Research & Policy Briefs From the World Bank Malaysia Hub No. 24, July 2019 Structural Change Rediscovered: The Role of Human and Physical Capital Pedro M. G. Martins Changes in economic structures are an essential feature of development. But how are these brought about? A growing body of evidence underscores the transformative role that human and physical capital can have in enhancing and sustaining economic growth. Investments in education, health, and economic infrastructure need to be at the core of national policy agendas. Introduction productivity growth. In this regard, industrial upgrading (such as through improvements in factor endowments) plays a key role in facilitating structural It has long been established that structural change is a key characteristic and change; see Lin (2011) and Ju, Lin, and Wang (2015). driver of economic and social development (Kuznets 1966). However, the concept is not well understood and is notably under-researched. Structural Structural change is important not only for boosting productivity and change emphasizes the central role of productivity in the development average incomes, but can also ensure that economic gains are equitably process from a specific viewpoint: while productivity improvements within distributed across society—because workers in a “lagging” sector are unlikely economic sectors are crucial to support economic growth, changes in to experience significant increases in living standards. This insight is economic structures are essential to ensure that those efforts can be particularly important given concerns that recent economic growth patterns enhanced and sustained. Much of the research to date has focused on have not been inclusive or sustainable. For example, Loayza and Raddatz assessing the pace and pattern of structural change across countries and (2010) find evidence that the composition of economic growth matters for regions. A few studies have attempted to uncover an even more pertinent gap poverty reduction, with strong contributions from unskilled labor-intensive in knowledge about the process: what are the key determinants of structural sectors, such as agriculture, construction, and manufacturing. In sum, change and how can they be enhanced? empirical studies focusing on structural change can provide valuable complementary insights to the mainstream economic growth literature. Neoclassical and Structuralist Perspectives on Economic Growth Rediscovering Structural Change: Concepts and Methodologies The literature on economic growth has largely relied on theoretical models underpinned by an aggregate production function, such as Solow's The early literature on structural change dates to the 1950s and 1960s. neoclassical growth model. These one-sector models emphasize the role of Kuznets (1957), Chenery (1960), and Chenery and Taylor (1968) uncover economy-wide factor accumulation and productivity—as they imply strong important stylized facts on the relationship between a country's economic homogeneity across economic sectors. They have provided a theoretical structure—namely, production—and its income level. This literature posits foundation for countless empirical studies investigating the determinants of that structural change is a key feature and driver of economic development. economic growth through econometric methods and growth accounting In fact, the historical experience of developed and emerging economies frameworks. In particular, the seminal work of Barro (1991) on cross-country confirms that sustained economic development requires structural change. growth regressions opened a vast and prolific field of empirical research. Although there is no universally agreed definition of structural change, it However, several studies have shown that the empirical results tend to be is possible to categorize existing perspectives into three broad groups: (1) a sensitive to model specification, sample data, and estimation method; for very narrow (production) focus; (2) a narrow (productivity) focus; and (3) a example, see Levine and Renelt (1992) and Pritchett and Summers (2014). broad (socioeconomic) focus. The first group assesses structural change This lack of robustness might be partly due to one-sector models not merely in terms of shifts in the structure of output (see, for example, accounting for the large heterogeneity across sectors that is characteristic of Dabla-Norris et al. 2013). Structural change happens when the economy shifts developing economies. The aggregation of heterogeneous sectors in toward the production of goods and services associated with higher value cross-country growth regressions can have a considerable impact on added, which in turn stimulates economic growth. This usually entails a inference, as Eberhardt and Teal (2013) demonstrate. relative decline in agriculture and a concomitant increase in the share of These critiques have contributed to a renewed interest in dual-economy industry and/or services. It is implicitly assumed that the market will models and the role of structural change in the growth process (McMillan and automatically and efficiently facilitate any required reallocation of resources Heady 2014). While one-sector growth models were originally conceived with across sectors, including capital, labor, and land. developed