Impact evaluation of a Rural Business Program using field experiment data.
详细信息
- 作者：Toledo Torres ; Patricia Eliana.
- 学历：Doctor
- 年：2011
- 导师：Carter, Michael,eadvisor
- 毕业院校：The University of Wisconsin
- ISBN：9781267223333
- CBH：3501169
- Country：USA
- 语种：English
- FileSize：6686746
- Pages：135
文摘
The first chapter describes a sample-experimental design of a field experiment developed to evaluate the impact of a Rural Business Program in Nicaragua RBP). By using the baseline dataset, we analyze how the results of the implementation of the protocol of assignment as well as the variations that the program experienced during the first two years may have affected the internal and external validity of the experimental data. For a two-side complier subsample, we found no significant differences for the household expenditures distribution of the treatment and the control groups. However, there are significant differences across treatment and control groups in tenure status of land and farmers age. If these variables affect the dynamic of household expenditures, a traditional difference-in-difference estimator would be biased. It is not straightforward that the sample represents the population of participants. However, there were requirements that every farmer had to fulfill to participate in the program, so farmers out of the sample frame are similar to those included. Around 80% of sampled participants are above the poverty line. Then, it is not clear that the impact of the RBP program may apply if the program were developed with less wealthier farmers. The second chapter estimates the impact of RBP on households expenditures. The unconditional average effect is not significantly different from zero. However, we found evidence that the RBP had heterogeneous effects. Average treatment effect varies positively with the initial stock of installations for farmers who had recent credit history previous to their participation in the program. In contrast, we found that the average treatment effect conditional to the initial stock of installation is negative for farmers without credit history. We hypothesize that lack of credit history would be a signaling of overconfident farmers who overestimate their ability to develop the business plan, so they would perform equally or worse than without the program. However, the negative estimated impact could be also associated with the failure of the parallel-trend assumption.