UC Berkeley

This dissertation consists of three contributions to the rapidly growing applied microeconomics literature. In these three chapters, my coauthors and I use rigorous identification and large administrative datasets to answer three interesting yet challenging questions.

The first chapter studies the impact of public sector managers on office productivity. I use novel Italian administrative data containing a homogeneous measure of tasks to construct an output-based measure of the productivity of public offices. This is the ideal setting to isolate the contribution of managers to office performance as all sites are subject to the same rules, workers produce a homogeneous product, and there are virtually no differences in physical capital across offices. Exploiting quasi-experimental variation in the rotation of managers across offices, I find that managers explain 9% of the total variation in productivity, about one third as much as the permanent component of productivity associated with different offices. A one-standard-deviation increase in managerial talent is associated with a 10% increase in office productivity. I explore what makes for a good manager and show that the rise in productivity associated with the arrival of a more productive manager is mainly driven by the exit of older white-collar workers (who appear to retire when the more productive manager takes over). The estimates from my productivity model imply that an optimal social allocation assigns the best managers to the largest and most productive offices. If top-level bureaucrats were reassigned on this basis, overall productivity would increase by at least 6.9%.

The second chapter is based on joint work with David Card and David Silver. We study how hospital treatment varies — often with little connection to medical needs – in the context of childbirth. In particular, we focus on low-risk first births, where csection rates vary enormously across hospitals, and where policymakers have focused much of their attention in calls for reducing unnecessary c-sections. We find that proximity to hospitals with high average c-section rates leads to more cesarean deliveries, fewer vaginal births after prolonged labor, and higher average Apgar scores. Infants born in these hospitals are less likely to be readmitted in the year after birth, but more likely to visit the emergency department for a respiratory-related problem. They also have lower mortality rates, driven by a reduction in the joint probability of prolonged labor and subsequent death. A stylized cost-benefit analysis suggests that re-allocating births to high c-section hospitals could lead to net social benefits.

The third chapter studies the impact of the fight against the Mafia on Italian firms and workers. The impact of organized crime on local economies is inherently hard to measure due to (i) the lack of exogenous variation in criminal activity and (ii) poor data quality. I address the first challenge by exploiting the timing of the dismissal of public elected officials due to mafia infiltration as a source of plausibly exogenous variation. I address the latter by using large administrative datasets on the universe of private sector employees and public procurement auctions. I document that the three external commissioners (who replace the dismissed official) temporarily cut public investment and public procurement. Moreover, connected firms experience a sharp reduction in the probability of winning a call for bidders after the takeover. Higher exit rates and the lower number of public procurement contracts awarded only partially explain this sharp drop. Overall my findings are consistent with stigma being associated with connected firms and lasting longer than takeover itself. I do not find any evidence that this policy negatively impacts local formal businesses and workers. More specifically, I find suggestive evidence of higher churning in the economy, and I do not detect any negative impact on local workers in terms of employment.