economies in mind, it can be argued that structural (dual-sector) The second group evaluates structural change in terms of labor shifts models provide a better representation of developing economies. Temple from lower-productivity sectors to higher-productivity sectors (see, for (2005), for instance, asserts that dual-economy models should take center example, McMillan, Rodrik, and Verduzco-Gallo 2014). This reallocation of stage in the analysis of economic growth in developing countries. Recent labor raises workers’ productivity, which helps accelerate aggregate examples include studies by Herrendorf, Rogerson, and Valentinyi (2014), who productivity and thus economic growth. This dynamic is typically assessed by develop a multisector extension of the one-sector growth model that is decomposing output per worker growth into “between-sector” effects (a consistent with the stylized facts of structural change; Duarte and Restuccia proxy for structural change) and “within-sector” effects (labor productivity (2010), who use a general equilibrium model to investigate the role of sectoral improvements within a specific sector). labor productivity in explaining structural change; and Temple and Wößmann (2006), who develop empirical growth models suitable for dual economies. While the same sectoral patterns are expected, the explicit focus is on labor productivity rather than production alone. This emphasis on labor These models assume the coexistence of a relatively “advanced” sector productivity stems from the observation that changes in the structure of and a “lagging” sector in the economy, such as modern versus traditional, employment often lag shifts in production. The reallocation of labor (toward industry versus agriculture, capitalist versus subsistence, or formal versus more productive sectors) is necessary for structural change to happen, but informal (Fields 2007). More importantly, they acknowledge that productivity resource reallocations may not always induce structural change—especially if gaps across sectors are an important source of economic growth through the they occur within sectors or firms. Within-sector improvements are often reallocation of factors of production (Lewis 1954). These gaps can be seen as achieved through enhanced skills, complementary capital, improved allocative inefficiencies and thus opportunities to catalyze growth. technology, better management practices, and resource reallocations. The reallocation of labor across sectors assumes particular importance in The third group goes beyond changes in the economic structure—such as these models. Nonetheless, these employment shifts ought to be production and employment—by also considering changes in other aspects of complemented by greater sectoral dynamism to ensure that sectors with society (see, for example, Martins 2019). For instance, structural change may above-average productivity can absorb workers without sacrificing entail a demographic transition (through lower fertility rates), changes in Affiliation: The World Bank. E-mail address: pmartins@worldbank.org Acknowledgement: Norman Loayza provided useful comments and suggestions to this brief. Nancy Morrison provided editorial assistance. Objective and disclaimer: Research & Policy Briefs synthesize existing research and data to shed light on a useful and interesting question for policy debate. Research & Policy Briefs carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions are entirely those of the authors. They do not necessarily represent the views of the World Bank Group, its Executive Directors, or the governments they represent. Structural Change Rediscovered: The Role of Human and Physical Capital Figure 1. Structure of Output and Employment Value Added (%), 2013 Employment (%), 2013 Agriculture Mining & Utilities Manufacturing Agriculture Mining & Utilities Manufacturing Construction Commerce Transport Construction Commerce Transport Other services Other services 0 20 40 60 80 100 0 20 40 60 80 100 Africa Africa Asia Asia LAC LAC Other Other Source: Martins (2019). Note: Economic sectors correspond to the following ISIC revision 3.1 sections (in parentheses): agriculture & fishing (A and B); mining & utilities (C and E); manufacturing (D); construction (F); commerce (G and H); transport (I); and other services (J–P). LAC = Latin America and the Caribbean. labor participation (through changing social preferences), and a spatial even below) the aggregate average, which limits the impact of labor reorganization of the population (through rural-urban migration). Some of reallocations. Since the highest-productivity sectors tend to be capital these additional insights can be provided by decomposing output per capita intensive and are thus less able to absorb large numbers of workers, the key growth into three components: changes in output per worker (that is, labor for accelerating structural change might be in increasing the dynamism of productivity); changes in the employment rate; and changes in the relative manufacturing, commerce, and other services. size of the working-age population (that is, demographic change). The first component can then be further decomposed into the traditional Pace of Structural Change within-sector and between-sector productivity effects. A stepwise