Prosecutor Elections and Police Accountability

With Haritz Garro


Media: The Washington Post, Pod Save the People

Prosecutors play an important role in holding police accountable by determining whether or not an officer has broken the law. At the same time, prosecutors and police officers work together closely, raising concern over conflicts of interest. We study the effect of prosecutor turnover on the number of deaths caused by police officers. Using data from 2,315 district attorney elections in the United States, we find that the election of a new district attorney leads to a 17% reduction in the number of deaths caused by police officers. For close elections, deaths decrease by 40% (0.3 fewer deaths per year) after a new district attorney ousts an incumbent. We observe no corresponding changes in crime, arrests, assaults on police officers, or deaths of police officers. The effects are significant regardless of the political party of the newly elected district attorney. We find suggestive evidence in favor of increased police accountability and uncertainty about the district attorney’s type as mechanisms.

Do police maximize arrests or minimize crime? Evidence from racial profiling in U.S. cities

R&R at Journal of the European Economic Association


It is difficult to identify sources of discrimination in police stop and search data. In part, this is due to uncertainty over the objective of discretionary police stops: do officers aim to maximize arrests or to minimize crime? In this paper, I compare theoretical predictions implied by these two objectives to data from U.S. cities. Empirical evidence is consistent with a model of arrest maximization and inconsistent with a model of crime minimization. The findings support the validity of existing tests for discrimination that rely on the assumption that police officers maximize arrests.

Crossing the District Line: Border Mismatch and Targeted Redistribution


Online Appendix

Electoral district borders regularly cross the borders of local governments. At the same time, legislatures allocate resources using transfers to local governments. Political parties may try to target these transfers in order to win elections, but can only do so imperfectly because of border mismatch. This border mismatch creates inequality: otherwise similar local governments receive different transfers depending on the district map. To show this, I incorporate border mismatch into a model of political competition and test the predictions using data on transfers from U.S. states to counties. The results demonstrate a novel link between redistricting and voter welfare.

Polling Place Locations and the Cost of Voting

With Gaurav Bagwe and Juan Margitic


Surprisingly little is known about the location of polling places across the United States and their effect on turnout. To fill this gap, we acquire voter registration data, voting history data, and polling locations for over 15 million voters from Pennsylvania and Georgia. Using a precinct border discontinuity design, we find small average effects of a voter’s distance to the polling place on turnout, but considerable heterogeneity. A one mile increase in distance to polling place decreases the likelihood of voting by up to 1.2 p.p., but by up to 27.6 p.p. in areas where eligible voters rely on public transportation to go to work. The availability of no excuse vote by mail may help to substantially attenuate the reduction in turnout caused by distance to the polling place.

Works in Progress

Gerrymandering with Swing Voters (with Laurent Bouton, Micael Castanheira, and Garance Genicot)