Hansen, BenjaminWilson, Melissa2022-10-042022-10-042022-10-04https://hdl.handle.net/1794/27632It is evident based on recent news articles and social media discussions that racial bias in police action is currently at the forefront of public interest in the U.S. Whether police operate outside of what is considered fair under our justice system can be challenging to estimate. This paper analyzes the incidence of racial bias in traffic stops by city police departments in 9 cities across the country. Under the “veil of darkness,” police cannot determine the race of a driver prior to pulling them over, based on the hypothesis of Grogger and Ridgeway (2012). I take their method a step further in order to address an issue that may cause a bias in their results. Utilizing the Stanford Open Policing data, I employ a regression discontinuity design around the start of daylight savings time in order to make an accurate comparison between daylight and nighttime stops drawn from the same distribution of drivers. I find little evidence of racial disparities in police stops with no significance for black drivers and significance for Hispanic drivers that is not fully robust to functional forms. This indicates that daylight times do not affect the proportion of stops of minority drivers and racial disparities are not affected by visible lighting. I posit that this might be due to a flaw in the “veil of darkness” hypothesis, rather than a lack of racialdiscrimination. Next, I test the effect of marijuana decriminalization in Illinois on racial disparities in arrests for marijuana possession in Chicago and provide evidence to support that the disparity is driven by racial prejudice. I use drug arrest data with amount in possession reported to determine if racial discrimination affects police decision-making at varying severity levels differentially. By showing there is a larger racial disparity in arrest rates at lower contraband levels than at higher levels, I provide evidence that over half of the disparity is driven by officer taste-based discrimination. More notably, I conduct an Interrupted Time Series estimation using a large unique Chicago Police Department dataset to show that marijuana decriminalization led to a substantial drop in the racial disparity for marijuana-related arrests in Chicago. Additionally, there is a shift in the trend to be slightly positive, driven by arrests of black individuals over the decriminalized amount. This implies a shift in resources to target higher severity drug crime activity, but still disproportionately affects black individuals. This motivates policy decisions to decriminalize marijuana and, subsequently, other minor crimes that disproportionately affect minority groups in order to reduce racial disparities in arrests. Almost universally, drug crimes carry sanctions which vary across weight. A Beckerian model of crime has sharp predictions that those carrying drugs should attempt to sort in response to these crimes. However, police discretion can also vary how the actual weights up recorded. We investigate these competing factors using administrative records from marijuana possession arrests in Chicago. Using a bunching approach, changes in sanction thresholds, and variation in officer-suspect race matches, we test both how decriminalization and officer-race matching affect bunching. This dissertation includes unpublished co-authored material with Benjamin Hansen.en-USAll Rights Reserved.Applied EconometricsCrime EconomicsMicroeconomicsRacial DisparitiesElectronic Thesis or Dissertation