Shapley decomposition approach can be employed to calculate the proportion of Most empirical studies seeking to measure the pace of structural change output per capita growth that can be attributed to each of these components. decompose output per worker growth into within-sector and between-sector For further technical details, see World Bank (2012); for a practical effects; examples include McMillan, Rodrik, and Verduzco-Gallo (2014); application, see Martins (2019). McMillan and Harttgen (2014); Timmer, de Vries, and de Vries (2014); UNCTAD (2014); and Roncolato and Kucera (2014). Martins (2019) provides The recent emphasis on structural change has led to a rapidly expanding additional insights on demographic change and employment rates, as well as body of theoretical and empirical work. Datasets have been compiled to a subregional perspective. document regional patterns, with varying degrees of sectoral disaggregation and country coverage. However, most studies have small country samples and Overall, the evidence suggests that within-sector productivity effects are there have been very few attempts to empirically assess the determinants of the strongest driver of aggregate labor productivity growth. Nonetheless, the structural change in developing countries. Martins (2019) attempts to bridge contribution of structural change is often sizeable and growing in importance. this gap by constructing a comprehensive dataset comprising 169 countries This is particularly noticeable since the early 2000s, when economic and conducting an original empirical exercise on the determinants of performance improved in many developing countries. For instance, Martins structural change. (2019) finds that within-sector productivity improvements played a major role in accelerating output per capita growth in most of the 13 subregions The Pace and Pattern of Structural Change spanning the developing world that were analyzed—and was consistently larger than between-sector effects in all subregions but one (Figure 3). Economic Structure and Labor Productivity However, structural change was also a key contributor to the improved Economic structures vary considerably across regions (Figure 1). In 2013, economic performance in several subregions, especially in Eastern and manufacturing contributed 26 percent of total gross value added (GVA) in Western Africa, as well as Central, Eastern, and Southern Asia. Some studies Asia, but only 11 percent in Africa. Other services accounted for 52 percent of report negative structural change for some regions—that is, labor total GVA in developed countries, but less than 30 percent in Africa and Asia. reallocations toward a lower-productivity sector—but mostly in earlier The disparities are even starker regarding employment. Agriculture employed periods. For example, McMillan, Rodrik, and Verduzco-Gallo (2014) find more than half of Africa's workers, but accounted for less than 5 percent of growth-reducing structural change in Africa and Latin America during the total employment in developed countries. Economic structures also vary 1990–2005 period; McMillan and Harttgen (2014) suggest the same for Latin substantially within regions. As noted in the early literature on structural America in 2000–2005; as do Timmer, de Vries, and de Vries (2014) for Latin change, differences in economic structure are partly responsible for the large America in 1990–2010. A range of factors might explain these discrepancies, income gaps observed across countries. such as differences in country samples, time frames, level of sectoral aggregation, data sources, and empirical methodologies. For instance, The concept of structural change is intrinsically linked to labor McMillan, Rodrik, and Verduzco-Gallo (2014) include only nine countries from productivity, which is usually measured by GVA per worker. Agriculture is Africa and nine countries from Latin America when reporting negative often found to be the sector with the lowest labor productivity level, while structural change in these regions. labor productivity gaps appear to be negatively related to income levels—as larger gaps are found in poorer regions (such as Africa). The share of Changes in demographic structures were found to have a positive effect employment in agriculture declined in all regions between 1991 and on per capita growth, while the impact of employment rates tends to vary 2013—especially in Asia, where it dropped by 21 percentage points (Figure 2). considerably (Martins 2019). However, these two components play a Ideally, agricultural labor should move to sectors that have above-average relatively minor role when compared to productivity growth. Demographic (and growing) levels of labor productivity. Hence, sectors with rising trends were particularly important in Asian subregions, while employment employment shares should have “relative labor productivity” above zero. rates were relatively more important in Latin America. The positive impact of Between 1991 and 2013, labor shifted mainly toward other services— except demographic change is unlikely to be automatic because investments in skills for Asia, where employment moved toward construction, commerce, and also and adequate job creation are required to seize a demographic dividend. The other services. Regions