POVERTY & EQUITY NOTES FEBRUARY 2020 · NUMBER 19 Structural Transformation in Sub-Saharan Africa Gildas Deudibe, Joshua Merfeld, Justin Ndoutamou, and David Newhouse Economic development is typically accompanied by the movement of labor from agriculture to the non-agricultural sector, a pattern commonly referred to as structural transformation. This note aims to better understand the ongoing structural transformation in Sub-Saharan Africa). It finds that: (i) The structural transformation is occurring more slowly, and is much less variable across countries, than prevailing estimates suggest. (ii) There is a weak relationship between initial agricultural employment shares and the pace of transformation, suggesting little convergence across regions. (iii) Movement out of agricultural employment is clearly but only modestly correlated with poverty reduction. Economic development is typically accompanied by estimates from SSAPOV in the same years. We restrict large movements of labor out of agriculture into attention to countries with greater than 15 million non-farm employment, a pattern commonly referred people, for which there are two separate SSAPOV to as the structural transformation. Non-farm surveys that ask about the sector of primary employment tends to be less risky, less prone to employment over the previous seven days, the same underemployment, and more productive than recall period on which the ILO estimates are based. agricultural employment. Movement out of agriculture and into non-farm employment is often a There is significant heterogeneity in the speed of the sign of both economic development and welfare structural transformation across SSA countries. The improvements. However, commonly used measures variance in the ILO modeled estimates is particularly of structural transformation come from modeled striking, ranging from an annual decrease of 4.4 estimates – not household survey data – and little is percentage points of the share of agricultural known about the accuracy of these estimates. employment in Burkina Faso to a 1.3 percentage Because of this, it is not entirely clear how much the point per year increase in Uganda. The range of current structural transformation is reducing poverty. SSAPOV estimates is considerably smaller, as the standard deviation for the ILO estimates is more than Table 1 compares modeled estimates from the 40 percent smaller than that of the SSAPOV estimates International Labor Organization (ILO) with estimates computed from SSAPOV, a set of harmonized On average, the SSAPOV estimates are more than 0.5 nationally representative household surveys from percentage points per year smaller in magnitude Sub-Saharan Africa (SSA). The ILO estimates mainly than the ILO estimates, equivalent to a 60 percent rely on GDP growth to project changes in sectoral decline in magnitude. (Table 1, bottom row) The employment patterns since the last survey. SSAPOV, SSAPOV estimates suggest that the structural in contrast, is based solely on surveys and is transformation is proceeding much more slowly than frequently updated, because it is the source of the the ILO estimates would suggest. The country- Bank’s official poverty estimates for SSA countries. specific estimates differ greatly in some cases, as the discrepancy between the SSAPOV estimates and the We compare ILO estimates in the change of ILO estimates exceeds one percentage point in eight agricultural employment with household-level survey countries. The direction of the travel by the SSAPOV Table 1: A Snapshot of Structural Transformation in large SSA countries country year1 year2 Share of employment in agricultural ILO estimates of share of employment in sector from SSAPOV agricultural sector Year 1 Year 2 Annual Year 1 Year 2 Annual reduction reduction Angola 2008 2014 46.9% 45.2% -0.3 p.p. 44.6% 49.4% 0.8 p.p. Burkina Faso 2009 2014 85.1% 82.4% -0.4 p.p. 52.2% 30.4% -4.4 p.p. Cameroon 2007 2014 64.0% 50.3% -2.0 p.p. 59.8% 47.6% -1.7 p.p. Congo Dem. Rep. 2004 2012 73.4% 78.9% 0.7 p.p. 72.4% 70.7% -0.2 p.p. Ethiopia 2010 2015 72.6% 73.5% 0.2 p.p. 73.9% 68.9% -1.0 p.p. Ghana 2012 2016 49.0% 46.2% -0.7 p.p. 46.8% 34.7% -3.0 p.p. Kenya 2005 2015 66.3% 52.3% -1.4 p.p. 61.1% 58.3% -0.3 p.p. Liberia 2014 2016 20.4% 18.5% -1.0 p.p. 45.7% 46.5% 0.4 p.p. Madagascar 2005 2012 81.6% 81.4% 0.0 p.p. 82.0% 68.9% -1.9 p.p. Mozambique 2008 2014 83.3% 77.1% -1.0 p.p. 77.8% 73.0% -0.8 p.p. Malawi 2010 2016 83.8% 81.3% -0.4 p.p. 73.3% 72.2% -0.2 p.p. Nigeria 2010 2012 48.0% 48.8% 0.4 p.p. 40.8% 39.3% -0.7 p.p. Uganda 2012 2016 76.5% 79.9% 0.9 p.p. 66.1% 71.4% 1.3 p.p. South Africa 2014 2016 4.7% 5.4% 0.3 p.p. 4.7% 5.6% 0.5 p.p. Zambia 2010 2015 69.9% 61.0% -1.8 p.p. 64.2% 54.7% -1.9 p.p. Average -0.35 pp -0.88 pp Standard deviation 0.87 pp 1.51 pp Source: SSAPOV database, Sub-Saharan Africa Team for Statistical Development, World Bank, Washington DC and World Development Indicators (which reports ILO estimates). reinforces the importance of distinguishing between and the ILO estimates differs for a number of modeled and actual estimates when examining countries, including Angola, Liberia, Congo, Ethiopia, changes in sectoral employment composition. and Nigeria. We next consider the extent to which rates of Figure 1 compares the ILO and SSAPOV estimates agricultural employment are converging across graphically. The blue line is the 45-degree line, regions. One possibility is that urbanization is representing perfect agreement between the occurring in peri-urban areas with a medium level of SSAPOV and ILO estimates. Zambia, Cameroon, initial agricultural employment, leading to increased Mozambique, Malawi, South Africa, and Uganda are polarization in agricultural employment between located relatively close to this line. However, the remote rural areas and growing cities. Figure 2 sheds other countries lie quite far from it, especially Burkina light on the extent of convergence in agricultural Faso, Ghana, and Madagascar. The green line employment shares across countries in SSA, by using represents the actual correlation between the two the SSAPOV data to compare the speed of structural estimators, which is r=0.371. Squaring the correlation transformation with initial shares