WPS7380 Policy Research Working Paper 7380 Diversification, Growth, and Volatility in Asia Chris Papageorgiou Nikola Spatafora Ke Wang East Asia and the Pacific Region Office of the Chief Economist July 2015 Policy Research Working Paper 7380 Abstract Economic development critically involves diversification entry into completely new products. In addition, devel- and structural transformation—that is, the continued, oping Asia has on average benefited significantly from dynamic reallocation of resources from less productive to quality upgrading, helping it capitalize on already exist- more productive sectors and activities. This paper docu- ing comparative advantages. Yet, agricultural and natural ments that, over an extended period, developing Asia has resources tend to have lower potential for quality upgrad- on average been particularly successful in diversifying its ing than manufactures. Therefore, for lower-income exports, particularly in comparison with Sub-Saharan “frontier” countries, diversification into products with Africa. Much of the progress has occurred through diver- longer “quality ladders” may be a necessary first step sification along the ‘extensive margin,’ that is, through before large gains from quality improvement can be reaped. This paper is a product of the Office of the Chief Economist, East Asia and the Pacific Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at nspatafora@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. 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. Produced by the Research Support Team Diversification, Growth, and Volatility in Asia Chris Papageorgiou, Nikola Spatafora, and Ke Wang1 JEL Classification Codes: F14, L15. Keywords: Structual Transformation; Export Diversifcation and Quality; Growth; Volatility; Asian Countries. Authors’ E-Mail Addresses: cpapageorgiou@imf.org; nspatafora@worldbank.org; kwang3@imf.org. 1 Corresponding author: Nikola Spatafora. We thank Doug Addison, Andrew Berg, Frederico Gil Sander, Cathy Pattillo, and Alfred Schipke for useful comments on earlier drafts. A version of this paper is forthcoming in Schipke, ed., Frontier and Developing Asia: The Next Generation of Emerging Markets. 2 Contents Page I. Introduction ............................................................................................................................3  II. How Is Diversification Measured? ........................................................................................4  III. Diversification, Growth and Volatility ................................................................................5  IV. Patterns of Diversification ...................................................................................................6  V. Patterns of Quality Upgrading ..............................................................................................7  VI. Country Case Studies...........................................................................................................9  VII. Conclusions ......................................................................................................................10  Appendix I. Definitions of Main Indices .................................................................................20  A. Herfindahl Index .....................................................................................................22  B. Theil Index ..............................................................................................................22  C. Product Quality .......................................................................................................23  Appendix II. Region Definition ...............................................................................................26  Appendix III. Case Studies ......................................................................................................27  3 I. INTRODUCTION Limited diversification in exports and broader economic structure has long been an underlying characteristic of many developing economies. Yet some have shown a remarkable economic transformation, especially over the past two decades. In particular, it has been argued that developing Asia has benefited significantly from diversification. This paper examines the claim with a comprehensive look at the facts, employing newly developed data sets covering diversification in both external trade and domestic production. The paper focuses on two key questions. First, is diversification crucial to sustaining growth and reducing volatility? Put differently, does concentration in sectors with limited scope for productivity growth and quality upgrading, such as primary commodities, result in less broad- based and sustainable growth? And does lack of diversification increase exposure to adverse external shocks and macroeconomic instability? Second, what precisely does diversification, in both external trade and the broader domestic economy, involve? How is it linked to broader structural transformation, including the process of quality upgrading? And which countries and regions have been more successful in promoting diversification? Throughout, our focus is on Asia, and in particular on two groups of countries: “Emerging Asia”, comprising those economies generally classified as emerging markets; and “Frontier Asia”, comprising some economies that are still at lower income levels, but have experienced rapid growth and, in most cases, demonstrated a fair degree of macroeconomic stability over an extended period of time.2 2 Frontier Asia comprises: Bangladesh, Bhutan, Cambodia, Lao PDR, Maldives, Mongolia, Myanmar, Nepal, Papua New Guinea, and Vietnam. Emerging Asia comprises: China, India, Indonesia, Malaysia, the Philippines, Sri Lanka, and Thailand. 