value-added measures (using f.o.b. prices, in current USD) are provided over 11 years (2000, 2006, and 2011) and placed next to peer countries in order to provide a relevant context for these countries’ GVC integration. The peers include: 3 Backward linkages in the Francois et al. (2013) database serve as a reasonable proxy for the domestic value added embodied in exports, as the share of re-imported intermediates is generally negligible. This analysis draws on input-output data available from the GTAP dataset. 4 This is the condensed version of Eora with countries that have more than 26 sectors in their input-output or supply-use tables having their accounts simplified. However, this does not apply to Botswana, Lesotho, Namibia and Swaziland, which all have just 26 sectors even in the expanded Eora database. 5 In comparative analysis with the WIOD dataset, Eora was found to provide broadly similar results when calculating foreign and domestic value added, albeit with a slight upward bias (which is to be expected as the greater number of highly heterogeneous developing countries, many of which have been subsumed in WIOD’s rest of world matrix) (UNCTAD 2013). 4  Southern African Development Community (SADC) neighbors with resource-rich economies: Mozambique, Tanzania and Zambia  Other African countries that have been reasonably successful at integrating into GVCs: Kenya, Mauritius, and Rwanda.  A selection of Asian and South American low- and middle-income countries with economic and/or geographical structures that are similar to one or more SACU country: Argentina, Bolivia, Cambodia, Chile, Lao PDR, Paraguay, Peru, and Thailand. Key sectors for each country are also analyzed. Here selection drew on whether the relevant sector was tradable, as well as how significant they were as export sectors (an overview of this is provided in Section 3.1 and Appendix 3). Based on Eora’s 26-sector classification system6, the following fourteen sectors were selected for closer consideration for each of the SACU countries: 1. Agriculture 2. Fishing 3. Electrical and machinery 4. Financial intermediation and business services 5. Food and beverages 6. Hotels and restaurants 7. Metal products 8. Other manufacturing 9. Petroleum, chemicals, and non-metallic minerals 10. Post and telecommunications 11. Textiles and wearing apparel 12. Transport 13. Transport equipment 14. Wood and paper 2.3. Assessment of reliability of Eora data at the national and sectoral level In most cases Eora results provide a reasonably accurate estimate for key indicators of GVC competitiveness at the country level, and of relative performance of different sectors, both within a country and in relation to comparator countries. Further, they are likely to provide a largely reliable approximation of the sectoral decomposition of value added and the direction of value-added trade, as well as trends for these indicators over an 11-year time period (especially as all three data-points are derived from the same source and methodology). Thus, in the absence of national input-output tables (IOTs) and supply-use tables (SUTs), and despite uncertainties (particularly at the sectoral level), Eora provides the best available and internationally comparable dataset for calculating key value-added trade indicators, though they are best complemented by more nuanced sectoral analysis drawing on alternative methodologies (e.g. producer surveys, firm-level analysis, and case studies). However, a few caveats relating particularly to the accuracy of the Eora data are in order. Firstly, it is important to bear in mind that Eora’s MRIO tables are modeled based on existing sources – national accounts data, Comtrade import and export data, among others – when national input-output or supply- use tables were not available. Further, in order to achieve the MRIO’s overall balancing requirements, raw data on imports and exports has frequently been adjusted, with the overall focus on representing large data items and fulfilling balancing conditions for large countries. Eora’s optimization approach attempts to strike a balance between the frequent conflicts between country-wise total exports and 6 Correspondence to ISIC classifications can be found in Appendix 3. 5 imports and trade balances but this can lead to substantial uncertainties, particularly for small values (such as for the four SACU countries Botswana, Lesotho, Namibia and Swaziland that have not developed national input-output tables). Given these slight imbalances, total gross exports are at times marginally larger and/or smaller than the value added constituting it, and the sum of domestic and foreign value added as a share of gross exports is generally not exactly equal to total gross exports at the national and particularly sectoral level. As a result, indicators of domestic, foreign and indirect value as shares of exports use total exported value added as the denominator, rather than gross exports. This better reflect the relative importance of foreign and domestic contributions to exported value added7. In order to provide an indicative assessment of Eora’s reliability, data on exports and imports were compared to results from the Comtrade database (using ISIC Rev. 3, which was also used to create the correspondence for Eora). Furthermore the standard deviation of each sector’s non-exported output is also presented. This can be found in Appendix 4. Based on this comparative analysis, Eora results presented in this report should be interpreted with some caution.8 Particularly, due to concerns about data quality for the SACU economies, as well as to a lesser extent the price fluctuations for mineral products, the mining and quarrying sector was omitted from analysis and dropped from the MRIO, with subsequent analysis based on a 25 sector summarized table for each country9. 7 The Eora Frequently Asked Questions provide an explanation for the causes of these imbalances: “data on country‐wise total exports and imports fundamentally conflicts with global trade balances. One cannot achieve a balanced global multi‐region input‐ output table whilst at the same time respecting data on exports and imports. This means that in a real MRIO table, either balancing conditions must be violated or raw data mis‐represented.” Under ideal balancing conditions, national ratios of Gross National Expenditure + exports versus Gross Domestic Product + imports should be 1, i.e. in an IO table the total of all inputs to a sector a given sector (i.e. the column values) should equal the total value of that sector’s outputs (i.e. the row values). However, due to data conflicts, this is in most cases a few per cent more or less than 1. 8 Eora developers state that “results will generally be uncertain at the sectoral level and for small sectors, but not necessarily uncertain for small countries, especially not for small countries with high-quality IO data” (Lenzen et al. 2013, p. 39). 9 In the case of Botswana, however, diamonds are classified under manufacturing and is, therefore, included in the analysis 6 3. SACU EXPORTS AND STRUCTURAL INTEGRATION IN GLOBAL TRADE NETWORKS Before analyzing value added trade and participation in GVCs, it is worth profiling briefly the nature of exports for the SACU countries and their integration into global trade networks. Moreover, as the analysis of value added trade remains an inexact science, particularly for smaller countries (i.e BLNS). Therefore, there it is useful to get a broad picture of potential GVC participation from the available aggregate trade data. 3.1. Overall trade integration Trade openness – or trade share of GDP – is a standard measure to assess the importance of trade to a country’s economy, and by extension, its integration with global markets. Figure 1 shows trade openness plotted against national wealth (measured as the log of GDP per capita). Traded shares of GDP increase as countries grow wealthier, although regardless of income level small countries tend to have a larger traded share of GDP than large ones – this is because large countries trade more internally, while small countries tend not to have sufficiently large domestic markets. Figure 1 shows that most SACU countries trade above the level that their incomes alone would predict, with Lesotho and to a lesser degree Swaziland among the most trade-dependent countries in the world. Given they have relatively similar populations as Lesotho and Swaziland and large mineral exports, Botswana and Namibia are actually substantially less integrated into global trade. South Africa, meanwhile records a relatively low degree of trade openness in regional terms, but still remains above the level of many of its peers, including China, Brazil, Colombia, and Peru. Overall the region sits in the middle between the highly integrated East Asian economies and the poorly integrated South American ones. Figure 1: Global comparison of trade openness by national income level Source: Authors based on data from WDI 3.2. Overview of main traded sectors The SACU countries can be characterized as having two types of exporters. First, Botswana and Namibia, rely heavily on the mining sector (especially diamonds in the case of Botswana and Namibia) and the exports of raw materials (especially agricultural goods and food/beverages in the case of Namibia). Second, Swaziland and Lesotho have narrow but well-developed industries that drive most export earnings: in the case of Swaziland it is sugar and (related) concentrated beverage syrups; in Lesotho it is apparels and textile. South Africa sits somewhere in the middle, with a large share of exports in mining 7 (iron ore, gold, platinum, diamonds) but also a well-developed agricultural and manufacturing export sector. Table 1 provides a breakdown of exports by broad industry classification for each country. Thinking about traditional GVCs (vertically integrated production), the most relevant sectors are manufacturing as well as some within agriculture. Here we see that while Botswana and Namibia appear to have large shares of exports in manufacturing, this is mainly explained by the classification of diamonds in the manufacturing sector; removing this tells a very different story. In the case of Botswana, diamonds and crude materials account for more than 85% of total exports; in Namibia it is around 45%. Namibia does, however, have substantial exports in manufacturing and machinery. South Africa is also fairly skewed toward commodity exports, although it still has a large share of manufacturing exports. Swaziland is highly concentrated in food (sugar) and processed sugar in the form of Coca-Cola syrup (classified under chemicals). Lesotho, by contrast, looks quite different from the rest of the region, with more than 60% of exports in manufacturing. Table 1: Structure of exports by SACU country (2013) – by SITC(2) section (1-digit classification) Botswana Lesotho Namibia South Africa Swaziland Food and live animals 2.2% 4.1% 19.9% 7.8% 25.4% Beverages and 0.1% 0.1% 3.7% 1.7% 0.3% tobacco Crude materials, 9.3% 2.5% 19.7% 18.1% 9.3% inedible, excl fuel Mineral 0.4% 0.1% 1.2% 10.8% 4.2% fuels,lubricants Animal and vegetable 0.0% 0.0% 0.1% 0.3% 0.0% oils,fats Chemicals and related 0.8% 0.1% 0.7% 7.1% 40.7% products Natural resource / 12.8% 6.9% 45.3% 45.6% 79.8% commodity subtotal Manufactured goods- 82.7% 35.6% 34.1% 25.0% 4.3% by materials Machinery and 2.7% 8.5% 16.4% 18.8% 5.0% transport equipment Miscellaneous 0.7% 48.9% 2.0% 3.0% 10.0% manufactured articles Manufacturing subtotal 86.1% 93.1% 52.5% 46.7% 19.3% Est manufacturing subtotal – excl diamonds 10.1% 61.0% 28.6% 41.7% 18.2% Commodities not 1.1% 0.0% 2.2% 7.6% 0.8% elsewhere specified TOTAL 100% 100% 100% 100% 100% Source: Authors based on data from UN Comtrade (via WITS) Turning to imports, Table 2 shows that manufactured goods account for a much larger share of imports than for exports in the region (with the exception of Lesotho). Botswana’s figures are again skewed by diamond imports (for aggregation and trading); excluding these, Botswana again appears to be significantly less integrated in manufacturing trade in relative terms than other countries in the region. Other notable differences across countries include the much higher share of machinery and transport equipment imports in South Africa and Namibia and the much higher share of food imports in Lesotho and Swaziland. 8 Table 2: Structure of imports by SACU country (2013) – by SITC(2) section (1-digit classification) Botswana Lesotho Namibia South Africa Swaziland Food and live animals 8.0% 18.2% 9.3% 4.9% 15.0% Beverages and tobacco 1.5% 3.2% 3.3% 0.8% 2.1% Crude materials, inedible, excl fuel 1.7% 2.1% 7.4% 2.1% 1.7% Mineral fuels,lubricants 0.5% 1.0% 0.4% 0.8% 0.7% Animal and vegetable oils,fats 17.3% 16.3% 10.0% 21.7% 18.6% Chemicals and related products 6.5% 8.4% 7.9% 10.4% 14.1% Natural resource / commodity subtotal 35.6% 49.4% 38.3% 40.7% 52.2% Manufactured goods- by materials 37.1% 21.8% 19.2% 10.6% 17.2% Machinery and transport equipment 19.7% 17.5% 32.9% 34.2% 19.4% Miscellaneous manufactured articles 6.1% 11.0% 9.4% 8.5% 10.9% Manufacturing subtotal 62.9% 50.3% 61.5% 53.3% 47.5% Est manufacturing subtotal – excl diamonds 34.0% 47.6% 57.8% 51.6% 44.9% Commodities not elsewhere specified 1.6% 0.3% 0.1% 6.1% 0.4% TOTAL 100% 100% 100% 100% 100% Source: Authors based on data from UN Comtrade (via WITS) 3.3. Proxying GVC integration: trade in intermediates Going beyond the broad sectoral classifications, we can also look at trade trends with respect to how goods are normally used – e.g. as inputs into another production process (i.e. intermediate goods) or as end products for businesses or consumers (i.e. consumption goods). The trade in intermediates is fundamental to GVCs, who basic concept is “importing to export” or I2E as Baldwin and Lopez-Gonzales (2013) call it. One country (for example South Africa) exports parts that are incorporated in the exports of another country (for example Germany). This single flow of intermediate goods is the basis of two key measures of supply chain integration, which help understanding better the role of a country in GVCs: on the sales side, it indicates that a country’s exporters are selling into a GVC. On the sourcing side, it indicates that the country is buying from a GVC. Patterns on the buying side provide information on the source of technology transfer and the type of GVCs a country is likely to join. This ultimately affects the growth of domestic value added since it affects the nature of the intra-firm know-how applied via GVCs. Patterns on the selling side indicate, instead, the likely exposure to demand shocks. We can distinguish three types of buyer roles in GVCs: for production of intermediate inputs in the value chain, for final production destined as exports, and for assembly. The main supplier functions are also three: supply of turnkey components, supply of other inputs, and supply of primary inputs (see Figure 2). 