The Case for Using Project-Level Data to Study Aid Distribution and Impact

Finer-grained aid information is helping scholars gain greater leverage on questions related to aid allocation and effectiveness.

February 20, 2012
Bradley C. Parks

This post first appeared on The Monkey Cage and is re-published here.

Earlier this month, President Obama hosted a Google+ Hangout session following his State of the Union address, in which he discussed and defended U.S. foreign aid policy. Fielding a question from a homeless veteran, Obama responded, “We only spend about 1 percent of our budget on foreign aid. But it pays off in a lot of ways…Aside from it being the right thing to do…it’s also important to make sure that people understand this is part of our overall security strategy.”

Since 9/11, the USG has promoted foreign assistance as a useful tool for combating global terrorism. Indeed, the case for foreign aid is often made on the basis of its presumed efficacy in preventing terrorism. But, until recently, the evidence supporting these claims was rather flimsy.

Formal models of the aid-terrorism relationship suggest that aid may prevent terrorism when it is targeted in ways that promote human capital through education (Azam and Thelen 2008; Bueno de Mesquita 2005). However, many of these theoretical arguments have not been subjected to careful empirical scrutiny because of insufficiently granular data.

A recent article by Joseph Young and Michael Findley seeks to correct these weaknesses.  AidData’s detailed activity coding methodology allows the authors to disaggregate aid figures by project purpose. In their analysis, Young and Findley include separate measures of education aid to test the specific argument that aid targeted to education may prevent terrorism. They are also able to examine the potentially substitutable effects of general budget aid, health aid, and aid tied to counterterrorism.

Here is a brief summary of their findings:

Does foreign aid reduce terrorism? We examine whether foreign aid decreases terrorism by analyzing whether aid targeted toward certain sectors is more effective than others. We use the most comprehensive databases on foreign aid and transnational terrorism—AidData and ITERATE—to provide a series of statistical tests. Our results show that foreign aid decreases terrorism especially when targeted toward sectors, such as education, health, civil society, and conflict prevention. These sector-level results indicate that foreign aid can be an effective instrument in fighting terrorism if allocated in appropriate ways.

Young and Findley’s article demonstrates that finer-grained aid information is helping scholars gain greater leverage on questions related to aid allocation and effectiveness. A November 2011 special issue of World Development also features nearly a dozen empirical studies that rely on the project-level information contained in AidData.

A relatively new initiative to geo-reference the physical locations of individual aid projects has also opened up exciting new avenues for research on the sub-national determinants of aid distribution and impact. Consider the map below, which documents the spatial distribution of violence in Afghanistan and then overlays the geographical locations of successful and unsuccessful World Bank projects. Contrary to the conventional wisdom, this map suggests that aid projects are not more likely to fail in conflict-affected areas. Indeed, many failed World Bank projects seem to cluster in the relatively less violent provinces north of Kabul. Readers can find more analysis of this issue at AidData’s blog, The First Tranche. New efforts to geo-code the universe of aid in individual countries also merit attention.

I recommend to readers of this blog who do aid-related empirical research. It contains a project-level database with more than 1 million individual development finance activities from 87 donor agencies to 200+ recipients from 1947 to 2011. And, while it includes the data in the OECD‘s Creditor Reporting System, it is not limited to official development assistance (ODA) flows. For example, it includes official loans from bilateral and multilateral lenders that do not meet the 25% concessionality threshold for ODA, flows to “non-ODA countries” (e.g. the United States, Russia), South-South cooperation activities, and other non-ODA technical cooperation activities. AidData also provides project-level data for many“non-traditional” donors, such as Brazil, India, South Africa, Poland, UAE, Saudi Arabia, and Kuwait. In addition to its project-level database, AidData maintains an extensive collection ofreplication datasets associated with published research on aid allocation and aid effectiveness. If you’d like to include one of your own aid-related replication datasets in this collection, you can let the AidData team know by sending an email to

Brad Parks is the Executive Director of AidData at William & Mary. He leads a team of over 30 program evaluators, policy analysts, and media and communication professionals who work with governments and international organizations to improve the ways in which overseas investments are targeted, monitored, and evaluated. He is also a Research Professor at William & Mary’s Global Research Institute.

Brad Parks is Research Faculty at the College of William and Mary and Co-Executive Director of AidData.