4 II. HOW IS DIVERSIFICATION MEASURED? Measures of economic diversification need to look beyond trade, to capture domestic sector diversification and the underlying dynamic process of structural transformation. Trade diversification and domestic diversification are in principle interlinked, the former reflecting diversification in the external sector, and the latter capturing diversification in the domestic production process across sectors. An underlying theme of this paper is that focusing on the entire structure of production paints a more comprehensive and illuminating picture. Therefore, the two dimensions of diversification are evaluated simultaneously, filling a gap in the existing literature, which has treated them independently. In addition, the analysis focuses on “diversification spurts,” that is, rapid, sustained, significant spells of diversification. Trade diversification can be achieved along several dimensions. First, diversification may occur across either products or trading partners. Second, product diversification may occur through the introduction of new product lines (the extensive margin), or a more balanced mix of existing exports (intensive margin). Finally, product-quality upgrading represents a slightly different notion and is evidenced by higher prices for existing exports. Our main data source for trade is an updated version of the UN–NBER data set, which harmonizes UN COMTRADE bilateral trade flow data at the 4-digit SITC (Rev. 1) level.3 However, while the existing literature typically focuses on the post-1988 period, this paper uses data extending back to 1962. The extended time dimension turns out to be greatly helpful in examining relationships more comprehensively. Analysis of domestic diversification in frontier economies required construction of a new data set. This paper examines diversification in sectoral output and the sectoral allocation of labor using data from existing and new sources. Existing data sets include measures of value added for 28 manufacturing sectors, during 1985–2010 (from UNIDO, 2011; 3-digit ISIC classification); and labor employment shares in nine economy-wide sectors, during 1969–2008 3 The dataset combines importer- and exporter-reported data from COMTRADE to maximize comprehensiveness, while ensuring internal consistency, using the methodology of Asmundson (forthcoming). 5 (from ILO, 2011; 1-digit classification). It is well known, however, that both of these data sets are quite limited in their coverage of frontier countries. For this reason, a new data set was constructed, covering 12 economy-wide sectors during 2000–2010, using country data compiled from IMF desk inputs (see below for further discussion). Appendix I provides greater details on the diversification indices and quality measures employed in this paper. Briefly, diversification is measured using the Theil index, which has the advantage of being decomposable into diversification along the extensive and intensive margins. Lower values of the index indicate greater diversification. Quality measures are based on individual products’ unit values (that is, trade prices), but with important adjustments for differences in production costs, as well as for selection bias in the composition of international trade. Appendix II sets out a full list of the countries and regions analyzed. III. DIVERSIFICATION, GROWTH, AND VOLATILITY We start by examining the evidence on the links between diversification and growth. One result stands out: diversification patterns and growth are clearly related, although the relationship displays much heterogeneity. In particular, greater diversification is on average associated with faster subsequent output growth (Figure 1). The relationship holds both for the sample as a whole, and for Asian countries alone. Adopting a multivariate regression approach, output growth remains significantly associated with both initial diversification and initial product quality measures, even after controlling for a variety of standard growth determinants (Table 1). This conclusion is in line with a large literature, including Singer (1950), Sachs and Warner (1995) on the “natural-resource curse”, and Hausmann and others (2007) on the links between growth and product sophistication. In a similar vein, diversification spurts (defined as in Papageorgiou and Spatafora, 2012) are associated with sharp subsequent growth accelerations (defined analogously to diversification spurts). This is especially true for non-fragile frontier countries. Conversely, growth accelerations are associated with subsequent increases in diversification among non-fragile frontier countries. 6 Next, we examine the links between diversification and volatility. Does diversification serve as a buffer against external shocks? Related, are diversification spurts associated with increased macroeconomic stability? The existing literature provides some evidence that countries with more diversified production structures tend to have lower volatility of output, consumption, and investment (Mobarak, 2005; Moore and Walkes, 2010). Further, product diversification can increase the resilience of frontier economies to external shocks (Koren and Tenreyro, 2007). A key channel is that diversification involves frontier economies shifting resources from sectors where prices are highly volatile and correlated, such as mining and agriculture, to less volatile and correlated sectors, such as manufacturing, resulting in greater stability. And indeed, the data show clearly that output volatility diminishes after diversification spurts (Figure 2). IV. PATTERNS OF DIVERSIFICATION Having established that diversification is indeed linked with macroeconomic performance, we now examine in greater details patterns of diversification, with a focus on identifying which regions and countries have made greater progress in achieving diversification. Overall, higher income per capita and development are broadly associated with greater trade diversification (Figure 3), at least until an economy reaches advanced-economy status (with GDP per capita of $25,000–$30,000; see also Cadot and others, 2011). The relationship holds for