9 Figure 2: Seller and buyer functions supply of turnkey components Supplier /Selling supply of other Function inputs supply of primary inputs for production of intermediates Buyer/Sourcing for production of Function final exports for assembly Source: Taglioni and Winkler (Forthcoming). Figure 3 provides an aggregate view on the share of intermediates in SACU trade. We can see that share of intermediates in exports varies substantially across countries and over time. With the notable exception of Lesotho, all SACU countries have more than half their exports in intermediates. Botswana records more than 90% of exports in intermediates, but this is skewed by the categorization of diamonds, which alone accounts for more than 70% of exports. Both Lesotho and Swaziland show substantial declines in the share of their exports in intermediate goods – in the case of Lesotho, the share of intermediates fell dramatically from almost 50% in 2000 to just 12.2% in 2012. But while intermediates tend to proxy for production chain integration, in the case of Swaziland and Lesotho, the decline in intermediates is actually likely to be the result of GVC integration. Specifically, it is the result of participation in the apparel GVC, where both countries specialize in final stage assembly. Indeed, as shown in Table 3, apparel exports – where Lesotho, in particular, became concentrated in the 2000s, are largely consumer products (rather than intermediates) and Lesotho and Swaziland have the lowest share the intermediates in exports among all peer countries. In the case of South Africa, the share of intermediates in total exports fell slightly between 2000 and 2012. This comes despite a significant increase in the share of intermediates exported in the transport equipment sector (up from 13% to 22%). Similarly, intermediate exports have grown rapidly in the food and beverages sector (up from 15% to 23%), but South Africa still sells more consumer food and beverages products than most peers, perhaps indicative of a strong GVC position in this sector. South Africa’s virtual exit from global GVCs in apparel and footwear may be evidenced by the decline in intermediate exports from close to 10% in 2000 to just 1.6% in 2012. Botswana’s high share of intermediate exports is clearly distorted by diamonds. The sector-specific data in Table 3 shows that Botswana has among the lowest share of intermediate exports across virtually all sectors. Namibia shows a similar trend, with the one exception being their small electronics sector, which experienced a strong shift toward intermediate products (while Botswana’s exhibited the opposite shift). 10 Figure 3: Intermediates as a share of gross exports and imports (2000, 2012) Exports Imports Source: Authors based on data from UN Comtrade (via WITS) Integration matters as much for imports as exports. Data in Figure 3 shows that the majority of peer countries increased the share of imports in intermediate goods. In SACU, by contrast, only South Africa increased imports of intermediates. The other four countries experienced declining relative imports of intermediates, some significantly so: Botswana’s intermediates fell from 60.8% of imports in 2000 to just 42.9% in 2012; Swaziland’s fell from 53.8% to 47.9%. Overall, SACU countries, with the exception of Lesotho, come out lower than most peer countries in their reliance on imported intermediates. Again, the situation varies substantially by sector, although Botswana and Namibia fall well below the peer average in every sector10. Table 3: Intermediates as a share of gross exports (2012)- selected sectors Agriculture Food & Apparel & Machinery Electronics Transport Beverages Footwear Equipt Turkey 25.9 70.8 2.4 46.2 36.0 21.3 Thailand 13.9 16.0 2.0 62.9 63.9 30.0 Peru 54.2 82.9 0.1 41.0 55.8 32.5 Chile 12.3 17.9 11.2 40.1 38.7 31.7 Brazil 65.0 55.7 1.2 46.8 34.8 21.6 Argentina 66.2 72.2 5.6 55.1 65.6 23.4 Mauritius 66.0 84.6 0.0 20.3 43.9 24.8 Swaziland 7.4 70.8 0.1 8.9 75.0 2.1 South Africa 21.2 23.0 1.6 30.9 47.9 22.0 Namibia 9.4 12.6 3.1 27.2 42.9 21.2 10 The only exception being machinery for Namibia. 11 Agriculture Food & Apparel & Machinery Electronics Transport Beverages Footwear Equipt Lesotho 3.0 31.8 0.0 70.6 93.9 85.3 Botswana 9.4 21.9 1.4 27.0 17.5 8.8 Source: Authors based on data from UN Comtrade (via WITS) Table 4: Intermediates as a share of gross imports (2012)- selected sectors Agriculture Food & Apparel & Machinery Electronics Transport Beverages Footwear Equipt Turkey 79.2 70.2 4.9 30.6 40.5 27.0 Thailand 46.3 56.8 6.0 52.8 77.5 67.8 Peru 73.4 42.6 8.4 35.2 35.2 13.2 Chile 51.5 41.5 3.2 32.2 31.1 11.2 Brazil 60.8 26.8 6.1 48.0 67.4 50.5 Argentina 41.4 15.2 2.0 35.9 38.1 39.0 Mauritius 28.1 25.1 5.7 28.3 26.5 11.5 Swaziland 75.6 3.7 3.2 23.2 25.9 14.8 South Africa 55.4 41.8 3.0 39.6 33.0 21.2 Namibia 30.0 28.5 1.7 45.3 41.7 15.7 Lesotho 89.8 52.5 0.7 21.6 22.4 28.3 Botswana 46.5 28.2 1.5 25.0 38.0 12.6 Source: Authors based on data from UN Comtrade (via WITS) 3.4. Positioning in global trade networks To get a further sense of the region’s links into global trade networks, we can look at measures of network centrality. In Table 5, we report two measures of local centrality along with two measures of global centrality:  Local centrality refers only to the first order links of each country (neighbors), namely outward “Node Degree” and “Node Strength”, where the former measures the centrality in terms of the number of markets reached by SACU countries and the latter the intensity of exports.  Global centrality measures describe the characteristics of nodes’ neighborhood, and in particular assess the extent to which a country trades with partners that are themselves important exporters. “Out-closeness” is a measure of how close a node is with respect to all other nodes, in terms of intensity of export relations. It therefore provides a measure of the relevance of links. “Eigenvector centrality”11 stresses the relevance of nodes, i.e. it is important to assess if a node is connected to central players or to peripheral ones. Specifically, a node's eigenvector centrality is determined by the eigenvector centrality of its neighbors, so that their centrality is also taken into account. In general, countries displaying high value of eigenvector centrality are the ones which are connected to many other countries which are, in turn, connected to many others. The largest values correspond to countries in large and cohesive (high-density) sub-networks. 11 Bonacich (1972) 12 Table 5 presents the results of network centrality measures for SACU for overall trade intermediates and four key sectors that tend to be traded in GVCs: agri-food, apparel, automotive, and electronics12. In terms of local centrality, the results indicate that SACU ranked 14th out of 216 countries included in the Comtrade database in 2010 in the Node Degree Index and 39th in the Node Strength Index, suggesting that the region is the top-trading partner for a sufficiently large number of exporters but the intensity of trade volumes is only moderate. It has also improved its relative importance since 2000, while relative strength declined slightly. In terms of global centrality, SACU still ranks well but somewhat lower than for local centrality, with an Out-Closeness ranking of 41 (down 6 places from a decade earlier) and an Eigenvector Centrality ranking of 34. Sectoral rankings are broadly in line with the overall ranking, although the region ranks notably lower in apparel in both local and global centrality and notably higher in electronics (as well as agri-food global centrality). Table 6 compares South Africa’s centrality indexes in 2010 with those of 20 peer countries. The results for total trade suggest that South Africa performs relatively well, but with the exception of Node Degree, tends to be at the lower end of comparisons with BRICS and East Asian peers. Table 5: Centrality Ranking for SACU, World Ranking Local Centrality Global Centrality Node Degree Node Strength Out-Closeness Eigenvector Index Index Index Centrality Index Agri-food (2010) 16 42 23 24 Apparel (2010) 17 49 51 47 Automotive (2010) 14 39 41 34 Electronics (2010) 15 25 18 19 Total Intermediates (2010) 14 (4) 39 (3) 41 (6) 34 (--) Total Intermediates (2000) 18 36 35 34 Data Source: UN Comtrade. Table 6: Centrality Measures for South Africa and Peer Countries, World Ranking 2010, Total Intermediates Local Centrality Global Centrality Node Degree Node Strength Out-Closeness Eigenvector Index Index Index Centrality Index 14 39 41 34 SACU (3rd of 20 peers) (9th of 20 peers) (8th of 20 peers) (6th of 20 peers) Brazil 17 28 38 35 Russia 25 23 40 36 India 5 22 14 14 China 5 1 1 1 Kenya 40 79 75 73 Mauritius 54 112 91 80 Rwanda 83 156 185 142 Tanzania 46 101 87 83 12Table 5 and Table 6 report rankings rather than absolute values of centrality measures in order to simplify comparability across different indexes. 13 Local Centrality Global Centrality Node Degree Node Strength Out-Closeness Eigenvector Index Index Index Centrality Index Zambia 61 115 132 115 Cambodia 82 117 97 134 Malaysia 11 8 11 11 Thailand 12 12 13 17 Vietnam 44 34 33 45 Argentina 35 43 67 54 Bolivia 64 104 106 113 Chile 49 47 61 59 Colombia 40 51 64 61 Peru 50 58 78 74 Turkey 16 31 21 24 Data Source: UN Comtrade. Finally, network representations help visualize the complexity and heterogeneity of actors and trade links in GVCs. Figure 4 and Figure 5, visualize the network reporting the strongest flow for each node. The most connected countries – the central nodes, as they are the main trade partner for several countries – are the “roots” of the tree, distinguished from the peripheral countries – the “leaves.” The size of the node reflects a country’s strength or centrality in the network. The thickness of links reflects the weight of the value added relation. Larger bilateral trade flows are portrayed by closer distances between nodes. Ideally, we would like to present the network using trade in value added data. However, such data are not available yet for a sufficient number of countries. Instead, we present the network for trade in consumption13 goods (Figure 4) and intermediates (Figure 5). In the case of SACU14, trade in consumption goods provides a sense of where the region sells its end products and where it is most linked into GVCs in terms of its backward integration (i.e. where it sources from). Here, we see the strongest links are with the EU trade network based around Germany. Figure 4 shows that SACU (mainly South Africa) is an important node linking some other regional economies into the European trade network. It also shows that while SACU is a significant player in the European regional trade network, it is has less direct links into Germany than trade partners like Eastern Europe and Turkey (as illustrated by its positioning further from Germany) and its level of trade is lower (as illustrated by thinner connecting lines). The story is a bit different for intermediates (Figure 5), where China is demonstrably more central to trade for more countries than in consumer products. Here, SACU is primarily linked into the Chinese production network. Although the region is notably distant from the core of the Chinese network, it appears to play an even stronger role in intermediates as a regional node for Southern Africa. Figure 4: Minimal spanning tree: trade in consumption goods (2010), SACU in red 13The selection of consumption and intermediate goods is based on the UN Broad Economic Categories (BEC) classification which assigns goods to their final use, namely capital goods, consumption goods, and intermediate goods. 14 The dataset of trade networks includes SACU as a single data point; no individual country data is available 14 s a c u Source: Santoni and Taglioni (2014). Data: CEPII, BACI Dataset. Figure 5: Minimal spanning tree: trade in intermediate goods (2010), SACU in red s a c u Source: Santoni and Taglioni (2014). Data: CEPII, BACI Dataset. 15 4. STEPPING INTO GVCs: MEASURING EXPORT VALUE-ADDED IN SACU Section 3 of this note provided an initial overview of the scale and nature of integration of SACU countries in global trade networks. The analysis of trade in intermediate goods provided an initial view on SACU countries’ trade in the types of products that generally typify GVC trade. Up until this point, however, we have been considering only data on “gross trade” – i.e. we have not considered the “value added trade” that is at the heart of analysis of global value chains. In this section, the analysis hones in more on GVC trade by isolating value-added trade from our traditional (gross) trade figures; it is followed in the remaining sections by further analysis of the nature of GVC trade. 4.1. Why understanding (domestic) value added is important in studying GVCs One of the major implications of the growth of trade in fragmented global production networks is the inflation of aggregate export figures. This results from the double and triple counting of intermediates as they cross over national borders in the process of coming together to form an end product. For example, a Korean semiconductor that is incorporated into an ipod will be counted as a Korean export when it is shipped to Thailand to be assembled into an internal drive, and then again in Thailand’s exports as the drive is shipped to China for final assembly, and then again from China as it is exported as a finished ipod. As a result, understanding a country’s gross exports as well as exports of domestic value-added is important for understanding trade performance and GVC participation. It provides an insight into the critical issue of how trade performance contributes to the domestic economy in terms of output, industry linkages, and employment. Indeed, what matters most for a country is not gross exports (which may include a significant share of foreign value added via imported inputs) but the domestic value added (DVA) embodied in gross exports. Figure 6 exemplifies the decomposition of gross exports for the auto industry. Domestic value added consists of value added created in the auto industry, value added created in other sectors supplying the auto industry, and of re-imported intermediates (which have been previously exported). Figure 6: From gross exports to domestic value added: decomposition of gross exports in the auto industry Source: Taglioni and Winkler (2014), based on Baldwin and Lopez-Gonzales (2013). A country’s ability to benefit from GVCs is best shown by the evolution of its DVA embodied in gross exports over time (see Box 2). At the industry level, DVA consists of value added created in a specific 16 industry itself, value added created in other domestic sectors supplying this industry, as well as previously exported intermediates re-imported from abroad for use in a given industry. In simple terms, an increase in DVA embodied in gross exports over time signifies greater value addition within the country itself. As a function of productivity, it is associated with a country’s breadth, variety and sophistication of tasks and activities (Taglioni and Winkler 2014a). Beyond the likely welfare and employment implications, this also has a broader significance for trade policy.15 Box 2: Domestic and foreign value added: substitutes or complements? Operating in GVCs is fundamentally about global trade integration – this means not only exporting within production chains but also making use of imported parts and components. Thus, while nominal domestic value added (DVA) is ultimately the measure of aggregate success for any country, the level of foreign value added (FVA) embodied in a country’s exports is an important measure of GVC integration. Mathematically, in percentage terms, DVA and FVA are substitutes – if a country increases its FVA from 40% to 50% of exports that necessarily means that DVA has declined from 60% to 50%. But that does not mean that maximizing DVA share should be the primary goal, particularly if it achieved through an import substitution strategy that results in producers having to accept lower quality or higher priced inputs (or simply inputs that are incompatible with those required by GVC-oriented buyers). In this case, maximizing DVA share comes at the expense of total volumes, as domestic producers may experience declining competitiveness in global markets and may be unable to participate in GVCs. From a dynamic perspective, therefore, DVA and FVA can be seen as complements. Access to quality and cost effective imported inputs raises firm competitiveness, resulting in higher exports and therefore higher nominal DVA. Over time, technology spillovers from imported inputs may also result in some goods and services becoming competitively produced in domestic markets, leading to a productive substitution of imports for domestic supply, and potentially even higher DVA share. 