Impact Evaluation of Aid-funded Projects

Abstract

The importance of quantitative impact evaluation of development programs using statistical or econometric methods has been increasingly recognized among researchers, foreign aid practitioners, and policy makers. Accordingly, such impact evaluation has been widely implemented. This paper aims to overview research and practices on impact evaluation to readers who are not familiar with statistics or econometrics. First, this paper discusses why simple comparison of outcomes of participants before and after the program or outcomes of participants and non-participants does not necessarily lead to impact evaluation. Then, I explain two leading methods of impact evaluation, randomized experiments and propensity score matching. Most importantly, the two methods eliminate, or at least alleviate, biases in the estimate of the effect of a program due to arbitrary selection of participants in the program. I also present as an example impact evaluation of Japanese aid-funded technical assistance projects in the Indonesian foundry industry using propensity score matching. This example shows that impact evaluation can lead to cost-benefit analysis of the program and suggestion of better program designs. Finally, I provide suggestions to the evaluation system of Japanese aid-funded programs.