also shed a considerable amount of manufacturing impact of employment rates varies considerably across countries, partly jobs, especially developed countries. Overall, the sectors where employment because declines may suggest an investment in future generations, as young 2 is expanding the most have labor productivity levels that are similar to (or are people stay longer in education. Research & Policy Brief No.24 Figure 2. Changes in Employment and Labor Productivity Gaps Employment shares (pp change), 1991-2013 Relative labour productivity, 2013 Agriculture Mining & Utilities Manufacturing Agriculture Mining & Utilities Manufacturing Construction Commerce Transport Construction Commerce Transport Other services Other services -20 -10 0 10 20 -2 -0 -2 -4 -6 Africa Africa Asia Asia LAC LAC Other Other Source: Martins (2019). Note: Relative labor productivity is calculated as the natural logarithm of the ratio of sectoral productivity to aggregate productivity. Large productivity gaps are represented by wider bar areas. If the width of a bar measures 1 unit, then the sector's productivity is 10 times higher than the average (if positive) or one-tenth of the average (if negative). “pp” stands for percentage point. Patterns of Structural Change entrepreneurship, creativity, and dynamism—thus affecting both labor supply and demand. Health outcomes can also be important, since good In terms of sectoral dynamics, services are often identified as the main driver physical health and cognitive functions are key for workers to seize better of economic performance and the key catalyst for structural job opportunities. Physical capital can also be critical to enhance structural change—especially as labor typically moves from agriculture to services. change, since improving basic infrastructure—such as energy, water and However, the services sector is highly heterogeneous; it comprises both sanitation, transport, and telecommunications—can significantly enhance dynamic modern activities and informal activities with limited productivity a country's competitiveness. Finally, good governance and strong growth potential. Although many studies provide some level of institutions can provide a more conducive environment for accelerating disaggregation—sometimes four or five services subsectors—further scrutiny structural change. is needed. Martins (2019) finds that the share of employment in agriculture has dropped in all subregions, while the largest relative increases in There are few empirical studies on the determinants of structural change, employment has occurred in services. Because agriculture generally has the and even fewer using estimates on the labor reallocation effect as a proxy for lowest level of labor productivity, the reallocation of workers from agriculture structural change. McMillan, Rodrik, and Verduzco-Gallo (2014) do this for a to other sectors leads to positive structural change—which helps boost relatively small cross-sectional dataset that can explain differences across aggregate productivity and thus economic growth. In fact, there is a clear countries. Their findings suggest that a higher share of employment in negative relationship between agricultural employment and average incomes, agriculture, a lower share of raw materials in total exports, an undervalued both within and across regions. It also seems that the faster labor moves out exchange rate, and greater labor market flexibility all contribute to of agriculture, the larger is the increase in output per capita, suggesting that growth-enhancing structural change. accelerated economic development depends on the rate at which production resources are reallocated to more efficient uses. Despite this, raising Martins (2019) uses a large panel dataset that can also explain agricultural productivity remains crucial for eradicating poverty—through accelerations in the pace of structural change—that is, changes within structural change within the sector (see Divanbeigi, Paustian, and Loayza countries. The study uses principal components analysis (PCA) to isolate 2016). Moreover, manufacturing is found to have had a limited impact in common elements in highly correlated determinants and a panel fixed-effects enhancing structural change, even though it has provided strong estimator to improve robustness. The findings provide strong evidence of the contributions to within-sector productivity growth—especially in Asia. This is key role of human and physical capital—namely, education and partly because employment in manufacturing has been declining in relative infrastructure—in enhancing structural change. As expected, the initial share terms. The sector could play a more important role if employment and labor of employment in agriculture is positively related to structural change, given productivity are simultaneously increased. the greater scope to benefit from productivity gaps. In addition, regional estimates provide further insights. For instance, the Drivers of Change: Human and Physical Capital expansion of secondary education appears to be relevant in Africa, while The pace of structural change can be influenced by several plausible tertiary education is pertinent in Asia and Latin America, which is not factors—many of which emerging from the theoretical and empirical