4.2. Is domestic value added increasing in SACU countries? All five SACU countries have been able to grow their DVA in gross exports, albeit at different rates. Figure 7 examines compound annual growth rates of DVA embodied in gross exports16 in the five SACU countries as well as their comparators both from 2000 to 2011. Among the five SACU countries, Lesotho has recorded the largest annual increase at 13% followed by South Africa and Namibia (12%) and Swaziland (11%). Lesotho also performs well in relation to its comparators. Only Bolivia, Lao PDR, and Zambia have seen a larger increase in DVA over the eleven-year period. The performance of the four leading SACU countries is comparable to that of Argentina, Paraguay, and Rwanda. By contrast, Botswana lags behind significantly at just 5% CAGR – in fact, Botswana’s growth in DVA is the lowest among all 19 countries. 15 Forexample, this information could be valuable in determining the effect of a country’s currency appreciation on exports or in predicting the impact of exogenous shocks on welfare or employment. 16 Exports are measured in nominal US$ 17 Figure 7: Compound annual growth rate of domestic value added embodied in gross exports, 2000-2011 Source: Own computations using Eora database Figure 8: DVA embodied in gross exports as a share of GDP, 2000 and 2011 Source: DVA data own computations from Eora database, GDP data from World Bank World Development Indicators 18 Perhaps not surprisingly, some of the fastest rates of DVA growth, such as Lesotho and Rwanda, occurred in those economies that started exports from a very low base and which also have a low share of DVA embodied in gross exports as a share of GDP (Figure 8). Here, SACU countries still show relatively weak performance relative to peers. Only South Africa (increase from 16% to 18%) is among the leading countries in the group of comparators, though it is far eclipsed by Thailand (where exported DVA as share of GDP grew from 38% to 44%). Botswana saw a decline in its DVA as a share of GDP to 4%, with only Rwanda having a lower rate. Lesotho, Namibia and Swaziland saw marginal increases from a low base suggesting that their largest export sectors (which experience strong DVA growth) may have relatively low rates of domestic value addition. Interestingly, the largest increases between 2000 and 2011 were registered in Zambia, Thailand, Argentina, and Bolivia – all of which are relatively large producers of metals and/or commercial agriculture. To start understanding what may be behind the DVA performance in a country, it can be useful to compare trends in DVA with trends in the development of gross exports (the latter reflects the standard measure of exports). This is show in Figure 9. Here, we see that growth in DVA has largely tracked gross export growth for the five SACU economies, as well as for most comparators. However, for both Botswana and Lesotho gross export growth has been faster than DVA growth (by 2% and 3%, respectively), suggesting that exports from sectors with lower rates of domestic value addition have been growing more rapidly. This may be indicative of increasing GVC participation, but it may also simply reflect changing composition in the export basket (see Box 3). Figure 9: Compound annual growth rate of domestic value added embodied in gross exports, and of gross exports, 2000-2011 Source: Own computations using Eora database 19 Looking at the share of DVA in gross exports (in addition to their relative growth) is also critical to understanding the nature of exports and of GVC integration. Figure 10 highlights significant differences among SACU countries in this regard, as well as significant changes over time in some countries. While Lesotho (52%) has the lowest share of DVA among the entire set of peers and Swaziland is among the lowest (63%), the rest of the region has DVA shares of exports between 70% and 85%, generally in line with the peer countries, although somewhat on the lower end. All BLNS countries had declining DVA as share of exports – in Lesotho’s case a striking 16 percentage point decrease, and 7 percentage points for Botswana. The 11-year time period 2000 to 2011 largely covers Lesotho’s integration into the textile and apparel GVC (facilitated in part by AGOA and its third-country fabric provision), most likely explaining the scale of the change over the time period. Among comparators, Argentina and Tanzania, and to a lesser degree Kenya also saw substantial decline in DVA as share of exports, while Lao PDR, Cambodia, Thailand, and Zambia saw significant increases. Figure 10: DVA embodied in gross exports as share of gross exports, 2000 and 2011 Source: Own computations using Eora database Box 3: The challenge of interpreting DVA results As noted earlier, what matters ultimately for a country is growing DVA (in nominal terms) over time, regardless of the relative share of DVA. But from the perspective of understanding GVC participation and performance, the interpretation of high or low levels of DVA growth is not always that obvious. While we want to see increasing DVA, rapid integration into global production chains is likely to result in lower DVA as a share of gross exports. In fact, evidence of decreasing DVA a share of gross exports can be indicative of participation in longer and more 20 sophisticated value chains in which more imported value added is in turn being re-exported (Taglioni and Winkler 2014b). While increasing DVA may reflect growth in the services economy, which tends to have short value chains and high values of exported DVA as a share of exports. Moreover, DVA growth is affected by important factors that are not linked directly to GVCs (at least not the variety of GVCs associated with vertically fragmented production). Most notably for SACU and other developing countries with large agriculture and natural resource exports, growth in commodity exports and changing global commodity prices will shape DVA measures significantly (for example, extractives exports may have high DVA, despite weak links to domestic labor markets and supply chains). Increasing DVA can also signify increasing quality of exports (higher unit prices) regardless of whether these exports are within GVCs. And so changing sectoral composition of the export basket will have a significant impact on the DVA measure. The figure below provides a basic overview of the some of the different situations that may explain various DVA outcomes, based around: i) the level of DVA to gross exports; and ii) the growth of DVA to gross exports. This underscores the importance of going beyond the aggregate analysis to understand better the factors shaping DVA performance, and the degree to which they are shaped by GVC participation and position. At minimum, it highlights the need to look at data at the sectoral level. Beyond this, assessing sectoral structure and performance at a qualitative level is likely to be important in order to interpret the results effectively. Note that in the case of the analysis presented in this note, minerals exports have been excluded, so at least some of the effect of natural resources exports on the DVA figures is controlled for in the results presented here. 4.3. Where is DVA growth coming from? – sectoral assessment Performance at the sectoral level has been highly heterogeneous and can reveal a bit more about the particular sources of DVA growth. Table 7 summarizes for each of the five SACU countries the five non- mining sectors contributing the most to overall DVA for the year 2011, including the sectoral compound annual growth as well as the DVA in exports as a share of total exports. In terms of their overall DVA in exports, it is striking – if not entirely surprising – how much more DVA as share of total exports is contributed by services sectors. Examining individual countries, this disaggregation suggests that Botswana’s overall stagnation is in part attributable to slower growth in manufacturing sectors. Similarly, the pace of Lesotho’s DVA growth is greatest in services sectors, although these are starting from a very low base. In Namibia, the leading sources of DVA growth are predominantly in food processing and manufacturing. South Africa, on the other hand, has a relatively balanced level of growth across its leading sectors. This sectoral disaggregation by country is expanded in Appendix 5. 21 Table 7: DVA in exports for 2011, compound annual DVA growth, and DVA in exports as share of exports for 5 largest sectors by DVA, 2000-2011 Largest sector (left to Hotels & Transport Other Education Food & Total right) Restaurants manuf. & Health Beverages Total DVA in exports (in Bots- 114,227 101,622 53,883 49,804 40,461 543,139 $1,000) and CAGR wana (11%) (9%) (6%) (11%) (8%) (5%) (2000-11) DVA in exports as % of 79.7% 74.3% 63.9% 88.7% 60.1% 72.3% sectoral exports (2011) Largest sector (left to Textiles & Education Post & Financial Transport Total right) Apparel & Health Telecomm. Intermed. Total DVA in exports (in 45,755 17,971 13,150 10,673 7,324 179,161 Lesotho $1,000) and CAGR (10%) (12%) (22%) (18%) (14%) (13%) (2000-11) DVA in exports as % of 53.5% 43.4% 77.5% 63.6% 84.1% 52.3% sectoral exports (2011) Largest sector (left to Petroleum Textiles & Food & Electrical & right) & Agriculture Wearing Total Beverages Machinery Chemicals Apparel Total DVA in exports (in Namibia 489,912 133,363 71,785 65,986 57,525 1,283,849 $1,000) and CAGR (12%) (9%) (15%) (14%) (15%) (12%) (2000-11) DVA in exports as % of 73.3% 68.6% 61.1% 87.9% 66.6% 71.9% sectoral exports (2011) Largest sector (left to Petroleum Metal Financial Electrical & right) & Transport Total products Intermed. Machinery Chemicals South Total DVA in exports (in 12,724,103 10,513,070 10,086,014 7,566,665 6,669,546 71,327,567 Africa $1,000) and CAGR (13%) (13%) (10%) (14%) (13%) (12%) (2000-11) DVA in exports as % of 83.7% 93.9% 79.4% 87.9% 77.6% 82.9% sectoral exports (2011) Largest sector (left to Food & Electrical & Hotels & Transport Agriculture Total right) Beverages Machinery Restaurants Total DVA in exports (in Swazi- 98,087 78,682 41,032 35,711 30,488 489,622 $1,000) and CAGR land (11%) (13%) (15.0%) (16%) (17%) (11%) (2000-11) DVA in exports as % of 61.3% 49.5% 54.1% 86.5% 61.4% 58.3% sectoral exports (2011) Source: Own computations using Eora database Figure 11 compares DVA growth rates for two sectors that are important in most SACU countries – Agriculture and Food & Beverages – with those of peers. For the first of these – Agriculture – the SACU countries are, with the exception of Botswana, towards the middle, with DVA growth ranging from 9% (Lesotho) to 16% (Swaziland). Looking over the most recent five years (2006-2011, not shown in the figure below), Lesotho’s growth of 30% far exceeds all other countries. The highest performers in agriculture tended to be the least developed comparator countries, Zambia (24%) and Lao PDR (22%), with the middle-income countries in the sample generally achieving growth rates between 5% and 15%. For the Food and Beverages sector, the SACU countries again perform in the middle, with Namibia and South Africa having the highest overall performance (12%). While Lesotho and Botswana lag at 8% annual 22 growth. Lesotho’s growth in the 2006-2011 period (not shown) was again substantial (26% CAGR). Again the highest performers are the less developed comparators Zambia (28%), Bolivia (19%), and Lao PDR (16%); but low income countries also showed the lowest growth, including Cambodia (3%) and Tanzania (4%). Figure 11: Compound annual growth rate of sectoral domestic value added embodied in gross exports for the Agriculture and Food & Beverages sectors, 2000-11 Source: Own computations using Eora database Complementing the sectoral DVA growth with data on DVA share of total exports over time in these sectors can provide further insights into GVC developments. Figure 12 shows these results for the Agriculture and Food & Beverages sectors. In Agriculture, Swaziland saw a large increase in DVA share of gross exports, from 75% to 86%. This was the second largest change in any country and, if the data is correct, it would most likely reflect changing composition of agricultural exports. Lao PDR experienced the largest growth in DVA share (15 percentage points), while Zambia and Cambodia also showed large growth in its DVA share. The SACU countries all tended to be in the lower half of the comparator group in terms of DVA share, indicating that their agricultural exports make greater use of imported inputs. For the Food & Beverages sector, South Africa has by far the highest share of DVA among the SACU countries (at 85% in 2011), while the rest of the region showed relatively low levels of DVA. Both Botswana and Lesotho experienced large declines in DVA share over the decade and Swaziland, while experiencing growth, showed the second lowest DVA share among all peers (at 61%, just after Botswana at 60%). The findings suggest that domestic processing in BLNS increasingly relies on imported inputs, most likely (in the case of this sector) from intra-regional sources. 