the sample as a whole. It also holds between and within countries (that is, when the figure is restricted to show the pure cross-sectional or time-series variation); in the latter case, the data set’s extended time dimension is critical to confirming the relationship. At the regional level, Western Europe is the most diversified region. However, Emerging and Frontier Asia have been rapidly catching up (Figure 4). Asia in general shows higher, and more rapidly growing, diversification than Sub-Saharan Africa and MENA countries, although the progress has slowed down after 1995. Increases in diversification have largely occurred along the extensive margin, that is, through entry into completely new products, although there has also been some progress along the intensive margin for Emerging Asia (Figure 5). Also, changes in trade diversification over time have been paralleled by decreases in the relative 7 importance of agricultural exports, and increases in the relative importance of manufactured exports, especially for Asian countries (Figure 6). Higher income levels are also associated with increasing diversification across trade partners— at least until advanced-economy status is reached. After 1995, Asia greatly diversified its trade across partners (Figure 7). Frontier economies in general, including in Sub-Saharan Africa, also have made progress in diversifying their exports across partners. The trend is especially clear when considering the extensive margin, with a significant increase in exports to completely new partners. This is related to the ongoing process of globalization and a clear shift in trade away from the European Union and toward Asia, and China in particular (see also Samake and Yang, 2011). The data also reveal that, within developing countries, greater income per capita is also associated with greater real-sector diversification, that is, diversification in the broader domestic economy. During the 2000s, both across all developing countries and within Frontier Asia, analysis of six key sectors shows that there was significant real diversification. In particular, the share of agriculture in output declined significantly. The gap was filled largely by nontradables such as construction, wholesale trade, and transportation, rather than by manufacturing (Figure 8). That said, there is significant cross-country variation, both in the magnitude of the resource shift out of agriculture and in the precise identity of the sectors that have expanded in its place. V. PATTERNS OF QUALITY UPGRADING Economic development is underpinned not just by new products and markets, but also by quality improvements to existing products. Producing higher-quality varieties, through the use of more physical- and human-capital intensive production techniques, helps build on existing comparative advantages. It can boost countries’ productivity and export revenues.4 Ongoing work is helping to develop a toolkit to answer key questions, including calculating an 4 See Schott (2004) for an early demonstration that product quality varies significantly and systematically across exporters. 8 economy’s export quality and how it has evolved over time, determining the current potential for quality upgrading, and analyzing whether diversification into new products is a pre- requisite for further quality upgrading. One robust conclusion is that both Emerging Asia and, more recently, Frontier Asia have on average enjoyed remarkable success in quality upgrading. That said, there remains significant cross-country variation. Our quality measures are based on individual products’ unit values (that is, trade prices). However, these unit values are adjusted to reflect differences in production costs, as well as selection bias in the composition of international trade. Quality estimates at the country level are then constructed as a geometric value-weighted mean of the quality estimates for individual products. For full details, see Appendix I, as well as Henn, Papageorgiou, and Spatafora (forthcoming). Among other benefits, these quality measures smooth much of the artificial volatility often observed in unit values. The data suggest some clear patterns. Higher incomes per capita are associated with greater export quality at the country level. The relationship holds both across all goods (Figure 9), and (even more clearly) within manufacturing, which has greater scope for differentiation. Quality upgrading is particularly marked as countries evolve from frontier status into middle-income economies. There is much heterogeneity in quality levels, even when controlling for income per capita. In particular, Emerging Asia has enjoyed immense success in quality upgrading since 1970 (Figure 10), whereas Frontier Asia only began the process in the early 2000s. Sub-Saharan Africa stands out as producing relatively low-quality goods. Focusing on Asia, some countries have converged or are continuing to converge to the world frontier. In other cases, convergence seems to have slowed since the mid–1990s (Figure 11). Overall, improvements in export quality are associated with growth takeoffs. Hence, Japan converged to the world frontier in the 1970s; the Republic of Korea’s convergence occurred between the 1970s and the early 1990s; China started its take-off in the late 1980s, and has since been converging very rapidly; and Vietnam’s convergence started in the 1990s. In Malaysia and Thailand, convergence was rapid but appears to have stalled before reaching the 9 world frontier. India seems to be converging but only slowly. Likewise, Bangladesh’s convergence is very slow, particularly given its large catch-up potential. Crucially, developing countries’ potential for quality upgrading does not appear to be limited by low demand for quality in their existing destination markets. Frontier economies do tend to serve markets that import lower-quality products (Figure 12). However, the differences are not substantial enough to act as a constraint