The Changing Structure of Africa’s Economies Maggie McMillan IFPRI/NBER/Tufts September 20, 2013 Based on joint work with Ken Harttgen, Dani Rodrik, Inigo Verduzco-Gallo and Sebastian Vollmer. Thanks to DFID/ESRC and the African Development Bank for financial support. 1
Outline 1. Motivation 2. Structural Change in Africa: Recent Evidence • McMillan&Rodrik 2011 • McMillan, Rodrik & Verduzco 2013 • McMillan 2013, Harttgen and Vollmer 2013 (AEO) 3. Structural Change in Africa: Using DHS Data • Harttgen, McMillan and Vollmer in Progress 4. Summary and Directions for Future Research 2
Motivation Most of Africa has been growing like gangbusters over the past decade. What is driving this growth? Commodity prices? Structural change? Something else? 3
Motivation: Commodity Prices? 250 Energy Agriculture Metals & Minerals Africa GDP World GDP 200 150 Index 2005=100 100 50 0 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Source: World Bank, authors' calculations 4
Motivation: Structural Change? Employment shares of 3 broad sectors Comparing sample from Duarte and Restuccia (2010) and African countries (sample from Jan 2013) Agriculture Industry 0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1 sEhi_ind sEhi_ag 6 7 8 9 10 6 7 8 9 10 Log GDP per capita (1990 International $ Log GDP per capita (1990 International $ D&R sample Africa sample D&R sample Africa sample Services (with fitted values) 0 .2 .4 .6 .8 1 sEhi_srv 6 7 8 9 10 Log GDP per capita (1990 International $ D&R sample Africa sample Note that Africa data measure sectoral share of total employment whereas D&R data measure share of total hours. Hours shares from Duarte and Restuccia (2010) cover 29 countries from 1950-2006 Their data were accessed 07/24/2012 from Duarte's website GDP from Maddison (2010) 5
Motivation: Why Should You Care? Understanding what is driving Africa’s growth is important for understanding both its’ sustainability and the likely distributional implications of this growth McMillan & Rodrik (2011) found that structural change in Africa had been growth reducing, but they focused on the period 1990-2005 and only 9 countries in Africa 6
Structural Transformation in Africa Decomposition of productivity growth by country group 1990-99 2000-10 within structural LAC LAC AFRICA AFRICA ASIA ASIA HI HI -1.00 0.00 1.00 2.00 3.00 4.00 -1.00 0.00 1.00 2.00 3.00 4.00 % change % change 7
Explaining the Reversal • 1990s still going through adjustment • Renewed commitment to agriculture and increasing agricultural productivity • Demographic trends – rural pop growth rates coming down • Political change – governments more accountable 9
Averages Hide Country Specific Heterogeneity • Structural change in Mauritius, a diversified economy , has been based on services. • In Nigeria, a resource-driven economy , changes in employment shares were tiny. • In Uganda, an emerging economy , structural change was significant and productivity grew in all sectors of the economy. • There was very limited but positive structural transformation in the pre-transition economy of Malawi 10
Mauritius: Diversified Economy Correlation Between Sectoral Productivity and Change in Employment Shares in Mauritius (2000-2007) Log of Sectoral Productivity/Total Productivity β = 2.5940; t-stat = 2.37 .2 min ter 0 man -.2 -.4 agr -.05 0 .05 .1 Change in Employment Share ( ∆ Emp. Share) Fitted values *Note: Size of circle represents employment share in 2000 **Note: β denotes coeff. of independent variable in regression equation: α + β∆ Emp. Share ln(p/P) = Source: Authors' calculations with data from Mauritius' CSO and UN National Accounts Statistics 11
Nigeria: Resource Driven Correlation Between Sectoral Productivity and Change in Employment Shares in Nigeria (1999-2009) Log of Sectoral Productivity/Total Productivity β = 85.2651; t-stat = 0.52 6 min 4 2 ter man 0 agr -2 -.01 -.005 0 .005 .01 Change in Employment Share ( ∆ Emp. Share) Fitted values *Note: Size of circle represents employment share in 1999 **Note: β denotes coeff. of independent variable in regression equation: α + β∆ Emp. Share ln(p/P) = Source: Authors' calculations with data from Adeyinka, Salau and Vollrath (2012) 12
Uganda: Emerging Economy Correlation Between Sectoral Productivity and Change in Employment Shares in Uganda (1999-2009) Log of Sectoral Productivity/Total Productivity β = 9.9173; t-stat = 7.92 1 ter man .5 0 min -.5 -1 agr -.1 -.05 0 .05 .1 Change in Employment Share ( ∆ Emp. Share) Fitted values *Note: Size of circle represents employment share in 1999 **Note: β denotes coeff. of independent variable in regression equation: α + β∆ Emp. Share ln(p/P) = Source: Authors' calculations with data from Uganda's Bureau of Statistics, IMF, and UN National Accounts Statistics 13