23 Figure 12: DVA embodied in gross exports as share of gross exports for the agriculture and food & beverages sectors, 2000 and 2011 Source: Own computations using Eora database Figure 13: Compound annual growth rate of sectoral domestic value added embodied in gross exports for the Hotels & Restaurants sector, 2000-2011 Source: Own computations using Eora database For tourism, proxied through the Eora “hotels and restaurants” sector (see Figure 13), Lesotho, Swaziland, and Namibia have all seen rapid growth (16%-20%), while South Africa and Botswana (13% and 11%, respectively) experienced relatively strong growth but still trailed most peers. In terms of DVA share of gross exports ( 24 Figure 14), however, there are stark differences in the region. South Africa shows the highest DVA among all peers (near 95%) while the rest of the region trails most peers. Swaziland (61%) and Lesotho (62%) have particularly low levels of domestic value added in tourism. Figure 14: DVA embodied in gross exports as share of gross exports for hotels & restaurant sector, 2000 and 2011 Source: Own computations using Eora database The textiles & apparel and transport equipment sectors (see Figure 15) provide a useful comparative overview of the manufacturing sectors that tend to be most GVC- oriented on a global basis. For textiles & apparel, the five SACU countries look broadly similar to East Asian and South American comparators and ahead of most African peers, with all five achieving annual compound DVA growth at or near 10% (Swaziland lags at 8% and Namibia leads at 15%, but from a small base). For the transport equipment sector, Swaziland (17% growth) has been the fastest growing country among all comparators, other than Kenya. Again, however, this comes from a very small base; Botswana has also experienced strong growth (14%) in DVA. The other three SACU countries all rank towards the middle of the table, with growth in DVA between 9% and 10%. In terms of DVA share of exports ( Figure 16), Botswana, Lesotho, and Namibia have seen their exported DVA as a share of total sectoral exports decline in both manufacturing sectors – in some cases quite substantially. For example, Lesotho’s DVA embodied in gross exports as a share of gross exports in textiles fell from 66% to 53% and Botswana’s from 71% to 60%, while Lesotho’s fell in transport equipment from 68% to 49% and Botswana’s from 64% to 54%. This suggests these value chains may have grown in length and complexity, with less value added in country, and/or that Lesotho and Botswana’s tasks have become lower value-added. South Africa and Swaziland in both sectors have remained reasonably consistent over the decade, although South Africa’s DVA share is substantially above that of Swaziland in both sectors. It is worth noting that relatively low DVA shares in these sectors may indicate greater GVC participation. For example, South Africa’s DVA 25 share in automotive (55%) is similar to that of countries like Thailand and Argentina, both of which have significant automotive sectors, while other countries with limited automotive sectors show higher DVA shares. Figure 15: Compound annual growth rate of sectoral domestic value added embodied in gross exports for the textiles & wearing apparel and transport equipment sectors, 2000-2011 Source: Own computations using Eora database Figure 16: DVA embodied in gross exports as share of gross exports for the textiles & wearing apparel and transport equipment sectors, 2000 and 2011 Source: Own computations using Eora database 26 5. GLOBAL VALUE CHAIN PARTICIPATION AND POSITIONING The analysis of trends in value-added trade give some insight into how SACU countries are integrating into GVCs, but the analysis still remains a step removed from actual GVC trade. In this section, we use the Eora database to calculate recently developed indicators on GVC participation and positioning. 5.1. GVC participation – introduction and overall index A country’s level of participation in GVCs can in part be assessed based on both its forward and backward integration:  Forward integration, or indirect value added (IVA) –refers to a country’s share of value added embodied in other countries’ exports – i.e. producing intermediates that you export to other countries, who will then add further value and export them as finished products or further stage intermediates.  Backward integration, or foreign value added (FVA) – is the share of foreign value added in embodied in a country’s exports – i.e. intermediate inputs imported from other countries that you then add value to and export as finished products or further stage intermediates. Both forward and backward integration matter, but neither should be inherently maximized. As discussed in Box 3, the products involved and the qualitative nature of the integration determine the benefits that accrue from it. Backward integration provides access to quality inputs, which contributes to downstream competitiveness; it also has significant potential to deliver productivity spillovers through access to global frontier technologies. As such, backward integration tends to be particularly important for developing countries as it links to a number of measures of structural transformation. But taken to the extreme, backward integration may crowd out local production and limit domestic value addition. Similarly, forward integration is an indicator of integration into value chains and also provides opportunities to benefit from technology spillovers. But the desirability of forward integration depends a lot on what is being exported and where you sit on the value added chain. High levels of forward integration in developing countries can often be associated with higher resource dependency and is negatively linked to measures of diversification and structural change (AfDB 2014).17 On the other hand, countries like the US and Japan have high forward integration by selling leading edge technologies (with high value added) into the early stage of global production processes. Figure 17 reports the GVC participation index for SACU countries and peers in 2000 and 2011, based on data from the Eora database. The GVC participation index combines the measures of forward and backward integration18, each of which will be elaborated on in more detail in the next section. The index is intended to indicate the extent to which a country participates in vertically integrated production (Koopman et al. 2010). The higher the foreign value added in gross exports and the higher the value of inputs exported to third countries and used in their exports, the higher the participation value. This tends to favor small, open economies, so it is perhaps unsurprising that Lesotho stands out as the most GVC- integrated of all peers, particularly given what we know about the development of its apparel sector in the 2000s. Lesotho’s GVC participation measure increased by more than 50% over the decade. Other countries that show significantly growing GVC participation include Tanzania, Rwanda, and Zambia. The 17 Domestic value added and foreign value added by definition should equal to the total sum of exports and thus - as the corollary of DVA - a declining share of domestic value added in total gross exports will by definition result in increased foreign value added as a share of exports. Due to the balancing considerations outlined earlier, this does not hold completely for the Eora dataset. 18 The index combines FVA and IVA, both as a share of gross exports 27 South American and (surprisingly) East Asian peers also experienced declining GVC participation. What is notable from the review of peer countries is that while those countries which are dependent on commodity exports fare well in terms of DVA, they perform less well when measuring GVC participation. Figure 17: GVC Participation Index 2000 and 2011 Source: Own computations using Eora database 5.2. What is driving GVC participation? – forward versus backward integration But while the GVC Participation is a useful initial indicator, what matters much more are the components that make up this index. Figure 18 reports backward and forward integration as a share of gross exports for SACU and peer countries in 2011, and gives a perspective on what is driving the broad measure of GVC participation. This is followed in Figure 19 with an illustration of growth in forward and backward integration over the decade. It shows that, overall, SACU countries tend to be slightly less forward integrated than peers and slightly more backward integrated; but with the exception of Lesotho, growth in both forward and backward integration is trailing many peers. Reviewing each country briefly:  Botswana: More forward integrated (41%) than backward (28%), although forward integration is likely to be distorted by diamonds, as is suggested by the very low growth over the past decade (1%). Backward integration is relatively high, but growing slowly.  Lesotho: Highest level of backward integration (48%) and highest growth (20%) among all peers; forward integration much lower (18%) and trailing most peers (although growing rapidly), reflecting focus on assembly stage of apparel manufacturing.  Namibia: Slightly above average level (28%) and growth (14%) of backward integration, while forward integration remains limited (19%).  South Africa: Below average backward integration (17%), which is common for larger countries (although Thailand’s backward integration is 28%); forward integration moderately high 33%. Growth is moderate in both FVA and IVA.  Swaziland: High level of backward integration (42%) but among lowest forward integration among peers (17%); growth in both FVA and IVA among the lowest of peers. 28 Figure 18: Foreign value added (left) and indirect value added (right) embodied in gross exports as share of gross exports, 2000 and 2011 Backward integration (FVA) Forward integration (IVA) Source: Own computations using Eora database Figure 19: Compound annual growth rate of foreign and indirect value added embodied in gross exports, 2000- 2011 Backward integration (FVA) Forward integration (IVA) Source: Own computations using Eora database 29 5.3. Sectoral drivers of GVC participation At the sectoral level, there is considerable heter ogeneity among these countries’ growth rates in foreign and and indirect value added (see Figure 20). For Botswana, foreign value added increased the most in transport equipment (18%), along with metal products and wood and paper (17% each). As was the case for DVA, FVA in financial intermediation declined substantially (-15%) while other manufacturing and agriculture were stagnant. In Namibia, the largest growth in foreign content was in the metal products (20%) and electrical machinery and transport sectors (19% each); financial intermediation also showed a sharp decline in FVA. For Lesotho, growth in foreign content was highest in services sectors (especially hotels and restaurants, post and telecommunications, and transport). Swaziland showed a similar pattern of strong growth in services FVA (with the exception of decline in financial intermediation), but showed much weaker growth in other areas. For South Africa, FVA increases were largest in agriculture and services sectors, while manufacturing FVA growth was modest. Financial intermediation and agriculture consistently shows the lowest FVA as a share of gross exports while transport equipment shows the highest foreign content (see Appendix 6). Other manufacturing sectors, including electrical and machinery and food and beverages also show relatively high foreign content across the region. Growth in value added exports embodied in third countries’ exports (indirect value added) came across a broad range of sectors, with agriculture (except Botswana) and services particularly strong. In Botswana, the most rapid growth in IVA came in transport and communications, followed by tourism, while most other sectors were stagnant or in decline. In Namibia, Swaziland, and Lesotho growth in IVA was relatively strong across the board, although the rates of growth were substantially higher in Lesotho. A similar story of broad sectoral IVA growth can be seen in South Africa. Figure 20: Compound annual growth of foreign value added and indirect value added by SACU country, by sector, 2000-2011 30 Source: Own computations using Eora database Looking at performance in individual sectors is useful but runs the risk of obscuring the bigger picture, particularly as sectors vary substantially in their contributions to GVC participation across the countries. Figure 21 aggregates sectors according to the OECD’s classification of technology content in sectors. This aggregation gives a better sense of where foreign content embodied in GVCs is coming from across the SACU countries and the type of sectors where its GVC exports are feeding into. This is critical for understanding the potential of generating spillovers from GVC participation. The data on foreign value added shows that Swaziland and South Africa (and to a lesser degree Namibia) gain most of their foreign value added embodied in exports from manufacturing sectors, with a substantial amount coming from high technology sectors. Indeed, these countries compare very favorably to peers, although trail substantially behind Thailand (driven by large FVA in the electrical and machinery sector) and Argentina (FVA in transport equipment). In the case of Swaziland, the high technology FVA is coming mainly from exports embodying foreign content in the electrical and machinery sector, while in South Africa it comes from both electrical and machinery and the transport equipment sectors. From the perspective of facilitating technology spillovers, Lesotho’s imported content is among the least favorable across all peer countries. 31 In terms of IVA, the story is similar, although the differences are less stark across countries. What is notable for SACU and just about all peers is that the profile of their forward contribution to GVCs is much less technology intense than their backward integration. This allows for some tentative conclusions about relative positioning in GVCs (see Section 5.5). That said, contribution to services exports of other countries appears to be significant, which may represent interesting opportunities for learning and upgrading. Figure 21: Sectoral contribution to FVA and IVA (2011) Source: Own computations using Eora database; OECD 5.4. Geographical drivers of GVC participation Understanding the direction of forward integration (the selling side of GVCs) is important for identifying potential sources of shocks a country may face (see Appendix 7) 1920. For South Africa, the most important destination is Germany, followed by the UK and Netherlands. This underscores the continuing reliance on European demand in GVC-oriented sectors. The BLNS does not register as a significant source of IVA for South Africa. For Botswana, Israel – presumably as a destination for diamonds and other minerals – makes up over 30% of Botswana’s IVA, followed by the UK, Germany and Norway (all of which grew their share from 1996 to 2011 at the expense of regional neighbors). For Lesotho, the largest share of IVA is in the “rest of world” sector, followed by Belgium and Germany. Namibia also shows a strong link to the European market, with IVA highest in Germany, France, the Netherlands and Belgium. Similarly, understanding the source of foreign value added (the buying side of GVCs) is important, in this case for identifying the source of technology transfer and the type of GVCs a country is likely to join. This ultimately affects the growth of domestic value added since it affects the nature of the intra-firm and arm’s length transfer of know-how and the country’s ability to absorb tacit knowledge (i.e. business models and all other types of knowledge that cannot be codified) and/or the knowledge embedded in the imported inputs. Data on FVA sources for SACU countries is also available in Appendix 7. For South Africa, FVA sources are spread across the three main global poles, with Germany as the largest contributor, followed by the US and China. The questions for South Africa (which requires further assessment) are: i) whether integration with more technologically advanced nations produces a premium in terms of growth of the domestic value added embedded in exports; and, ii) whether distance matters for the rate of growth. The latter is important since tacit knowledge is likely to flow more easily over shorter distances and – assuming that the latter produces more spillovers – distance and trade costs, 19 These figures are similar but not identical to those derived from the I2E index ( Baldwin and Gonzalez-Lopez 2013). 