on quality upgrading. Indeed, on average, the lower- income the exporter, the greater the gap between its export quality and the average quality of its trade partners’ imports. Likewise, in slow-converging countries, export quality is substantially lower than the average quality of their trade partners’ imports. All this suggests that policy should focus on creating a domestic environment broadly conducive to quality upgrading; lowering barriers to entry into higher-quality export markets constitutes a less urgent priority. VI. COUNTRY CASE STUDIES To obtain robust policy conclusions, it is critical to complement the above cross-country analysis of product diversification and quality upgrading with individual country case studies. To this end, Appendix III provides a more detailed discussion of the experience of Bangladesh, a frontier economy with income per capita well below $1,000, and Vietnam, a country that has achieved middle-income status. In addition, Pitt et al. (forthcoming) analyze developments in Tanzania, another frontier economy; Angola, the second largest oil exporter in Sub-Saharan Africa and a middle-income country still facing significant physical and human capital needs; and Malaysia, an emerging market whose income per capita has grown 20-fold over the past 40 years. Overall, these case studies provide some tentative evidence in favor of four main themes. First, analyzing the entire structure of production paints a more comprehensive and illuminating picture than focusing purely on external trade. Structural transformation may well be associated with significant diversification of domestic production, including of nontradables. Examining this may shed light on the underlying mechanisms and barriers to further transformation. 10 Second, diversification and structural transformation are often underpinned by reforms and policy measures that are general in scope. Macroeconomic stabilization is a clear example. But even microeconomic measures are often broad-based, focusing on improving the quantity and quality of infrastructure or essential business services, or on setting up a welcoming environment for foreign investors. It remains an open issue to what extent industry-focused and narrowly targeted measures have historically helped underpin diversification efforts. Third, effective policy measures come in “waves” and aim at exploiting the evolving comparative advantages of the economy in changing external conditions. The types of reforms underpinning diversification and structural transformation in the early stages of development are different from those required later on and need to be adapted to the external environment faced by the economy. Finally, the frequency with which new products are introduced and the rate at which they grow can indicate potential policy-driven bottlenecks. Little entry may indicate that barriers deter firms from exporting or experimenting. If survival rates are low, firms may face more obstacles than expected. If surviving firms cannot expand, they may have inadequate access to finance. VII. CONCLUSIONS One key message from this paper and related work is that economic development critically involves diversification and structural transformation—that is, the continued, dynamic reallocation of resources from less productive to more productive sectors and activities. This process involves not just external trade, but the broader economy. Success in this transformation process will lead to lower volatility and higher growth. However, there are major differences across regions and countries in the degree to which they have succeeded in diversifying and transforming their economies. Over an extended period, Asia has on average been particularly successful in diversifying its exports, particularly in comparison with Sub-Saharan Africa. Much of the progress has occurred through diversification along the ‘extensive margin,’ that is, through entry into completely new products. 11 Structural transformation crucially involves changes not only in the type, but also in the quality of goods produced. Emerging Asia has on average benefited significantly from quality upgrading, helping it capitalize on already existing comparative advantages. Yet, the potential for quality upgrading varies by product. Agricultural and natural resources tend to have lower potential for quality upgrading than manufactures. Therefore, for frontier countries, diversification into products with longer “quality ladders” may be a necessary first step before large gains from quality improvement can be reaped. Overall, development strategies must promote sustained resource reallocation, and encourage continued quality upgrading. Ongoing work is focused on identifying the specific bottlenecks to structural transformation. In particular, it will analyze in greater detail measures of product quality, and examine what policies are needed to promote diversification and to sustain quality upgrading. That said, case studies of individual countries have already yielded some important lessons. For instance, diversification and structural transformation are often underpinned by reforms and policy measures that are general in scope, rather than industry-focused and narrowly targeted. In addition, the types of reforms underpinning diversification and structural transformation in the early stages of development are different from those required later on. 12 Figure 1. Growth and Export Diversification, 1962–2010 10 All Countries Asian Countries 8 CHN 6 Real GDPPC Growth KOR Real GDPPC Growth 5 HKG SGP THA MYSVNM 4 T AU P ES SV N P IDN LAO N L D FA R T I A GB R N H U LKA IND A U S JPN K DN S WE U DE E CZ V HR E C H AUS 0 PNG MNG 2 NZL PHL NPL KHMBGD 0 2 2.5 3 3.5 4 4.5 -5 Theil Index 1 2 3 4 5 6 Theil Index Frontier Asia Other APD Countries Sources: Penn World Table 7.0, UN COMTRADE, and author calculations. Note: GDPPC = Gross Domestic Product Per Capita. AZE = Azerbaijan, etc. APD = Asia & Pacific. Figure 2. Volatility and Export Diversification 4.5 4.4 4.3 Volatility of GDP Per Capita, 4.2 in percent 4.1 4.0 3.9 3.8 3.7 3.6 Pre-episode Episode Post-episode Sources: UN COMTRADE, and author calculations. Note: Episode indicates diversification spurts. The procedure for identifying spurts is based on Berg et al. (2012). 