Corruption in the Spending of Development Aid: New Evidence from Contract-Level Data

Academic research is unclear about whether development aid fuels corruption in recipient countries. For example, Tavares (2003) and Okada & Samreth (2012) report a negative link between aid and corruption, Knack (2001) demonstrates a positive relationship; Ear (2007) finds no significant effect. This ambiguity reflects two major weaknesses in prior research. First, researchers have thus far lacked methods to measure corruption in the aid allocation process directly and have relied only on perceptions-based indicators at the national level. Second, with few exceptions (e.g. Charron 2011; Dalgaard & Olsson 2008; Svensson 2000), previous studies have only paid scant attention to contextual factors such as the nature of local political power dynamics. This is despite a growing body of work suggesting that – generally speaking – both the types of corruption that prevail and the effectiveness of corruption controls depend upon the political context (Rocha Menocal 2014; Booth & Unsworth 2014). Our paper addresses both of these problems. First, by mining procurement contracts funded by three major donor organizations for corruption “red flags”, we have developed a dataset that provides unprecedentedly accurate indicators of corruption risk in the spending of aid across the developing world. Second, we use this data to assess the effectiveness of donor regulatory reforms to curb corruption in different recipient-country environments. To test our hypotheses, we employ a multi-method, “nested analysis” research design that combines quantitative modes of inquiry with qualitative case studies from Sub-Saharan Africa, Latin America and Southeast Asia. The quantitative analysis uses a regression discontinuity design that exploits the fact new donor regulations were implemented at different points in time throughout the 1998-2016 period. Our findings indicate that the effect of anti-corruption tools varies widely depending on the political context in recipient countries. For example, “developmental patrimonialism” – characterized by relatively high levels of bureaucratic autonomy and administrative capacity – provides a conducive environment for donors’ anti-corruption reforms to work effectively, while kleptocratic regimes – distinguished by an informal privatization of bureaucratic organizations – undermine the intended effect of such reforms.