20 Here, (as for sectoral disaggregation), results should be treated with caution: in some cases statistical discrepancies (i.e. the RoW sector) is among these countries’ largest destination for intermediates. In SACU, this is the case with Lesotho. Further, trad e flows within the SACU region are very poorly tracked (including by Comtrade). 32 particularly those affecting the services sector, may matter for the ability of countries to boost domestic value added via GVCs. For the BLNS, the distance issue is less relevant – here, South Africa dominates as a source of intermediates. Where South Africa’s trade flows are measured in Eora, it is by far the largest exporter of FVA, making up 60% of FVA imported into Botswana, Swaziland, and Namibia (it is not included for Lesotho). For Botswana, the other main sources of FVA have traditionally been the Netherlands, US, and Germany, with China emerging as a leading FVA source in 2011. Re-exported intermediates from South Africa to Lesotho are not reported and as a result China and Chile, followed by India and Taiwan are the main sources of backward linkages. Namibia also sources over 70% of foreign value added from South Africa, followed by Germany and the US. As is the case for Botswana, China has become increasingly important as a provider of inputs while inter-regional sourcing (besides South Africa) remains minimal. Swaziland also depends on South Africa for the bulk of its FVA, followed by Germany and the US, though their share has been declining as China’s has increased. 5.5. Assessing GVC positioning As discussed earlier, what is ultimately more important than participating in GVCs is capturing value that facilitates sustainable growth and higher-quality employment. This depends, in part, on a country’s positioning within a GVC. It can be upstream (production of inputs at the beginning of the value chain) or downstream (production of goods and services towards the end of the value chain) depending on its specialization. Countries specialized in upstream activities produce the raw material or the intangibles involved at the beginning of the production process (e.g., research and design). Countries concentrated in downstream tasks specialize in the assembly of the final products or in customer services. Finally, countries involved in activities at the center of the value chain focus on the standardized labor-intensive manufacturing jobs. Again, it is not always obvious where a country would ideally want to be positioned – it depends very much on the value chain in question. For some value chains, most of the value is captured upstream, for others downstream; and in some cases both. Generally, speaking, mid-stream activities (“at the center of the value chain”, as described above) are least likely to be in a position to capture significant value. “Upstreamness” of a country’s specialization can be measured by its “distance to final demand” (Antràs, Chor, Fally and Hillberry, 2012) – i.e. the distance in terms of number of production stages between the production of good i in country c and final demand21. Evidence suggests that only a few countries have managed to move downstream. Most countries have increased their upstreamness because the overall length of value chains has increased with the fragmentation of production. Moreover, the offshoring process that lengthens GVCs tends to affect more the early stages of production, although a new wave of services offshoring has been taking place in recent years (Taglioni and Winkler, forthcoming). A final useful metric is to combine import and export upstreamness to compute the domestic length of the value chain. A positive gap indicates that exports are relatively more downstream (or “closer to final demand”) compared with imports. This is the case in economies where the manufacturing sector has been a key source of export-led growth, such as China, Japan, and Thailand. Conversely, a negative gap indicates that a country’s export profile is more upstream than its import profile. This is the case in economies whose GVC _ DISTci  1  cdij Ddj 21 d,j Measured as , Ddj is the distance to final demand in terms of number of production stages in country d and industry j. These are summed up over all country-industry (d,j) combinations who use inputs from industry i and country c using φcdij as weights. φcdij is the fraction of production from industry i in country c that is purchased as an intermediate good by industry j in country d. 33 exports are concentrated in agriculture products and primary commodities, such as Australia, and New Zealand. Or it may be the case that the country is a home to a sophisticated consumer market and therefore an intensive importer of finished consumer goods, rather than being a reflection of its exports, such as the United States. Figure 22 shows that most SACU countries export in upstream positions, relatively far from the final consumers, and show a “negative” domestic GVC length. The exceptions here are Botswana (distorted by the short value chain of diamonds) and Lesotho (final stage apparel assembly). Turkey, Thailand and Mauritius appear to show a pattern of importing relatively upstream and exporting at a more downstream stage, thus indicative of a positive domestic GVC length. By contrast, the South American peers export further upstream than SACU and all show negative domestic GVC length. Between 2000 and 2012, almost all peer countries’ exports became more upstream (probably as a result of increasing fragmentation of global production), but South Africa, Swaziland, and Namibia moved further upstream than most. Figure 22: Import and Export Upstreamness and Domestic GVC length Source: Authors calculations based on data from UN Comtrade (via WITS) The relative structure of value chains and country positioning obviously varies considerably by sector. Table 8 shows export upstreamness in key sectors in 2012. SACU countries appear to export slightly more downstream than peers in agriculture and food and beverage sectors and significantly more so in transport equipment, while they are positioned around the peer average in the other manufacturing sectors. No SACU countries stand out as being appreciably more upstream or downstream across the sectors, although Swaziland is quite a bit more upstream than SACU peers in food and beverages. Analogously, the data on domestic length of GVCs (Table 9), suggests the SACU countries have longer domestic chains in agriculture and agriprocessing, as well as transport equipment. But their import patterns in other manufacturing sectors means that SACU countries generally show negative domestic length in these sectors, indicating substantially shorter domestic value chains in manufacturing than in most of the peer countries. 34 Table 8: Export upstreamness, selected sectors (2012) Agriculture Food & Apparel & Machinery Electronics Transport Beverages Footwear Equipt Turkey 1.85 1.47 1.10 1.71 2.20 1.27 Thailand 2.07 1.77 1.11 1.93 2.26 1.29 Peru 1.77 2.02 1.06 1.73 2.32 1.38 Chile 1.85 1.52 1.11 1.69 1.89 1.34 Brazil 3.10 2.22 1.12 1.87 2.20 1.36 Argentina 3.02 2.69 1.16 1.91 2.18 1.17 Mauritius 2.20 2.00 1.06 1.50 2.27 1.76 Non-SACU avg 2.27 1.96 1.10 1.76 2.19 1.37 Swaziland 1.91 2.15 1.06 1.60 2.55 1.28 South Africa 2.01 1.58 1.32 1.85 2.12 1.17 Namibia 2.15 1.37 1.29 1.79 1.89 1.18 Lesotho 2.26 1.29 1.05 2.06 2.17 1.00 Botswana 2.10 1.33 1.10 1.68 2.21 1.22 Source: Authors based on data from UN Comtrade (via WITS) Table 9: Domestic GVC length, selected sectors (2012) Agriculture Food & Apparel & Machinery Electronics Transport Beverages Footwear Equipt Turkey 1.21 0.59 0.01 0.11 (0.08) 0.09 Thailand 0.52 0.41 0.18 (0.03) 0.14 0.30 Peru 1.10 0.35 0.05 (0.02) (0.37) (0.23) Chile 0.44 0.60 (0.02) (0.01) (0.09) (0.12) Brazil (0.66) (0.27) 0.01 0.02 0.02 0.06 Argentina (0.98) (0.94) 0.07 (0.02) (0.17) 0.19 Mauritius 0.08 (0.37) 0.05 0.32 (0.30) (0.62) Non-SACU avg 0.08 (0.04) 0.06 0.04 (0.13) (0.07) Swaziland 1.08 (0.30) 0.12 0.05 (0.69) 0.59 South Africa 0.53 0.30 (0.22) (0.10) (0.25) 0.08 Namibia 0.10 0.26 (0.15) (0.06) 0.09 0.02 Lesotho 0.05 (0.01) 0.21 (0.47) (0.29) 0.14 Botswana 0.25 0.27 0.02 0.00 (0.09) 0.01 Source: Authors based on data from UN Comtrade (via WITS) 35 6. SUMMARY CONCLUSIONS The increasing fragmentation of global production has the potential to offer substantial opportunities for the five SACU countries: Botswana, Lesotho, Namibia, South Africa, and Swaziland. Particularly for the four smaller SACU economies (the BLNS), proximity to and regional trade agreements with the “headquarter economy” South Africa both augment these opportunities and create risks of “crowding out”, with South Africa reaping the gains of any investments in the region. Overall the analysis in this note, which draws extensively on the Eora multi-region input-output database over an eleven-year time period and including fourteen comparator countries, suggests that the SACU region is moderately integrated into GVCs. But the scale and nature of this integration varies enormously by country and sector. Considering overall trade integration first, the region fares well – less integrated than East Asia but more so than might be expected given its peripheral location (and more so than South American peers). However, the nature of integration tends to be biased toward imports, with exports largely commodity dependent (with notable exceptions of South Africa, Lesotho, and Namibia). Taking the analysis a step further and looking at trade in intermediates, the picture for the region looks less optimistic. Here, SACU as whole has a lower share of intermediates in both exports and imports as compared to peer countries. Moreover, intermediates are declining as a share of exports and imports (with the exception of exports in Namibia and imports in South Africa). This highlights that the region’s apparent trade integration remains biased toward commodity exports and consumption imports, with potentially rather limited GVC participation. The analysis of network integration highlights that South Africa, at least, remains a moderately important player in global trade networks and a key regional hub both in consumption goods (where it is linked to the European market) and intermediates (where it is increasingly linked to China). This data, however, is based only on gross trade and not value-added trade and so South Africa’s global position in intermediates is likely to be biased strongly by commodity exports, obscuring its real participation in GVCs. Lesotho appears to have made the most strides in the region in terms of GVC participation, as evidenced by rapidly growing gross exports and DVA, and a declining ratio of DVA to gross exports. By contrast, Botswana fares poorly in measures of DVA growth and GVC integration, showing stagnant DVA and declining DVA share of gross exports. Swaziland also shows relative stagnation in its performance. South Africa and Namibia meanwhile show moderate performance, with South Africa in particular showing fairly strong levels of GVC participation and moderate growth in DVA and DVA share. From a sectoral perspective performance varies significantly, but what is most notable is that services sectors – and particularly transport and hotels and restaurants – have in general been growing more rapidly than manufactured goods. These results are also mirrored in the GVC participation index, for which Lesotho saw the largest increase between 2000 and 2011 (from 48% to 70%). Among SACU economies only Namibia improved in this time period (from 37% to 39%). However, all five countries are – broadly speaking – in the middle range compared to the list of peer countries. 36 The region also shows big differences across countries in terms of the nature of this integration. Overall, South Africa is the only country in the region showing relatively strong forward integration in GVCs; Botswana also shows relatively high forward integration, but excluding diamonds, forward integration is limited. Lesotho and Swaziland, by contrast, show the lowest forward integration among all peers, and Namibia is also among the least forward integrated countries in the comparison. Lesotho, South Africa, and to a lesser degree Namibia are well integrated backward into GVCs. Botswana and Swaziland perform less well here, and both of these countries show stagnant performance over time across both forward and backward integration. The direction of forward and backward integration shows the overwhelming dominance of South Africa as a source country of foreign content for the other four SACU economies, which suggests that they are more integrated into regional value chains than to global ones (see Keane, 2015). This may also have implications for access to global frontier technology, and therefore to productivity growth potential. China has grown substantially in significance, particularly as a source of foreign content. All countries, however, remain highly dependent on European markets for their forward participation in GVCs. Finally, the region appears to be positioned relatively upstream in GVCs, in particular South Africa, Swaziland, and Namibia (while Lesotho is downstream positioned in the apparel GVC). While the region appears to be better positioned than the South American peers, a clear gap is apparent with countries like Thailand, Turkey, and Mauritius, which import relatively upstream in GVCs and then export further downstream. By contrast, most SACU countries import further downstream than they export. In interpreting these results it is important to bear in mind the importance of the recent commodity super-cycle on the relative export value of goods and services for commodity-dependent exporters (including some of the SACU countries). This is likely to explain some of the trends observed, particularly in relation to growth in forward and indirect value added. These results – particularly in conjunction with existing sectoral studies on the export growth and diversification efforts of these economies – provide compelling evidence of the varied pace of GVC integration in recent years in the region, as well as the heterogeneity across sectors and source and destination countries. 37 References AfDB/OECD/UNDP (2014) ‘African Economic Outlook 2014: Global Value Chains and Africa’s Industrialisiation.’ Tunis: African Development Bank. Ahmad, N. (2013) ‘Estimating trade in value-added: why and how?’ in Elms, D. and Low, P. (eds.) Global value chains in a changing world. Geneva: WTO. Amador, J. and Cabral, S. (2014). “Global Value Chains: Surveying Drivers, Measures and Impacts”. Banco de Portugal Working Papers 3/2014. Antras, P., & Rossi-Hansberg, E. (2009). ‘Organizations and Trade;. Annual Review of Economics, 1, 43-64. Antràs, P., Chor, D., Fally, T., & Hillberry, R. (2012). “Measuring the upstreamness of production and trade flows”. National Bureau of Economic Research. Baldwin, R. (2006). “Globalisation: the great unbundling(s).” Economic Council of Finland, 20 (2006), 5-47. Baldwin, R., & Venables, A. J. (2013). Spiders and snakes: offshoring and agglomeration in the global economy. Journal of International Economics, 90(2), 245-254. Barnes, J. & Kaplinsky, R. (2000). Globalisation and the Death of the Local Firm? The Automobile Components Sector in South Africa. Regional Studies, 34(9), 797–812. Barnes, J., & Morris, M. (2008). Staying alive in the global automotive industry: what can developing economies learn from South Africa about linking into global automotive value chains?. The European Journal of Development Research, 20(1), 31-55. Barrientos, Stephanie, and Margareet Visser. 2012. South African Horticulture: Opportunities and Challenges for Economic and Social Upgrading in Value Chains. Capturing the Gains Working Paper 2012/12. http://www.capturingthegains.org/pdf/ctg-wp-2012-12.pdf Black, A., (2001). Globalisation and restructuring in the South African automotive industry. Journal of International Development, 13 (6), 1-12. Bonacich, P., (1972). Factoring and weighting approaches to status scores and clique identification. Journal of Mathematical Sociology 2 (1), 113–120 Daudin, G., Rifflart, C. and Schweisguth, D. (2011), ‘Who produces for whom in the world economy?’, Canadian Journal of Economics 44(4), 1403–1437 Foster, N., Stehrer, R., & de Vries, G. (2011). ‘Decomposing net trade in value added and the patterns of trade in factors.’ Francois, J., Manchin, M. & Tomberger, P. (2013) Services Linkages and the Value Added Content of Trade. World Bank Policy Research Working Paper No. 6432. World Bank, Washington, DC. Gereffi, G., Humphrey, J., & Sturgeon, T. (2005). The governance of global value chains. Review of international political economy, 12(1), 78-104. Görg, H. (2000). Fragmentation and trade: US inward processing trade in the EU. Weltwirtschaftliches Archiv, 136(3), 403-422. Grossman, G. M., & Rossi-Hansberg, E. (2006). ‘Trading tasks: a simple theory of offshoring,’ American Economic Review 98(5), 1978-1977. Hoekman, B. (2014) Supply Chains, Mega-Regionals and the WTO. London: Centre for Economic Policy Research publication (summarised at http://www.voxeu.org/article/supply-chains-mega-regionals-and- wto-new-cepr-book). Hummels, D., Ishii, J., & Yi, K. M. (2001). ‘The nature and growth of vertical specialization in world trade. Journal of international Economics, 54(1), 75-96. Johnson, R. C., & Noguera, G. (2012). ‘Accounting for intermediates: Production sharing and trade in value added. Journal of International Economics, 86(2), 224-236. 38 Keane, J. (2015) “Firms and value chains in Southern Africa” ODI Working Paper. Final Draft, January, 2015 (unpublished). Keane, J. (2014) “Global value chain analysis: what’s new, what’s different, what’s missing?” ODI Briefing Paper. Koopman, R., Wang, Z., & Wei, S. J. (2012). Tracing value-added and double counting in gross exports (No. w18579). National Bureau of Economic Research. Lenzen, M., Kanemoto, K., Moran, D., & Geschke, A. (2012). Mapping the structure of the world economy. Environmental science & technology, 46(15), 8374-8381. Lenzen, M., Moran, D., Kanemoto, K., & Geschke, A. (2013). Building Eora: a global multi-region input–output database at high country and sector resolution.Economic Systems Research, 25(1), 20-49. Leontief, W. (1970). Environmental repercussions and the economic structure: an input-output approach;. The review of economics and statistics, 262-271. Lin, J. (2011). How to Seize the 85 million Jobs Bonanza. July 27, 2011. http://blogs.worldbank.org/developmenttalk/how-to-seize-the-85-million-jobs-bonanza Miller, R. E., & Blair, P. D. (2009). Input-output analysis: foundations and extensions. Cambridge University Press. OECD (2012) ‘Trade in Value-Added: Concepts, Methodologies and Challenges’ OECD/WTO (2013) ‘Aid for Trade at a Glance 2013: Connecting to Value Chains’ Paris/Geneva: OECD/WTO. Porter, M. E. (1985). Competitive Advantage: Creating and Sustaining Superior Performance. New York.: Simon and Schuster.. Swenson, D. L. (2013). The Nature of Outsourcing Relationships: Evidence from OAP Prices. Economic Inquiry, 51(1), 181-197. Morris, M., Staritz, C., & Barnes, J. (2011). Value chain dynamics, local embeddedness, and upgrading in the clothing sectors of Lesotho and Swaziland. International Journal of Technological Learning, Innovation and Development, 4(1), 96-119. Sturgeon, T. J., & Memedović, O. (2011). Mapping global value chains: Intermediate goods trade and structural change in the world economy. United Nations Industrial Development Organization. Taglioni and Winkler 2014a “Making global value cCHAINSahins work for development” Making global value chains work for development. Economic premise ; no. 143. Washington DC ; World Bank Group. Taglioni and Winkler 2014b “Bulgaria’s Integration and Economic Upgrading in Global Value Chains”. Tukker, A., & Dietzenbacher, E. (2013). Global multiregional input–output frameworks: an introduction and outlook. Economic Systems Research, 25(1), 1-19.) UNCTAD (2013) World Investment Report 2013 Geneva: United Nations Conference on Trade and Development Winkler, D., & Milberg, W. (2012). ‘Bias in the ‘Proportionality Assumption’ Used in the Measurement of Offshoring’. World Economics, 13(4), 39-60. World Bank (2013) ‘Global Value Chain Outcomes Analysis for South Africa’ Washington, DC: International Trade Unit, World Bank. WTO Secretariat (2009) ‘Trade Policy Review: Botswana, Lesotho, Namibia, South Africa and Swaziland” WT/TPR/S/222. Geneva: World Trade Organisation. Yeats, A. J. (1998). Just how big is global production sharing? (No. 1871). World Bank, Development Research Group. 39 Appendices Appendix 1: Overview of Eora and other MRIOs A growing number of international and MRIO tables are allowing us to analyse sources and destinations of value that flows through GVCs, identifying the role of countries and industries in supply chains. Pioneered by Wassily Leontief in the 1940s, these were initially mostly applied to the analysis of national accounts for individual countries and were later on used primarily for environmental and ecological analysis (e.g. measuring responsibility for emissions) as well as in regional science (see Miller and Blair 2009 for an overview). In recent years, national account data has been integrated with bilateral trade data and information about sourcing of inputs at the industry-level to derive information on use of domestic and imported inputs, and in turn, where value is added along supply chains provided a new, easily accessible approach to measuring trade in GVCs (see Figure 23 for an overview). Figure 23: Main strands of empirical research on GVCs Note: The size of the circles represents the coverage of each measure relatively to the real size of the GVCs phenomenon in the world economy. Larger circles stand for higher coverage. Source: Amador and Cabral, p. 17 Of late, triggered by greater interest in understanding processes of international fragmentation and global supply chains, numerous empirical papers have been published using existing input-output databases (especially the forerunner of many recent MRIOs, the Global Trade Analysis Project, GTAP). Measuring trade in value added via MRIOs has a number of clear benefits (see Ahmad 2013). Firstly, it provides a crucial tool to understand a country’s actual industrial structure and the national and international inter-linkages of sectors for developing growth and development strategies, as well as trade and industrial policies. Secondly, it makes clear how particularly non-tariff barriers (including regulatory measures) can impact competitiveness and upstream producers. Thirdly, it can help policymakers better anticipate ex ante the potential impact of macroeconomic shocks. Fourthly, it can allow for calculations of the ‘job content’ of trade. Finally, environmentally extended MRIOs allow for an assessment of the impact of trade as it affects ecosystem services. 40 MRIO tables are usually constructed through harmonised national supply-use tables (SUTs) and/or input- output tables (IOTs). These show truncations between domestic industries supplemented by tables breaking down imports by users, and have increasingly become indispensible for relevant macroeconomic and trade policy analyses (for an overview see Ahmad 2013, Jones et al. 2013, Dietzenbacher et al. 2013, Amadou and Cabral 2014). 22 MRIO tables allow for analysis of value contribution along supply chains, enabling the analysis of sources and destinations of value that flow through GVCs (for recent examples see Johnson and Noguera, Lenzen et al. 2013, UNCTAD 2013). There are now a number of global input-output databases that vary significantly in terms of country, sectoral coverage, time span and approach (see Table 10 below). Table 10: Overview of main MRIO databases Name Countries Type Detail (I x Time Extensions Approach p)a Eora World MR Variable 1970-2010 Various, Create initial estimate, (over SUT/IOT (20-500 especially gather all data in original 180) environmental formats, formulate constraints; detect and judge inconsistencies; set routine; calculate global MR SUT/IOT EXIOPOL/ World (43 MR SUT 129 x 129 2000 and 30 emissions, Create SUT; split use into CREEA countries 2007 60 IEA, energy domestic and imported use; + RoW) carriers, detail and harmonise SUTs; water, land, use trade shares to estimate 80 resources implicit exports; confront with exports in SUT; RAS out differences; add extensions WIOD World (40 MR SUT 35 x 59 1995- Detailed Harmonise SUTs; create countries 2009, socio- bilateral trade database for + RoW) annually economic and goods and services; adopt environmental import shares to split use satellite into domestic and imported accounts use; trade information for RoW is used to reconcile bilateral trade shares; add extensions GTAP- World MR IOT 57 x 57 1990, 5 (GWP), land Harmonise trade; use IOTs to MRIO (129 1992, use, energy link trade sets; IOT balanced countries) 1995, volumes, with trade and 1997, migration macroeconomic data 2001, 2004, 2007 GRAM World (40 MR IOT 48 x 48 2000, 2004 Various Use harmonised OECD IOTs; countries neglect differences like i x i and p x p; use OECD bilateral trade database to link trade 22For examples of early work on decomposing trade into the value-added shares see Hummels et al. (2001), Daudin et al. (2006), Johnson and Noguera (2010) and Koopman et al. (2012). 41 IDE- Asia- MR IOT 56 x 56 1975 – Employment Harmonise IOTs based on JETRO Pacific (8: (1975); 78 x 2005 matrices cross-country survey 1975; 10: 78 (1985- (2000, 2005) information; link via trade; 1985- 1995); 76 x manual balancing to reduce 2005) 76 (2000, discrepancies within certain 2005) bounds a i – number of industries, p - number of products; SUT… supply-use tables, IOT… Input-Output tables Source: Tukker and Dietzenbacher (2013) The spate of recent MRIO tables greatly improves our understanding of how GVCs function in practice. They are, however, subject to a degree of uncertainty – as is common with any applications using accounts data and trade flow data – which is augmented when examining developing counties, where statistical capacities tend to be substantially worse. Moreover, MRIO tables also suffer from a deficiency faced in many other empirical approaches to trade flow analysis, namely they are not able to accurately assess the impact of services. They furthermore are subject to two assumptions that contribute to their uncertainty, the former of which particularly is relevant for analysing trade flows from developing countries (OECD 2012). Firstly they assume that all products (for export and domestic use) have the same import content (proportionality assumption – see Winkler and Milberg 2012) and secondly they assume a uniform use of inputs among all firms in a sector (homogeneity assumption). The Eora MRIO dataset, has recently been used for GVC-related analyses in a number of international reports (UNCTAD 2013, AfDB/OECD/UNDP 2014). It has several advantages to other databases (see Lenzen et al. 2012). These include: 1. It is disaggregated into 187 countries (including all SACU members), providing important advantages for assessing impacts of consumption and production on relatively poor countries; 2. It has a historical time series spanning 1990-2011 (soon to be extended to 1970-2011 and updated with an approximate two year delay) based on an iterative process using an initial year estimate for 2000, overlaying estimates for 1999 and 2001 respectively with new data and then re-balancing; 3. It includes tables of basic prices, as well as two margins (taxes on products and subsidies on products) with constant prices to be added soon; 4. To clarify levels of uncertainty, standard deviation estimates have been calculated for all MRIO events; 5. It is publically available at www.worldmrio.com In its construction, it was based on the principle of changing to the structure of the original data sources as little possible for the sake of transparency. Its matrices are based on the use of the following types of raw data (in order of assumed accuracy):  input–output (I–O) tables and main aggregates data from national statistical offices where these are available;  I–O compendia from Eurostat, IDE-JETRO, and OECD,  the UN National Accounts Main Aggregates Database ,  the UN National Accounts Official Data,  the UN Comtrade international trade database, and  the UN Servicetrade international trade database. 42 This makes it well suited for dynamic analysis of smaller developing countries not included in other datasets,23 however should be complemented by more nuanced sectoral analysis drawing on alternative methodologies. Appendix 2: Methodology for value-added analysis The calculation below of the different value added measures comes directly from Koopman et al. (2010) who provide a full decomposition of value-added exports in a single framework encompassing previous measures by Hummels et al. (2001) and Johnson and Noguera (2012), among others. They start with a standard input-output model where each country produces goods in N tradable sectors (Leontief 1970, Miller and Blair 2009): X = (I – A)-1Y = LY,… … X is the 2 × 1 gross output vector for both countries … A is the 2 × 2 coefficient matrix giving intermediate use of each country’s sector of goods produced in r. … I is a 2 × 2 identity matrix … Y is the 2 × 1 final demand vector for both countries … L is the 2 × 2 Leontief inverse matrix. Assuming the case of a 2-region, 1 sector example (i.e. N=1) consisting of Botswana (indexed B) and the rest of the world (indexed R), the two-country system can be written in block matrix notation as: [ ] = [ ][ ] Value added trade is calculated using V as a 1 × 2 row vector with each element representing the value added per unit industry output and each element in V gives the share of direct domestic value added in total output. One can then calculate the value-added share (VAS) matrix as: 0 ≡ = [ ][ ] = [ ] 0 The columns in denote Botswana’s domestic value-added share of domestically produced products in a particular sector at home. The columns of denote the share of the Rest of the World’s value- added in these goods produced in Botswana. The second set of N columns present value-added shares for production in the RoW for Botswana’s goods ( ) and RoW value added for goods produced in the RoW. The sum along each column must equal unity: + = + = u 23In comparative analysis (UNCTAD 2013) with the WIOD dataset, Eora was found to provide broadly similar results when calculating foreign and domestic value added, albeit with a slight upward bias (which is to be expected as the greater number of highly heterogeneous developing countries, many of which have been subsumed in WIOD’s rest of world matrix). 