Economics at your fingertips  

Measuring Racial Discrimination in Algorithms

David Arnold, Will Dobbie and Peter Hull ()

AEA Papers and Proceedings, 2021, vol. 111, 49-54

Abstract: Algorithmic decision-making can lead to discrimination against legally protected groups, but measuring such discrimination is often hampered by a fundamental selection challenge. We develop new quasi-experimental tools to overcome this challenge and measure algorithmic discrimination in pretrial bail decisions. We show that the selection challenge reduces to the challenge of measuring four moments, which can be estimated by extrapolating quasi-experimental variation across as-good-as-randomly assigned decision-makers. Estimates from New York City show that both a sophisticated machine learning algorithm and a simpler regression model discriminate against Black defendants even though defendant race and ethnicity are not included in the training data.

JEL-codes: J15 K40 (search for similar items in EconPapers)
Date: 2021
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (5) Track citations by RSS feed

Downloads: (external link) (application/pdf) (text/html) (application/pdf) (application/zip)
Access to full text is restricted to AEA members and institutional subscribers.

Related works:
Working Paper: Measuring Racial Discrimination in Algorithms (2020) Downloads
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link:

Ordering information: This journal article can be ordered from

DOI: 10.1257/pandp.20211080

Access Statistics for this article

AEA Papers and Proceedings is currently edited by William Johnson and Kelly Markel

More articles in AEA Papers and Proceedings from American Economic Association Contact information at EDIRC.
Bibliographic data for series maintained by Michael P. Albert ().

Page updated 2023-01-20
Handle: RePEc:aea:apandp:v:111:y:2021:p:49-54