13 Figure 3. Export Diversification and Real GDP Per Capita 6 5 Theil Index 3 4 2 1 0 20000 40000 60000 Real GDP Per Capita All Countries Nonparametric Fit Quadratic Fit Sources: Penn World Table 7.0, UN COMTRADE, and author calculations. Note: Each observation denotes one country-year combination. Real GDP Per Capita bases on 2000 constant U.S. dollars. Figure 4. Export diversification by Region, 1960–2010: Extensive Margin 4 3 Theil Extensive 21 0 1960 1970 1980 1990 2000 2010 Year Frontier Asia Emerging Asia Sub-Saharan Africa Western Europe Latin America & Caribbean Middle East & North Africa Sources: UN COMTRADE, and author calculations. 14 Figure 5. Export diversification by Region and Period: Intensive Margin 4 3 Theil Intensive 2 1 0 Emerging Asia Frontier Asia Sub-Saharan Africa Western Europe 1965-1970 2006-2010 Sources: UN COMTRADE, and author calculations. Figure 6. Manufacturing Exports Share, by Region and Period .8 Share of Manufacturing Exports .2 .4 0 .6 Emerging Asia Frontier Asia Sub-Saharan Africa Western Europe 1965-1970 2006-2010 Sources: UN COMTRADE, and author calculations. 15 Figure 7. Trade Diversification across Partners over Time 4 Theil Index on Partners 2 1 0 3 Emerging Asia Frontier Asia Sub-Saharan Africa Western Europe 1965-1970 2006-2010 Sources: UN COMTRADE; Author calculations. Figure 8. Real Sector Share of Frontier Asia, 2000–10 Agriculture Mining Manufacturing .4 .2 Real Sector Share 0 Construction Wholesale trade Transport and communications .4 .2 0 2000 2005 2010 2000 2005 2010 2000 2005 2010 year Frontier Asia All Developing Countries Graphs by sectorcode Sources: IMF, and author calculations. 16 Figure 9. Quality Index and GDP Per Capita, 1960–2010 Quality across A all Exports:Frontier Asia = 1) = 1) 1 percentile .8 percentile Quality(90th(90th .2 .4 .6 0 0 10000 20000 ,                                   30000 ,                                   40000 ,                                  , Exporter GDP per capita (2000 constant Dollars) U. S. dollars) Rest of World Frontier Asia Lowess Fit Sources: Penn World Table 7.0, UN COMTRADE, and author calculations. Figure 10. Manufacturing Quality Index by Region, 1960–2010 Sources: Penn World Table 7.0, UN COMTRADE, and author calculations. 17 Figure 11. Quality Convergence of Asian Countries, 1960–2010 1 Fast-convergenceFrontier Fast-Convergence Asia Frontier Asia Slow-Convergence Frontier Slow-convergence Frontier Asia Asia 1 .9 Quality (90th percentile = 1) .9 Quality (90th percentile = 1) .8 .8 .7 .7 .6 .6 .5 .5 .4 .4 1970 1980 1990 2000 2010 Year 1960 1970 1980 1990 2000 2010 Year Bhutan Bangladesh Cambodia Laos Nepal Papua New Guinea Maldives Vietnam Mongolia Fast-Convergence Emerging Fast-covergence Emerging Asia Asia Slow-convergence Emerging Slow-Convergence Emerging Asia Asia 1 1 .9 Quality (90th percentile = 1) .9 Quality (90th percentile = 1) .8 .8 .7 .7 .6 .6 .5 .5 .4 1960 1970 1980 1990 2000 2010 .4 Year China India 1960 1970 1980 1990 2000 2010 Year Indonesia Malaysia Sri Lanka Thailand Fiji Philippines Sources: Penn World Table 7.0, UN COMTRADE, and author calculations. 18 Figure 12. Export Quality by Region, 2009 Export Quality Relative to Destination Markets (World Frontier = 1) Emerging Asia Frontier Asia Sub-Saharan Africa Western Europe 0.6 0.7 0.8 0.9 1 Average Quality Demanded in Destination Countries Quality Exported Sources: Penn World Table 7.0, UN COMTRADE, and author calculations. 19 Table 1. Growth Regressions with Diversification and Quality Indices Growth Regression, GLS fixed effects All Countries East Asia South Asia VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) Lagged GDP -5.363*** -6.027*** -5.898*** -5.897*** -5.604*** -4.886*** -3.882 -7.313*** (0.439) (0.464) (0.454) (0.471) (1.697) (1.551) (3.488) (2.173) Education 0.124*** 0.139*** 0.146*** 0.137*** 0.236** 0.217** 0.203 0.208* (0.023) (0.023) (0.023) (0.023) (0.101) (0.100) (0.125) (0.112) Investment 3.599*** 3.523*** 3.513*** 3.374*** 4.520*** 4.324*** 4.046** 4.436*** (0.433) (0.429) (0.429) (0.436) (1.428) (1.408) (1.626) (1.401) Population growth -0.053 -0.194 -0.238 -0.118 0.869 0.670 -1.346 -1.759 (0.229) (0.227) (0.228) (0.227) (0.738) (0.736) (2.749) (2.248) Diversification Index -0.608** (0.279) Quality upgrade index 8.761*** 13.660* (2.124) (6.878) Quality Index, Agriculture 9.687*** 21.036** -8.572 (2.348) (8.361) (15.365) Quality Index, Manufacture 7.646*** 48.638** (2.485) (18.846) Constant 35.550*** 31.748*** 29.948*** 31.842*** 16.662* 6.508 22.728 0.110 (3.720) (3.534) (3.614) (3.570) (9.267) (9.840) (19.191) (18.537) Observations 790 789 789 789 75 75 46 46 R-squared 0.234 0.250 0.250 0.241 0.291 0.317 0.295 0.402 Number of countrycode 113 113 113 113 10 10 6 6 Source: Author calculations. Note: Both country and time fixed effects are used in the estimations. For Asian country groups, the estimated coefficients on the diversification index are not statistically significant. Standard errors in parentheses. *** denotes p < 0.01, ** denotes p < 0.05, * denotes p < 0.1. 20 REFERENCES Asmundson, Irena. In preparation. “More World Trade Flows: An Updating Methodology”, IMF Working Paper. Berg, A., J.D. Ostry, and J. Zettelmeyer. 2012."What makes growth sustained?" Journal of Development Economics 98 (2): 149-166. Cadot, O., C. Carrere, and V. Strauss-Kahn. 2011. “Export Diversification: What’s Behind the Hump?” Review of Economics and Statistic, Vol. 93, No. 2, pp. 590–605. Hallak, J.C., 2006, “Product Quality and the Direction of Trade,” Journal of International Economics Vol. 68, pp 238–65. Harrigan, J., X. Ma, and V. Shlychkov. 2011. “Export Prices of U.S. Firms,” NBER Working Paper No. 17706. Hausmann, R., J. Hwang, and D. Rodrik. 