Jose Luis (Pepe) Montiel Olea - Columbia
Empirical Bayes Counterfactuals for Police Homicides
Date: 10/22/2020 (Thu)
Time: 3:30pm- 5:00pm
Location: Seminar will be held on-site: Zoom
Organizer: Adam Rosen
Meeting Schedule: (Not currently open for scheduling. Please contact the seminar organizer listed above.)
All meetings will be held in the same location as the seminar unless otherwise noted.
3:30pm - Seminar Presentation (3:30pm to 5:00pm)
Additional Comments: "Empirical Bayes Counterfactuals for Police Homicides" (Joint with my Dan O'flaherty, Rajiv Sethi, Yun Li, and Wenda Ma) Abstract: This paper examines the variation across law enforcement agencies in the use of deadly force in the United States using a newly constructed data set. Methodologically, we develop an Empirical Bayes approach to estimate counterfactual parameters in a Poisson regression model for a short count panel on police homicides. We allow for changes in both observed covariates and unobserved agency-level characteristics. We report a number of counterfactual estimates for police homicides involving pairwise comparisons across agencies, changes in agency size, and changes in the level of reported violent crime. Our results suggest that i) significant variation across agencies exists in the propensity to use deadly force, ii) that this variation cannot be easily accounted for by counterfactual changes in crime rates and policing intensity, iii) and that these counterfactual changes can have qualitatively different effects for different agencies.