“The Predicament of Establishing Persistence: Slavery and Human Capital in Africa”, 2021-11-23 (; similar):
We investigate the impact of historic slave trade on contemporary educational outcomes in Africa by replicating the empirical approach in 2008 and 2011. We show that slavery’s long-term legacy for literacy depends on how spatial effects are accounted for.
In cross-country regressions, exposure to historic slave trade negatively predicts contemporary literacy. However, within countries, individuals whose ethnic ancestors were historically more exposed to slave exports, have higher education levels today compared to individuals from ethnicities less exposed to slave trade in the past.
We argue that these somewhat puzzling findings resonate with emerging critiques of persistence studies that link historical variables with long-run development outcomes.
[Keywords: slave trade, literacy, life expectancy, persistence]
…Contrary to the cross-country evidence, our subnational estimations that use data on individual survey respondents from 2 contemporary waves of Afrobarometer survey (2005 and 2008) posit a different empirical pattern. We find that, controlling for country fixed effects, the measure of slave exports has a positive and statistically-significant effect on contemporary educational attainment. This effectively implies that, within countries, individuals whose ethnic ancestors were more intensely exposed to slave trade in the past have systematically lower levels of education today (relative to individuals whose ancestors had less exposure to slavery). Splitting the sample into subregions within Africa and re-estimating the main specification, we demonstrate that the positive impact of slavery is mainly driven by coastal countries. Recognizing the possibility that historic exposure to slave trade could determine the location of Christian missions and, thereby, education (2016; 2010; 2021), our individual-level regressions consistently include an ethnic group’s exposure to Christian missions and the disease environment through malaria prevalence. Following Nunn and Wantchekon, we also include a battery of individual, ethnicity, and colonial-level controls. The result also survives after directly controlling in the model for the distance of an individual’s ethnic group to the coast during slave trade, to the Saharan trade routes, and to historical reliance on fishing.
While these findings might appear as puzzling or counter-intuitive, they resonate with emerging concerns on persistence studies that the recent literature in historical political economy has highlighted (2021; 2019). An important concern underlined by these critiques relates to spatial or geography-related factors. Effects of historical variables are relative to where the comparison units are located, how they are defined, and the pattern of spatial dependence (2020). Taking cue from this, 2021 re-estimate the main specification in some prominent persistence papers and show that the inclusion of World Bank’s regional classifications as additional controls weakens the results. This helps to demonstrate that persistence papers can be sensitive to spatial dependence, manifested in this instance through “variation due to being in the same part of the world” (p. 58). Another important concern with persistence studies is the “compression of history” that emanates from regressing a historical variable on outcomes measured several centuries later (2008). This is especially evident in the case of slavery which predated the colonial period. Considerable time has elapsed between the initial exposure to slave trade and modern day outcomes. It is important to determine what might have happened in the intervening periods. There are at least 3 major time spans that are important for assessing slavery’s impact on education: the pre-colonial exposure to slave trade, colonialism, and the creation of national borders for modern African states. Our results are consistent with the suggestion that what happens after independence is important for appreciating slavery’s long-run impact on education. Overall, our empirical replications of 2 influential works on slavery by Nathan Nunn in the context of education leave some nontrivial implications for the persistence literature, and highlight the importance of both spatial and temporal factors