2007. "What You Export Matters," Journal of Economic Growth, Springer 12 (1) (March): 1–25. Henn, C., C. Papageorgiou, and N. Spatafora. 2013. “Export Quality in Developing Countries”, IMF Working Paper, International Monetary Fund, Washington, DC. International Labour Organization (ILO). 2011. Yearbook of Labor Statistics. Geneva: ILO. Koren, M., and S. Tenreyro. 2007. “Volatility and Development,” Quarterly Journal of Economics 122: 243–87. Mobarak, A.M. 2005. “Democracy, Volatility, and Economic Development.” Review of Economics and Statistics 87: 348–61. Moore, W., and C. Walkes. 2010. “Does Industrial Concentration Impact on the Relationship between Policies and Volatility?” International Review of Applied Economics 24: 179–202. 21 Papageorgiou, C., and N. Spatafora. 2012. “Economic Diversification in LICs: Stylized Facts and Macroeconomic Implications.” IMF Staff Discussion Note 12/13, International Monetary Fund, Washington, DC. Pitt, A., T.S. Choi, N. Duma, N. Gigineishvili, and S. Rosa. In preparation. “Economic Diversification: Experiences and Policy Lessons from Five Case Studies.” IMF Working Paper, International Monetary Fund, Washington, DC. Sachs, J.D., and A.M. Warner. 1995. “Natural Resource Abundance and Economic Growth,” NBER Working Papers 5398, National Bureau of Economic Research, Cambridge, Mass. Samake, I., and Y. Yang, 2011, “Low-Income Countries’ BRIC Linkage: Are There Growth Spillovers?” IMF Working Paper 11/267, International Monetary Fund, Washington, DC. Schott, P., 2004, “Across-Product versus Within-Product Specialization in International Trade,” Quarterly Journal of Economics 119: 647–78. Singer, H., 1950, “US Foreign Investment in Underdeveloped Areas: The Distribution of Gains between Investing and Borrowing Countries.” American Economic Review 40: 473– 85. United Nations Industrial Development Organization (UNIDO), 2011, Industrial Statistics Database. Vienna: UNIDO. 22 APPENDIX I. DEFINITIONS OF MAIN INDICES A. Herfindahl Index As a starting point, we measure diversification using the Herfindahl index. The value of Herfindahl index, for any given country i and time period t, equals the sum of squares of export shares (in total exports), where the summation is across all goods j in the set Jit of categories which the country exports: HFIit = jJit (Xijt / kJit Xikt)2 where Xijt equals the value of exports by country i of good j at time t. This is an inverse measure of diversification which ranges from maximum of 1 (no diversification: all exports lie in a single category) down to 0 (full diversification: each category contains a negligible fraction of the country's exports). B. Theil Index We calculate the overall, within, and between Theil indices following the definitions and methods used in Cadot et al (2011). We first create dummy variables to define each product as “Traditional,” “New,” or “Non-traded.” Traditional products are goods that were exported at the beginning of the sample, and non-traded goods have zero exports for the entire sample. Thus, for each country and product, the dummy values for traditional and non-traded remain constant across all years of our sample. For each country/year/product group, products classified as “new” must have been non-traded in at least the two previous years and then exported in the two following years. Thus, the dummy values for new products may change over time. The overall Theil index is a sum of the within and between components. The between Theil index is calculated for each country/year pair as: TB = ∑k (Nk/N) (µk/µ) ln(µk/µ) where k represents each group (traditional, new, and non-traded), Nk is the total number of products exported in each group. µk/µ is the relative mean of exports in each group. 23 The within Theil index for each country/year pair is: TW = ∑k (Nk/N) (µk/µ) {(1/Nk) ∑i∈Ik (xi/µk) ln(xi/µk)} C. Product Quality Our methodology measures quality based on unit values, but with important adjustments for differences in production costs and for selection bias in the composition of international trade. Henn, Papageorgiou, and Spatafora (forthcoming) provide full details of the methodology. Briefly, we employ a modified version of Hallak (2006), which sidesteps data limitations to achieve maximum country and time coverage.5 As a first step, for any given product, the trade price (equivalently, unit value) pmxt is assumed to be determined by the following relationship: ln ln ln ln , (1) where the subscripts m, x, and t denote, respectively, importer, exporter, and time period. Prices reflect three factors. First, quality θmxt. Second, exporter income per capita yxt; this is meant to capture cross-country variations in production costs systematically related to income. With high-income countries typically being capital-abundant, we would expect 0 for capital- intensive sectors and 0 for labor-intensive sectors.6 Third, the (great circle) distance between importer and exporter, Distmx. This accounts for selection bias: typically, the composition of exports to more distant destinations is tilted towards higher-priced goods, because of higher shipping costs.7 5 The key difference is that we directly use unit values at the SITC 4-digit level, whereas Hallak gathers unit values at the 10-digit level and then normalizes them into a price index for each 2-digit “sector.” 6 This approach builds on Schott (2004), who showed that unit values for any given product vary systematically with exporter relative factor endowments, as proxied by GDP per capita. 