WPS8336 Policy Research Working Paper 8336 Structural Change in West Africa A Tale of Gain and Loss Fiseha Haile Macroeconomics, Trade and Investment Global Practice Group February 2018 Policy Research Working Paper 8336 Abstract Economic growth in Benin, Burkina Faso, and Côte d’Ivo- loss in Benin. However, structural change made a smaller ire occurred in tandem with a rapid exodus of labor out contribution in Côte d’Ivoire. Within-sector productivity of agriculture. This paper investigates the contribution (or loss generally held back growth. The pattern of structural lack thereof ) of within- and between-sector productivity change observed in Benin, Burkina Faso, and Côte d’Ivo- changes to overall productivity growth and output per ire starkly contrasts with that of Asia, where within-sector capita growth since 2000. Productivity growth was rela- productivity gains were preponderant and dynamic struc- tively significant in Burkina Faso, modest in Benin, and tural change was the norm rather than the exception. The in the negative territory in Côte d’Ivoire. The results show bulk of Benin, Burkina Faso, and Côte d’Ivoire’s displaced that static structural change drove growth in Burkina Faso agricultural workers moved into still-low productiv- and Benin, although it was partly offset by a dynamic ity service activities, as is typical of the African sample. This paper is a product of the Macroeconomics, Trade and Investment Global Practice Group. 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 author may be contacted at fgebregziabher@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 Structural Change in West Africa: A Tale of Gain and Loss‡ Fiseha Haile The World Bank Group JEL: O11, O47, O55. Keywords: Structural change, labor productivity, growth, Benin, Burkina Faso, Côte d’Ivoire ‡This paper has been prepared under the overall guidance of Lars Christian Moller (Practice Manager, World Bank) and Jacques Morrisset (Lead Economist and Program Leader, World Bank). Thanks are due to Michael Weber (Senior Economist, World Bank) for kindly providing survey data on employment for Burkina Faso. The author would also like to thank Patrick Premand (World Bank) for kindly sharing the labor market data used for the World Bank (2017) report “Towards Better Employment and Productive Inclusion of the Poor in Côte d’Ivoire.” David Cal MacWilliam, Boulel Toure, Samba Ba, and Marim Diop kindly provided assistance with data collection. Useful comments were received from Jacques Morrisset. The views expressed in this paper are solely those of the author and do not represent the opinions of the Word Bank Group.     1. Introduction Despite wide-ranging external and domestic headwinds, Benin, Burkina Faso, and, more recently, Côte d’Ivoire have registered respectable economic growth. Specifically, Benin’s real GDP growth averaged 4 percent per annum during 2005-2016, which recently became more robust and more resilient to shocks. Similarly, Burkina Faso has posted sustained higher and more stable growth since the CFA franc devaluation in 1994, which held steady at about 6 percent. By contrast, Côte d’Ivoire fared much worse due to a protracted civil conflict that lasted from 2002 to 2011, with growth hitting rock bottom at about 0.4 percent per annum (-1.9 percent in per capita terms). However, economic growth rebounded sharply with the return of political stability and the ensuing policies that bolstered economic recovery, averaging 9 percent in 2012-2016, although the economy has yet to recover ground lost during the crisis.2 However, the three West African economies face substantial development challenges. Poverty is deeply entrenched and much remains to be done to accelerate growth and put a significant dent in poverty. The ($1.25 per person per day) poverty rate in Benin, Burkina, and Côte d’Ivoire (hereafter jointly referred to as BBC), albeit receding, remains stubborn at about 51 percent, 44 percent, and 35 percent,3 respectively. This partly reflects that marginally productive activities have propelled much of the economic growth in BBC. Over 80 percent of the labor force in Burkina Faso and Côte d’Ivoire, and over 90 percent of that in Benin remain stranded in the informal sector, which is characterized by the conspicuous absence of productive and ‘decent’ wage employment.4 There is thus a pressing need to create millions of productive and better-paying jobs for the youth as well as for the surplus labor from agriculture. However, this would be difficult, if not impossible, without a structural change that lifts workers out of low-productivity agriculture and informal activities to high-productivity industries in the modern, non-agricultural parts of the economy (McMillan et al., 2014). In addition, despite agriculture’s lingering dominance, burgeoning informal services constitute the hallmark of these economies. Benin has experienced phenomenal change in its structure of employment, while Burkina Faso and Côte d’Ivoire witnessed a relatively smaller shift. Agriculture’s share of total employment plunged by 18 percent in Benin (between 2006 and 2015), by 9 percent in Burkina Faso (1998 to 2014), and by 10 percent in Côte d’Ivoire (2002 to 2014). The majority of the labor force shed by agriculture, however, ended up in low-productivity and largely informal trade and distribution services rather than formal manufacturing and modern service industries. In fact, the employment share of manufacturing has barely changed for decades, while its output share declined. This may not bode well for the sustainability of growth as manufacturing has the potential to absorb the bulk of an economy’s low-skilled workforce at a large productivity premium (Rodrik, 2013a, 2014).5 In addition, manufacturing is a powerful source of learning by doing (Agénor and Dinh, 2013) and remains the centerpiece of catch-up development (Jones and Olken, 2008). Large inter-sectoral productivity gaps, however, indicate the presence of enormous potential for growth via structural change. Agriculture typically has the lowest labor productivity. In Burkina Faso, for instance, productivity in manufacturing is about 10 times higher than productivity in agriculture, and productivity in construction is almost 14 times as large. Growth happens due to either productivity improvements within-sectors or resource reallocation from low- to high-productivity activities or both. While the first necessitates a steady accumulation of fundamentals (physical capital, skills, and institutional capabilities) across the entire gamut of economic activities, the second requires speedy                                                                   2Note, however, that the extraordinary post-conflict growth resurgence partly reflects the protracted economic decline that preceded the political stability. 3The poverty rates for Benin and Burkina Faso correspond to the year 2011 while the poverty figure for Côte d’Ivoire is for 2008. The data are admittedly a bit old, but these are the latest information available on poverty rates in these countries. 4The information on informal employment for Burkina Faso comes from the World Bank’s International Income Distribution Dataset (I2D2), for Côte d’Ivoire from the 2014 labor force survey (ENSETE 2014), and for Benin from World Bank (2016). 5Labor-intensive manufacturing generally requires skills not too different from those required by agriculture and represents an important source of more productive employment opportunity for the huge pool of workers in agriculture (Dinh et al., 2012). 2       movement of resources to sectors operating at higher productivity levels (Rodrik, 2013b).6 Industrialization-led structural change can solely fuel rapid growth, as applied in Asia.7 By contrast, investments in fundamentals take time to bear fruit, tend to be costly, and mostly yield steady, albeit slow, growth, typical of the West. Structural change is thus purportedly an easier and a more viable way of achieving sustained rapid growth during early stages of development.8 In a nutshell, the paper seeks to address the following main questions: What were the overriding trends underlying output, employment, and labor productivity in BBC? How do these trends compare with those of Asian economies (both when they were at BBC’s stages of economic development and during their respective growth spurts) and SSA comparators? What has been the pace and nature of structural change in BBC and what similarities and differences would stand out from comparisons with Asian and SSA benchmark countries? How much of productivity growth and overall economic growth in the three West African counties can be traced to structural change and within-sector productivity gains/losses? Has structural change been growth-enhancing (with labor moving from low- to high- productivity sectors) or growth-reducing? Has structural change been driven by static or dynamic reallocation effects? What are the relative contributions of demographic changes and employment rate to per capita growth? We find that growth was driven by static structural change and modest within-sector productivity growth, despite heterogeneity across countries and over time. Specifically, growth in Benin during 2006-2015 was almost entirely due to static structural change, while a dynamic loss and negative “within” effect reduced growth. Between 1998 and 2014, Burkina Faso registered growth per capita of 3 percent, over 80 percent of which was due to structural change. The contribution of the static reallocation effect stood at 1.6 percent, roughly half of the growth per capita. In contrast, within-sector changes played a marginal role. Côte d’Ivoire’s meager growth in 2002-2014 was underpinned by a higher employment rate and, to a lesser extent, (static) structural change, while within-sector productivity exhibited a secular decline. In all three countries, demographic changes made a positive contribution to growth. Finally, we show that, unlike the case with fast-growing Asian economies, structural change in BBC has been characterized by the sheer absence of labor movements to high- productivity sectors. BBC, however, exhibit patterns more or less similar to their African counterparts. The rest of the paper is organized as follows: Section 2 provides an overview of developments in the level, growth, and composition of output, employment, and labor productivity in BBC and benchmarks these trends against those of selected Asian and African economies. Section 3 discusses the decomposition method used in this paper. Section 4 thoroughly discusses the empirical results for BBC while putting them in perspective through comparisons with selected benchmark countries. Finally, Section 5 winds up the paper with some concluding remarks. 2. Output, Employment, and Productivity Trends: An Overview This section provides an overview of the trends underlying output, employment, and labor productivity in Benin, Burkina Faso, and Côte d’Ivoire since 2000. We seek to gain some insight into the nature and pace of structural change in these economies by examining the levels, growth, and composition of production, employment and labor productivity. Annual data on value added by sectors were obtained from the respective countries’ national accounts data set. The gross and sectoral value added (GVA) data generally span the period since the mid-1990s. However, the analysis uses a slightly truncated sample period due to the paucity of data on sectoral employment.                                                                   6It should be borne in mind, however, that both processes are not mutually exclusive and reinforce each other. 7However, an economy cannot grow indefinitely solely based on a shift of resources from low- to high-productivity activities. In other words, growth based on structural change will eventually peter out and, beyond a certain point, it should be supported by investment in fundamentals to be sustained. 8 It has been shown that the differential performance between successful and unsuccessful countries can be largely attributed to the pace at which growth-enhancing structural change takes place (McMillan et al., 2014; Timmer et al., 2014). 