43 In order to then determine domestic and imported content shares of each country’s production and trade at the sector level, use gross exports as weight, letting ∗ = ∑ = ∑( + ) 0 = [ ∗ ] , and 0 ∗ ̂=[(∗ ) E 0 ] 0 (∗ ) Where: … E is a 2 × matrix ̂ is a 2 × 2 diagonal matrix … The value-added share by source country for each sector can then be calculated as: _E ̂ = [ ̂ ≡ E ] Each element is the total upstream direct and indirect value added by source country and sector in gross exports for each sector (thus also including the specific sector itself). Looking at aggregates and 2 countries (i.e. in the example above) there is no need to define sectoral shares, so one can just use 0 = [ ∗ ] 0 ∗ and in turn for value-added by source in gross exports: ∗ ∗ VAS_E= VLE = [ ] ∗ ∗ The diagonal elements represent the domestic value-added of each country’s exports; off-diagonal elements give the foreign value-added embodied in each country’s exports24: ∗ DV= [ ] ∗ ∗ FV = [ ] ∗ Assuming 3 countries (Botswana, Swaziland [indexed S] and RoW) and N sectors, production, value-added share and sources of value-added in gross exports are as follows: X = (I – A)-1Y = LY,… VAS = VL VAS_E = VLE 24Generalization of Hummels, Ishii and Yi (2001) vertical specialization measure as the this only caputres foreign value added in gross exports when only one country’s intermediate goods are used abroad. 44 X and Y are 3 × 1 vectors; A and L are 3 × 3; V and VAS are 3 × 3 matrices; E is a 3 × 3 and VAS_E is 3 × 3. For aggregate measures, all results continue to hold – can be expressed just by replacing relevant weighting matrix. Complexity arises from intermediate inputs that cross multiple borders (derivation via application of expression for inverse of a partitioned matrix). As before, value-added shares can be applied to gross exports to produce VAS_E (3 × 3). ∗ ∗ ∗ VAS_E = [ ∗ ∗ ∗ ] ∗ ∗ ∗ Here the sum of off-diagonal elements along a column is the measure of foreign value-added embodied in a country’s gross exports, i.e.: = ∑≠ , or for this 3-coutnry case: = + The sum of off-diagonal elements along a row provides information on a country’s value-added embodied as intermediate inputs in third countries’ gross exports (i.e. indirect value added): = ∑≠ , Domestic value in gross exports is, as above: = Sum of all DV and FV should add up to gross exports To capture country’s position (upstream/downstream) it makes sense to compare its exports of intermediates used by other countries, with that country’s use of imported intermediates in the same sectors. If it lies upstream in the global value-chain, it participates in producing inputs for others – then its IV share of gross exports will exceed its FV share: _ = + 45 Appendix 3: Correspondence between Eora sectors and ISIC Rev.3 Eora sector ISIC Rev.3 correspondence Agriculture 1,2 Fishing 5 Mining and quarrying 10, 11, 12, 13, 14 Food and beverages 15, 16 Textiles and wearing apparel 17, 18, 19 Wood and paper 20, 21, 22 Petroleum, chemical and non-metallic mineral products 23, 24, 25, 26 Metal products 27, 28 Electrical and machinery 29, 30, 31, 32, 33 Transport equipment 34, 35 Other manufacturing 36 Recycling 37 Electricity, gas and water 40, 41 Construction 45 Maintenance and repair 50 Wholesale trade 51 Retail trade 52 Hotels and restaurants 55 Transport 60, 61, 62, 63 Post and telecommunications 64 Financial intermediation and business activities 65, 66, 67, 70, 71, 72, 73, 74 Public administration 75 Education, health and other services 80, 85, 90, 91, 92, 93 Private households 95 Other 99 46 Appendix 4: Exports and imports for SACU countries - Comparing Eora and Comtrade This section provides an overview of changes in each of the five country’s aggregate and sectoral exports between 2001 and 2011 drawing on the Eora database, as this is the source of the value-added analysis. For Botswana, the Eora data (Table 11) shows a particularly strong increase in the significance of service exports in the economy. While the Other Manufacturing sector (which is likely to include diamonds) was the largest in 2001, the two largest export sectors in 2011 were Transport and Hotels and Restaurants. Eora data suggests a strong increase in both exports and imports in the given time period, with the former growing at almost 11% annually and the latter at over 30%. Besides the Other Manufacturing sector, the most important non-services sector is Food and Beverages, which has seen considerable growth (20.8% p.a.). In terms of imports, the largest sector is the Petroleum, Chemical and Non-Metallic Mineral Products sector, which has also experienced substantial growth since 2001, followed by Electrical and Machinery. Financial Intermediation and Business Services is the largest services import sector. What is most striking when one compares Eora trade data to Comtrade’s data for Botswana (Table 12) is the fact that Other Mining and Quarrying, which is by far the largest export sector in Comtrade (more than US$8 billion in exports and 15 times as many exports as any other sector) is completely underrepresented in Eora, both in terms of trade value and relative significance (it is 6 th with approximately 45 million in exports). This discrepancy inevitably skews Botswana’s results for any kind of value-added trade analysis. In terms of imports, the overlap is more significant with the EORA Petroleum sector aggregated from ISIC sectors 23-27 on Comtrade. Likewise, while Comtrade does not have information on the four services sectors featured as top ten imports in Eora (Financial Intermediation and Business Services, Public Administration, Education, Health and Other Services and Construction), there is an approximate correspondence between some of the other top-ten import sectors if one aggregates ISIC sectors to the 26-sector Eora format. Table 11: Top 10 import and export sources (Eora) for Botswana, 2011 Exports Imports Sector Value (in Annual Sector Value (in Annual $1000) growth since $1000) growth since 2001 (%) 2001 (%) Petroleum, Chemical and 19.5 19.5 Transport 174,082 Non-Metallic Mineral 690,397 Products Hotels and 168,985 22.0 Electrical and Machinery 546,937 22.0 Restaurants Other Financial Intermediation 100,843 0.2 334,514 0.2 Manufacturing and Business Activities Food & Beverages 79,724 20.8 Transport Equipment 268,340 20.8 Education, Health 60,420 13.7% Public Administration 267,401 13.7% and Other Services Electrical and Education, Health and 45,284 0.28 217,366 0.28 Machinery Other Services Mining and 42,807 2.00 Construction 207,161 2.00 Quarrying Transport 42,651 3.36 Food & Beverages 193,044 3.36 Equipment 47 Textiles and 39,598 2.38 Metal Products 180,521 2.38 Wearing Apparel Post and 35,272 1.99 Transport 156,526 1.99 Telecommunications Source: Own computations using Eora database Table 12: Top 10 import and export sectors (Comtrade) for Botswana, 2011 Exports Imports Sector Value ( in Sector Value ( in $1000) $1000) Manufacture of Machinery Other Mining and Quarrying 8,266,573 and Equipment 201,976 Manufacture of Fabricated Manufacture of furniture 553,748 Metal Products 169,873 Manufacture of Basic Metals 456,757 Other Mining and Quarrying 166,646 Manufacture of Electrical Mining of Metal Ores 53,225 Machinery 156,209 Manufacture of Radio and Manufacture of Motor Vehicles 45,465 Television 88,072 Manufacture of Food Products and Beverages 32,947 Manufacture of Chemicals 72,558 Manufacture of Motor Manufacture of Chemicals 15,777 Vehicles 71,880 Manufacture of Machinery and Manufacture of Wearing Equipment 12,743 Apparel 57,139 Manufacture of Coke, Refined Petroleum 12,115 Manufacture of furniture 52,994 Manufacture of Textiles 11,395 Manufacture of Basic Metals 52,359 Source: Comtrade Lesotho’s top export sector – Textiles and Wearing Apparel – far exceeds flows from any other goods sector according to Eora data (Table 6)(Table 13) with the subsequent sectors predominantly consisting of services (most prominently Transport, Public Administration and Hotels and Restaurants). These sectors have also grown significantly faster than the textiles and apparel sector. However, the largest export sector according to Comtrade, Other Mining and Quarrying, is again far less significant on Eora (Table 13) in Eora its approximate equivalent sector, Mining and Quarrying (which also includes four other ISIC sectors) has an export volume of approximately US$ 14 million on Eora compared to US$716 million for Comtrade. Eora reports the largest import sectors to be predominantly services (Public Administration, Financial Intermediation, and Transport) – in all likelihood a product of its dependence on South African imports. Leading goods imports are listed as Wholesale Trade (which is not more clearly specified), Electrical and Machinery and Petroleum, Chemical and Non-Metallic Mineral Products. Again, the discrepancy to Comtrade is rather striking– here textiles are by far the largest import product, followed – more consistently with results from the Eora database- by “Manufacture of Chemicals” and “Manufacture of Radio and Television.” 48 Table 13: Top 10 import and export sectors (Eora) for Lesotho, 2011 Exports Imports Sector Value ( in Annual Sector Value ( in Annual $1000) growth since $1000) Growth since 2001 (%) 2001 Textiles and Wearing 97,848 20.6 Public Administration 121,309 32.5 Apparel Financial Intermediation Transport 47,040 34.5 109,822 19.4 and Business Activities Public Administration 19,718 97.3 Transport 100,507 66.3 Hotels and 18,618 62.5 Wholesale Trade 100,335 49.6 Restaurants Education, Health and 18,523 66.7 Electrical and Machinery 70,980 42.5 Other Services Education, Health and Wholesale Trade 18,314 67.9 70,781 20.4 Other Services Petroleum, Chemical Post and 17,936 53.3 and Non-Metallic 52,624 41.2 Telecommunications Mineral Products Post and Construction 17,287 74.1 49,698 68.0 Telecommunications Re-export & Re-import 16,254 71.7 Transport Equipment 46,955 54.8 Retail trade 15,826 7.07 Construction 45,716 25.5 Source: Own computations using Eora database Table 14: Top 10 import and export sectors (Comtrade) for Lesotho, 2011 Exports Imports Sector Value ( in Sector Value ( in $1000) $1000) Other Mining and Quarrying 716,373 Manufacture of Textiles 111,462 Manufacture of Wearing Apparel 274,034 Manufacture of Chemicals 25,968 Manufacture of Radio and Manufacture of Textiles 81,575 Television 22,577 Agriculture, Hunting and Manufacture of furniture 34,138 Related Sectors 13,273 Manufacture of Machinery Manufacture of Electrical Machinery 11,498 and Equipment 12,790 Manufacture of Food Products and Beverages 2,224 Manufacture of Medical 12,168 Manufacture of Other Transport Manufacture of Electrical Equipment 763 Machinery 6,984 Manufacture of Rubber and Plastics 323 Publishing, Printing… 5,646 Manufacture of Paper and Paper Manufacture of Food Products 265 Products and Beverages 5,466 Manufacture of Rubber and Manufacture of Chemicals 172 Plastics 4,995 Source: Comtrade 49 For Namibia, the largest three export sectors according to Eora are Food and Beverages, Petroleum, Chemical and Non-Metallic Mineral Products, Mining and Quarrying and Transport (Table 15). Here particularly mining and transport have seen remarkable growth over the past years (46% p.a. and 39% p.a., respectively, since 2001). Subsequent sectors are primarily manufacturing, including Electrical and Machinery, Transport Equipment and Textiles and Wearing Apparel. Comtrade data again has Other Mining and Quarrying as the top export sector (Table 9), which broadly reflects the importance of Mining and Quarrying in the Eora rankings. The subsequent sectors on Comtrade are Manufacture of Basic Metals, Manufacture of Food Products and Manufacture of Coke and Refined Petroleum. In terms of Namibia’s imports, the leading imports according to Eora are Electrical and Machinery, Petroleum, Chemical and Non-Metallic Mineral Products and Financial Intermediation. Again there is a substantial discrepancy to Comtrade, where the leading import is again related to mining (Mining of Metal Ores) followed by Manufacture of Machinery and Manufacture of Food Products (which is ranked 6th according to Eora). Table 15: Top 10 import and export sectors (Eora) for Namibia, 2011 Exports Imports Sector Value ( in Annual Sector Value ( in Annual $1000) growth since $1000) Growth since 2001 (%) 2001 Food & Beverages 738,877 25.2 Electrical and Machinery 439,343 34.5 Petroleum, Chemical Petroleum, Chemical and Non-Metallic 255,705 25.7 and Non-Metallic 390,857 22.7 Mineral Products Mineral Products Financial Intermediation Mining and Quarrying 233,262 46.7 307,464 25.9 and Business Activities Transport 174,492 38.8 Public Administration 259,543 35.5 Electrical and 130,698 55.2 Transport Equipment 219,818 34.4 Machinery Education, Health and Transport Equipment 97,785 20.3 204,362 29.0 Other Services Textiles and Wearing 94,898 35.8 Food & Beverages 196,442 33.3 Apparel Agriculture 86,970 27.8 Construction 166,123 31.3 Metal Products 74,061 42.1 Transport 141,606 33.5 Hotels and 70,476 41.3 Metal Products 138,940 32.8 Restaurants Total 2,266,669 31.0 Total 3,202,782 30.4 Source: Own computations using Eora database Table 16: Top 10 import and export sectors (Comtrade) for Namibia, 2011 Exports Imports Sector Value ( in Sector Value ( in $1000) $1000) Other Mining and Quarrying 1,622,257 Mining of Metal Ores 639,250 50 Manufacture of Machinery and Manufacture of Basic Metals 1,419,880 Equipment 271,041 Manufacture of Food Products and Manufacture of Food Products Beverages 942,701 and Beverages 159,416 Manufacture of Coke, Refined Petroleum 713,334 Manufacture of Motor Vehicles 153,705 Mining of Uranium and Thorium Ores 194,633 Manufacture of Chemicals 129,420 Manufacture of furniture 177,684 Manufacture of Textiles 80,778 Manufacture of Radio and Agriculture, Hunting and Related Sectors 123,012 Television 68,904 Manufacture of Coke, Refined Manufacture of Basic Metals 80,510 Petroleum 68,679 Fishing 52,698 Manufacture of Basic Metals 59,134 Manufacture of Fabricated Metal Manufacture of Electrical Products 52,660 Machinery 50,365 Source: Comtrade For South Africa, the largest export sectors according to Eora for 2011 were – broadly speaking – in mining and minerals, including the ‘Other mining products’ and ‘Iron and steel products’, followed by ‘Coal and lignite and non-ferrous metals’ (Table 17). The largest non-mineral-related sector was ‘Transport services’ followed by ‘Other business services’.25 This mirrors the results in Comtrade (Table 18), where the largest sectors were ‘Manufacture of basic metals’ and ‘Mining of metal ores’, followed by ‘Manufacture of motor vehicles’, which likely captures ‘Transport services’ and the 9th ranked sector, ‘Motor vehicle parts’. The other top export sectors are either extractives, such as ‘Basic chemical products’ (6th), ‘Petroleum products’ (8th) as well as agricultural products (7th). These are all represented in Comtrade as leading sectors. Among the leading imports, Eora’s top sectors are ‘Motor vehicles’, followed by the two services sectors ‘General government’ and ‘Communications’, and then ‘Other mining’ and ‘Transport services’. In Comtrade the leading import sector is ‘Extraction of crude petroleum’ and ‘Manufacture of machinery and equipment’ followed by ‘Manufacture of chemicals and chemical products’ and ‘Manufacture of motor vehicles’. While the ranking differs, both Eora and Comtrade demonstrate the centrality of the extractive and transport sectors both as inputs into further production and into final demand. Table 17: Top 10 import and export sectors (Eora) for South Africa, 2011 Exports Imports Annual Annual Value ( in Value ( in Sector growth since Sector Growth since $1000) $1000) 2001 (%) 2001 Other mining 29.2 25.7 products 15,651,790 Motor vehicles 6,873,574 Iron and steel 27.1 27.7 products 11,870,540 General Government 4,451,320 Coal and lignite 15.0 24.8 products 4,057,202 Communications 2,078,012 25As Eora could draw on South Africa’s own input-output table, there are more sectors available than for the other SACU countries. However, the analysis in the subsequent sections draws on the summarized 26-sector summary (25, excluding mining and quarrying). 