UC Berkeley

This dissertation combines research on three topics in applied empirical economics. The first paper, which is based on joint work with Patrick Johnston, examines the effect of development projects on civil conflict. The second paper estimates the effect of subsidized employment on the happiness of the unemployed. The third paper, based on joint work with Santiago Guerrero, analyzes the effect of restrictions to alcohol accessibility on Marijuana use.

The first paper develops a theoretical model of bargaining and conflict in the context of development projects. The model predicts that development projects cause an increase in violent conflict if governments cannot (1) ensure the project's success in the face of insurgent opposition and (2) credibly commit to honoring agreements reached before the start of the project. The model is tested by estimating the causal effect of a large development program on conflict casualties in the Philippines. Identification is based on a regression discontinuity design that exploits an arbitrary poverty threshold used to assign eligibility for the program. Consistent with the model's predictions, we find that eligible municipalities suffered a substantial increase in casualties, which lasts only for the duration of the project and is split evenly between government troops and insurgents.

The second paper estimates the causal effect of a type of subsidized employment projects - Germany's \emph{Arbeitsbeschaffungsmassnahmen} - on self-reported happiness. Results from matching and fixed effects estimators suggest that subsidized employment has a large and statistically significant positive effect on the happiness of individuals who would otherwise have been unemployed. Detailed panel data on pre- and post-project happiness suggests that this effect can neither be explained by self-selection of happier individuals into employment nor by the higher incomes of the employed. This suggests that subsidized employment programs are more effective at increasing the happiness of the unemployed than an increase in unemployment benefits.

The third paper estimates the effect of the Minimum Legal Drinking Age of 21 years on Marijuana use. The casual effect of this law is estimated through a regression discontinuity design that compares Marijuana use among individuals just below and just above age 21. We find a significant drop in Marijuana use at age 21, which suggests that individuals substitute between alcohol and Marijuana. Policies that restrict alcohol accessibility are therefore likely to have the unintended consequence of increasing Marijuana use.