Not So Black and White: Uncovering Racial Bias from Systematically Misreported Trooper Reports

Abstract

Highway police officers, or troopers, may misreport the race of people that they engage with in order to evade detection of racial bias. I propose a new test of racial bias in the presence of misreporting that is well-suited to explore the rich heterogeneity in bias behavior. Using a unique event in Texas where troopers were caught deliberately misreporting minority motorists as white, I find bias against all minority motorists, but especially against Hispanic motorists. I estimate bias for each trooper and find that over 30% of troopers were engaging in this behavior. Using my trooper level measure of bias, I identify causal relationships between bias and labor outcomes using a panel data set of trooper employment outcomes. I show misreporting was used effectively to evade detection of bias, with bias having no effect on labor market outcomes when the misreporting was possible. I find that after a rule change to trooper stop recording policy in response to the misreporting led to negative labor outcomes for biased troopers, specifically, lower rates of promotion and had lower salary growth. I further test how individual trooper bias changes in response to changes in peer, demographic compositions. In particular, black or Hispanic troopers are sensitive to changes in peer composition, while white troopers are unaffected.