UC Berkeley

This dissertation comprises three studies on development economics and state capacity. In the first two chapters, developed through a partnership with the Tax Authority in Honduras, I examine the design and enforcement of tax policies when state capacity is low. State-provided public goods are crucial for economic development, and collecting taxes is necessary to finance the state. The first study evaluates the impact of a policy aimed at increasing tax payments from large corporations and highlights the importance of third-party information on enforcement. The second study reports the result of a large-scale experiment with taxpayers, aimed at estimating the compliance effects of informing taxpayers about the Tax Authority’s knowledge on their transactions. The final chapter focuses on another dimension of state capacity and development: the selection of state personnel that are key to the delivery of public services. In that chapter, I study whether the impersonal examinations used to select judges in Brazil are predictive about their performance on the job.

In chapter 1, co-authored with Felipe Lobel and Pedro Zúniga, I study how corporations in Honduras responded to the introduction of a minimum income tax, a provision that taxes firms on revenue when reported profits are low. Minimum taxes are recommended to tax authorities in lower-income countries by the International Monetary Fund and figure prominently in current debates on international tax cooperation. Evidence on their impact on tax collection and behavioral responses of firms is nonetheless still scarce. Using non- linearities in the tax schedule introduced by the policy and tools from the bunching literature, I show that the reaction of corporations reveals substantial evasion under profit taxation and a large elasticity of reported revenue. I also document that the enforcement environment faced by firms is important in determining these responses, highlighting the fact that the trade-offs between policy tools change with enforcement capacity.

In chapter 2, co-authored with Giselle Del Carmen and Edgardo Espinal, I use experimental methods to further explore the use of third-party information to enforce tax compliance. The use of letters and emails to directly influence the behavior of taxpayers is popular among tax authorities and has been shown to positively impact tax compliance. While the marginal cost of communication interventions is very small, targeting the most responsive taxpayers can be important: informing taxpayers about knowledge of their wrongdoing and not following up by punishing non-compliers erodes the credibility of the tax authority. I present results of an experiment that randomized approximately 32,000 taxpayers into receiving information about the tax authorities’ knowledge on their revenues before the income tax filing deadline. Across a range of outcomes, we estimate that our intervention had on average a null effect on compliance. The only subset of taxpayers to consistently react positively were those deemed to be of ex-ante low-risk of noncompliance. Using rich administrative data on past behavior of taxpayers and a causal forest algorithm, I find responsiveness to the intervention and predicted risk are negatively correlated. I conclude the chapter with a discussion of the implications of those findings for targeting of compliance tools.

In chapter 3, co-authored with Ricardo Dahis and Laura Schiavon, I approach another crucial dimension of the link between state capacity and economic development: the selection of public personnel. In the absence of strong incentives, public service delivery crucially depends on bureaucrat selection. Despite being widely adopted by governments, it is unclear whether civil service examinations can predict job performance. I investigate this question focusing on state judges in Brazil. Exploring monthly data on judicial output and cross-court movement, I estimate that judges account for a substantial share of the observed variation in the number of cases disposed. Using a novel data set on admission examinations, we show that judges with higher grades perform better than lower-ranked peers. These results suggest competitive examinations can be an effective way to screen candidates that will better deliver public services.