3       Data on employment9 come from several waves of standardized and harmonized household and labor force surveys (LFS).10 Specifically, aggregate and sectoral employment data for Benin covering 2006, 2010, and 2015 were mainly taken from three waves of Integrated Modular Survey on Household Living Conditions (EMICoV 2007, EMICoV 2011, and EMICoV 2015) and complemented with data from the World Bank’s International Income Distribution Data Set (I2D2).11 For Burkina Faso, sectoral employment data for 1998, 2005, 2009, and 2014 were extracted from the World Bank’s I2D2 database. The analysis for Côte d’Ivoire relies on standardized and comparable data on employment constructed based on information from the 2002 and 2008 household living standard surveys ‘Enquête Niveau de Vie des Ménages’ (ENV 2002 and ENV 2008) and the 2014 Labor Force Survey (ENSETE12 2014).13 To mitigate potential biases from data inconsistencies across surveys, the analysis focuses on major sectors. For all countries and periods, workers were categorized across sectors according to the latest revision of the International Standard Industry Classification (ISIC Rev. 4), which bears some distinctive features compared to earlier revisions. For instance, while fishing was classified as a separate economic activity in ISIC Rev. 3 and Rev. 3.1, it falls under the rubric of ‘agriculture, hunting and forestry’ in the latest ISIC revision. We were unable to regroup a handful of sub-sectors due to a dearth of sectoral employment data. However, these economic activities represented only a negligible proportion of total employment and thus are unlikely to drive the empirical results. Specifically, the decomposition analysis considers the following main sectors: agriculture (including fishing, hunting, and forestry), mining and quarrying (henceforth referred to as ‘mining’), manufacturing, utilities (electricity and water), construction, commerce (comprising wholesale and retail trade, and hotels and restaurants), transport and communications (hereafter ‘transport’), finance, and ‘other services’ (including public administration, education, health, real estate, renting and business activities, and community, social, and personal services). These sectors are grouped under ‘other services’ as they are relatively heterogeneous and thus not of particular interest here. In addition, as noted above, using more aggregated data helps dampen any potential bias associated with definitional changes across surveys. We use data covering 2006 to 2015 for Benin, 1998 to 2014 for Burkina Faso, and 2002 to 2014 for Côte d’Ivoire. Special emphasis is, however, given to developments over the past decade because BBC generally experienced more robust and stable economic growth in recent years. Recent surveys generally use the latest ISIC revision and thus definitional discrepancies between surveys tend to be substantially smaller for the last decade compared to the 1990s and early 2000s. We could not use alternative time periods given the scantiness of data on sectoral employment. In other words, the cutoff points corresponding to each of the sub-periods above were determined by data availability. This paper also benchmarks BBC against selected Asian and SSA economies using data from the Groningen Growth and Development Center (GGDC) database. We compare employment and                                                                   9Data on the number of workers by sector are based on the broadest employment concept, comprising self-employed, family- workers and other informal workers. 10It should be pointed out upfront that we are fully cognizant of the potential shortcomings of the data on employment for the three countries. However, these are the only data available at our disposal and thus we rely on these imperfect figures. As shown later in this paper, analysis based on these data yields interesting insights about the pace and nature of structural change in BBC. 11We opted for the data from EMICoV given that it covers a longer period of time. However, the EMICoV data have been found to be generally consistent with the I2D2 data set. The employment data for 2015 had some aberrant observations which likely represent inexplicable measurement errors and need to be corrected prior to the empirical analysis. Thus, having checked our data against alternative sources and consulted several studies, we made some adjustments to the data to ensure consistency over time. 12ENSETE stands for Enquête nationale sur la situation de l'emploi et du travail des enfants (National survey on the employment and child labor situation). 13It is important to note that these data are fairly consistent with those from the I2D2 data set at least for periods for which I2D2 data are available for Côte d’Ivoire. However, we use these employment data since they provide a more comprehensive and more recent picture of the labor market situation in Côte d’Ivoire. 4       productivity trends as well as the pattern of structural change in BBC with those in five Asian countries (China, the Republic of Korea, Thailand, Indonesia, and Malaysia),14 representing economies that set a good development precedent and that BBC may aspire to emulate, and five SSA economies (Ethiopia, Tanzania, Ghana, Senegal, and Botswana) roughly considered as SSA peers. Data dating as far back as 1960 allow us to compare recent trends in BBC with the golden age of Asia’s growth performance. The updated and extended GGDC database15 provides annual data on value added and employment disaggregated into 10 broad sectors16 (Timmer et al., 2015). We transformed these data into purchasing power parity (PPP) dollars to put the productivity performance of BBC in a comparative perspective. Note that the benchmarking analysis is not meant to be an exhaustive endeavor; however, the main conclusions of this paper were found to be robust to expanding the set of benchmark countries with more Asian and African economies that represent fair comparators. 2.1 Trends in Gross Value Added and Sectoral Contributions The structure of output in the three West African economies is broadly similar. GVA by sector of economic activity for Benin, Burkina Faso, and Cote d’Ivoire is shown in Figure 1 and Table 11. The share of agriculture in total GVA stood at 27 percent in Benin (in 2006), 33 percent in Burkina Faso (1998), and 24 percent in Côte d’Ivoire (2002) and declined by 5 percentage points (ppts), 7 ppts, and 3 ppts over the period until 2015, respectively. Manufacturing’s share of output plunged in all three countries. Benin and Burkina Faso saw a particularly striking drop of about 5 ppts between 2005 and 2015. In contrast, commerce’s share in GVA remained stagnant at about 12-14 percent in Benin and Côte d’Ivoire, and at 11-12 percent in Burkina Faso. Services has superseded agriculture as the largest contributor to the national economy and currently contributes over 50 percent of total output in BBC. Real GVA increased in all economies, despite variations across countries and over time. Of the three countries, Burkina Faso experienced the highest and most stable output growth since the late 2000s. GVA in Benin and Burkina Faso increased by about 36 percent and 48 percent between 2005 and 2015, respectively. Côte d’Ivoire, however, experienced a slight increase, particularly during the 2000s, owing to a decade-long political crisis that took a heavy toll on the economy. Economic activity rebounded in the early 2010s as evidenced by the nearly 40 percent GVA growth since 2011. On a sector-specific basis, output growth in the agricultural sector stands at around 2 percent in Benin and Côte d’Ivoire, and 4 percent in Burkina Faso. Despite its less-than-stellar growth, agriculture accounted for most of the increase in total output due to its large weight. Some sectors (notably finance, utility, and mining) saw hefty growth. Manufacturing expanded at an annual average rate of 2 percent in Burkina Faso and Côte d’Ivoire, while it grew only meagerly in Benin. 2.2 Trends in Employment To facilitate understanding of employment trends, we first briefly highlight the key demographic trends.17 Benin’s population was about 7.5 million in 2006 and increased to nearly 10 million in 2015. The working-age population increased from 3.8 million (about 53 percent of total population) in 2006 to 4.5 million (54 percent) in 2010 and 5.3 million (55 percent) in 2015. The population of Burkina Faso increased from 11 million in 1998 to about 13 million in 2005 and 18 million in 2014. Similarly, the working-age population was about 5 million (about 51 percent) in 1998 and increased to 6 million                                                                   14These countries constitute fair comparators since BBC generally envision transformation into a semi-industrialized economy in the foreseeable future, a path well-trodden by the Asian economies in our sample. 15 The GGDC data on value added come from national accounts while the employment data were constructed using data on total employment levels and sectoral distribution from household surveys and data on growth in employment between census years from labor force surveys. 16Note that, to ensure comparability with our analysis, we merged two of the service sectors (government services and community, social, and personal services) into a single one, namely ‘other services’, reducing the total number of sectors to nine. 17It is, however, important to bear in mind that significant data discrepancies may exist across different sources. For instance, total population and working age population from Census data might not conform to those from LFS reports. To ensure consistency between demographic statistics and employment data, we use information on both variables form survey reports. However, the focus should be on the broader picture rather than on the exact numerical figures. 5       (51 percent) in 2005 and about 9 million (52 percent) in 2014. Côte d’Ivoire’s population surged from about 17 million in 2002 to 22 million in 2014, while the country’s working-age population rose from about 9 million (53 percent) in 2002 to 12 million (55 percent) in 2014. The substantial increase in population, combined with the only slight increase in the proportion of the working-age population, testifies to BBC’s youthful age structure. Turning to the main labor market indicators, the labor force participation rate (LFPR) declined significantly in two of the three economies. In Benin, the LFPR (the proportion of the economically active among the working-age population) shrank from 73 percent in 2006 to 69 percent in 2015, while it dropped from 87 percent in 2005 to 81 percent in 2014 in Burkina Faso. By contrast, Côte d’Ivoire had a significant increase in LFPR of roughly 7 ppts during 2002-2014.18 Similarly, the employment rate (the proportion of those employed among the working age population) dropped slightly in Benin and Côte d’Ivoire, while it remained at about the same level in Burkina Faso.19 With respect to sectoral composition of employment, Benin and Côte d’Ivoire have seen dramatic change in their structure of employment while Burkina Faso saw only a moderate shift. Figure 2 and Table 12 show the sectoral composition of employment. In Benin, agriculture’s share of total employment declined steeply from nearly 60 percent in 2006 to about 42 percent in 2015. Commerce and, to a lesser extent, ‘other services’ absorbed most of the workforce that left agriculture. Similarly, in Côte d’Ivoire, the largest loss in employment (of about 10 ppts) was registered in agriculture, while the largest gain was recorded in commerce. Among the three countries, Burkina Faso saw comparatively smaller change in employment structure. Between 1998 and 2014, the decline in the employment share of agriculture stood at about 9 ppts. Manufacturing’s share of employment remained almost unchanged over the past 15 years, at 7-8 percent in Benin, 2-3 percent in Burkina Faso, and 4-5 percent in Côte d’Ivoire. In general, the primary absorber of the workforce leaving agriculture has been low-productivity non-tradable services activities.20 The prevalence of informal services remains cause for concern as they are less likely to be an engine of sustained growth in BBC. Turning to employment growth, agriculture represents the least expanded sector while commerce grew substantially, despite variation across countries. In all three economies, agriculture saw the lowest employment growth, at about 2 percent in Burkina Faso and Côte d’Ivoire, while Benin had a slight contraction in employment.21 Employment in commerce, transport, and ‘other services’ generally experienced strong growth. In addition, expansion in the mining and construction sectors played a significant role in Burkina Faso and Côte d’Ivoire. However, some of the rapidly growing sectors still account for a small proportion of employment and thus are unlikely to absorb a significant portion of the labor force leaving agriculture as well as the hundreds of thousands of youth entering the labor market every year. Most of the jobs created over the last decade were in the urban (informal) services sector, reflecting the rarity of productive job creation in BBC. Compared to Asian economies, BBC exhibit distinct employment trends. The labor shed by BBC’s agricultural sector moved predominantly to commerce, which has the second-lowest level of productivity. This contrasts with the Asian experience of moving labor out of agriculture to high- productivity manufacturing industries. The share of manufacturing in GDP stood at 10-20 percent in Korea, China, Thailand, and Malaysia during the mid-1970s, which surged to 31-37 percent in the 2010s.22 Manufacturing also absorbed a substantial and growing share of the labor force, ranging from                                                                   18The significantly lower LFPR in the year 2002 may, however, be capturing anomalies due to the political crisis that stifled economic activity starting in the early 2000s, thereby resulting in a decrease in the participation rate. 19It is noteworthy that the decline in participation and employment rates does not necessarily represent an adverse development as it may simply reflect the rise in student population in these countries driven by rapid expansion of education.    20Manufacturing tends to exhibit stronger productivity growth. Although the dynamic within manufacturing tends to play a less significant role at an early stage of development, it constitutes a galvanizing force as the sector becomes larger (Rodrik, 2013a). 21In Benin, most of the sectors registered significant growth in employment during 2006-2015, while they experienced only lackluster expansion in the 2010s. 22The values corresponding to mid-1970s and early 2010s represent averages for 1970-1975 and 2005-2010 respectively. We use five-year averages because we might end up picking anomalies by using annual observations. 