Urbanization in SubSaharan Africa (SSA) Vernon Henderson: Brown University Mark Roberts, Uwe Deichmann: World Bank Adam Storeygard: Tufts University May 2013
Literature s claim: Africa is different Urbanization without growth: Fay and Opal (2000), Response to climate change (rainfall): Barrios et al (2006): more pronounced in Africa Urbanization without industrialization Industrialization vs. consumer cities: Jedwab (2012), Gollin, Jedwab and Vollrath (2103) Private sector (farmers) spend rents in local towns Rainfall and satellite (night lights) data Agricultural and natural resource rents captured by the state Redistributed to cities (general or elites)
Our findings Africa is like the rest of the world in an important way Human capital growth drives urbanization in Africa and elsewhere, as modeled in the (theoretical growth) literature Galor et al. (1993, 2009); Lucas (2004); Henderson & Wang (2005) Effects by sector Agricultural price shocks have expected effects in SSA: deter urbanization Unexpected in rest of world
Our findings (continued) Effects by sector Modern manufacturing: positive OECD income shocks to countryspecific historical trade paths spur urbanization Effect is conditional on level of development. SAS less developed (lower base educational attainment) New work just on SSA: Look at subnational level Positive rainfall shocks induce rural population growth Agricultural price shocks: to be done
Cross section: % urban vs GDPpc or schooling < Mean(Global GDP pc) GDP pc = PWT7.0 < Mean(Global GDP pc) GDP pc Urban share 0.2.4.6.8 1 'ZWE' 'ZAR' 'URY' 'CHL' 'JOR' 'BRA' 'PER' 'COL''PAN' 'DZA' 'BGR' 'DOM' 'TUR' 'MYS' 'MNG' 'BOL' 'ECU' 'TUN' 'COG' 'SLV' 'PRY' 'ZAF' 'BWA' 'GMB' 'NIC' 'MAR' 'AGO' 'SYR' 'GHA' 'CIV' 'HTI'CMR' 'HND' 'FJI' 'ALB' 'ROM' 'LBR' 'NGA' 'PHL' 'IDN' 'GTM' 'CHN' 'GNB''BEN' 'SEN' 'MRT' 'EGY' 'MUS' 'SOM''CAF' 'TGO' 'SLE' 'ZMB' 'NAM' 'GIN' 'MLI' 'PAK' 'MDG' 'MOZ' 'SDN' 'LAO' 'BTN' 'THA' 'VNM' 'IND' 'BFA''TZA' 'LSO' 'BGD' 'GUY' 'AFG' 'KEN' 'TCD' 'SWZ' 'NER' 'MWI' 'ETH' 'RWA' 'KHM' 'UGA' 'NPL' 'LKA' 'BDI' 'PNG' 'MLT' 'KOR' Urban share 0.2.4.6.8 1 'MLT' 'URY' 'CHL' 'BRA' 'JOR' 'COL' 'PER' 'BGR' 'PAN' 'TUR' 'DZA' 'MYS' 'DOM' 'ECU' 'SLV' 'TUN' 'BOL' 'COG'ZAF' 'PRY' 'BWA' 'SYR' 'GMB' 'MAR''NIC' 'CIV'CMR' 'HND' 'HTI' 'FJI' 'ALB' 'GHA' 'ROM' 'GTM' 'LBR' 'IDN' 'CHN' 'PHL' 'BEN' 'MRT' 'SEN' 'MUS' 'EGY' 'SLE' 'CAF' 'ZMB' 'NAM' 'TGO' 'ZWE' 'MLI' 'PAK' 'ZAR' 'MOZ' 'SDN' 'LAO''THA' 'VNM' 'IND' 'TZA' 'LSO' 'BGD' 'GUY' 'AFG' 'KEN' 'SWZ' 'NER' 'RWA' 'KHM' 'MWI' 'UGA''NPL' 'LKA' 'BDI' 'PNG' 'MNG' 'KOR' 5 6 7 8 9 10 ln(gdp per capita) 0 2 4 6 Aver yr school (high + college) SSA NSSA SSA NSSA Development and urbanization Correlation with measured GDP : Two problems Correlation between two outcomes SSA measurement: Young (2009), Henderson, Storeygard & Weil (2012) Economic fundamental : human capital growth Education: avg, years of high school and college of the adult (25+) population