7 Hallak (2006) uses distance to the United States instead of distance to the importer, because it only focuses on prices of exports to the United States. Harrigan, Ma, and Shlychkov (2011) find that the 24 Next, we specify a quality-augmented gravity equation. This equation is specified separately for each product, because preference for quality and trade costs may vary across products: ln ln ln (2) ImFE and ExFE denote, respectively, importer and exporter fixed effects. Distance is as defined above. The matrix is a set of standard trade determinants from the gravity literature.8 The exporter-specific quality parameter is , which enters interacted with the importer’s income per capita . If 0, then greater income increases the “demand for quality”. The estimation equation is obtained by substituting observables for the unobservable quality parameter in the gravity equation. Rearranging equation (1) for ln , and substituting into (2), yields: ln ln ln ln ln ln ln (3) where , , , and ′ ln . This equation is estimated separately for each of the 851 products in the dataset, yielding 851 sets of coefficients. We obtain estimates by two stage least squares. is a component of , so that the regressor ln ln is correlated with the disturbance term ′ . We therefore use ln ln as an instrument for ln ln . Where a unit value for the correlation between export prices and distance is due to a composition, or “Washington apples,” effect. They also find that U.S. firms charge higher prices to larger and richer markets. 8 It includes indicator variables for a common border, a common language, the existence of a preferential trade agreement, a colonial relationship, and a common colonizer. 25 preceding year is not available (for instance, because the good was not traded), we use the unit value in the closest available preceding year, going back up to 5 years.9 The regression results are used to calculate a comprehensive set of quality estimates. Rearranging (1) and using the estimated coefficients, quality is calculated as the unit value adjusted for differences in production costs and for the selection bias stemming from relative distance: ln ′ ln ′ ln ′ ln (4) As is standard, quality and importers’ taste for quality δ are not separately identified.10 The quality estimates are then aggregated into a multi-level database. The estimation yields quality estimates for more than 20 million product-exporter-importer-year combinations. To enable cross-product comparisons, all quality estimates are first normalized by their 90th percentile in the relevant product-year combination. The resulting quality values typically range between 0 and 1.2. The quality estimates are then aggregated, using current trade values as weights, to higher-level sectors (SITC 4-, 3-, 2-, and 1-digit, as well as country-level totals).11 At each aggregation step, the normalization to the 90th percentile is repeated. Aggregations are also produced based on the BEC classification, as well as for 3 broad sectors (agriculture, non-agricultural commodities, and manufactures). To allow for easy comparisons with unit values, the latter are also normalized with the 90th percentile set equal to unity. 9 If unit values are not available in any of the preceding five years, the observation is excluded from the estimation. 10 The preference for quality parameter δ will also vary by sector. Therefore, when we aggregate quality estimates across sectors, the aggregation will necessarily also aggregate across these heterogeneous preference for quality parameters. 11 Changes in the higher-level (including country-level) quality estimates in general reflect both quality changes within disaggregated sectors, and reallocation across sectors with different quality levels. If the composition of exports is shifting toward product lines characterized by low quality levels, it is quite possible for the quality of any given product to be rising sharply, but country-level quality to rise slowly (or indeed decline). 26 APPENDIX II. REGION DEFINITIONS Frontier Asia Emerging Asia Sub‐Saharan Africa Western Europe Bangladesh Brunei Darussalam Angola Austria Bhutan China Benin Belgium Cambodia Fiji Botswana Croatia Lao PDR India Burkina Faso Cyprus Maldives Indonesia Burundi Denmark Mongolia Malaysia Cameroon Finland Myanmar Marshall Islands Cape Verde France Nepal Micronesia Central African Republic Germany Papua New Guinea Philippines Chad Greece Vietnam Sri Lanka Comoros Iceland Thailand Congo, Dem. Rep. Ireland Tuvalu Congo, Rep. Israel Cote d'Ivoire Italy Equatorial Guinea Luxembourg Eritrea Malta Ethiopia(excludes Eritrea) Netherlands Gabon Norway Gambia, The Portugal Ghana Slovak Republic Guinea Slovenia Guinea-Bissau Spain Kenya Sweden Lesotho Switzerland Liberia United Kingdom Madagascar Malawi Mali Mauritania Mauritius Mozambique Namibia Niger Nigeria Rwanda Senegal Seychelles Sierra Leone South Africa Sudan Swaziland Sao Tome and Principe Tanzania Togo Uganda Zambia Zimbabwe Somalia 27 APPENDIX III. CASE STUDIES This section uses case studies to illustrate the lessons from structural transformation at different stages of development. Specifically, we focus on two countries: Bangladesh, with income per capita well below $1,000; and Vietnam, by now well on its way to emerging market status, and which is representative of countries that are successfully diversifying, or have successfully diversified, their economies. Bangladesh illustrates that initial diversification success, to be sustained, requires a combination of further reforms. Diversification in Bangladesh was largely triggered by external factors such as the introductions of the multi-fiber agreement (MFA) and the generalized system of preferences in the 1970s. These spurred development of the ready-made garments industry. As a result, Bangladesh shifted rapidly away from traditional agricultural and jute products towards manufacturing (Figure A1). Combined with the rise in output from wholesale and retail trade, this contributed to a steady increase in output diversification. Now, however, with ready-made garments accounting for 80 percent of total exports, Bangladesh’s output diversification has seemingly peaked, although as a low cost producer scope remains for further gains through increases in global garment market shares. Attempts to move beyond garments or to increase their quality have been hindered by a lack of supportive reforms. Challenges include poor governance and the high cost of doing business as a result of scarce electricity supplies, severe infrastructure bottlenecks, weak contract enforcement, and expensive credit provision. While such factors did not hinder diversification and inward FDI in the 1990s and early 2000s, they may now be preventing further progress. 