surprising given current levels of educational achievement. The share of literature mentioned earlier. Among these are countries’ initial conditions. mining in total output has a negative impact on Africa’s structural change. For instance, given the large productivity gaps usually observed between Moreover, the exchange rate seems to matter in Latin America and the agriculture and other economic sectors, countries with a high share of political regime in developed countries. Dabla-Norris et al. (2013) investigate employment in agriculture would have greater scope to benefit from the determinants of sectoral output shares in agriculture, manufacturing, and employment reallocations. Moreover, resource-rich countries may have services, rather than structural change as measured by the employment limited incentives to diversify their economic structures, especially when reallocation effect. However, the authors also conclude that human and high demand and prices for natural resources reinforce their comparative physical capital are important for structural change, in addition to a set of advantage and specialization. Macroeconomic instability—measured initial conditions such as natural resource dominance. through fiscal deficits, public debt, inflation, and current account deficits— Overall, these findings complement the existing evidence on the broader may undermine structural change through greater economic uncertainty. link between productivity and economic growth, such as Kim, Loayza, and Openness to international trade and a competitive real exchange rate may Meza-Cuadra (2016). Loayza (2016) also shows that capital accumulation facilitate structural change if they lead to output and employment growth contributes to structural change—in this case, defined as a shift from in higher-productivity sectors. High real interest rates undermine credit informality to formality. expansion to the private sector, which may in turn restrict production and employment growth in dynamic sectors. Given that employment plays a Policy Implications central role in inducing structural change, human capital is likely to be of vital importance. Workers need improved skills to gain access to McMillan, Rodrik, and Sepúlveda (2017) posit that neoclassical and higher-productivity jobs, while skills and knowledge are also key to promote dual-economy models offer complementary perspectives on economic 3 Structural Change Rediscovered: The Role of Human and Physical Capital Figure 3. Contribution to Output per capita Growth Contribution to GVA per capita growth (%, 1991-2002) Contribution to GVA per capita growth (%, 2002-2013) Within-sector Between-sector Employment Within-sector Between-sector Employment Demography Demography -2.0 0.0 2.0 4.0 6.0 8.0 -2.0 0.0 2.0 4.0 6.0 8.0 Eastern Africa Eastern Africa Middle Africa Middle Africa Northern Africa Northern Africa Southern Africa Southern Africa Western Africa Western Africa Central Asia Central Asia Eastern Asia Eastern Asia South-Eastern Asia South-Eastern Asia Southern Asia Southern Asia Western Asia Western Asia Caribbean Caribbean Central America Central America South America South America Source: Martins 2019. Note: Results from Shapley decompositions. Countries are grouped according to the United Nations (UN) regional classification. growth: with neoclassical models explaining the growth process within History suggests that governments have played an important role in modern sectors, mainly through broad-based physical and human capital facilitating structural change, and analytical frameworks have been accumulation, and dual-economy models explaining relationships and flows developed to identify sectors with latent comparative advantage (Lin and across sectors, through policies that ensure that resources flow to modern Monga 2010). Whether broad-based improvements in human and physical higher-productivity activities. McMillan, Rodrik, and Sepúlveda (2017) also capitals are less (or more) important than targeted measures is difficult to claim that it is possible to have rapid structural change without significant assess. Sector-specific policies are not easily captured, and even if they were, improvements in the “fundamentals”—defined as infrastructure, education, the sectors being targeted often vary from country to country. It is plausible and institutions. that the unexplained variation in structural change within countries could be accounted by (unobserved) sector-specific interventions. However, there is growing evidence to suggest that physical and human In sum, there is still much scope for accelerating structural change. Labor capital also play an important part in promoting structural change. Stronger productivity gaps and employment shares in agriculture remain high in investments in education, health, and economic infrastructure can thus several parts of the world. While the past two decades have been facilitate (growth-enhancing) labor reallocations, which is key to long-term unquestionably positive for developing countries, it is vital to improve the economic and social progress. Overall, this supports the growing emphasis pace of structural change to fully seize its benefits. The key message that is being placed on human capital, such as through the recently launched emerging from recent research is that investments in education and Human Capital Project (HCP). Announced in 2017 by the World Bank Group, economic infrastructure are critical to accelerating structural change—and the HCP is a program of advocacy, measurement, and analytical work to raise thus aggregate productivity. Hence, the policy recommendations arising public awareness and accelerate (more and better) investments in people for from both neoclassical and dual-economy models may not be as different as greater equity and economic growth. one may expect. References Barro, R. 1991. “Economic Growth in a Cross-section of Countries.” Quarterly Journal of Economics 106 Loayza, N. 2016. “Informality in the Process of Development and Growth.” The World Economy 39 (12): (2): 407–43. 