Growth rate of urban share.02 0.02.04.06 'LBR' 'ZAR' Long differences: 19702010 < global mean(gdp pc) GDP pc = (PPP, PWT7.0) 'BWA' 'RWA' 'MOZ' 'BTN' 'BDI' 'BFA' 'NPL' 'BGD' 'AGO' 'TZA' 'LAO' 'GMB''LSO' 'MRT' 'GNB' 'IDN' 'HTI' 'MWI'BEN' 'CMR' 'MDG' 'GIN' 'KEN' 'MLI' 'ZWE' 'NGA' 'NER' 'AFG' 'UGA' 'SWZ' 'CIV' 'TCD' 'ETH' 'SDN' 'MYS' 'KOR' 'TGO' 'SLE' 'NAM' 'GHA' 'DZA' 'SOM' 'BOL' 'HND' 'TUR' 'SLV' 'PRY' 'COG' 'ECU' 'MAR' 'ALB' 'DOM' 'CAF' 'BRA' 'SEN' 'JOR' 'FJI' 'IND' 'GTM' 'PAK' 'MNG' 'BGR' 'ZMB' 'ZAF' 'COL' 'PAN' 'THA' 'VNM' 'PHL' 'TUN' 'PER' 'NIC' 'KHM' 'PNG' 'SYR''ROM' 'URY' 'CHL' 'GUY' 'EGY' 'MUS' 'MLT' 'LKA' 'CHN'.05 0.05.1 Growth rate of GDP pc SSA NSSA Growth rate of urban share.02 0.02.04.06 < global mean(gdp pc) GDP pc = (PPP, PWT7.0) 'BWA' 'RWA' 'MOZ' 'BDI' 'NPL' 'BGD' 'TZA' 'LAO' 'IDN' 'LSO' 'CHN' 'GMB' 'MWI' 'CMR' 'BEN' 'MRT' 'HTI' 'KEN' 'MLI' 'ZWE''SWZ' 'UGA' 'KOR' 'MYS' 'SDN' 'AFG' 'NER' 'GHA''LBR' 'TUR' 'NAM' 'CIV' 'DZA' 'ALB' 'BOL' 'DOM''ECU' 'HND' 'VNM' 'SLE' 'BRA' 'PAN' 'PRY' 'SLV' 'THA' 'COG' 'FJI' 'IND' 'MAR' 'PHL' 'MNG' 'BGR' 'GTM' 'JOR' 'SEN' 'CAF' 'KHM' 'ZAF' 'COL' 'PAK' 'TUN' 'PNG' 'ROM' 'PER' 'ZMB' 'CHL' 'NIC''SYR' 'URY' 'MLT' 'ZAR' 'MUS' 'GUY' 'EGY' 'LKA' 'TGO'.02.04.06.08.1 Growth rate of aver yrs school (high + college) SSA NSSA constant Growth Adj R 2 N Growth in GDP per capita 0.0158*** (0.0013) 0.0545 (0.0688) 0.003 86 Growth in effective technology 0.0028 (0.0033) 0.2817*** (0.0841) 0.171 76
Sector composition and urbanization 2010 crosssection (< Mean(Global GDP pc)) Urban share 0.2.4.6.8 1 'MLT' 'URY' 'CHL' 'KOR' 'JOR' 'BRA' 'PER' 'PAN' 'COL' 'BGR' 'MYS' 'DZA' 'DOM' 'TUR' 'TUN' 'ECU' 'BOL' 'MNG' 'BWA' 'ZAF' 'COG' 'SLV' 'PRY' 'AGO' 'MAR' 'NIC'SYR' 'GMB' 'ROM' 'HND' 'FJI' 'ALB' 'CHN' 'GTM' 'CIV' 'GHA' 'PHL''IDN' 'MUS' 'EGY' 'SEN''MRT' 'NAM' 'ZMB' 'ZWE' 'GIN' 'BTN' 'PAK' 'THA' 'SDN' 'IND' 'MDG' 'LAO' 'MOZ' 'LSO' 'BGD' 'GUY' 'VNM' 'TZA' 'TCD' 'KEN' 'SWZ' 'AFG' 'RWA' 'KHM' 'LKA' 'UGA' 'MWI' 'NPL' 'PNG' 'ZAR' 'SLE' 'ETH' 'CAF' 'LBR' 0.2.4.6 Value added in agriculture (% GDP) SSA NSSA Crosssection: SSA has no relationship to agriculture OR manufacturing Bad measurement in general? Long difference: no relationship for SSA or elsewhere
Strategy in estimation Analyzing conditions affecting urbanization SHOCKS: Exogenous and wellmeasured Price shocks for agriculture and natural resources World prices; African exports measured by source imports of other typically more developed countries Relative price changes. Relative price changes can mean windfall (surplus) gains How are surpluses spent? Go to farmers (reduce urbanization) Ceilings on prices paid to farmers, surplus spent on urban public goods, urban food subsidies etc. (increase urbanization) Surplus goes to rural landowning elites (not SSA): invest in cities?» Depends on ownership regimes, inequality, capital markets Rainfall shocks: improve productivity
