Research

Job Market Paper 

No Money Bail, No Problems? Evidence from an Automatic Release Program

Abstract:  Are the effects of money bail on pretrial misconduct large enough to justify its costs? Money bail advocates argue that its usage is critical for averting misconduct, while skeptics counter that its effects are small and not worth the human costs of pretrial detention. I examine these tradeoffs using a program in Kentucky that eliminated financial bail for a subset of low-level cases. I use administrative data and a stacked differences-in-differences approach to estimate the program’s effects on misconduct and detention. The program decreased the rate of financial bail by 50.5 p.p. Even though initial bail amounts were only $360 on average, the program increased the rate of one-day pretrial release by 13.7 p.p., indicating an inability to put up even small bail amounts initially barred release. The program increased failure to appear in court by 3.3 p.p. and had no detectable effect on pretrial rearrest, with the data ruling out even modest sized increases. The effects on pretrial misconduct are mainly driven by reduced usage of money bail, which requires pretrial payments, rather than reduced usage of unsecured bail, which threatens future financial obligations, suggesting that these threats may not effectively deter pretrial misconduct. Taken together, the results imply that one instance of pretrial misconduct needs to be at least 18 times as costly as one day in detention for money bail to justify its costs.

Press: Phenomenal World

Working Papers

If You Give a Judge a Risk Score: Evidence from Kentucky Bail Decisions

Abstract: Prediction tools, such as risk scores, often serve as decision aids rather than entirely overriding human discretion. How does the implementation of risk scores impact human decisions? I study a policy change that recommended no money bail for defendants below a set risk threshold. The policy decreased the money bail rate for defendants with risk scores slightly lower than the risk threshold but did not change rates for defendants with scores slightly higher than the threshold. As such, the policy changed the overall rate of money bail rather than simply shifting its allocation across cases. I argue this is consistent with a model where prediction-based recommendations shift the costs of mistakes from individual decision-makers (judges) to the social planners who set recommendations. Black defendants benefit less than white defendants with identical risk scores from the no money bail recommendation. Disparate benefits by race are driven by variation in behavior between (rather than within) judges – judges who see more white defendants are more likely to comply with the no money bail recommendation. The relationship between defendant population and judicial behavior is not explained by judge covariates. Results are consistent with a model where recommendations decrease the costs of mistakes more in racially homogeneous populations.

Press: WIRED, Axios


After The Burning: The Economic Effects of the 1921 Tulsa Race Massacre
(with Jeremy Cook, James Feigenbaum, Laura Kincaide, Jason Long, and Nathan Nunn)

Abstract: The 1921 Tulsa Race Massacre resulted in the looting, burning, and leveling of 35 square blocks of a once-thriving Black neighborhood. Not only did this lead to severe economic loss, but the massacre also sent a warning to Black individuals across the country that similar events were possible in their communities. We examine the economic consequences of the massacre for Black populations in Tulsa and across the United States. We find that for the Black population of Tulsa, in the two decades that followed, the massacre led to declines in home ownership and occupational status. Outside of Tulsa, we find that the massacre also reduced home ownership. These effects were strongest in communities that were more exposed to newspaper coverage of the massacre or communities that, like Tulsa, had high levels of racial segregation. Examining effects after 1940, we find that the direct negative effects of the massacre on the home ownership of Black Tulsans, as well as the spillover effects working through newspaper coverage, persist and actually widen in the second half of the 20th Century.

Press: The New York Times, TIME Magazine, CNN, CNBC, NBER Digest, Black Enterprise, The Root, Business Insider


Does Bias Training Change Bail Decisions?

Abstract: Bias training programs are a popular low-cost approach to addressing racial disparities in the labor market as well as the criminal justice system. Despite their growing prevalence, the magnitude and permanence of their effects on high-stakes decision-making are uncertain. I investigate the effects of a brief bias training program on racial gaps in bail outcomes. The training was provided to a subset of judges who set bail but not others. Leveraging differential treatment over judges as well as time variation, I provide estimates of the training effect using difference-in-difference and triple differences approaches. In the month immediately following training, racial gaps in bail decisions decreased. However, the initial effects dissipated in the second and third months after training, suggesting that brief trainings are unlikely change judicial decision-making in the long run.

Older Papers

Uncorking Expert Reviews with Social Media: A Case Study Served with Wine
(with Peter Pedroni and Stephen Sheppard)