Which incentives work and when? An experimental analysis of incentives for trainers

One conjecture in the theory of incentives is that incentives based on broader outcomes may be better at motivating agents than incentives based on narrow measures. We designed an experiment to test these hypotheses using a ""prospective randomized evaluation procedure"" (PREP). We then apply PREP to training programs as typically funded by donors of economic development assistance. We randomly assigned 274 participating entrepreneurs in the Philippines to one of 26, simultaneous, one-day, training classes in marketing. Trainers were given cash incentives based on the average score of their ""students"" on a standardized test containing an alternative number of questions, which were randomly assigned to each class. We then examined outcomes based on student satisfaction ratings of the trainer. Our results suggest that incentives based on broad outcomes are more effective than incentives based on narrow outcomes. We conclude with ways to improve our approach as well as with a discussion of the implications for using prospective randomized evaluation for improving the evaluation of donor projects.