51 Non-ferrous metals 9,317,438 27.1 Other mining 3,828,292 27.6 Transport services 7,911,638 31.1 Trade 3,747,478 26.9 Other business 28.4 25.0 services 7,841,567 Transport services 3,159,089 Basic chemical 25.0 24.0 products 7,369,986 Agriculture 2,472,700 Agricultural products 5,407,014 27.1 Iron and steel 2,130,715 18.5 Petroleum products 2,518,601 7.2 Buildings 1,921,877 29.3 Other manufacturing 2,445,980 30 Petroleum 1,756,022 20 Motor vehicles parts 2,233,046 20 Insurance 1,739,630 20 Source: Own computations using Eora database Table 18: Top 10 import and export sectors (Comtrade) for South Africa, 2011 Exports Imports Sector Value ( in Sector Value ( in $1000) $1000) Manufacture of basic metals 32,237,822 Extraction of crude petroleum 14,300,212 Manufacture of machinery Mining of metal ores 14,164,483 and equipment 12,711,015 Manufacture of motor vehicles, Manufacture of chemicals and trains, … 9,020,687 chemical products 10,645,394 Manufacture of motor Mining of coal and lignite 7,525,758 vehicles, trains, … 8,920,932 Manufacture of machinery and Manufacture of coke, refined equipment 6,896,916 petroleum 6,4613,79 Manufacture of chemicals and chemical Manufacture of food products products 6,886,717 and beverages 5,530,841 Manufacture of food products and Manufacture of radio, beverages 5,083,184 television,… 5,307,494 Agriculture, hunting and related sectors 4,165,987 Manufacture of basic metals 3,554,794 Manufacture of electrical Manufacture of coke, refined petroleum 3,439,863 machinery 3,092,348 Manufacture of fabricated metal Manufacture of office, products 1,900,436 accounting… 2,717,723 Source: Comtrade Swaziland’s largest export sectors are Food and Beverages, followed by Electrical and Machinery and Transport, Petroleum, Chemical and Non-Metallic Mineral Products and Hotels and Restaurants (Table 19). Together with Education and Health, both Transport and Hotels and Restaurants have been among the fastest growing sectors (up to 50% p.a.), indicative of the increasing importance of the services economy. This is broadly consistent with Comtrade data, where Manufacture of Food Products and Beverages is the largest sector, ahead of Manufacture of Chemicals and Manufacture of Machinery and Equipment. The other goods in Swaziland’s list of ten most significant exports on Eora – Agriculture, Metal Products, Textiles and Wearing Apparel and Wood and Paper – are all represented in the top 10 according to Comtrade data. 52 The largest sources of imports according to Eora are Petroleum, Chemical and Non-Metallic Mineral Products, Electrical and Machinery and Financial Intermediation. This is followed by two other services sectors: Public Administration and Education and Health. When comparing Eora import data for Swaziland to Comtrade (Table 20), this does create numerous discrepancies. Comtrade data has Manufacture of Motor Vehicles, Manufacture of Chemicals, Manufacture of Textiles and Manufacture of Machinery and Equipment as the top goods sectors. As Manufacture of Chemicals is one of four sectors subsumed in Eora’s Petroleum, Chemical and Non-Metallic Mineral Products sector and Transport Equipment ranks sixth in the Eora ranking (and third among goods sectors) there is a certain degree of overlap, though as before the omission of certain sectors of particular importance in the Comtrade rankings (e.g. textiles) is conspicuous. Table 19: Top 10 import and export sectors (Eora) for Swaziland, 2011 Exports Imports Annual Annual Value ( in Value ( in Sector growth since Sector Growth $1000) $1000) 2001 (%) since 2001 Petroleum, Chemical and Food & Beverages 211,043 15.5 Non-Metallic Mineral 260,906 12.3 Products Electrical and 194,634 27.3 Electrical and Machinery 230,333 20.0 Machinery Financial Intermediation Transport 183,346 47.1 163,488 17.2 and Business Activities Petroleum, Chemical and Non-Metallic 88,480 2.1 Public Administration 118,285 18.4 Mineral Products Hotels and Education, Health and 87,129 50.9 102,794 18.7 Restaurants Other Services Agriculture 56,498 21.7 Transport Equipment 102,745 18.7 Metal Products 56,463 37.7 Food & Beverages 97,073 21.1 Education, Health and 49,298 45.2 Construction 92,147 18.1 Other Services Textiles and Wearing 47,988 13.8 Metal Products 79,513 18.8 Apparel Wood and Paper 47,966 1.9 Transport 76,134 20.5 1,228,800 1,761,590 Total 20.7 Total 17.5 Source: Own computations using Eora database Table 20: Top 10 import and export sectors (Comtrade) for Swaziland, 2011 Exports Imports Sector Value ( in Sector Value ( in $1000) $1000) Manufacture of Food Products and Manufacture of Motor Beverages 456,054 Vehicles 61,422 Manufacture of Chemicals 235,535 Manufacture of Chemicals 50,189 Manufacture of Machinery and Equipment 79,342 Manufacture of Textiles 32,355 53 Agriculture, Hunting and Related Manufacture of Machinery Sectors 73,965 and Equipment 21,693 Manufacture of Other Manufacture of Wearing Apparel 62,402 Transport Equipment 21,617 Agriculture, Hunting and Manufacture of Basic Metals 45,162 Related Sectors 18,007 Manufacture of Coke, Refined Manufacture of Food Products Petroleum 25,899 and Beverages 17,968 Manufacture of Radio and Television 25,384 Manufacture of furniture 16,387 Manufacture of Radio and Manufacture of Textiles 22,490 Television 14,595 Manufacture of Medical 12,711 Manufacture of Basic Metals 14,582 Source: Comtrade As the above analysis shows, there exist substantial discrepancies between Eora and Comtrade, even when going beyond sectoral trade volumes and examining just the relative significance of different sectors in terms of total imports/exports. Trade data for SACU countries is notoriously flawed so there is little guarantee that either source provides an accurate estimate of ‘true’ trade volumes (and initial cursory analysis of trade data on the SACU statistical portal provides a further set of contradictions). However, the processes required to create the contiguous Eora database have likely come at the expense of precision for less globally significant economies, including SACU countries. In this context, Table 22 through Table 25 provides a helpful overview not only of total non-exported output across time for each of the 15 tradable sectors (23 for South Africa as it has a more sophisticated IO table), calculated by adding output produced by a sector for use in any domestic sector as an intermediate and for domestic final consumption (but not for export), as well as the standard deviation. This provides some indication of the reliability of estimates. As can be seen, standard deviations vary across sectors, countries, and time26. For example, standard deviations tend to be significantly lower for services sectors than for mining and quarrying, while the level of uncertainty for Swaziland (over 60% for 2000 values and 30% for 2011) is far in excess of those of the other four SACU countries, where in aggregate the standard deviation tends to be range from a fraction of 1% (South Africa) to 5% of the result (Namibia and Lesotho). For each country adherence reports are available describing in greater detail which data sources have been respected most and least in the final outcome. Table 21: Non-exported output and standard deviation of 15 key sectors for Botswana, 2000 and 2011 2011 2000 Sector Value Standard Value ( in $1000) Standard (in $1000) deviation deviation (in $1000) (in $1000) Agriculture 383,261 3,516 121,917 87 Electrical and Machinery 1,500,841 4,936 556,254 2 26 According to the Eora website, standard deviations are calculated where different data sources exist and assigning quality scores to these. Then, “conflicting data points along with quality scores are run through optimization software which produces a quality-weighted result.” If two conflicting estimates exist, the researchers first ascertaine – either from written documentation, or from data source provider interviews – the reliability of the data source, and then assign standard deviations to the two or more values, with the optimiser then producing a quality-weighted final result. 54 Financial Intermediation and Business Activities 5,554,148 2,023 1,895,242 0 Fishing 26,225 43 6,610 91 Food & Beverages 628,833 1,714 234,667 5 Hotels and Restaurants 777,537 7,247 211,257 141 Metal Products 476,045 2,663 200,577 22 Mining and Quarrying 1,016,865 99,065 169,162 458 Other Manufacturing 205,582 105 47,222 63 Petroleum, Chemical and Non-Metallic Mineral Products 1,363,878 5,119 532,774 4 Post and Telecommunications 836,964 741 306,151 7 Textiles and Wearing Apparel 131,175 704 54,872 135 Transport 731,332 665 266,167 52 Transport Equipment 699,045 2,180 272,529 12 Wood and Paper 426,097 1,770 176,108 19 Total (all sectors) 26,532,650 179,446 8,759,349 14,862 Source: Own computations using Eora database Table 22: Non-exported output and standard deviation of 15 key sectors for Lesotho, 2000 and 2011 2011 2000 Sector Value Standard Value ( in $1000) Standard (in $1000) deviation deviation (in $1000) (in $1000) Agriculture 170,660 2,851 35,617 191 Electrical and Machinery 328,269 6,915 102,207 344 Financial Intermediation and Business Activities 1,053,422 592 444,168 22 Fishing 17,352 211 2,180 531 Food & Beverages 211,462 2,449 62,134 206 Hotels and Restaurants 179,279 322 63,539 63 Metal Products 130,875 3,643 36,935 559 Mining and Quarrying 111,166 5,988 12,394 3,756 Other Manufacturing 57,897 263 16,017 286 Petroleum, Chemical and Non-Metallic Mineral Products 294,126 7,025 88,163 683 Post and Telecommunications 157,451 208 58,118 0 Textiles and Wearing Apparel 26,076 772 8,333 391 Transport 173,330 284 58,689 0 Transport Equipment 152,713 3,122 45,366 182 Wood and Paper 131,753 2,443 35,614 580 Total (all sectors) 5,532,083 54,329 1,876,050 16,929 Source: Own computations using Eora database 55 Table 23: Non-exported output and standard deviation of 15 key sectors for Namibia, 2000 and 2011 2011 2000 Sector Value Standard Value ( in $1000) Standard (in $1000) deviation deviation (in $1000) (in $1000) Agriculture 782,552 100,970 187,154 78 Electrical and Machinery 1,058,158 10,669 345,435 11 Financial Intermediation and Business Activities 4,698,246 7,761 1,548,393 2 Fishing 23,578 471 5,453 161 Food & Beverages 458,088 3,700 156,539 0 Hotels and Restaurants 753,301 17,679 214,359 3 Metal Products 342,036 5,558 122,712 18 Mining and Quarrying 743,865 175,197 120,141 76 Other Manufacturing 163,039 297 48,532 127 Petroleum, Chemical and Non-Metallic Mineral Products 836,127 10,796 287,976 2 Post and Telecommunications 835,131 40,803 242,667 0 Textiles and Wearing Apparel 89,628 1,297 34,701 160 Transport 779,543 30,113 238,494 1 Transport Equipment 467,252 4,768 150,500 81 Wood and Paper 380,157 3,667 121,817 65 Total (all sectors) 21,838,240 578,107 6,484,898 12,582 Source: Own computations using Eora database Table 24: Non-exported output and standard deviation of 23 key sectors for South Africa, 2000 and 2011 2011 2000 Sector Value Standard Value ( in $1000) Standard (in $1000) deviation deviation (in $1000) (in $1000) Agricultural products 21,650,518 328 7,437,639 1,037 Beverages and tobacco products 12,445,538 27 4,262,259 484 Dairy products 2,961,020 20 1,018,344 698 Fruit and vegetables products 1,918,177 12 667,194 830 FSIM 11,598,011 624 3,728,123 2,445 Furniture 2,733,789 4 939,597 2,001 Gold and uranium ore products 2,820,449 40 935,523 2,239 Iron and steel products 7,838,554 210 2,751,794 1,346 Leather products 465,829 71 129,146 732 Meat products 8,053,766 68 2,757,444 634 56 Motor vehicles 19,069,797 82 6,099,955 745 Motor vehicles parts 4,960,422 71 1,686,151 926 Non-ferrous metals 3,435,876 136 1,273,067 1,342 Other business services 17,902,473 242 5,945,001 2,497 Other mining products 10,170,479 259 3,812,418 1,095 Paper products 3,422,100 116 1,138,586 872 Petroleum products 10,198,927 144 2,920,779 1,434 Plastic products 4,424,651 252 1,479,190 1,794 Radio and television products 3,189,169 28 1,079,260 523 Textile products 2,040,989 131 635,060 868 Transport services 34,676,479 144 11,483,357 1,280 Wearing apparel 3,673,105 2 1,261,429 1,957 Wood products 3,015,569 112 976,779 767 Total (all sectors) 659,938,975 7,683 214,167,952 119,236 Source: Own computations using Eora database Table 25: Non-exported output and standard deviation of 15 key sectors for Swaziland, 2000 and 2011 2011 2000 Sector Value Standard Value ( in Standard (in $1000) $1000) deviation deviation (in $1000) (in $1000) Agriculture 192,194 53,592 51,421 41,878 Electrical and Machinery 392,319 228,382 147,627 173,158 Financial Intermediation and Business Activities 1,874,520 1,007,495 759,029 738,247 Fishing 13,651 1,486 2,734 1,468 Food & Beverages 242,711 85,074 80,651 65,776 Hotels and Restaurants 267,420 6,870 100,527 5,717 Metal Products 155,291 120,112 61,008 92,320 Mining and Quarrying 106,815 7,335 31,937 6,139 Other Manufacturing 65,290 5,654 22,159 4,823 Petroleum, Chemical and Non-Metallic Mineral Products 472,864 229,321 161,275 173,757 Post and Telecommunications 267,443 90,012 103,364 69,520 Textiles and Wearing Apparel 43,261 29,318 14,830 23,373 Transport 322,730 67,656 126,048 52,527 Transport Equipment 217,904 106,899 79,748 82,321 Wood and Paper 157,423 79,804 37,817 61,909 Total (all sectors) 7,539,815 2,288,715 2,865,782 1,727,781 Source: Own computations using Eora database 57 Appendix 5: Growth of DVA embodied in gross exports, by sector Figure 24: Compound annual growth rate of DVA embodied in gross exports by sector, 2000-2011 58 Source: Own computations using Eora database 59 Appendix 6: FVA in exports as a share of gross exports, by sector Figure 25: Foreign value added in exports by sector, 2000 and 2011 60 61 Source: Own computations using Eora database 62 Appendix 7: Forward and backward integration by partner country Figure 26: Foreign and indirect value added in exports by source and destination, 2011 63 64 Source: Own computations using Eora database 65