Call Us: US - +1 845 478 5244 | UK - +44 20 7193 7850 | AUS - +61 2 8005 4826

An empirical perspective of American Economic

. The test is based on the historical evidence that, by increasing the accountability of local chiefs, precolonial centralization had two main benefits: it reduced local tyranny (i.e., the extent to which local chiefs could abuse their masses) and it fostered the coordination between the chiefs of different districts (e.g., Apter 1961; Tosh 1978). Thus, the “local accountability” view implies that the impact of precolonial centralization on public goods provision should depend on the severity of local tyranny and on the severity of the coordination problem. As a proxy for the severity of local tyranny, we use another dimension of African ethnic groups coded in our anthropological dataset: the degree of social stratification at the local level. Absent the accountability mechanisms of precolonial centralization, it is precisely in stratified societies that local chiefs are likely to be tyrannical and hamper public goods provision. Thus, the “local accountability” view predicts that precolonial centralization should disproportionately boost public goods provision in stratified, as opposed to egalitarian, societies. The “local accountability” view further predicts that the extra benefit of precolonial centralization in stratified groups should be smaller for public goods for which the coordination problem is especially severe. Indeed, in the provision of such goods, centralization should greatly help both egalitarian and stratified societies. We test these predictions and find that, consistent with the “local accountability” view, precolonial centralization boosts public goods provision more in stratified than in egalitarian groups. Crucially, precolonial centralization affects public goods depending on the severity of the coordination problem, which we proxy by the amount of geographic spillovers. For high spillovers goods, such as roads and immunization, centralization benefits both stratified and egalitarian groups. In contrast, for education and infant mortality, centralization benefits stratified groups, but not egalitarian ones. Because interdistrict coordination is much less important for the public goods behind these latter outcomes (local schools and clinics), centralization does not foster the provision of these goods in egalitarian groups (where local tyranny is small). Not only these results are consistent with the “local accountability” view; they are also hard to reconcile with the two alternative views that centralized groups were simply more “advanced” or that their institutions only improved national politics. A general version of these views predicts that—either for their advancement or better national politics—centralized groups should uniformly enjoy more public goods, irrespective of local stratification. Although in more nuanced versions of these hypotheses local stratification may matter, they still cannot explain the different patterns obtained for high and low spillovers goods. In line with the “local accountability” view, the role of geographic spillovers signals that precolonial centralization helped colonial and postcolonial national governments to foster coordination among local chiefs.