How Is Foreign Aid Spent ?

It is notoriously difficult to measure the causal impact of foreign aid on the economy. The “micro-macro paradox” (Paul Mosley 1986) renders it impossible to add up the effects of individual aid projects, since foreign aid is fungible. Thus, researchers are left to conduct cross-country analyses to capture the effect of aid on economic growth and other outcomes net of the recipient governments’ domestic budget reshuffling. This large literature has produced mixed results and much disagreement, largely due to different interpretations over causal inference (David Roodman 2007). After all, since donors may reward countries for good performance—or bail out basket cases—concerns of endogeneity are justified. Not surprisingly, the literature has employed a number of instrumentation strategies (see Raghuram G. Rajan and Arvind Subramanian 2008). Existing instrumental variable approaches use the literature on the determinants of aid (e.g., Alberto Alesina and David Dollar 2000) to isolate variables that predict foreign aid, broadly, and then use the best candidates to predict aid in a two-stage aid-on-growth regression. But each existing instrument for aid (see Peter Boone 1996; Henrik Hansen and Finn Tarp 2001) can be criticized for one of three broad reasons: it is highly collinear with aid itself (e.g., lagged aid, lagged aid squared); it stands a good chance of not being truly exogenous to the economy