6       about 15 percent in Thailand to roughly 30 percent in Korea during the 2000s, representing a marked increase compared to 7-10 percent in the mid-1970s. These trends accelerated in the ensuing decades of Asia’s unprecedented growth.23 The employment trends in BBC are, however, similar to those observed in SSA countries. 2.3 Trends in Labor Productivity As structural change is inherently intertwined with the movement of labor across sectors and resulting changes in labor productivities, we now turn to discussing productivity trends. Figure 3 and Table 3 show trends in labor productivity (as measured by GVA per worker), productivity growth, and employment elasticities.24 Benin and Burkina Faso had modest productivity growth of about 9 and 13 percent between 2005 and 2015, respectively. In Côte d’Ivoire, overall productivity contracted by 3 percent during 2008-2014, indicating that the country has yet to catch up with its productivity levels in the late 1990s. The overall trends, however, mask significant changes over time. Productivity growth in Benin was larger in 2010-2015. Burkina Faso posted a 30 percent productivity growth in 1998-2005 while productivity growth was quite smaller since the mid-2000s. In the case of Côte d’Ivoire, productivity decline slowed down since 2008, driven by the rebound in economic activity that followed the return to political stability in 2012. Despite the overall limited changes in aggregate labor productivity, there exist significant inter- sectoral productivity gaps. In Benin, each agricultural worker produced CFAF 0.5 million in 2015, while the average worker in manufacturing and construction was roughly 4 and 6 times more productive. Finance workers, on average, produced nearly 50 times more than their agricultural counterparts. Similarly, in Burkina Faso, productivity in manufacturing is about 10 times higher than productivity in agriculture, and productivity in construction is almost 14 times as large. Utilities posted the highest labor productivity, which was about 96 percent of agricultural productivity. Turning to Côte d’Ivoire, we observe fairly similar trends. Labor in manufacturing is 9 times more productive than that in agriculture. Finance, mining, and utilities have even greater productivity levels than agriculture. In general, the manufacturing-agriculture productivity ratio in Burkina Faso and Côte d’Ivoire is substantially higher than the African average of about 2.3 (McMillan et al., 2014), signifying significant potential for structural change-induced productivity improvements. Finance, utilities, and mining featured strong labor productivity growth. However, high average labor productivity in these capital-intensive sectors may simply indicate that the labor share of value added is quite small.25 Put slightly differently, although these sectors operate at very high productivity, they employ only a tiny proportion of the labor force and thus are unlikely to generate much employment so as to absorb BBC’s surplus and largely unskilled labor. In addition, the already high and rising productivity implies that fewer workers will find employment in these sectors as the economies grow. In contrast, the sectors that absorbed the vast majority of the labor force that left agriculture, notably                                                                   23 In Asia, services expanded significantly and absorbed much of the labor force rather late in the development process, after industrialization had run its course and at much higher levels of productivity. Asian economies shifted into services-oriented economy once they accumulated the human capital and other fundamental capabilities necessary for transforming those services into high-productivity activities (Rodrik, 2013b). 24Employment elasticities measure the percentage change in employment associated with the percentage change in GDP the higher the elasticity, the more employment-intensive economic growth is. In addition to evaluating economic performance through a ‘productivity’ lens, employment elasticities constitute a good indicator of an economy’s capacity to generate employment. When economic growth is positive, employment elasticities between 0 and 1 indicate the presence of both productivity growth and employment growth (Kapsos, 2005- cited in Martins (2014)) the higher the elasticity the more employment-intensive growth is, while lower elasticity values indicate strong labor productivity growth. Elasticities exceeding 1 are combined outcomes of positive employment growth and negative productivity growth. In contrast, a negatively elasticity value serves as a signal for negative employment growth and positive productivity growth. 25 Assuming a Cobb Douglas production function, the marginal productivity of labor is given by the product of average productivity and the share of labor in value added. This reflects that productivity comparison might be misleading if labor shares significantly differ across sectors. For instance, high average productivity in a capital-intensive sector may simply reflect that the labor share is low. 7       commerce, experienced negative productivity growth, reflecting their limited capacities to add new workers while maintaining productivity levels. Only a few sectors have seen both productivity growth and employment growth. The employment elasticities in the last two columns of Appendix Table 3 shed light on the extent to which economic growth has been employment-intensive. In all three economies, only a handful of sectors experienced simultaneous growth in productivity and employment those with elasticities between 0 and 1. However, the overall elasticity for Burkina Faso is around 0.7 percent for 2005-2014, which falls within the ‘ideal’ range and suggests that employment growth happened without associated slowdown in productivity. From an economy-wide perspective, Burkina Faso is the only country that has consistently achieved employment growth without a concomitant productivity decline. Benin saw simultaneous growth in productivity and employment only during 2010-2015, while positive employment growth in 2006-2010 was at the expense of a much lower productivity growth. Unlike Burkina Faso and Benin, overall employment elasticity was persistently negative for Côte d’Ivoire, reflecting the fact that none of the sectors had employment boost without associated productivity bust. Overall employment elasticities, however, hide substantial sectoral heterogeneity. Agriculture shows negative elasticity in Benin, reflecting that the sectoral shift in employment away from agriculture has improved agricultural productivity. Employment growth in the agricultural sector coincided with positive productivity growth in Burkina Faso, while it entailed productivity slowdown in Côte d’Ivoire. The transport sector in Benin and Burkina Faso witnessed the strongest employment response to production growth; however, it accounts for a small proportion of aggregate employment. In general, above-unity elasticities for many of the sectors in BBC indicate the presence of positive employment growth but resultant challenges to sustaining productivity growth.26 However, the economy-wide impact of sectoral labor productivities hinges on the relative weight of the most dynamic sectors. Given the large inter-sectoral differences in the degree of capital intensity, comparison of productivity levels may be more meaningful across sectors with similar potential to absorb labor. Figure 4 plots labor productivity in each sector against the sectoral shares of employment for the start and end points of the sample periods. The area of each column represents sectoral GVAs scaled by total employment (GVA ⁄L ∗ L ⁄L GVA ⁄L) and provides a powerful illustration of the relative size and productivity level of each sector. Agriculture, commerce, and ‘other services’ represent the three most important sectors, despite differences across countries. Comparing the width of the columns (measuring sectoral labor shares) for the two periods, we see significant changes for Benin and Côte d’Ivoire, while the modest change for Burkina Faso reflects relatively static employment shares. The high-productivity sectors employ a negligible portion of the labor force, while the rapidly expanding sectors maintain very low productivity. Specifically, agriculture’s share of employment declined significantly in Benin and Côte d’Ivoire, which was accompanied by productivity improvements. In contrast, Burkina Faso experienced smaller changes in the relative size and productivity of its agricultural sector. In Benin and Côte d’Ivoire, commerce and ‘other services’ expanded substantially in terms of employment shares; however, this was followed by a significant productivity decline. To sum up, perhaps the most striking feature of the three economies is the sheer absence of labor movements to relatively high-productivity and more dynamic sectors. Figure 5 plots changes in employment shares against the relative productivity in each sector (as measured by the log of the share of end-of-period sectoral productivity in total productivity). Growth-enhancing structural change                                                                   26 However, it should be borne in mind that employment growth is also associated with population (and thus labor force) growth. Despite significant employment growth, one of the key challenges being faced by many African economies, including BBC, is the growing prevalence of the informal economy, which is typically characterized by low productivity and stagnant wages. Therefore, it is important that the positive trend in employment growth is complemented with data on the quality of employment, which would likely provide a more nuanced picture of the labor market situation in these countries. 8       suggests positive correlation between the direction of labor flows and (end-of-period) labor productivity in individual sectors. The relative size of each sector (measured by employment share) is indicated by the circles in the scatter plots. According to the traditional path of structural change and for BBC’s income level, one expects agriculture to fall in the bottom-left quadrant (low productivity and declining labor share) and the relatively more dynamic sectors in the top-right quadrant (relatively high labor productivity and a rising labor share). In all three cases, there are some signs that structural change was growth-enhancing, although to a smaller extent in Benin and Côte d’Ivoire. The sector with the lowest labor productivity, namely agriculture, experienced the largest relative loss in employment. In Burkina Faso, the relative sizes of the non-agricultural sectors are still small, indicating the presence of sizable room for productivity-enhancing structural change. Low-productivity activities largely absorbed the labor displaced from agriculture, suggesting a limited role for growth-enhancing structural change. For Benin and Côte d’Ivoire, commerce is located in the bottom-right quadrant, indicating that the drastic expansion in the employment share of this sector was ensued by a rapid decline in the already low level of productivity. Commerce features a high level of informality and is the second least productive sector. ‘Other services’ is another rapidly expanding sector, which was again a relatively unproductive non-tradable sector. We see that the slope for Benin and Côte d’Ivoire is much flatter than that for Burkina Faso, suggesting a very small growth-enhancing role for structural change in these two economies. To be sure, the employment shift away from agriculture led to an improvement in agricultural as well as aggregate labor productivity; however, the impact would have been stronger if labor had relocated to more productive sectors. Table 1. Labor productivity (LP) in selected Asian and African economies Economy- Coef. of Sector with Sector with Compound wide variation highest Lowest annual LP* of LP productivity* productivity* growth of Sector LP Sector LP economy- wide LP** Asia Korea, Rep. 48477.8 0.053 PU 372158 Agri 20083 7.6% Malaysia 37367.2 0.107 Min 872603 Con 13793 6.9% Thailand 16277.7 0.113 Min 533843 Agri 5401 6.8% China 14463.8 0.093 Fin 97604 Agri 4209 15.5% Indonesia 3419.9 0.069 Min 90830 Agri 980 7.9% Africa Botswana 50703.8 0.121 Min 718869 Agri 4409 11.3% Ghana 11068.0 0.081 PU 42264 Trade 5709 24.7 Senegal 4877.0 0.173 PU 367608 Agri 1726 4.4% Benin 3921.9 0.112 Fin 50022 Trade 1769 4.5% Ethiopia 3781.5 0.126 PU 52034 Agri 2417 10.8% Tanzania 3570.2 0.100 Con 24194 Agri 1498 14.8% Côte d’Ivoire 3379.6 0.227 Min 9164 Agri 68.2 1.4% Burkina Faso 2894.5 0.204 Min 2249 Agri 35.8 6.7% Note: All numbers are for 2010 unless otherwise stated. *2000 PPP $. All numbers are for 2005. ‘Agri’ stands for Agriculture, hunting, forestry and fishing; ‘Min’ for Mining and quarrying; ‘Man’ for Manufacturing; ‘PU’ for Public utilities (electricity, gas and water); ‘Con’ for Construction; ‘Trade’ Wholesale and retail trade, hotels and restaurants; ‘Tran’ for Transport, storage and communications; and ‘Fin’ for Finance, insurance, real estate and business services. *Compound annual growth rate of economy- wide productivity (1990-2010). Source: Author’s computation based on data described in Section 2. Finally, we benchmark the pattern of structural change in BBC against those in the Asian and SSA countries. Table 1 shows economy-wide and sectoral (highest and lowest) labor productivity levels and country-specific productivity growth. Average labor productivity ranges from $2,895 (at 2005 PPP dollars) for Burkina Faso to $48,478 for Korea. China has the fastest productivity growth, while the lowest productivity growth corresponds to Côte d’Ivoire. Similarly, Benin and Burkina Faso are located at lower rungs of productivity