Base Specification in crosscountry specification Sample all countries in 1970 with less than mean income and pop. over 300,000. N=76 Pooled first differences: 7080, 8090, 9000, 0010. Δurban% it = b 0 Δeduc it + b 1 Δln(pop it ) + b 2 ΔX it + T t + Δe it Interact SSA with everything Can we claim causality? Potential omitted variables Population: Improved (unobserved) urban vs. rural health clinics in cities draw more people in, and may also reduce fertility Education: timing helps. Today s relative improvement in urban schools are a draw but do not change adult education No valid instruments
Base specification Δln(total nat. population) Δln(post prim. educ) Average years of hs & college per adult ln(land area) ln(landarea)*δln(nat.pop.) SSA* Δln(total nat. pop.) SSA* Δln (post prim. educ) [1] [2] [3] 8.457** (4.032) 2.892** (1.356) 3.364 (6.435) 2.290 (3.816) 7.166** (3.105) 3.491*** (1.268) 42.86** (19.65) 3.515*** (1.264) 0.739** (0.293) 3.006* (1.587) Rsquared 0.100 0.118 0.126 Interpretation: 1 sd increase in Δeducation leads to a 1.05 point increase in Δshare urban (mean of 4.7)
Shocks to agriculture Shocks to international agricultural prices (not wood products) for k= 1 to n products (Collier and Goderis, 2009) n a jt k = 1 kt USAt kj,1962 9 k = 1 n kj,1962 9 1 [ ], 1 PI = p CPI a = PS = EXSH ln PI, jt j,1962 9 jt a EXSH Share of exports in GDP kj,1962 9 Share of commodity k in export group Price shock PS jt l, t 10 l based on changes in world prices 10 year change, lagged 5 years, and smoothed 3 yrs for beginning and end Rainfall shocks (country level)
Agricultural Price and Rainfall Shocks (1) SE Δln(nat. population) 67.82 (48.14) ln(land area) 0.969 (0.80) ln(land area)*δln(nat. pop.) 4.645 (3.79) Δln(postprimary educ.) 4.151** (1.62) Δln(rainfall) 2.448 (2.00) Agricultural price shock 9.022** (4.12) SSA*Δln(rainfall) 2.170 (3.69) SSA*Ag. price shock 16.61*** (5.11) Controls: time effects & SSA interacted with all covariates Observations 260 Rsquared 0.202
Nonfood agriculture Δln(nat. population) 65.98 (47.32) ln(land area) 0.911 (0.774) ln(land area)*δln(nat. pop.) 4.439 (3.713) Δln(postprimary educ.) 4.205*** (1.565) Δln(rainfall) 2.503 (1.914) Agricultural price shock 48.97 (32.44) SSA*Δln(rainfall) 2.386 (3.581) SSA*Ag. price shock 59.45* (34.42) Observations 260 SE palm oil, coffee, cocoa, linseed oil, wool, tobacco, cattle hides, copra, sisal, rubber, tea and cotton Consistent with full ag. results Rsquared 0.209
Agricultural Rent Results Why differ for nonssa? SSA: enhance rural sector (includes locally produced services and goods in rural sector) NonSSA: Capture by state or statesanctioned institutions: Urban bias Land owners invest surpluses in urban capital (vs SSA with communal owned land) Other explanations Geography: rainfall, Ramankutty s agricultural potential, latitude, share tropical: NO
Subnational rainfall/aridity Long term reductions in rainfall or aridity reduce long term rural populations (Long difference). Compare rainfall history of early generation with that of later generation Impacts in migration literature (Henry, Schoumaker, Beauchemin, 2004 and Beauchemin and Bocquier, 2004) Relocate to urban areas Relocate to other rural areas (or out of country) Add: Change activities in rural areas Short term reductions cause similar changes which may be temporary or permanent (panel)
