Long-term effects of trade liberalization on regional dynamics: Evidence from South Africa Paulo Bastos University of Lisbon, REM and CEPR Nicolas Santos World Bank
Motivation • Growing evidence that tariff liberalization and trade shocks have sizable impacts on local labor markets (Topolova 2007, 2010; Kovak 2013; Autor, Dorn and Hanson 2013) • Evidence on long-term effects of trade shocks on regional dynamics is still limited • Dix-Carneiro and Kovak (2017) find sizable long-term impacts of import tariff liberalization on regional dynamics in Brazil: - Tariff reduction led to prolonged declines in formal employment - Impact of tariff changes on regional earnings 20 years after liberalization was three times larger than the effect after 10 years - Inconsistent with spatial arbitrage - Consistent with limited labor mobility, slow capital adjustment, agglomeration economies
This paper • Examine medium to long-term impacts of trade liberalization on regional dynamics in South Africa • Legacy of apartheid makes the issue especially important in South Africa • Document heterogeneity of effects of trade liberalization across regions with stronger presence of previously marginalized groups
Trade liberalization
Commodity prices
Need to account for this, but mining accounts for small share of employment (less than 8%)
The homeland system during apartheid • 13% of the land for 50% of the population • Massive forced relocation: 1950s-1980s millions forced from their homes and resettled in homelands • 4 homelands declared independent states (Transkei 1976, Bophuthatswana 1977, Venda 1979, Ciskei 1981). Other 6 under limited selfgovernment in preparation for independence • Apartheid ended in 1994
Data • Community-level data from 1996, 2001 and 2011 census. Georeferenced community profiles, with aggregates on: - Demographics: age, race, education, etc - Labor market: employment, income, industry, etc • South Africa divided in 50 districts, 234 municipalities, 21k communities • Use municipal-level data to define local labor markets (examine robustness at districtlevel) • Use community-level data to identify presence of former homelands inside municipality
Homelands no longer exist, but differentials persist
Econometric method ∆đ?‘Œ! = Îą + Î˛âˆ†đ?‘Šđ?‘‡! + Îłâˆ†Controls! + Îľ!" ∆đ?‘Šđ?‘‡! = change in municipality employment-weighted tariff Defined as the summation of the change in sectoral tariff changes weighted by initial share of labor allocated to the industry in the district in 1993 Look at effects on: - Employment - Income per capita Exploit heterogeneous effects by municipality depending on dominant population group
Summary statistics Table 1. Summary statistics, municipality-level data 1996 Population Black population (% of total) White population (% of total) No education (% of total) Male (% of total) Average age Employed population (% of total) Log income per capita Obs.
Mean 168852,8 72,49 9,57 26,13 48,13 25,30 62,58 8,84 234
SD 343983,6 32,61 8,83 11,93 3,27 2,33 18,09 0,81
2001 Mean 188014,9 74,49 7,92 24,31 47,80 26,55 56,30 9,11 234
SD 401764,7 31,88 7,28 11,88 2,59 2,31 16,49 0,82
2011 Mean 178887,1 75,72 6,89 9,15 48,22 27,76 65,93 9,79 234
SD 422251,2 30,16 7,14 4,05 2,31 2,44 11,34 0,53
Econometric results, baseline
Econometric results, controls
Econometric results, excluding mining sector
Econometric results, excluding mining sector, with controls
Econometric results, district-level estimates with controls
Econometric results, heterogeneity homelands
Econometric results, heterogeneity black
Econometric results, heterogeneity unskilled
Econometric results, heterogeneity unskilled
Summary • We examined long-term impacts of trade liberalization on regional dynamics in South Africa • Legacy of apartheid makes the issue especially important in South Africa • Found evidence of sizable negative impacts of tariff liberalization on local employment and income per capita • Documented heterogeneity of long-term effects of trade liberalization across regions with marginalized and non-marginalized groups - negative effects stronger in municipalities with presence of homelands - negative effects stronger in municipalities with strong presence of black population - these results do not appear to be driven by skill: effects weaker among unskilled
Discussion • Trade liberalization leads to welfare gains on average • But may cause job and income losses for some locations and population groups • Municipalities/districts more exposed to tariff cuts observed employment and income losses • In the South Africa context, labor mobility across space was limited (much less than what was anticipated when apartheid ended) • What can be done to compensate losers? - place-based policies (strongly justified in South Africa context) - income re-distribution (both regional and across households) - policies to facilitate mobility and integration of previously marginalized racial groups