growth. In most countries, agriculture has the lowest 9       productivity level, followed by wholesale and retail trade, while mining and public utilities maintain the highest productivity levels. We note that productivity gaps are typical characteristics of underdevelopment. Economy-wide productivity gaps (as measured by the coefficient of variation of the log of sectoral labor productivities) tend to decline with sustained economic growth. Inter-sectoral productivity gaps are also the widest for the poorest countries in our sample. These indicate that structural change plays an important role in promoting convergence, both within and across countries. Altogether, unlike BBC, the Asian economies show a particularly clear-cut case of growth-enhancing structural change. Figures 6 and 7 plot the (end-of-period) relative productivity of sectors against the change in their employment shares for the Asian and SSA countries. China, Thailand, and Indonesia represent distinct cases of growth-enhancing structural change, as evidenced by the sharply positive slopes of their scatter plots. In these economies, labor largely moved out of agriculture to higher- productivity activities, notably manufacturing and tradable services.27 However, not all Asian countries exhibit this kind of pattern.28 However, strong “within” productivity growth in Korea has given a boost to overall productivity. Turning to our African sample, Ethiopia, Tanzania, and Botswana show growth-enhancing structural change, while productivity gain from structural change is relatively weaker in Senegal and Ghana. In general, the pattern of structural change in Benin and Côte d’Ivoire looks more like those of Korea and Ghana. Burkina Faso exhibits growth-enhancing structural change; however, the slower pace of structural transformation and missing contribution of high-productivity manufacturing sets it apart from the typical pattern observed in Asia. 3. Decomposition of Productivity and GDP Growth In this section, we decompose GDP per capita growth into changes in labor productivity, employment rate, and demographic structure both at the aggregate and sectoral level. Next, we decompose productivity growth into within-sector and between-sector productivity changes. In addition, we split the structural change component into static gains (or losses) in productivity (due to labor shifts from below- to above-average productivity level sectors) (or vice versa) and dynamic gains (or losses) in productivity (due to relocation of workers from below- to above-average productivity growth sectors) (or vice versa). Finally, putting the pieces together, we disentangle the sources of per capita growth by tracing the contributions of changes in within-sector productivity, structural change (static and dynamic), employment rate, and population structure. Most studies29 on structural change solely focus on decomposition of productivity growth, while the handful of studies that also decompose GDP growth fall short of decomposing structural change into its static and dynamic elements. In this paper, we adopt insights from Timmer et al. (2014) to quantify the portion of productivity growth and economic growth that can be attributed to static and dynamic effects. 3.1 Defining the Contribution of Structural Change The empirical analysis in this paper employs the Shapley decomposition method.30 As touched upon above, we move beyond the standard decomposition of productivity growth and rather pursue a broader approach that allows us to unravel the contributions of the above-mentioned contributory                                                                   27In China, however, the expansion of low-productivity service activities has dragged down economy-wide productivity growth. 28Korea particularly represents an anomaly as it saw its high-productivity manufacturing industries shrink in favor of relatively low-productivity service activities; hence the much flatter slope of the Korea plot. 29See, for instance, McMillan et al. (2014), McMillan and Harttgen (2014), Timmer and de Vries (2009), and Timmer et al. (2014). 30See World Bank (2012) for detailed technical discussions and illustrations of the Shapley decomposition technique. The estimates in this paper were obtained using the Job Generation and Growth (JoGGs) Decomposition tool an Excel-based software. The Shapley decomposition first computes the marginal effect on output per capita growth of the sequential elimination of each of the contributory factors, and then the growth contribution of each factor is derived as the average of its marginal contributions in all possible elimination sequences. In other words, the growth contribution of a variable is interpreted as its impact on output per capita growth had the remaining contributing factors remained unchanged. 10       factors to per capita GDP growth. The following equation for output per capita is the basic building block of our decomposition analysis: Y Y E A = . . (1) N E A N y = p . e . ω where Y stands for total output, N for total population, A for total working-age population, E for total employment, Y⁄E for output per worker (average labor productivity) (hereafter denoted with p), E⁄A for the proportion of the working-age population that is employed (the employment rate) (denoted with e), A⁄N for the share of the labor force in total population (the relative size of the working-age population) (denoted with ω), and y for output per capita. The fractions of growth associated with changes in the aforementioned three components can be computed based on an equation for output per capita growth: 1 1 ∆ = ∆ 3 6 Contributions of changes in output per worker 1 1 ∆ 3 6 Contributions of changes in the employment rate ∆ (2) Contributions of changes in the demographic structure where the three terms in the equation represent the respective contributions of changes in output per worker, the employment rate and the demographic component. The beauty of this decomposition methodology is that it is inherently additive where each component can be interpreted as the per capita growth consistent with a counterfactual scenario, in which the only observed change corresponds to the particular component in question while all else remains unchanged. For instance, such an exercise can be used to estimate the growth contribution of a change in output per worker in sector i had demographics, employment rate, and productivity in the rest of the sectors remained unchanged. The contributions of the three components can be expressed as a share of output per capita growth by dividing each of the terms in Eq. (2) by y. Denoting the fraction of growth explained by each component with ̅ ̅, and , output per capita growth can now be rewritten as: ∆ ∆ ∆ ∆ ̅ ̅ (3) where the bar represents the proportion of output per capita growth than can be linked to changes in output per worker, the employment rate and the share in total population of working age. The next step is to decompose output per worker (labor productivity) into changes in output per worker within-sectors and sectoral employment shifts: 11       4) Changes in aggregate output per worker can now be decomposed into within-sector and between- sector effects: , , , , ∆ ∆ ∆ (5) 2 2 Within-sector Between- sector component component where and refer to economy-wide and sectoral labor productivity levels, respectively. The ∆ operator denotes the change in productivity or employment shares between and 0. The first term in the equation captures the change in overall labor productivity that can be ascribed to productivity changes within sector , which is referred to as the “within” component. The second term in the equation corresponds to change in output per worker due to inter-sectoral shifts in employment the more workers move to high-productivity sectors, the larger the proportion of this component would be. A negative value for this term would, however, indicate that shifts in employment across sectors were on the whole detrimental to overall productivity growth. Stepwise decomposition approach Sectoral contributions to changes in the employment rate can then be calculated by further decomposing aggregate growth in employment (∆e) by sectors: ∆ ∆ (6) 12       where ∆ ∆ E ⁄A represents the change in employment in sector as a share of total working age population. Dividing both sides of Eq. (6) by , ∆ ⁄ captures the fraction of aggregate employment rate change that can be attributed to changes in employment in sector . Next, the structural change component is further decomposed into static and dynamic reallocation effects. We split the structural change term into contributions due to reallocation of workers to above- average productivity level sectors (static reallocation effect) and contributions arising from labor reallocation to above average productivity growth sectors (dynamic reallocation effect). The decomposition tool developed in Timmer et al. (2014), namely the shift-share method, uses base periods for employment shares and productivity levels instead of period averages as in our decomposition method (see Eq. (5)). However, we use insights from this method to further disaggregate the between-sector component in Eq. (5) into static and dynamic effects. We use average values for both employment shares and productivity levels rather than base periods as in Timmer et al. (2014). It should be noted, however, that the main findings of this paper stand up well to using the standard shift-share method instead of the Shapley decomposition technique. Figure 1 shows a schematic presentation of the stepwise decomposition approach adopted in this paper. Finally, taking these slight modifications into account, the change in aggregate productivity, ∆ , in Eq. (5) can be respecified as follows: , , , ∆ (7) 2 2 2 Within-sector Static reallocation Dynamic reallocation component effect effect where the superscript t 0 and t T represent initial and final period, respectively. Eq. (7) comprises three terms. The first term represents within-sector productivity changes. The “within” effect is the weighted sum of productivity growth within individual sectors, where the weights are the averages of the employment shares of each sector in the two periods. The “within” effect is positive (negative) when the weighted change in productivity is positive (negative). The second term measures the portion of overall productivity changes explained by labor reallocation across sectors, which is positive if changes in employment shares are positively correlated with productivity levels. The third term represents the joint effect of changes employment shares and sectoral productivity growth. It is positive (negative) if workers are relocating to sectors that are experiencing positive (negative) productivity growth. In other words, the second term indicates whether workers moved to above- average productivity level sectors (static effect) while the third term shows whether productivity growth is higher in sectors that expanded in terms of employment shares (dynamic effect). 4. Results 4.1 Decomposition at the Aggregate Level 4.1.1 Productivity, Employment Rate, and Demographic Changes Economic growth in Benin has been driven by strong productivity growth and demographic changes. As set out above, we first untangle the fractions of GDP per capita growth that are linked with changes in productivity, employment rate, and demographic structure. Table 2 presents the results. We find that about 90 percent of the 50 percent increase in GVA per capita during 2006-2015 was on account of increased labor productivity.31 Demographic changes were also important, accounting for 70                                                                   31 The interpretation is that had the employment rate and demographic structure remained unchanged, higher productivity would increase GVA per capita by around 90 percent. 