Racial profiling in traffic stops results in worse policing, study finds

A police officer pulled over driver for traffic violation
A police officer talks to a driver he pulled over for speeding. Photo credit: Kali9 for Getty Images

It’s well-documented that police are far more likely to search Black and Latino drivers than whites during routine traffic stops. These disparities are not only socially divisive, but they result in inefficient policing, a new study from UC Berkeley’s Haas School of Business has found.

In fact, if police were to search all drivers they stop with the same frequency, they would find slightly more illegal items than they do now, said Asst. Prof. Conrad Miller, a labor economist who co-authored the working paper with Benjamin Feigenberg of the University of Illinois at Chicago. 

“What we wanted to know is, are these disparities in search rates justified in some way? Is there a clear cost to restricting them?” said Miller. “If the point of police searches is to recover contraband and have fewer illegal items on the street, then the answer to both questions is ‘no.’”

What we wanted to know is, are these disparities in search rates justified in some way? Is there a clear cost to restricting them? If the point of police searches is to recover contraband and have fewer illegal items on the street, then the answer to both questions is ‘no.’ —Conrad Miller

Traffic stops are the most common interaction between police and the public, accounting for roughly 40% of all contacts. Miller and Feigenberg analyzed comprehensive data on 5 million traffic stops and 53,000 searches by Texas Highway Patrol troopers between 2009 and 2015, and found stark racial disparities.

Comparing drivers stopped for speeding along the same stretches of highway at the same time of day, Blacks were searched 2.7 times as often as whites, and Latinos 1.7 times as often. Yet though white drivers were searched less frequently than the other groups, they were the most likely to have contraband: searches of Blacks were 15% less likely and searches of Latinos were 30% less likely to turn up anything illegal.

The disparities in search rates decreased somewhat when the researchers controlled for criminal history at the time of the stop. Yet among motorists with no criminal history—either prior to the stop or at the end of the six-year data period—Black drivers were nearly three times more likely to be searched, and Latino drivers were nearly twice as likely to be searched after being pulled over. These patterns were true no matter the race or ethnicity of the trooper. 

These wildly divergent search rates have been previously documented, but the reasons for them have been long debated. For economists, who are focused on efficiency, the important question is not determining average search rates, but whether police are setting different thresholds for searching Black and Latino drivers. Perhaps racial disparities in search rates simply result from police using their expertise to target the most suspicious people, the argument goes. If police were forced to search everyone at equal rates, they might face a higher bar in searching non-white drivers, and would be stymied in doing their jobs. 

Miller and Feigenberg address this argument head on, and refute it. If troopers in the field are able to accurately gauge who is more or less likely to be carrying contraband, they would see diminishing returns as they search more people—since presumably they’d be searching more who are only borderline-suspicious. But this is not the case, the authors found. While some troopers search far more people than others, those who search twice as often find twice as much contraband. In other words, the share of searches that yield contraband remains about the same and does not taper off, no matter how many people are searched. 

“It’s been argued that troopers can discern who is more suspicious, and then they move on down the curve, but it doesn’t appear the decisions they’re making are that sophisticated,” Miller said.

It’s been argued that troopers can discern who is more suspicious, and then they move on down the curve, but it doesn’t appear the decisions they’re making are that sophisticated.”

What would happen if police searched everyone at the same rate? The researchers built a model and projected that without increasing their overall search rate, police would find about 1.5% more contraband—not a big increase, but also not the drop that some have feared if profiling were to be eliminated. Meanwhile, searches of Black drivers would be cut nearly in half, searches of Latino drivers would drop slightly, and searches of white drivers would increase by about 1/3. 

“What  we found suggests that in a world where people were searched at equal rates, you’d have less racial disparity and there wouldn’t be a cost in terms of increased crime,” Miller said. This has other important implications in the current debate over use of force, he added. “It’s more likely that force will be used if a search is conducted, so decreasing searches of Black and Latino motorists decreases the risk that things go badly and force is used.”

Back