Does the heterogeneity of project implementers affect the program participation of beneficiaries? : Evidence from rural Cambodia

Using the dataset collected for assessment of a post-harvest technology project in rural Cambodia, we focused on the heterogeneous social preferences of project implementers, often overlooked in the literature of Randomized Controlled Trials (RCT). Our study focuses on the gimplementer effect h on program participation for the treated farmers. We show the possibility that heterogeneous program participation of ordinary farmers across the treated villages could be induced due to heterogeneity in the characteristics of project staff. In particular, we show the altruism of project staff, measured by the dictator game, consistently increases participation and the number of participations in the training sessions of beneficiaries. This type of heterogeneity in project staffs f preferences across treatment sites might yield noises in mean effects estimated using RCT methods conducted at a certain cluster level, which undermines the external validity of the estimated results. While RCT methods are very powerful tools for many program and policy evaluations, we cannot emphasize too much the importance of the way how an actual project is implemented.