13       percent of the observed growth, suggesting that fewer dependents per each working age adult (lower dependency ratio) has resulted in an increase in output per capita. We found qualitatively similar results for 2010-2015; however, changes in the demographic structure played a far less significant role. The substantial productivity gains were, however, partially offset by a negative impact from the lower employment rate, implying that economic growth in the last decade was ‘job-less’. Table 2: Aggregate decomposition of output per capita growth Benin GVA per capita % Contribution of (Total growth, Output per Employment Demographic %) Worker rate change ∆(y/n) ∆(y/e) ∆(e/a) ∆(a/n) 2010-2015 10.34 118.85 -44.11 25.26 2006-2015 9.65 88.99 -56.81 67.82 Burkina Faso GVA per capita % Contribution of (Total growth, Output per Employment Demographic %) Worker rate change ∆(y/n) ∆(y/e) ∆(e/a) ∆(a/n) 1998-2005 29.27 144.17 -8.29 -35.88 2005-2014 40.40 59.94 -30.84 70.89 1998-2014 69.67 99.27 -20.51 21.25 Côte d’Ivoire GVA per capita % Contribution of (Total growth, %) Output per Employment Demographic worker rate change ∆(y/n) ∆(y/e) ∆(e/a) ∆(a/n) 2002-2008 -3.43 -154.08 58.95 -4.86 2008-2014 14.69 -21.74 106.42 15.32 2002-2014 10.76 -81.99 163.08 18.90 Note: y stands for total output, n for total population, a for total working-age population, e for total employment, a/n for the share of the labor force in total population (the working-age population), e/a for the proportion of the working-age population that is employed (the employment rate), and y/e for output per worker (average labor productivity). Source: Author’s estimation based on data described in Section 2. Similarly, sustained strong growth in Burkina Faso has been largely due to productivity growth. Much of the growth in GVA per capita in 1998-2014 can be explained by labor productivity growth. Specifically, productivity growth explains 99 percent of the 70 percent overall increase in GVA per capita during this period. Positive demographic developments also contributed to growth, albeit to a much smaller degree. The lower employment rate was a drag on growth. Interesting insights emerge from breaking down the sample into two sub-periods. We observe that stronger productivity growth explains most of the 30 percent growth in output per capita during 1998-2005, whereas changes in demographic structure and employment rate reduced growth. However, while growth was primarily propelled by productivity growth in 1998-2005, it was mainly driven by demographic changes in 2005- 2014, although productivity growth was still an important contributor. Overall, similar to Benin, lack of employment generation remains the Achilles’ heel of Burkina Faso’s economy. Côte d’Ivoire’s growth experience is diametrically opposed to those of Benin and Burkina Faso. Unlike these two countries, Côte d’Ivoire saw large negative labor productivity growth since the early 2000s, which was mainly caused by the civil war from 2002-2004, and the post-election crisis and armed conflict in the early 2010s. Productivity growth rebounded since 2012; however, it has not yet caught up with pre-crisis levels. An increase in employment rate was the largest contributor to the 11 percent growth over 2002-2014, while demographic changes made a negligible contribution. Splitting the sample into subperiods gives rise to qualitatively similar results. Productivity growth and 14       demographic changes were the key engines of growth in Benin and Burkina Faso, while lack of employment generation capacity held back growth. The reverse holds true in the case of Côte d’Ivoire. 4.1.2 Decomposing Aggregate Productivity Growth Labor productivity growth can be achieved in one of two ways. First, productivity growth can stem from improvements within sectors due to increased capital accumulation, technological change, or reduction of allocative inefficiencies. Second, productivity can be boosted by moving labor across sectors, namely, from low-productivity to high-productivity sectors. In addition, productivity gains from labor reallocation would have stronger economy-wide effects if labor moved from below-average to above-average productivity growth sectors. Thus, we examine the contributions of within-sector changes and structural change (static versus dynamic effects) to productivity growth in BBC. In Benin, productivity growth since the mid-2000s has been largely due to structural change while within-sector productivity gains featured more prominently in recent years. Figures 7 decomposes labor productivity growth into within-sector changes and structural change (as well as the static and dynamic subcomponents) for Benin, Burkina Faso, and Côte d’Ivoire.32 In Benin, the modest productivity growth during 2006-2015 was generally driven by structural change. Structural change accounted for 180 percent of the overall productivity growth during 2006-2010, most of which was offset by the sharp productivity decline within manufacturing, commerce, and ‘other services’. In particular, productivity growth was mostly on account of static gains. With respect to sectoral trends, agriculture’s productivity growth is due to “within” gains and static structural change, whereas it experienced a dynamic loss. Turning to 2010-2015, we find that only a small proportion of productivity growth can be attributed to structural change. The productivity gain in this period is mostly attributed to within productivity gains achieved in services and manufacturing. In Burkina Faso, we note that structural change accounted for most of the overall growth in labor productivity during 1998-2014. Structural change explains approximately 84 percent of the overall productivity growth. In particular, static effects dominated productivity gains form structural change. In general, within-sector productivity improvements made a relatively modest contribution to productivity growth. However, productivity growth was held back by sizable productivity loss in manufacturing and other industrial activities. Agriculture experienced a static loss, which partly offset productivity improvements from dynamic gain and within-sector productivity increase. A look at the results for the subperiods suggests that structural change made an even more significant contribution to overall productivity growth in 2005-2015, a good portion of which was dampened by productivity loss in manufacturing and other industry. We find qualitatively similar results for 1998-2005. Côte d’Ivoire’s productivity plunge between 2002 and 2014 was mainly due to within-sector productivity loss, while structural change contributed positively, albeit slightly. Roughly 60 percent of the overall productivity growth during this period can be explained by structural change. In particular, static reallocation gains, notably in the agricultural sector, was responsible for almost all of the productivity improvements due to structural change. This was, however, more than offset by the secular decline in productivity within sectors, particularly in the commerce sector. In fact, the 2000s in Côte d’Ivoire represents a “lost decade” with productivity in the majority of the sectors stuck in the negative territory. Services and, to a lesser extent, manufacturing experienced the largest loss in productivity. Splitting the sample reveals that the “within” effect is virtually identical in the two subperiods. But the contribution of structural change went up from 35 percent in 2002-2008 to 130 percent in 2008-2014, a remarkable reversal over the course of a few years.                                                                   32 The figures for Burkina Faso and Côte d’Ivoire show only results for the full sample since these results are more or less similar to those for the subperiods. In the case of Benin, there exists a stark difference between the two subperiods with respect to the major drivers of productivity growth. 15       4.2 Sectoral Growth Decomposition: The Role of Structural Change The overriding objective of this section is to examine the contributions of structural change and within- sector productivity changes to GDP growth per capita. Specifically, we decompose per capita growth into within-sector productivity changes, demographic effect, employment rate, and structural change. We then split structural change into productivity changes arising from static and dynamic reallocation effects. As mentioned above, the static component becomes positive (negative) when sectors with above-average productivity levels increase (decrease) their share in total employment. Similarly, the dynamic component will be positive (negative) if sectors with above-average productivity growth experienced an increase (decrease) in their employment share. Figure 3 shows the decomposition 16       results for BBC both for the full sample and the most recent periods. Tables 3-8 report the same results but only for the full sample, while the remaining results are relegated to Tables 4-6. Figure 3. Shapley Decomposition of GDP per capita growth Benin: 2006-2015 Benin: 2010-2015 Employment rate Employment rate Demographic effect Demographic effect Within-sector Within-sector Structural change- Structural change- Dynamic Dynamic Structural change- Structural change- Static Static -1.5% -0.5% 0.5% 1.5% 2.5% -1.5% -0.5% 0.5% 1.5% 2.5% Burkina Faso: 1998-2014 Burkina Faso: 2005-2014 Employment rate Employment rate Demographic effect Demographic effect Within-sector Within-sector Structural change- Structural change- Dynamic Dynamic Structural change- Structural change-Static Static -1.0% -0.5% 0.0% 0.5% 1.0% 1.5% 2.0% -1.0% 0.0% 1.0% 2.0% Cote d'Ivoire: 2002-2014 Cote d'Ivoire: 2008-2014 Employment rate Employment rate Demographic effect Demographic effect Within-sector Within-sector Structural change- Structural change- Dynamic Dynamic Structural change- Structural change-Static Static -1.5% -1.0% -0.5% 0.0% 0.5% 1.0% 1.5% 2.0% -2.0% -1.0% 0.0% 1.0% 2.0% 3.0% Source: Author’s estimation based on data described in Section 2. In Benin, the empirical results suggest that the modest output per capita growth during 2006-2015 has been mostly attributed to static structural change. In contrast, within-sector productivity and dynamic structural change acted in the opposite direction. Specifically, annualizing growth rates, structural change and demographic changes accounted for 1.7 ppts and 0.7 ppts of the 1.1 percent per capita growth. However, these were partly neutralized by a decrease in productivity within sectors (-0.8 ppts) and lower employment rate (-0.6 ppts). Interesting insights emerge from further breaking down structural change. Over 130 percent of the structural change-induced growth can be attributed to static 17       reallocation gains. The dynamic effect is, however, heavily negative (-0.6 ppts), suggesting that relocation of the workforce from ‘high’ productivity growth to low productivity growth sectors (i.e. reverse structural transformation) has been a drag on economic growth. Sectoral decomposition reveals that agriculture, commerce, and ‘other services’ were the major contributors to structural change and within-sector productivity changes. Agriculture made significant positive contribution to the “within” component, which was more than offset by productivity decline in ‘other services’, commerce, and manufacturing. A larger proportion of agriculture’s positive growth contribution was due to an increase in productivity and static gains. As agriculture has productivity level below the economy-wide average, the drop in its share of employment led to higher per capita growth. However, the decline in agriculture’s employment rate was so large that it reduced per capita growth. Manufacturing made a negative contribution to growth due to a marked decline in productivity. As opposed to this, commerce contributed negligibly. Despite absorbing most of the labor that left agriculture, the sharp productivity drop in commerce more than offset the growth gain due to employment growth, thereby causing static productivity loss. Altogether, despite negative within-sector productivity growth in 2006-2015, Benin’s economy grew by 4.5 percent mainly via reducing allocative inefficiencies. Table 3 presents the percent contribution of each sector to the structural change term. We see that agriculture has made the largest contribution to overall inter-sectoral productivity growth (about 60 percent), followed by other services (44 percent). Commerce contributed significantly but negatively because it absorbed most of the workforce displaced from agriculture while at the same time experiencing a steep decline in output per worker. However, a look at the more recent period, 2010-2015, reveals newly emerging trends with respect to the key drivers of growth. “Within” gains accounted for almost all of the average annual growth per capita of about 2 percent. Unlike 2006-2010, static structural change explains only 20 percent of the per capita growth during 2010-2015. This period was also marked by dynamic productivity loss, albeit to a relatively limited extent. Substantially higher “within” productivity gain has been a distinctive feature of the 2010s, unlike the 2000s during which structural change dominated while “within” productivity loss reduced growth. Table 3. Benin: Sectoral decomposition of GDP per capita growth: 2006-2015 (Annual percentage changes) % Contribution of Within-sector Employment Inter-sectoral Total productivity rate Static Dynamic total (%) Agriculture 0.69 -2.32 1.67 -0.65 1.02 -0.61 Mining -0.01 0.00 0.00 0.00 0.00 0.00 Utilities 0.08 0.00 0.01 0.00 0.01 0.09 Manufacturing -0.60 0.01 0.11 -0.03 0.07 -0.51 Construction -0.07 0.04 0.11 0.00 0.11 0.08 Commerce -0.53 0.98 -0.47 0.09 -0.47 -0.02 Transport 0.03 0.12 0.18 0.00 0.18 0.33 Finance 0.47 0.00 0.04 0.00 0.04 0.51 Other services -0.83 0.57 0.75 -0.04 0.75 0.49 Subtotals -0.77 -0.61 2.36 -0.64 1.73 0.35 Demographic component 0.73 Annual % change in GVA per capita 1.07 Source: Author’s estimation based on data described in Section 2. During 1998-2014, Burkina Faso registered per capita growth of almost 3 percent per annum, over 80 percent of which was accounted for by structural change. In particular, the contribution of static structural change stood at 1.6 percent, roughly half of the per capita growth in this period. This does not come as much of a surprise in view of the fact that labor moved predominately from very low- productivity agriculture to higher productivity sectors, including ‘other services’ and commerce. In 18       addition, the reallocation of workers to above-average productivity growth sectors made positive contributions to growth. However, only a small portion of growth in 1998-2014 can be traced to within-sector productivity growth. The remaining portion is explained by the lower employment rate and an increase in the share of working-age population. Table 4. Sectoral contributions to Inter-sectoral shifts (2006-2015) Direction of Contribution to employment inter-sectoral share shift shifts (%) Agriculture - 59.16 Mining + 0.27 Utilities + 0.45 Manufacturing + 4.29 Construction + 6.53 Commerce + -27.04 Transport + 10.42 Finance + 2.39 Other services + 43.62 Total 100 Source: Author’s estimation based on data described in Section 2. The sectoral decomposition results suggest that ‘other services’, agriculture, and transport together explained nearly 70 percent of the