Facilitating Police Reform: Body Cameras, Use of Force, and Law Enforcement Outcomes [Click for Draft], Journal of Public Economics, Accepted.
Strengthening Police Oversight: Impacts of Misconduct Investigators on Police Officer Behavior (with Andrew Jordan), Journal of Policy Analysis and Management, March 2025.
Measuring Police Performance: Public Attitudes Expressed in Twitter. AEA Papers and Proceedings, May 2022
Negotiation, Employer Demand, and Gender Disparities (with Ratchanon Chotiputsilp and Clémentine Van Effenterre) [Click for Draft]
Abstract: We study how candidates' negotiation behavior affects employer demand using a two-sided field experiment in which 493 professional recruiters evaluated over 7,000 job applications and counteroffers and advised on final compensation packages. While on average female candidates are not disadvantaged during the negotiation stage, gender differences emerge under more intensive negotiation. Women who increase their counteroffers above the target range see their rejection rates rise by 2.7 percentage points more than men do, and their likelihood of being recommended for a concurrently open senior position falls by 6.8 percentage points more than men. Men gain and women lose from more intensive negotiation, and recruiters' beliefs about productivity-related traits explain only small part of these differences. Finally, we show that pay transparency mitigates these non-salary disparities in negotiation.
Do Teams Perform Differently Under Black and Hispanic Leaders? Evidence from the Chicago Police Department (with Matthew Gudgeon and Andrew Jordan) [Click for Draft]
[Reject and Resubmit at Journal of Labor Economics]
Abstract: In an effort to improve policing and enhance public trust in minority communities, police departments have sought to increase racial diversity among their leadership. Yet, evidence on whether this makes a difference is lacking. In this paper, we ask whether team-level outcomes such as total arrests differ when minority police supervisors are on duty. We do so in the context of the Chicago Police Department (CPD), where we exploit the annual, pre-determined, rotational nature of CPD's operations calendar to compare team-level outcomes (within the same unit and watch) on days when Black and Hispanic police lieutenants (LTs) are predicted to be on duty as compared to days when they have a day off (and another LT is working). Underlying crime conditions are balanced across these two sets of days. We find that fewer arrests take place on days when Black or Hispanic LTs are predicted to be on duty. The reductions in arrests are concentrated in arrests related to low-level crimes, such as disorderly conduct arrests, which decline by 17% of the mean. However, we see no change in arrest quality, as proxied for by court records, and cannot rule out potentially meaningful declines in arrests for more severe offenses. Differences in subordinate officer composition do not explain our results. Instead, reductions in arrests appear to be driven by watch operations LTs, who are stationed in the district headquarters and in charge of approving arrests.Together, these results suggest that supervisors can influence the policing practices of their subordinates, and open up avenues for interventions involving supervisors.
Job Rotation and Worker Performance (with Brighton Chotiputsilp, Mitchell Hoffman, and Shing-Yi Wang) [Click for Draft]
[Accepted based on Pre-results Review, Journal of Development Economics; Final publication venue is unknown.]
Abstract: How should firms allocate workers to jobs? A standard approach is for firms to assign workers to jobs and to train workers for one job. An alternative approach is for firms to rotate workers to different jobs while providing training for multiple jobs. This study investigates the impact of job rotation on worker allocation and efficiency within a garment manufacturing firm in Asia. In this randomized controlled trial, the control group follows the standard, pre-existing practice at the firm, receiving training for and assignment to one job. In the treatment group, workers undergo training for multiple jobs. After training, preferences of both workers and managers guide permanent job assignments using a version of the deferred acceptance algorithm. We examine how the treatment affects performance, employee turnover, work satisfaction, and job preference discovery.
The Cost of Salary Negotiation: An Experimental Evaluation with Job Seekers in Thailand (with Brighton Chotiputsilp) [Click for Draft]
Abstract: The persistent gender pay gap is often linked to women avoiding competitive activities, particularly salary negotiation. Despite initiatives aimed at promoting negotiation skills, the impact of such skills on labor market outcomes remains unclear. In our field experiment with real job seekers, we introduced two interventions: negotiation training combined with the provision of market wage information and the provision of wage information alone. Following the intervention, we evaluated salaries and non-salary benefits through an obfuscated survey. Although the training increased negotiation frequency, it resulted in reduced salaries and employment propensity, particularly for women. The most consistent explanation for these declines appears to be social backlash against negotiation.