28 Figure A1. Bangladesh: Concentration of Output and Composition of Exports Bangladesh Output Concentration Composition of Exports (In percent) (In percent) Sources: Country authorities and author calculations. In contrast, Vietnam’s experience shows that “waves” of supportive reforms can sustain diversification and structural transformation. The first wave of reforms during the 1980s opened new areas of activity to the private sector by reducing barriers to entry and expansion. Domestic prices, external trade and access to foreign exchange were liberalized; the rationing system largely abolished; subsidies significantly cut back; and inflation reduced. In agriculture, individual land-use rights were recognized, production freed from state-set quotas, and collective assets privatized. As a result, agriculture expanded, rising to almost half of total exports in 1995, and also diversified into cash crops, such as coffee and marine and forestry products (Figure A2). In a second wave of reforms, during the 1990s, liberalization of FDI helped develop other sectors. Initially, FDI was concentrated in the oil sector, but real estate (including hotels), food processing and heavy and light industry gained importance. FDI helped Vietnam integrate into emerging global supply chains, and gradually diversify its output and exports from textiles to footwear and electronics. This product diversification was accompanied by a diversification of trade partners, first from the Commonwealth of Independent States (CIS) to Asia, and then towards Europe and the United States 29 Figure A2. Vietnam: Diversification of Exports and Composition of GDP Vietnam: Export Concentration 1/ Vietnam: Composition of GDP 0.40 (Index) 50 (In percent of total) 16 Agriculture and fisheries Services 45 14 0.35 Light industries 40 Industry and  12 Total exports 35 construction 0.30 Agriculture , 30 10 forestry and  0.25 25 fisheries 8 20 6 0.20 GDP per capita  15 (in hundred $, RHS) 4 10 0.15 2 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 5 0 0 1/ Sum of squares of individual product shares.  A lower number  1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 indicates greater diversity. Sources: Country authorities, and author calculations. Diversification in frontier economies depends crucially on the frequency with which new products are introduced, the likelihood that they will survive, and their growth prospects. Initial trade diversification in frontier economies is mainly driven by entry into new products (the extensive margin). In the above two countries, over 1990–2011, there were significant differences (over time and across countries) in three key measures of the extensive margin: (i) the number of new product varieties introduced in a given year,12 (ii) the survival rates of new varieties, (iii) and the growth rates of surviving varieties. Over time, such differences can cumulate into large differences in overall exports. Differences in these measures underline the case studies’ different experiences. Vietnam showed significant new entry and reductions over time in the relative importance of incumbent varieties (Figure A3). Vietnam in particular stood out as having a high probability of survival of new varieties. Bangladesh had less experimentation and also less growth in surviving varieties, accounting for its current, unusually high concentration. 12 Here, a variety is defined as a specific product exported to a specific country as in Asmundson (forthcoming). 30 Figure A3. Export Experimentation Growth in Varieties Exported Share of Export Value from (1990 = 1) Incumbent Varieties 20 1.0 18 0.9 16 0.8 14 0.7 BGD 12 0.6 10 0.5 8 VNM 0.4 BGD 6 0.3 4 0.2 VNM 2 0.1 0 0.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Sources: UN COMTRADE, and author calculations. Overall, these case studies provide some tentative evidence in favor of four main themes. First, analyzing the entire structure of production paints a more comprehensive and illuminating picture than focusing purely on external trade. Structural transformation may well be associated with significant diversification of domestic production, including of non-tradable. Analyzing this may shed light on the underlying mechanisms and barriers to further transformation. Second, diversification and structural transformation are often underpinned by reforms and policy measures that are general in scope. Macroeconomic stabilization is a clear example. But even microeconomic measures are often broad-based, focusing on improving the quantity and quality of infrastructure or essential business services, or on setting up a welcoming environment for foreign investors. It remains an open issue to what extent industry-focused and narrowly targeted measures have historically helped underpin diversification efforts. Third, effective policy measures come in “waves” and aim at exploiting the evolving comparative advantages of the economy in changing external conditions. The types of reforms underpinning diversification and structural transformation in the early stages of development are different from those required later on and need to be adapted to the external environment faced by the economy. 31 Finally, the frequency with which new products are introduced, and the rate at which they grow, can indicate potential policy-driven bottlenecks. Little entry may indicate that barriers deter firms from exporting or experimenting. If survival rates are low, firms may face more obstacles than expected. If surviving firms cannot expand, they may have inadequate access to finance. This type of analysis suggests directions for further study.