structural change-induced growth. ‘Other services’ contributed by about 0.8 percent, nearly 30 percent of per capita growth. Agriculture’s small (negative) contribution is the combined outcome of a static loss and dynamic gain.33 The remaining sectors are generally characterized by static gains but no dynamic gains/losses. In contrast, commerce made only meager contributions, despite employing over 50 percent of the workforce that left agriculture. Manufacturing made barely any contribution to growth since the late 1990s. Table 5 shows sectoral contributions to structural change. ‘Other services’ was the largest contributor, followed by agriculture and commerce. Table 5. Burkina Faso: Sectoral decomposition of GDP per capita growth: 1998-2014 (Annual percentage changes) % Contribution of Within-sector Employment Inter-sectoral Total productivity rate Static Dynamic total (%) Agriculture 0.56 -1.20 -0.38 0.83 0.45 -0.19 Mining 0.21 0.03 0.18 0.00 0.18 0.42 Utilities -0.50 0.05 0.27 -0.01 0.26 -0.19 Manufacturing 0.15 0.00 -0.05 0.00 -0.04 0.11 Construction -0.19 0.04 0.26 0.00 0.26 0.12 Commerce -0.03 0.33 0.18 0.00 0.18 0.48 Transport 0.06 0.04 0.42 0.00 0.42 0.52 Finance 0.25 -0.01 -0.03 0.00 -0.03 0.21 Other services -0.03 0.11 0.77 0.00 0.77 0.85 Subtotals 0.47 -0.60 1.63 0.82 2.45 2.32 Demographic component 0.63 Annual % change in GVA per capita 2.94 Source: Author’s estimation based on data described in Section 2. Structural change played an important positive role during 2005-2014, but it has been dominated by static gains. Static structural change contributed by 1.74 ppts, roughly 70 percent of the 2.5 percent growth per capita, while dynamic gains made a insignificant contribution. By contrast, a decline in                                                                   33 Dynamic productivity gain in agriculture reflects the fact that this sector shrank in terms of employment while productivity growth is below-average. Conversely, some of the labor that left agriculture moved to sectors that experienced higher productivity growth. 19       within-sector productivity and the lower employment rate have had a negative impact on growth. The contribution of demographic changes is, however, much larger in this period. We also note some distinctive features with respect to sectoral growth contributions. The contribution of agriculture to static structural change turned positive. All major sectors posted negative productivity growth. Altogether, unlike the period 1998-2005 when structural change and “within” gains together accounted for almost all of growth, they explained only 30 percent of per capita growth in 2005-2014. Table 6. Burkina Faso: Sectoral contributions to inter-sectoral shifts 1998-2014 Direction of Contribution to employment inter-sectoral share shift shifts (%) Agriculture - 18.53 Mining + 7.25 Manufacturing + 10.64 Utilities - -1.83 Construction + 10.75 Transport + 7.24 Commerce + 17.05 Finance - -1.16 Other services + 31.52 Total 100 Source: Author’s estimation based on data described in Section 2. In the case of Côte d’Ivoire, the very small per capita growth in 2002-2014 was due to the higher employment rate and, to a much lesser extent, structural change, while persistent decline in “within” productivity reduced growth. Of the 0.9 percent annual average growth in output per capita, acceleration in the employment rate contributed around 1.46 ppts while structural change made up 0.45 ppts. The substantial contribution of the employment rate, combined with the strongly negative productivity growth, suggests that increased production in Côte d’Ivoire since the early 2000s was not based on productivity increases but instead on increased accumulation of factors such as labor. The within component, however, made a significant negative contribution, offsetting almost all of the gains due to the employment rate and structural change. Table 7. Côte d’Ivoire: Sectoral decomposition of GDP per capita growth: 2002-2014 (Annual percentage changes) % Contribution of Within-sector Employment Inter-sectoral Total productivity rate Static Dynamic total (%) Agriculture -0.08 -0.07 0.49 0.06 0.56 0.41 Mining -0.03 0.01 0.08 0.00 0.08 0.06 Utilities -0.28 0.05 -0.04 -0.01 -0.04 -0.27 Manufacturing 0.33 -0.01 -0.09 0.00 -0.08 0.24 Construction -0.15 0.05 0.03 0.00 0.03 -0.06 Transport -0.19 0.08 0.10 0.00 0.09 -0.02 Commerce -0.81 0.98 -0.24 0.06 -0.19 -0.01 Finance 0.08 -0.01 -0.16 0.00 -0.16 -0.09 Other services -0.06 0.37 0.16 0.00 0.15 0.46 Subtotals -1.19 1.46 0.35 0.11 0.45 0.73 Demographic component 0.17 Annual % change in GVA per capita 0.896 Source: Author’s estimation based on data described in Section 2. Agriculture accounted for most of the growth attributed to structural change, while the higher employment rate could be largely explained by changes in commerce and ‘other services’. In addition, 20       the decline in within-sector productivity was driven by the commerce, utilities, and transport sectors. In particular, commerce accounted for 70 percent of the negative within-sector productivity growth. Manufacturing was the only sector that experienced relatively significant and positive productivity growth. All in all, ‘other services’, agriculture, and manufacturing explain most of the per capita growth since the early 2000s. Finally, the percent contribution of each sector to structural change is shown in Table 8. Agriculture played the most important role. Commerce and ‘other services’ contributed significantly, while manufacturing had a slight negative contribution. Table 8. Côte d’Ivoire: Sectoral contributions to inter-sectoral shifts 2002-2014 Direction of Contribution to employment inter-sectoral share shift shifts (%) Agriculture - 123.02 Mining + 17.74 Manufacturing - -9.42 Utilities - -18.02 Construction + 7.21 Transport + 20.86 Commerce + -40.82 Finance - -34.48 Other services + 33.93 Total 100 Source: Author’s estimation based on data described in Section 2. We found qualitatively similar results for 2008-2014. A third of Côte d’Ivoire’s average annual per capita growth in this period was due to structural change. In particular, static reallocation gains contributed over 20 percent of output per capita growth. The results suggest that improved economic performance since the late 2000s still has been driven by accelerated factor accumulation, while growth gains from structural change remained subdued. In addition, negative within-sector productivity growth held back growth. In general, the decomposition results indicate that plummeting within-sector productivity has been mainly responsible for the growth contraction in Côte d’Ivoire, but the movement of workers to higher productivity sectors contributed positively. A stocktaking of the experiences of BBC drives home several points, both commonalities and distinctive features. Economic growth in these economies was generally accompanied by substantial relocation of workers across sectors, despite the incommensurate changes in their output structures. Despite strong growth for about two decades, the pace of structural change in Burkina Faso lagged far behind those in the other two countries. Structural change played an important positive role in Benin and Burkina Faso, accounting for the bulk of (labor) productivity growth and GDP per capita growth. Although BBC experienced somewhat negligible or negative productivity growth within sectors, reduction of allocative inefficiencies was an important engine of growth in BBC. However, structural change has been dominated by static reallocation gains. The vast majority of workers leaving agriculture ended up in market services industries, such as retail trade and distribution, which are marked by low productivity and high informality. The missing contribution from manufacturing is one of the most notable features of these economies. In Benin and Côte d’Ivoire, within productivity loss reduced per capita growth while within gains contributed positively to Burkina Faso’s growth. In all three countries, the largest economic activity, namely agriculture, has registered comparatively low productivity increases while the second largest sector, commerce, has experienced productivity decline, which was most significant for Benin and Côte d’Ivoire. However, Benin took advantage of static productivity gains from labor reallocation earlier than Burkina Faso with a faster urbanization rate in the 1990s and early 2000s, although this 21       has faltered in recent years. Burkina Faso experienced productivity gains from shift of employment more recently, more slowly. Côte d’Ivoire in turn has yet to unleash significant productivity-enhancing structural change. In contrast to Burkina Faso and Côte d’Ivoire, which experienced dynamic gains, albeit very small, Benin has seen a dynamic loss that offset some of the static gains in productivity. In general, the pattern of structural change observed in BBC stands at variance with the Asian experience. In Asia, dynamic structural change has been the norm rather than the exception and dynamic losses were hardly observed. In contrast, productivity-enhancing structural change that shifts workers from low-productivity agriculture and informal services to dynamic high-productivity growth activities is yet to be unleashed in BBC. To place these results in context, we performed similar analysis for the Asian and African benchmark countries. Figure 8 summarizes the key findings. Three periods are considered to ensure meaningful comparisons: 1960-1975, 1975-1990, and 1990-2010. For the SSA countries, the analysis focuses on the period 1990-2010. We first note that within-sector productivity gains and structural change have been the main drivers of economic growth in Asia. At its peak, structural change accounted for about 35 percent of growth in Korea (1975-1990), 28 percent in China (1975-1990), and 31 percent in Indonesia (1990-2010). Thailand experienced a particularly phenomenal magnitude of structural change, accounting for about 60 percent of growth. Malaysia represents an anomaly and is the only Asian country in our sample that experienced growth-reducing structural change in 1975-1990, although this was reversed in 1990- 2010.34 We also note that the strong “within” productivity growth in Asia has been mostly due to pronounced productivity gains in manufacturing industries, particularly during their high-growth periods, and, to a lesser extent, agriculture. Service activities, and in particular commerce, also registered noticeable productive growth in the Asian economies, particularly in Indonesia (during 1990-2010) and Thailand (1975-1990). As opposed to this, the two largest sectors in BBC, namely agriculture and commerce, reported low within productivity gains. Table 9. Demographic trends: BBC and Selected Asian and SSA economies Population growth Working-age population Urban employment (% total population) (% total employment) 1975 1990 2010 1975 1990 2010 1975 1990 2010 Asia China 1.8 1.5 0.5 55.9 65.8 74.3 22.8 39.9 63.3 Indonesia 2.5 1.8 1.3 54.0 59.8 66.2 27.4 33.3 45.3 Korea, Rep. 2.0 1.5 0.5 58.4 69.4 72.7 55.5 81.3 93.1 Malaysia 2.3 2.8 1.6 54.1 59.3 67.7 54.3 72.4 86.4 Thailand 2.6 1.4 0.2 54.0 65.3 71.9 27.0 33.9 61.7 Africa Benin* 2.4 3.4 2.9 51.9 50.7 53.7 40.6 55.0 57.9 Burkina Faso‡ 1.9 2.7 3.0 52.8 49.5 51.2 11.9 15.7 20.8 Côte d’Ivoireǂ 4.6 3.5 2.2 52.6 52.5 53.4 37.2 39.7 47.5 Ethiopia 2.2 3.4 2.6 52.3 50.6 52.3 9.1 11.0 24.9 Ghana 2.3 2.7 2.5 51.4 53.4 57.4 46.1 46.3 58.4 Senegal 2.8 3.1 2.9 52.6 49.9 53.2 28.7 34.1 48.6 Tanzania 3.2 3.17 3.17 51.0 51.3 51.8 10.9 13.8 28.3 Note: *The data on the share of urban employment for Benin correspond to the periods 2006, 2010, and 2015. ‡ The data on the share of urban employment for Burkina Faso are for the periods 1998, 2005, and 2014. ǂThe data on the share of urban employment for Côte d’Ivoire are for the periods 2002, 2008, and 2014. Source: Author’s computation based on data described in Section 2.                                                                   34 However, for some of the countries, there was a limited role for structural change since the 1990s, which likely reflects the purported disappearance of inter-sectoral productivity gaps over the course of development. Although many of the Asian economies have experienced considerable structural change, with labor moving from manufacturing to service industries, this on its own has made little contribution to productivity growth. Overall productivity performance in these countries largely depends on how productivity fares in each individual sector.  