Aridity is precipitation divided by potential evapotranspiration (PET) from UNEP (1992). PET is calculated using the Thornthwaite (1948) method (Willmott et al, 1985)
Change in nonagricultural activity in rural sector in SSA Share of agriculture in total local employment (males in DHS) Country Beginning year share Ending year share Source Kenya 0.57 ( 93) 0.46 ( 09) DHS South Africa 0.28 ( 96) 0.16 ( 07) IPUMS Benin 0.85 ( 96) 0.72 ( 06) DHS Malawi 0.91 ( 87) 0.66 ( 08) IPUMS Niger 0.87 ( 92) 0.57 ( 06) DHS Senegal 0.75 ( 92) 0.54 ( 05) DHS Cameroon 0.79 ( 91) 0.75 ( 04) DHS Chad 0.92 ( 96) 0.90 ( 04) DHS BurkinaFaso 0.94 ( 93) 0.90 ( 03) DHS
Organization of data At district level (192) in 16 countries annualized growth rate in rural population and in share urban From Population census volumes Differencing helps with variations between countries in definitions of urban Issue of changing definitions within countries and redrawing of subnational boundaries over time. (1) LD: 1st and last census year (at least 17 years apart) (2) Panel (N=479): first difference each adjoining census pair
Basic formulation LD r ij1st last j ij1st last a = [ln A ln A ] / L r ij1 st last ij, t _ lag18 ijt L _ lag18 j ij1st last α A L j ij, t _ lag18 J = α + a : annualized growth of rural population : country fixed effect J : Average of aridity in the 18 years before t : number of years between 1st and last census Panel: A ij,_ t lag 3 ; Countryyear fixed effects; standard deviation in aridity between t and tl jt
Annualized growth rate in aridity Long difference: all districts Panel: just arid countries (A 5070 <0.95) Annualized growth in rural population 0.903** (0.446) Annualized growth rate in share urban 0.520 (0.679) Annualized growth in rural population 0.413** (0.205) Annualized growth rate in share urban 1.045* (0.586) [rainfall] {arid A 5070 <0.95} Standard deviation {s.d.} in A between censuses s.d. *annualized growth rate [1.017**] {0.961*} [0.614] {0.854}.030 (.093) 3.374*** (1.107) 0.193 (0.141) 5.968 (6.863) controls Country fixed effects Countryyear fixed effects observations 192 {92} 181 197 197 R squared 0.179 0.343 0.292 0.379 Long difference: Mean (s.d.)of rural pop growth 0.023 (.018) Mean (s.d.) of aridity growth: 0.0031 (.0049) Panel: Mean (s.d.)of rural pop growth 0.022 (.019) Mean (s.d.) of aridity growth: 0.0022 (.014) Mean (s.d.) of aridity variation: 0.103 (.057)
Conclusions Africa like rest of world in terms of relationship between changes in urbanization and changes in effective technology Africa differs in reaction to changes in agricultural price shocks NonSSA: promote urbanization SSA: Agriculture: deters urbanization (some to farmers, some to elites (?)) Positive rainfall shocks: Encourage rural population growth Urbanization effect is weaker