Research

Working Papers

Striking the Balance: Human Discretion and Algorithmic Insights in Parole Supervision Decision-Making
with Nicholas Powell
Updated: May 2023

In this paper, we examine the interplay between predictive algorithms and hu- man discretion in determining parole supervision levels. Adopting a method- ological approach centered on the random assignment of parole officers at spe- cific risk score thresholds—particularly at junctures where parolees transition between various supervision levels—we investigate the impact of officers’ deci- sions to deviate from algorithmic recommendations on recidivism rates. Our findings reveal that professional adjustments to higher supervision levels consis- tently lead to reduced recidivism rates, while adjustments to lower supervision levels don’t display a significant effect. This underscores the pivotal role of strategic resource allocation in parole supervision, indicating that harsh over- rides can be resource-optimal in effectively lowering recidivism. Conversely, lenient overrides maintain stable recidivism rates without necessitating inten- sified supervision. Additionally, the study contributes to the ongoing discourse on the role of human intervention in algorithmic recommendations within the criminal justice system.

Discrimination and Constraints: Evidence from The Voice
Updated: November 2022

Gender discrimination in the hiring process is one significant factor contributing to labor market disparities. However, there is little evidence on the extent to which gender bias by hiring managers is responsible for these disparities. In this paper, I exploit a unique dataset of blind auditions of The Voice television show as an experiment to identify own-gender bias in the selection process. The  first televised stage audition, in which four noteworthy recording artists are coaches, listens to the contestants “blindly” (chairs facing away from the stage) to avoid seeing the contestant. Using a difference-in-differences estimation strategy, a coach (hiring person) is demonstrably exogenous with respect to the artist’s gender, I  find that artists are 4.5 percentage points (11 percent) more likely to be selected when they are the recipients of an opposite-gender coach. I also utilize the machine-learning approach in Athey et al. (2018) to include heterogeneity from team gender composition, order of performance, and failure rates of the coaches. The  findings of offer a new perspective to enrich past research on gender discrimination, shedding light on the instances of gender bias variation by the gender of the decision maker and team gender composition.

Revisiting Unilateral Divorce, Health, and Crime
with Muhammad Salman Khalid and Morgan Stockham
Updated: May 2022

Stevenson and Wolfers (2006) theorized that unilateral divorce laws shifted power from one party to a distribution of power across both parties in the marriage, effectively providing an application of Coasian Bargaining. Utilizing a two-way fixed effects (TWFE) difference-in-difference estimator, they find that unilateral divorce had significantly reduced suicide rates, domestic violence, and intimate homicide. Innovations in econometric theory have raised concerns regarding the use of TWFE with differential timing in the treatment variable, leading to biased estimation. We revisit Stevenson and Wolfers (2006) with more modern estimators for suicide and intimate homicide rates and utilize appropriate estimators to examine the effect of unilateral divorce laws on suicides and intimate partner homicide rates. In contrast to the original research, we do not find significant effects of unilateral divorce on suicide and intimate homicide rates when controlling for covariates.

Pandemic Safeguards and Household Safety

with G. DeAngelo, Y.Le, S.Cunningham, and R.Thornton

A flurry of research has examined the effect of COVID-19-related policies on family violence with considerable variation in approach, often producing conflicting results. In this paper, we provide four main contributions to the existing literature. First, we utilize up-to-date estimation methods developed by Goodman-Bacon (2021) and Callaway and Sant’Anna (2020) to account for the differential timing in implementing COVID-19 policies and compare our estimates with traditional two-way fixed effects. Second, we use the most comprehensive data from the United States from 30 jurisdictions across 18 states to ensure that our conclusions are not reached due to data selection issues. Third, we evaluate three COVID policies: shelter-in-place, school closures, and daycare closures. Fourth, we use two measures of family violence: adult domestic violence and child violence. We find that school closure significantly doubled the number of child abuse calls per day from the mean. However, daycare closure significantly reduced 1.2 calls of child abuse calls per day. We detect no effect for shelter-in-place or daycare closure orders and document a reversal of our estimates’ direction when using Callaway and Sant’Anna (2020). to measure the impact of daycare closure relative to a two-way fixed effect.

Works in Progress

Effect of AI-driven Recommendation System on Worker Productivity and Service Quality

with Josie Xiao

I have utilized a field experiment to answer the question of how machine learning-driven integration into call center operations impacts organizational productivity and work performance. This technological integration aims to help Community Resource Advisors (CRA) to search for referrals with less time and more accuracy. Though the tool is built to enhance search accuracy and better outcomes, one may worry about unintended consequences. Therefore, we propose a randomized control trial to understand the effectiveness of a technological nudge on work performance, superficially the work that connects the population in need to social programs. Specifically, we aim to understand the effect of implementing a smart recommendation system on the quality of work for an organization that relies on memorization, experience, and on-spot decision-making.

Online Advertisement and Human Trafficking

with Beata Luczywek