22       However, BBC exhibit a pattern similar to that observed in SSA (Figure 9). Structural change has played an important role in the five African economies, but it has been driven static gains. Dynamic gains contributed significantly only in few counties and in some periods. Similar to BBC, within- sector productivity gains made a small contribution to growth. In general, Burkina Faso’s experience is broadly similar to those of Ethiopia and Tanzania in that static gains were dominant while the “within” effect was either small or negative. Benin exhibits patterns similar to Ghana and Senegal, with productivity growth mainly explained by “within” gains while the contribution of structural change was quite modest. To sum up, BBC did not experience growth-reducing structural change, unlike some African nations, but neither did they witness significant labor movements to high- productivity sectors, unlike the rapidly growing Asian economies. Our results are generally consistent with Timmer et al. (2014), who find that structural change contributed positively to Africa’s stronger growth since the 1990s but was dominated by static structural change. Finally, we briefly compare our results with those from other studies that include BBC.35 Our results for Benin and Burkina Faso are generally consistent with McMillan et al. (2014) and McMillan and Harttgen (2014), who show that structural change accounted for nearly half of the productivity growth in Africa since 2000.36 In a similar vein, Timmer et al. (2014) estimate the contribution of structural change to labor productivity growth at about 54 percent. UNCTAD (2014) similarly shows that structural change explains a third of the labor productivity growth in developing countries. However, Timmer and de Vries (2009) and Kucera and Roncolato (2012) find that within-sector productivity gains explain a greater portion of productivity growth than structural change, which is strikingly different from our key result that within-sector productivity changes at best made a negligible contribution and at worst reduced growth in BBC. Our findings, excluding those for Côte d’Ivoire, generally resonate with Timmer et al. (2014), who conclude that structural change has played a stronger role in African economies in recent decades but was dominated by static gains. Our results for Benin indicating static gains but dynamic losses are particularly in line with Timmer et al. (2014). 5. Concluding Remarks This paper documented the pace and nature of structural change in Benin, Burkina Faso, and Côte d’Ivoire (hereafter BBC). Specifically, we used data organized into nine sectors to shed light on output, employment, and labor productivity trends since 2000. Several key features stand out: (i) Burkina Faso staged the highest and most stable output growth, Benin posted modest growth, while Côte d’Ivoire experienced meager growth; (ii) Agriculture remains the largest sector, despite the moderate decline in its output share; (iii) Benin and Côte d’Ivoire have had drastic changes in their structure of employment while Burkina Faso experienced a smaller shift. Agriculture exhibited the largest loss in employment while low-productivity commerce burgeoned. Manufacturing’s share, however, remained stagnant for decades.; (iv) BBC have seen a youth bulge, indicating potential for a significant demographic dividend; (v) Participation and employment rates subsided in Benin and Burkina Faso while both increased in Côte d’Ivoire; and (vi) Labor productivity increased strongly in Burkina Faso, modestly in Benin, while it has been in the negative territory in Côte d’Ivoire. We found evidence suggesting that growth in Benin and Burkina Faso occurred on the back of static structural change, while structural change made a smaller contribution to growth in Côte d’Ivoire. However, in Benin, static gains waned in recent years and were partly offset by a dynamic loss in productivity. Specifically, during 2010-2015, within-sector productivity gains accounted for a                                                                   35Itis important to note that these studies do not use the same decomposition method we employed in the present analysis. Other sources of discrepancies between our findings and those of the above-mentioned studies include: country samples, time period covered by the analysis, level of sectoral disaggregation, and data sources. Note also that only very few of these studies disaggregate structural change into its static and dynamic sub-components. 36More specifically, McMillan et al. (2014) show that the bulk of the growth differentials between Africa and Latin America, on the one hand, and Asia, on the other, can be accounted for by differences in the pattern of structural change with labor moving from low-to high-productivity sectors in Asia, but in the opposite direction in Africa and Latin America. 23       staggering 2.5 ppts, static structural change yielded 0.4 ppts, whereas a dynamic loss reduced growth by about 0.4 ppts. In Burkina Faso, over 80 percent of the 3 percent growth per capita in 1998-2014 can be explained by structural change, the bulk of which has been due to static gains. Côte d’Ivoire’s very modest economic growth since the early 2000s has been driven by the increased employment rate. In general, lower within-sector productivity reduced growth in BBC. Demographic changes contributed positively in BBC, reflecting the potential for a substantial demographic dividend.37 Benchmarking analysis reveals that, unlike the case with BBC, growth in Asian countries has been propped up by within-sector productivity improvements and structural change. The pattern of structural change observed in BBC is particularly characterized by the sheer absence of labor movements to high-productivity sectors. BBC seem to be bypassing the industrialization stage that propelled growth in East Asia. However, this is not a feature unique to BBC but one that is well underway in the rest of SSA. For the most part, SSA economies are experiencing falling manufacturing shares in employment and output in recent decades. Asian economies brought about unprecedented economic success by pursuing the well-trodden path of industrialization first (shifting workers from agriculture to high-productivity manufacturing) and then deindustrialization (with workers leaving manufacturing and moving into services). Western countries generally pursued this strategy to climb up the rungs of the economic ladder. Manufacturing, at its peak, employed 25-45 percent of the labor force in countries like the United Kingdom and the United States before deindustrialization set in. Similarly, Asian countries like China, Vietnam, and Korea experienced rapid manufacturing expansion in their respective take-off periods, with manufacturing employment peaking at about 30 percent in Korea in the 1990s (Lin, 2011). The challenge of achieving sustained rapid growth in BBC based on a service-driven growth model is three-pronged. First, the most rapidly expanding sectors are relatively unproductive non-tradable service sectors such as wholesale and retail trade, which exhibit prevalent and growing informality. Second, although some modern service industries (such as finance and IT) tend to be high-productivity and tradable, they are typically skill- and capital-intensive and thus lack the capacity to absorb the surplus predominantly unskilled labor force. Third, most of the remaining service activities either lack technological dynamism or are inherently non-tradable. However, there is a growing contention that the classic path of catch-up via manufacturing-led industrialization may be inaccessible for developing countries today for a wide array of reasons.38 The newly-emerging unorthodox view of structural change emphasizes the contribution of services (Ghani and O’Connell, 2014) and productivity increases within sectors and within firms to growth and job creation. Technological advances, the argument goes, expand the tradability of services and thus low-income countries could grow faster via exploiting their revealed comparative advantage within services. However, according to Amirapu and Subramanian (2015), any sector can lead an economy to rapid and sustained growth if it fulfills the following five important criteria: (i) a high level of productivity, (ii) “dynamic” productivity growth, (iii) expansion of the sector in terms of its use of inputs, (iv) comparative advantage, or alignment between resource requirements of the sector and resource endowments of the country, and (v) exportability. In BBC, only some service activities (including                                                                   37However, the economic benefits of a demographic dividend do not transpire automatically, and it is critical to equip the youth with adequate skills and to support activities in which the bulk of unskilled workers can be productively employed than in subsistence agriculture and informal services. 38These reasons include the following. To begin with, East Asian countries, notably China and its successors such as Vietnam and Cambodia, pose formidable competitive challenges to developing countries aspiring to jump-start structural change through unleashing light manufacturing, particularly in light of globalization and the ensuing reduction in trade barriers virtually elsewhere. In addition, new trade rules are narrowing the room for industrial policies– local content requirements, subsidies, import restrictions, which were aggressively deployed by Asian countries during their respective growth heydays (Rodrik, 2013b, 2014). Further, technological development in recent decades have made manufacturing itself more capital- and skill-intensive than in the past, making poor economies less-positioned to exploit opportunities in some manufacturing industries and reducing the scope for labor absorption in the sector. There are, however, counter‐arguments to this claim. BBC could possibly diversity into manufacturing by capitalizing on their rich natural resources, low cost of labor, access to sizable domestic and regional markets, and privileged access to high-income markets for exports, all of which could offset the low labor productivity of these economies relative to their Asian competitors (Dinh et al., 2014; Zafar, 2016). See also Rodrik (2013) and Amirapu and Subramanian (2015). 24       notably finance, insurance, and real estate) tend to feature many of the virtues conventionally associated with manufacturing (such as high productivity and unconditional convergence in labor productivity). However, these sectors are all too skill-intensive and thus unaligned with the comparative advantages of these countries at current stages of development. By contrast, BBC’s rapidly expanding sectors, particularly trade and distribution services, fail to exhibit most of these features, which casts serious doubt on the sustainability of the existing growth model.39 Nonetheless, the exceptionally high degree of inter-sectoral labor mobility, particularly in Benin and Côte d’Ivoire, offers a tremendous opportunity for lifting growth through structural change. Sustained rapid growth in BBC would require reversing current trends and significantly improving productivity within sectors, notably in the agriculture and service activities. Although there is no magic bullet to achieve sustained high growth, and the path towards prosperity tends to be context-specific, theory and experience suggest that long-term growth is driven by increased productivity. As Nobel Laureate Paul Krugman famously quipped, “Productivity is not everything, but in the long-run it is almost everything.” Given the large size of the agriculture and services sectors, it is imperative that considerable endeavors are geared toward improving the productivity of these sectors. Agriculture remains to the largest contributor to total output and is the biggest employer in all three economies, while the service sector is the largest in terms of contribution to output and the second largest employer. As the vast majority of BBC’s labor force will continue to work in agriculture and services, substantial labor productivity improvements in these sectors are critical toward improving living standards. It is also critical to support relatively high-productivity services industries in which substantial numbers of the unskilled workers leaving agriculture can be productively employed. References Agénor, Pierre-Richard, and Hinh T. Dinh (2013). ‘Public Policy and Industrial Transformation in the Process of Development’. Research Working Paper 6405. Washington, DC: World Bank. Amirapu, A. and Subramanian, A. (2015). Manufacturing or Services? An Indian Illustration of a Development Dilemma. Center for Global Development, Working Paper 409. Dinh, H. T., Palmade, V., Chandra, V., and Cossar, F. (2012). Light Manufacturing in Africa: Targeted Policies to Enhance Private Investment and Create Jobs. Washington, DC: World Bank. Dossou, A. S., Sinzogan, J.-Y, Mensah, S. (2005). Economic growth in Benin: lost opportunities. In The political economy of economic growth in Africa, Ndulu, B. J., O.Connell, S., Azam, J. P., Bates, R. H., Fosu, A. K., Gunning, J.W., Njinkeu, D. (Eds.). Cambridge: Cambridge University Press. Gala, P. (2007). Real exchange rate levels and economic development: Theoretical analysis and econometric evidence. Cambridge Journal of Economics, 32(2): 273–288. Ghani, E., and O’Connell, S. (2014). Can service be a growth escalator in low income countries? Policy Research Working Paper 6971. Washington, D.C.: World Bank. Haile, F. (2016). Growth and Structural Change in Tanzania: Retrospect and Prospect. Background paper prepared for the World Bank’s Systematic Country Diagnostics for Tanzania. World Bank: Dar es Salaam. International Monetary Fund. (2016). Côte d’Ivoire: Article IV consultation 2014. IMF Country Report No.16/147. June. Washington, D.C.: IMF Jones, B., and Olken, B. (2008). The anatomy of start-stop growth. Review of Economics and Statistics, 90(3): 582-587. Kapsos, S. (2005). The employment intensity of growth: Trends and macroeconomic determinants. International Labour Office, Employment Strategy Paper 2005/12.                                                                   39It should be noted that an expansion of services does not necessarily stand in the way of successful structural transformation and growth so long as an economy has accumulated human capital and other fundamental capabilities necessary for making these services sectors high-productivity activities. However, these are features that developing economies would typically have later in the development process, i.e. in post-industrialization stages (Rodrik, 2013b; Amirapu and Subramanian, 2015). 