UC Berkeley

This dissertation is comprised of a three-part study in labor and development economics. Each chapter analyzes one aspect of the labor market. I study the effects of work experience, the importance of household structure and arrangement, and how potential job seekers (politicians) respond to implicit prices. All chapters are unified by a focus on the allocation rules associated with government work or programs and their secondary effects. Chapters 1 and 3 consider directly how selection processes affect who works with the government and their returns. Chapter 2 focuses on marginal program recipients given allocation design. Together, the studies show the persistence of externalities from large-scale public policies and the importance of allocation-specific rules in selection.

The first chapter examines the career effects of receiving a temporary, managerial position from the government. To do so, we exploit the random allocation of contracts to manage construction projects to licensed individuals in the Dominican Republic. We link this randomization to detailed administrative data on both individual outcomes and information about the firms created by individuals. People who randomly receive these positions shift away from being employees in the private sector and start new firms about 25% more often than the lottery non-winners. This firm creation happens when they finish their randomly assigned position, and the newly created firms are higher quality than firms created by non-winners. We also show that individuals with a high desire for these positions are relatively more likely to create firms as a result of receiving the positions. This application behavior has implications for the redesign of this program and potentially other programs aimed at increasing entrepreneurship. 

The second chapter investigates the effect of having a parent or sibling incarcerated on a child's long-term development. Despite high incarceration rates in the United States and the plausibly large effects on close relatives, these externalities are unknown. We use the random assignment of judges to defendants in Ohio to generate quasi-random variation in having a family member incarcerated. We find the incarceration of a parent reduces criminal justice system involvement over the life course and increases the likelihood to live in higher socioeconomic status neighborhoods as an adult. Parental incarceration has no statistically significant effect on education nor on other risky behaviors such as teen pregnancy. The incarceration of a sibling also reduces criminal involvement. We further explore the mechanisms behind these effects.

Chapter three tests whether reducing illicit rents available to politicians changes the composition of elected local officials in India. We study the randomized allocation of a biometric authentication system that past research showed increased the share of government disbursements received by beneficiaries. This policy likely reduced the share of rents captured by political officials and increased wages for low-skill workers. A simple model shows the decrease in potential illicit rents and increase in low-skill wages have differing predictions, highlighting the need for empirical work. We show that the program reduced the education levels of the head of the local government, consistent with illicit rents being a driver to seek political office, and find no impact on electoral competition.