1856–1916. Chenery, H. 1960. “Patterns of Industrial Growth.” American Economic Review 50 (3): 624–53. Loayza, N., and C. Raddatz. 2010. “The Composition of Growth Matters for Poverty Alleviation.” Journal Chenery, H., and L. Taylor. 1968. “Development Patterns: Among Countries and over Time.” Review of of Development Economics 93 (1): 137–51. Economics and Statistics 50 (4): 391–416. Martins, P. 2019. “Structural Change: Pace, Patterns and Determinants.” Review of Development Dabla-Norris, E., A. Thomas, R. Garcia-Verdu, and Y. Chen. 2013. “Benchmarking Structural Transformation Economics 23 (1): 1–32. across the World.” IMF Working Paper 13/176, International Monetary Fund, Washington, DC. McMillan, M., and D. Headey. 2014. “Introduction–Understanding Structural Transformation in Africa.” Divanbeigi, R., N. Paustian, and N. Loayza. 2016. “Structural Transformation of the Agricultural Sector: A World Development 63 (November): 1–10. Primer.” Research & Policy Brief 2, World Bank. McMillan, M., and K. Harttgen. 2 014. “What I s D riving t he African Growth Miracle?” N BER Working Duarte, M., and D. Restuccia. 2010. “The Role of the Structural Transformation in Aggregate Productivity.” Paper 20077, National Bureau of Economic Research, Cambridge, MA. Quarterly Journal of Economics 125 (1): 129–73. McMillan, M., D. Rodrik, and C. Sepúlveda. 2017. “Structural Change, Fundamentals, and Growth: A Eberhardt, M., and F. Teal. 2013. “Structural Change and Cross-country Growth Empirics.” World Bank Framework and Case Studies.” NBER Working Paper 23378, National Bureau of Economic Economic Review 27 (2): 229–71. Research, Cambridge, MA. Fields, G. 2007. “Dual Economy.” ILR School Working Paper 17, School of Industrial and Labor Relations, Cornell University, Ithaca, NY. McMillan, M., D. Rodrik, and I. Verduzco-Gallo. 2014. “Globalization, Structural Change, and Productivity Growth, with an Update on Africa.” World Development 63 (2014): 11–32. Herrendorf, B., R. Rogerson, and A. Valentinyi. 2014. “Growth and Structural Transformation.” In Handbook of Economic Growth, Vol. 2, edited by P. Aghion and S. Durlauf, 855–941. Amsterdam: North-Holland. Pritchett, L., and L. Summers. 2014. “Asiaphoria Meets Regression to the Mean.” NBER Working Paper 20573, National Bureau of Economic Research, Cambridge, MA. Ju, J., J. Lin, and Y. Wang. 2015. “Endowment Structures, Industrial Dynamics, and Economic Growth.” Journal of Monetary Economics 76 (November): 244–63. Roncolato, L., and D. Kucera. 2014. “Structural Drivers of Productivity and Employment Growth: A Kim, Y., N. Loayza, and C. Meza-Cuadra. 2016. “Productivity as the Key to Economic Growth and Decomposition Analysis for 81 countries.” Cambridge Journal of Economics 38 (2): 399–424. Development.” Research & Policy Brief 3, World Bank. Temple, J. 2005. “Dual Economy Models: A Primer for Growth Economists.” The Manchester School 73 Kuznets, S. 1957. “Quantitative Aspects of the Economic Growth of Nations II. Industrial Distribution of (4): 435–78. National Product and Labor Force.” Economic Development and Cultural Change 5 (4, Supplement): 1–111. Temple, J., and L. Wößmann. 2006. “Dualism and Cross-country Growth Regressions.” Journal of --------. 1966. Modern Economic Growth. New Haven, CT: Yale University Press. Economic Growth 11 (3): 187–228. Levine, R., and D. Renelt. 1992. “A Sensitivity Analysis of Cross-Country Growth Regressions.” American Timmer, M., G. de Vries, and K. de Vries. 2014. “Patterns of Structural Change in Developing Countries.” Economic Review 82 (4): 942–63. GGDC Research Memorandum 149, Groningen Growth and Development Centre. Lewis, A. 1954. “Economic Development with Unlimited Supplies of Labour.” The Manchester School 28 UNCTAD (United Nations Conference on Trade and Development). 2014. The Least Developed Countries (2): 139–91. Report 2014–Growth with Structural Transformation: A Post- 2015 Development Agenda. New Lin, J. 2011. “New Structural Economics: A Framework for Rethinking Development.” The World Bank York and Geneva: United Nations. Research Observer 26 (2): 193–221. World Bank. 2012. Job Generation and Growth Decomposition Tool: Understanding the Sectoral Pattern Lin, J., and C. Monga. 2010. “Growth Identification and Facilitation: The Role of the State in the Dynamics of of Growth and its Employment and Productivity Intensity. Reference Manual and User's Guide, Structural Change.” World Bank Policy Research Working Paper 5313, The World Bank, Washington, DC. Version 1.0. Washington, DC: World Bank Group. 4