25       Khan, A. (2001). Employment policies for poverty reduction. International Labour Office, Issues in Employment and Poverty, Discussion Paper 1. Kucera, D. and l. Roncolato (2012) ‘Structure Matters: Sectoral Drivers of Growth and the Labour Productivity-Employment Relationship’, International Labour Office, Research Paper 3. Lin, J. (2011). From flying geese to leading dragons: New opportunities and strategies for structural transformation in developing countries. Policy Research Working Paper 5702, World Bank, Washington, DC. Martins, P. (2014). Structural change in Ethiopia – An employment perspective. World Bank Policy Research Working Paper 6749. Washington, D.C.: World Bank. McMillan, M., and Harttgen, K. (2014). What is Driving the ‘African Growth Miracle’? African Development Bank, Working Paper 209. McMillan, M., Rodrik, D., and Verduzco-Gallo, I. (2014). Globalization, structural change, and productivity growth. World Development, 63: 11-32. Rodrik, D. (2013a). Unconditional convergence in manufacturing. Quarterly Journal of Economics, 128(1): 165-204. Rodrik, D. (2013b). Structural Change, Fundamentals, and Growth: An Overview. Mimeo Princeton University. Available at http://drodrik.scholar.harvard.edu/publications/structural-change- fundamentals-and-growth-overview. Rodrik, D. (2014). An African growth miracle? NBER Working Paper No. 20188, June. Washington, D.C.: NBER. Rodrik, D. (2015). Premature deindustrialization. NBER Working Paper No. 20935. Washington, D.C.: NBER. Timmer, M., and de Vries, G. (2009). Structural change and growth accelerations in Asia and Latin America: A new Sectoral data set. Cliometrica, 3(2): 165-190. Timmer, P., and Akkus, S. (2008). The structural transformation as a pathway out of poverty: Analytics, empirics and politics. Working Paper 150. Center for Global Development. Timmer, M., de Vries, G., and de Vries, K. (2014). Patterns of structural change in developing countries. Research Memorandum 149, Groningen Growth and Development Centre. Timmer, M. P., de Vries, G. J., & de Vries, K. (2015). Patterns of Structural Change in Developing Countries. In J. Weiss, & M. Tribe (Eds.), Routledge Handbook of Industry and Development. (pp. 65-83). Routledge. UNCTAD (2014). Least Developed Countries Report 2014 – Growth with Structural Transformation: A Post-2015 Development Agenda. Geneva: United Nations Conference on Trade and Development. Zafar, A (Ed.). (2016). Industrialization in the CFA zone in Sub-Saharan Africa: The Last Frontier. Washington, D.C.: The World Bank. World Bank (1994). Adjustment in Africa: Reforms, results and the road ahead. Oxford: Oxford University Press. World Bank 2009. Benin: Constraints to Growth and Potential for Diversification and Innovation: Country Economic Memorandum, Report 48233-B. Washington, D.C.: World Bank. World Bank (2012). Job Generation and Growth Decomposition Tool: Understanding the Sectoral Pattern of Growth and its Employment and Productivity Intensity. Reference Manual and User’s Guide, Version 1.0. Washington, D.C.: World Bank. World Bank (2015). Côte d’Ivoire: From crisis to sustained growth priorities for ending poverty and boosting shared prosperity. Systematic Country Diagnostic. Washington, D.C.: World Bank. World Bank (2017, forthcoming). Benin: Systematic Country Diagnostic, Concept Note. Washington, D.C.: World Bank. 26       Appendix Appendix Table 1: Gross Value Added (GVA) by Sector Benin GVA by sector GVA by sector GVA by sector (Constant 2007 CFAF billion) (% Total GVA) (Annual growth, %) 2006- 2010- 2006- 2006 2010 2015 2006 2010 2015 2010 2015 2015 Agriculture 662.4 727.6 783.9 26.9 26.1 22.3 2.3 1.5 1.9 Mining 11.5 11.8 14.3 0.5 0.4 0.4 0.6 3.8 2.4 Manufacturing 495.0 410.8 515.9 20.1 14.7 14.7 -4.7 4.6 0.5 Utilities 10.7 17.2 40.2 0.4 0.6 1.1 11.9 17.0 14.7 Construction 192.8 220.6 280.8 7.8 7.9 8.0 3.4 4.8 4.2 Commerce 332.8 348.6 473.0 13.5 12.5 13.5 1.2 6.1 3.9 Transport 190.2 267.1 352.5 7.7 9.6 10.0 8.5 5.5 6.9 Finance 36.6 100.8 195.8 1.5 3.6 5.6 25.3 13.3 18.6 Other services 526.8 681.4 854.0 21.4 24.5 24.3 6.4 4.5 5.4 Total 2458.8 2785.9 3510.4 100.0 100.0 100.0 3.1 4.6 4.0 Burkina Faso GVA by sector GVA by sector GVA by sector (Constant 1999 CFAF billion) (% Total GVA) (Annual growth, %) 1998- 2005- 1998- 1998 2005 2014 1998 2005 2014 2005 2014 2014 Agriculture 533.3 751.2 989.4 32.9 31.6 25.9 4.9 3.1 3.9 Mining 7.8 12.7 187.7 0.5 0.5 4.9 6.9 29.9 19.8 Manufacturing 276.5 310.2 371.4 17.0 13.0 9.7 1.6 2.0 1.8 Utilities 26.2 39.2 86.7 1.6 1.6 2.3 5.7 8.8 7.5 Construction 106.5 138.4 222.3 6.6 5.8 5.8 3.7 5.3 4.6 Commerce 147.1 269.6 453.8 9.1 11.3 11.9 8.7 5.8 7.0 Transport 80.3 113.5 346.3 5.0 4.8 9.0 4.9 12.4 9.1 Finance 25.6 57.2 130.5 1.6 2.4 3.4 11.5 9.2 10.2 Other services 419.4 686.6 1039.3 25.8 28.9 27.2 7.0 4.6 5.7 Total 533.3 751.2 989.4 100.0 100.0 100.0 5.5 5.3 5.4 Côte d’Ivoire GVA by sector GVA by sector GVA by sector (Constant 2009 CFAF billion) (% Total GVA) (Annual growth, %) 2002- 2008- 2002- 2002 2008 2014 2002 2008 2014 2008 2014 2014 Agriculture 2245.8 2507.4 2797.6 23.8 24.6 20.8 1.84 1.83 1.83 Mining 289.9 561.1 540.6 3.1 5.5 4.0 11.00 -0.62 5.19 Manufacturing 1745.0 1841.1 2170.0 18.5 18.0 16.1 0.89 2.74 1.82 Utilities 51.9 69.4 459.3 0.6 0.7 3.4 4.84 31.49 18.16 Construction 231.0 163.7 232.1 2.5 1.6 1.7 -5.75 5.83 0.04 Commerce 1385.6 1280.9 1668.2 14.7 12.6 12.4 -1.01 3.94 1.47 Transport 1009.1 759.5 1473.0 10.7 7.4 10.9 -4.74 11.04 3.15 Finance 414.8 514.5 499.0 4.4 5.0 3.7 3.59 -0.51 1.54 Other services 2051.8 2503.9 3618.9 21.8 24.5 26.9 3.32 6.14 4.73 Total 9424.9 10201.5 13458.7 100.0 100.0 100.0 1.35 4.56 2.96 27       Appendix Table 2: Employment by sector Benin Employment by Employment by Employment by sector sector (thousands) sector (% total) (annual growth, %) 2006- 2010- 2006- 2006 2010 2015 2006 2010 2015 2010 2015 2015 Agriculture 1612.0 1432.1 1503.7 59.4 45.0 42.1 -2.8 1.0 -0.7 Mining 2.2 2.9 3.2 0.1 0.1 0.1 8.1 2.3 5.3 Utilities 4.6 5.8 6.8 0.2 0.2 0.2 6.3 3.4 5.2 Manufacturing 197.6 240.5 277.2 7.3 7.6 7.8 5.4 3.1 4.5 Construction 61.1 81.6 95.9 2.3 2.6 2.7 8.4 3.5 6.3 Commerce 518.4 882.7 1025.6 19.1 27.7 28.7 17.6 3.2 10.9 Transport 89.6 135.3 160.8 3.3 4.3 4.5 12.8 3.8 8.8 Finance 5.4 9.0 7.9 0.2 0.3 0.2 16.7 -2.4 5.2 Other services 222.1 393.1 487.4 8.2 12.4 13.7 19.3 4.8 13.3 Total 2712.9 3183.0 3568.5 100 100 100 4.3 2.4 3.5 Burkina Faso Employment by Employment by Employment by sector sector (thousands) sector (% total) (annual growth, %) 1998- 2005- 1998- 1998 2005 2014 1998 2005 2014 2005 2014 2014 Agriculture 4010.6 4336.3 5794.8 88.1 84.3 79.2 1.2 3.7 2.8 Mining 7.0 11.4 44.5 0.2 0.2 0.6 8.9 32.3 33.4 Manufacturing 102.6 103.0 228.1 2.3 2.0 3.1 0.1 13.5 7.6 Utilities 4.4 10.3 5.3 0.1 0.2 0.1 19.2 -5.4 1.3 Construction 30.1 34.8 91.4 0.7 0.7 1.2 2.3 18.0 12.7 Commerce 222.9 368.1 712.8 4.9 7.2 9.7 9.3 10.4 13.7 Transport 19.8 33.9 77.3 0.4 0.7 1.1 10.2 14.2 18.1 Finance 16.8 36.6 23.2 0.4 0.7 0.3 16.9 -4.1 2.4 Other services 136.9 210.2 343.7 3.0 4.1 4.7 7.7 7.1 9.4 Total 4551.1 5144.7 7321.2 100.0 100.0 100.0 1.9 4.7 3.8 Côte d’Ivoire Employment by Employment by Employment by sector sector (thousands) sector (% total) (annual growth, %) 2002- 2008- 2002- 2002 2008 2014 2002 2008 2014 2008 2014 2014 Agriculture 4233.1 4644.7 5490.6 62.8 60.3 52.5 1.4 2.0 1.9 Mining 9.8 7.6 19.9 0.1 0.1 0.2 -3.2 18.1 6.5 Manufacturing 318.4 391.7 475.7 4.7 5.1 4.5 3.3 2.4 3.1 Utilities 25.2 11.9 24.6 0.4 0.2 0.2 -7.5 11.8 -0.1 Construction 64.7 77.3 141.7 1.0 1.0 1.4 2.8 9.3 7.4 Transport 180.4 225.0 320.1 2.7 2.9 3.1 3.5 4.7 4.8 Commerce 1068.8 1389.8 2482.2 15.9 18.1 23.7 4.3 8.7 8.3 Finance 21.8 35.2 21.3 0.3 0.5 0.2 8.8 -4.4 -0.1 Other services 817.1 913.8 1485.0 12.1 11.9 14.2 1.7 6.9 5.1 Total 6739.2 7697.2 10461.3 100.0 100.0 100.0 2.0 4.0 3.5 28       Appendix Table 3. Labor productivity by sector Benin GVA per worker by GVA per worker by sector (constant CFAF, sector (annual Employment millions) growth, %) Elasticity 2006- 2010- 2006- 2006- 2010- 2006- 2006 2010 2015 2010 2015 2015 2010 2015 2015 Agriculture 0.41 0.51 0.52 5.9 0.52 3.0 -1.13 0.65 -0.37 Mining 5.31 4.12 4.47 -5.6 1.71 -1.8 12.35 0.55 1.96 Utilities 2.33 2.98 5.95 7.1 19.91 17.3 0.42 0.13 0.17 Manufacturing 2.51 1.71 1.86 -8.0 1.78 -2.9 -1.28 0.60 9.55 Construction 3.16 2.70 2.93 -3.6 1.66 -0.8 2.33 0.64 1.25 Commerce 0.64 0.39 0.46 -9.6 3.36 -3.1 14.80 0.45 2.32 Transport 2.12 1.97 2.19 -1.8 2.22 0.4 1.26 0.59 0.93 Finance 6.76 11.16 24.70 16.3 24.25 29.5 0.38 -0.13 0.11 Other services 2.37 1.73 1.75 -6.7 0.22 -2.9 2.62 0.95 1.92 Total 0.91 0.88 0.98 -0.9 2.48 0.01 1.30 0.47 0.74 Burkina Faso GVA per worker by GVA per worker by sector (constant CFAF, sector (annual Employment millions) growth, %) Elasticity 1998- 2005- 1998- 1998- 2005- 1998- 1998 2005 2014 2005 2014 2014 2005 2014 2014 Agriculture 0.133 0.173 0.171 4.3 -0.16 0.02 0.20 1.06 0.52 Mining 1.12 1.12 4.22 0.0 30.82 0.17 1.00 0.21 0.23 Manufacturing 2.69 3.01 1.63 1.7 -5.10 -0.02 0.03 6.15 3.56 Utilities 5.97 3.81 16.32 -5.2 36.46 0.11 2.71 -0.40 0.09 Construction 3.54 3.97 2.43 1.7 -4.31 -0.02 0.53 2.68 1.88 Commerce 0.66 0.73 0.64 1.6 -1.45 0.00 0.78 1.37 1.05 Transport 4.06 3.35 4.48 -2.5 3.76 0.01 1.72 0.62 0.88 Finance 1.52 1.56 5.61 0.3 28.82 0.17 0.96 -0.28 0.09 Other services 3.06 3.27 3.02 0.9 -0.82 0.00 0.84 1.24 1.02 Total 0.36 0.46 0.52 4.2 1.45 0.03 0.28 0.69 0.45 Côte d’Ivoire GVA per worker by GVA per worker by sector (constant CFAF, sector (annual Employment millions) growth, %) Elasticity 2002- 2008- 2002- 2002- 2008- 2002- 2002 2008 2014 2005 2014 2014 2005 2014 2014 Agriculture 0.53 0.54 0.51 0.3 -0.62 0.00 0.83 1.57 1.21 Mining 29.67 74.13 27.16 21.4 -7.04 -0.01 -0.24 -44.52 1.20 Manufacturing 5.48 4.70 4.56 -2.0 -0.33 -0.01 4.18 1.20 2.03 Utilities 2.06 5.82 18.65 26.0 24.47 0.50 -1.56 0.19 0.00 Construction 3.57 2.12 1.64 -5.8 -2.51 -0.03 -0.67 1.99 28.18 Transport 5.59 3.37 4.60 -5.7 4.04 -0.01 -1.00 0.45 1.69 Commerce 1.30 0.92 0.67 -4.1 -3.01 -0.03 -3.98 2.60 6.48 Finance 19.05 14.63 23.44 -3.3 6.69 0.01 2.56 13.11 -0.11 Other services 2.51 2.74 2.44 1.3 -1.23 0.00 0.54 1.40 1.07 Total 1.40 1.33 1.29 -0.7 -0.33 -0.01 1.73 1.12 1.29 29       Appendix Table 4. Benin: Sectoral decomposition of GDP per capita growth: 2010-2015 (Annual percentage changes) % Contribution of Within-sector Employment Inter-sectoral Total productivity rate Static Dynamic total (%) Agriculture 0.13 -1.00 0.32 -0.06 0.27 -0.60 Mining 0.01 0.00 0.00 0.00 0.00 0.01 Utilities 0.12 0.00 0.01 0.00 0.01 0.13 Manufacturing 0.26 -0.02 0.03 0.01 0.04 0.28 Construction 0.13 0.00 0.05 0.00 0.05 0.19 Commerce 0.42 -0.05 -0.05 -0.07 -0.11 0.26 Transport 0.22 0.01 0.06 0.01 0.07 0.30 Finance 0.78 -0.02 -0.24 0.00 -0.24 0.52 Other services 0.06 0.16 0.24 0.00 0.24 0.45 Subtotals 2.13 -0.91 0.42 -0.10 0.32 1.54 Demographic component 0.53 Annual % change in GVA per capita 2.07 Source: Author’s estimation based on data described in Section 2.   Appendix Table 5. Burkina Faso: Sectoral decomposition of GDP per capita growth: 2005-2014 (Annual percentage changes) % Contribution of Within-sector Employment Inter-sectoral Total productivity rate Static Dynamic total (%) Agriculture -0.05 -1.27 0.35 0.06 0.41 -0.91 Mining 0.32 0.04 0.21 0.00 0.21 0.58 Utilities -0.89 0.12 0.52 -0.01 0.51 -0.26 Manufacturing 0.43 -0.02 -0.31 0.00 -0.31 0.11 Construction -0.37 0.06 0.39 0.00 0.39 0.08 Commerce -0.20 0.25 0.13 0.00 0.12 0.18 Transport 0.24 0.04 0.34 0.00 0.34 0.63 Finance 0.52 -0.05 -0.31 0.00 -0.31 0.16 Other services -0.27 0.04 0.41 -0.01 0.40 0.18 Subtotals -0.26 -0.78 1.74 0.04 1.78 0.74 Demographic component 1.796 Annual % change in GVA per capita 2.53 Source: Author’s estimation based on data described in Section 2.   Appendix Table 6. Cote d’Ivoire: Sectoral decomposition of GDP per capita growth: 2008-2014 (Annual percentage changes) % Contribution of Within-sector Employment Inter-sectoral Total productivity rate Static Dynamic total (%) Agriculture -0.23 0.06 0.69 0.16 0.84 0.67 Mining -0.93 0.02 0.62 0.00 0.62 -0.29 Utilities -0.09 0.03 -0.24 0.00 -0.24 -0.31 Manufacturing 0.34 0.02 0.12 0.00 0.12 0.48 Construction -0.08 0.09 0.03 0.00 0.03 0.04 Transport 0.50 0.10 0.04 0.01 0.05 0.66 Commerce -0.71 1.56 -0.50 0.10 -0.40 0.45 Finance 0.40 -0.04 -0.62 0.00 -0.62 -0.25 Other services -0.54 0.76 0.44 -0.04 0.40 0.62 Subtotals -1.34 2.61 0.58 0.23 0.81 2.07 Demographic component 0.38 Annual % change in GVA per capita 2.45 Source: Author’s estimation based on data described in Section 2. 30       Appendix Figure 1: Sectoral GVA (% total GVA) and GVA growth 31       Appendix Figure 1. Sectoral employment shares and employment growth 32       Appendix Figure 2: Average labor productivity and productivity growth 33       Appendix Figure 4. Labor productivity gaps in Benin, Burkina Faso, and Côte d’Ivoire 34       Appendix Figure 5. Correlation between sector productivity and change in employment shares 35       Appendix Figure 6. Correlation between sector productivity and change in employment shares (1975-2010) 36       Appendix Figure 7. Correlation between sectoral productivity and change in employment shares (1975-2010) Tanzania 2.5 Log(Sectoral productivity/Total Con 2.0 Fin y = 6.99x + 1.08 Util Tran 1.5 Min productivity) R² = 0.27 Manuf 1.0 Commerce 0.5 Other services 0.0 -0.3 -0.2 -0.2 -0.1 -0.1 -0.5 0.0 0.1 0.1 0.2 Agri -1.0 -1.5 Change in employment share 37       Appendix Figure 8. Shapley decomposition of per capita growth in Asia 38       Appendix Figure 9. Shapley decomposition of per capita growth in SSA   Source: Author’s estimation based on data described in Section 2. 39