Malawi: Pre-Transition Correlation Between Sectoral Productivity and Change in Employment Shares in Malawi (1998-2005) Log of Sectoral Productivity/Total Productivity β = 43.9572; t-stat = 0.49 4 min 3 2 ter 1 man 0 agr -1 -.02 -.01 0 .01 .02 Change in Employment Share ( ∆ Emp. Share) Fitted values *Note: Size of circle represents employment share in 1998 **Note: β denotes coeff. of independent variable in regression equation: α + β∆ Emp. Share ln(p/P) = Source: Authors' calculations with data from Malawi's National Statistical Office, WDI 2010, and ILO's LABORSTA 14
Summarizing Results from Macro Data • Roughly half of Africa’s recent growth can be attributed to structural change • The expansion in services is only sustainable if commodity prices remain high • High skilled services cannot (now) be engine of growth in Africa – not enough skilled labor • Manufacturing has potential but is still very much lagging (Ethiopia shoes, Blue Skies Ghana) • Natural resources can facilitate structural change (Robinson, 2013) 15
Limitations of Macro Data • Differences in treatment of informality across countries (e.g. Kenya) • Differences in treatment of agriculture across countries (e.g. Botswana) • Limited availability of employment shares data (DFID/ESRC grant) • But even if national accounts data are perfect, the macro data ignores the following: – important within country heterogeneity in occupational structure and productivity. For example, across age groups (youth unemployment), across gender , across levels of education and across geographic location. – Only measures one standard of welfare, income. 16
Using DHS data to understand structural changes in Africa Harttgen, McMillan, Vollmer use occupational information from the Demographic and Health Surveys (DHS) to document levels and changes in occupations across countries and over time by socioeconomic characteristics. Occupations include: self-employed agriculture, agricultural employee, sales, clerical, services, professional, skillled and unskilled manual labor and unemployed. Importantly, surveys are consistent across countries and over time and take into account the seasonality of agriculture. Will compare outcome variables including health and education across occupation categories and over time within occupations to assess whether observed occupational changes are welfare enhancing. 17
DHS regions Source: Günther and Harttgen 2013. . 18
Changes in Occupational Structure Across Time 19
Socio-Economic Determinants of Occupational Structure: Full Sample (1) (2) (3) (4) (5) (6) (7) (8) Total sample Total sample Total sample Total sample Total sample Total sample Total sample Total sample Agriculture Clerical or (employee or Agriculture self Agriculture sales or Unskilled VARIABLES self employed) employed employee Professional service Skilled manual manual Not working No education 0.0811*** 0.0643*** 0.0169*** -0.0538*** -0.0446*** -0.0186*** -0.000868* 0.0355*** (0.00128) (0.00121) (0.000660) (0.000542) (0.00112) (0.000720) (0.000465) (0.00127) Age 15-24 -0.0490*** -0.0478*** -0.00124** -0.0348*** -0.0477*** -0.00608*** 0.000171 0.130*** (0.00117) (0.00111) (0.000560) (0.000457) (0.00102) (0.000677) (0.000467) (0.00123) Urban -0.359*** -0.301*** -0.0585*** 0.0468*** 0.173*** 0.0526*** 0.0300*** 0.0401*** (0.00107) (0.00102) (0.000507) (0.000691) (0.00123) (0.000824) (0.000590) (0.00124) Female -0.160*** -0.100*** -0.0593*** -0.0314*** 0.0903*** -0.0656*** -0.0210*** 0.185*** (0.00139) (0.00133) (0.000778) (0.000748) (0.00117) (0.000978) (0.000660) (0.00113) Log GDP per capita 0.0157*** 0.0281*** -0.0124*** 0.0368*** -0.0306*** 0.0639*** 0.0194*** -0.0763*** (0.00551) (0.00545) (0.00216) (0.00285) (0.00494) (0.00338) (0.00239) (0.00596) Polity IV score 0.00626*** 0.00406*** 0.00220*** 0.00177*** 0.00191*** 0.00116*** -0.00547*** -0.00612*** (0.000289) (0.000284) (0.000118) (0.000128) (0.000240) (0.000166) (0.000150) (0.000294) Observations 791085 791085 791085 791085 791085 791085 791085 791085 R-squared 0.310 0.327 0.192 0.065 0.131 0.047 0.054 0.241 Country FE YES YES YES YES YES YES YES YES Year FE YES YES YES YES YES YES YES YES Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 20
Recommend
More recommend
