Three Field Experiments in Development Economics.

Abstract

In the first chapter of this dissertation, I study the introduction of a bonus scheme with lottery payoffs at a large agricultural firm in Malawi. The “lottery bonus” was motivated by evidence from lab studies and by real world behavior that suggests that people often become risk-seeking when faced with small chances of large gains. Attendance- and performance-based bonuses were introduced for over 1,600 piece-rate workers. Attendance at work was higher under both schemes and that the effect of the lottery bonus on the probability of full weekly attendance was about twice as large as the effect of the fixed bonus. The larger effect on attendance appears to persist over time as workers gain experience with the lottery distribution. The second chapter, written jointly with Jason Kerwin, reports results from a field experiment with casual workers in Malawi designed to investigate how income timing and temptation affects spending decisions. Despite anecdotal evidence and suggestive survey data that in this study’s context market days increase the temptation to overspend, being paid at the site of the local market on Saturday compared to Friday did not strongly matter for expenditure levels or temptation spending. However, workers in the monthly group are much more likely than the weekly payment group to invest in a risk-free short-term “bond” that required a large payment and that was offered by the project in the week after the last payday. The third chapter, written jointly with Xavier Giné, Jessica Goldberg and Dean Yang, describes the results of a field experiment among Malawian farmers aimed at facilitating formal savings for agricultural inputs. Treated farmers were offered the opportunity to have their cash crop harvest proceeds deposited directly into new bank accounts in their own names, while farmers in the control group were paid in cash. The treatment led to higher savings and raised agricultural input usage. Because the treatment effect on savings was only a small fraction of the treatment effect on the value of agricultural inputs, mechanisms other than alleviation of savings constraints per se are needed to explain the treatment’s impact on input utilization.