Scholars have increasingly embraced fine-grained geospatial data to estimate the effects of aid programs at the subnational level. To date, however, scholars have yet to wrestle with a number of methodological issues arising from the use of these data, particularly when we move beyond coarse provincial (or administrative) level data to the level of aid disbursement itself. We require tools flexible enough to estimate aid effects when (1) the same aid disbursement is linked to multiple locations; (2) when the treatment and outcomes are measured at different levels or locations; and (3) where the distance of the presumed aid spillover, both spatially and temporally, are not known ex ante. We introduce a new method for estimating spatial treatment effects when fine-grained data are available for both the treatment and outcomes of interest. We demonstrate the utility of this approach using two empirical applications: the close of hundreds of International Security Assistance Force (ISAF) bases and the implementation of the Community Development Program (CDP) in violent areas of Afghanistan (2010-13). Drawing on multiple waves of a large-scale survey experiment, we examine how base closures and aid delivery affected support for the Afghan government relative to its Taliban enemy.
Funding: This research was supported by AidData at the College of William and Mary and the USAID Global Development Lab through cooperative agreement AID-OAA-A-12-00096. The views expressed here do not necessarily reflect the views of AidData, USAID, or the United States Government.