of agricultural implies that the ILO estimates explains just 14 employment. An important advantage of using the percent of the variation in the SSAPOV estimates. SSAPOV data is the ability to conduct sub-national This demonstrates the considerable difference analyses. Panel A presents a simple scatter plot and between the ILO and SSAPOV estimates, and line of best fit – weighted by population – at the February 2020 · Number 19 2 country level, Panel B at the country-region level, and Given the importance of this result, Table 2 presents Panel C at the country-district level. These three results from a set of related regressions. Columns one figures show that the level of measurement matters. and two present results at the country level, columns At the country level, the evidence suggests three through five at the country-region level, and divergence, with more agricultural countries actually columns six through eight at the country-district increasing their share of agricultural employment. level. Given that the sample contains only 11 countries with both requisite employment and Figure 1: SSAPOV vs ILO Estimates poverty data, there is insufficient data to draw a meaningful conclusion at the country level, whether the estimates are unweighted or weighted by country population (column two). Figure 2: Pace of Reduction by Initial Employment When examined across districts, however, this relationship greatly weakens and if anything, agricultural employment is converging. There is little evidence that districts with a medium amount of initial agricultural employment systematically saw greater reductions in agricultural employment. Columns three through five disaggregate the data to Furthermore, the overall linear point estimate is quite the region level, while columns six through eight small, indicating that the rate of convergence is slow. disaggregate the data to the district level. Our In other words, there is at best weak evidence that preferred specifications are columns five and eight, agricultural employment rates are converging across which isolate within-country variation in both districts in SSA at any level. variables. The coefficients are relatively similar in both columns. The coefficient in column eight is A final set of results explores an important dimension statistically significant but only moderately large. of structural transformation: poverty reduction. Because the SSAPOV data comprise a multitude of Within countries, a decrease in primary agricultural different indicators, they are well suited to analyzing employment of a percentage point in a district is how changes in agricultural employment correlate associated with a decrease in the poverty rate of with other household outcomes, like poverty. Figure approximately 0.24 percentage points. Reducing 3 presents scatter plots and lines of best fit for the agricultural employment can contribute to poverty relationship between the speed of structural reduction, but the quality of non-agricultural transformation and the speed of poverty reduction, employment matters, and increasing workers’ again at three separate levels of aggregation. In all productivity on and off the farm remains essential. three cases, reductions in agricultural employment are associated with reductions in poverty. February 2020 · Number 19 3 Table 2: Annual Change in Ag. Share and Poverty Rate DV: Change in (1) (2) (3) (4) (5) (6) (7) (8) poverty rate Country Country Region Region Region District District District (annual change) Annual change 1.000* 1.317 0.552* 0.515 0.285 0.258** 0.235 0.210** in ag. share (0.545) (0.856) (0.291) (0.430) (0.311) (0.117) (0.131) (0.069) Weighted by No Yes No Yes Yes No Yes Yes population? Country FE No No No No Yes No No Yes Observations 11 11 80 80 80 230 230 230 Note: Standard errors are in parentheses and are clustered at the country level (columns three through eight). Columns one and two are at the country level. Columns three through five are at the country-region level. Columns six through eight are at the country-district level. Source: SSAPOV database, Sub-Saharan Africa Team for Statistical Development, World Bank, Washington DC. * p<0.1 ** p<0.05 *** p<0.01 Figure 3: Changes in Ag. Share vs. Poverty Rate typically used. There is at best a weak relationship that reduction in agricultural employment is occurring in both urban and rural districts. Finally, the association between movement out of agriculture and poverty reduction is robust but only moderately strong. The evidence presented here is descriptive rather than causal, since choice of sector is itself a decision taken by workers. Future work can utilize these data to better understand which workers choose to work in agriculture and why. Additional analysis can examine how the quality of structural transformation varies across countries, by looking at how growth in different types of non-agricultural employment relates to poverty reduction. To recap, this note aims to better understand the ABOUT THE AUTHORS extent of structural transformation in Sub-Saharan Gildas Deudibe is a doctoral student at Clermont- Africa and its role in poverty reduction. The SSAPOV Auvergne University database is well-suited to this task, as it includes survey-based measures of both agricultural Joshua Merfeld is a faculty member at New York employment and poverty. The results show that the Univeristy’s Wagner School of Public Service. structural transformation is proceeding more slowly Justin Ndoutamou is a doctoral student at Laval on average, and is far less variable, than would be University. inferred from the ILO modeled estimates that are between the initial level of agricultural employment David Newhouse is a Senior Economist in the World and the pace of change across districts, suggesting Bank’s Poverty and Equity Global Practice. He can be contacted at: dnewhouse@worldbank.org This note series is intended to summarize good practices and key policy findings on Poverty-related topics. The views expressed in the notes are those of the authors and do not necessarily reflect those of the World Bank, its board or its member countries. Copies of these notes are available on www.worldbank.org/poverty February 2020 · Number 19 4