Working Paper: NBER ID: w28222
Authors: David Arnold; Will S. Dobbie; Peter Hull
Abstract: There is growing concern that the rise of algorithmic decision-making can lead to discrimination against legally protected groups, but measuring such algorithmic discrimination is often hampered by a fundamental selection challenge. We develop new quasi-experimental tools to overcome this challenge and measure algorithmic discrimination in the setting of pretrial bail decisions. We first show that the selection challenge reduces to the challenge of measuring four moments: the mean latent qualification of white and Black individuals and the race-specific covariance between qualification and the algorithm’s treatment recommendation. We then show how these four moments can be estimated by extrapolating quasi-experimental variation across as-good-as-randomly assigned decision-makers. Estimates from New York City show that a sophisticated machine learning algorithm discriminates against Black defendants, even though defendant race and ethnicity are not included in the training data. The algorithm recommends releasing white defendants before trial at an 8 percentage point (11 percent) higher rate than Black defendants with identical potential for pretrial misconduct, with this unwarranted disparity explaining 77 percent of the observed racial disparity in algorithmic recommendations. We find a similar level of algorithmic discrimination with regression-based recommendations, using a model inspired by a widely used pretrial risk assessment tool.
Keywords: algorithmic discrimination; pretrial bail; racial bias; machine learning
JEL Codes: C26; J15; K42
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| sophisticated machine learning algorithm discriminates against black defendants (J79) | recommends their release at an 8 percentage point (11%) lower rate than white defendants with identical potential for pretrial misconduct (K14) |
| unwarranted disparity accounts for 77% of the observed racial disparity in algorithmic recommendations (J70) | algorithm discriminates against black defendants (J70) |
| similar levels of discrimination in regression-based recommendations (C52) | algorithm discriminates against black defendants (J70) |
| algorithm's recommendations are influenced by race-specific covariances between qualification and treatment recommendations (C32) | algorithm discriminates against black defendants (J70) |