Becker Friedman Institute

Research Repository

Research. Insights. Impact. Advancing the Legacy of Chicago Economics.

Addressing Seasonality in Veil of Darkness Tests for Discrimination: An Instrumental Variables Approach

Veil of Darkness tests identify discrimination by exploiting seasonal variation in the timing of sunset to compare the rate that minorities are stopped by police at the same hour of the day in daylight versus darkness. Such tests operate under the presumption that race is more easily observed by police prior to traffic stops during daylight relative to darkness. This paper addresses concerns that seasonal variation in traffic patterns could bias Veil of Darkness tests. The conventional approach to addressing seasonality is to restrict the sample to a window around Daylight Savings Time (DST) changes when the time of sunset is abruptly changed by one hour twice a year. However, this restriction reduces the variation in the timing of sunset potentially exacerbating measurement error in daylight and may still fail to address seasonality. The latter point is due to the fact that a substantial fraction of the seasonal change in daylight hours occur in the fall and spring (near DST) because of the elliptical nature of earth’s orbit. Therefore, we consider an alternative to simply restricting the sample to fall and spring where we instead apply an instrumental variables and fuzzy regression discontinuity approach. Our approach allows us to isolate the treatment effect associated with one hour of additional daylight on the share of police stops that are of African-American motorists. We find larger racial differences in Texas highway patrol stops using the regression discontinuity approach as compared to the annual sample, even though traditional approaches using the DST sample yield smaller estimates than the annual sample. The larger estimates are robust to the fall DST change sample, addressing concerns that motorists are tired and more accident prone immediately after the spring DST change.

Jesse Kalinowski, Quinnipiac University
Matthew Ross, Wagner School of Public Service, New York University
Stephen L. Ross, University of Connecticut
Publication Date: 
April, 2019
Publication Status: 
Document Number: 
File Description: 
First version, April, 2019