Economics Research on Racial Disparities in Policing

The Issue:

Protests following the horrific death of George Floyd in Minneapolis have brought a new focus to racial disparities in U.S. policing. Administrative data and quantitative research has long documented large race disparities in arrests, traffic searches, speeding tickets, and use of force. For many years, economics research into racial disparities in policing has focused primarily on trying to understand the role that different forms of discrimination on the part of police officers contributes to disparate outcomes. Newer work is honing in more specifically on the issue of race disparities and racial bias in police use-of-force and on evaluating the effectiveness of different types of interventions and reforms to address police misconduct and race disparities in policing.

Large race disparities in U.S. policing mean that Black, white and Latino individuals have vastly different experiences interacting with police officers.

The Facts:

• There are stark racial disparities in policing outcomes in the United States. Black civilians consistently make up a disproportionate share of arrests: in 2018, for instance, 27% of all arrests were of Black civilians while Black civilians make up only about 13 percent of the total U.S. population, according to data collected from police departments by the Federal Bureau of Investigation (FBI). These data also show that Black civilians are arrested at over two times the rate of white civilians. The self-reported Police Public Contact Survey (PPCS) offers data from a smaller sample on police interactions beyond arrests. These data show Black individuals made up a much higher share of traffic stops and instances of use of force than would be accounted for by their share of the U.S. population in 2015, the latest available PPCS survey (see chart). While African-Americans and Hispanics were 13% and 17% of the U.S. population, respectively, they comprised 18% and 15% of self-reported street stops, 19% and 17% of arrests, and 20% and 21% of instances of use of force in 2015 according to PPCS data. While fatal police use-of-force incidents are not formally tracked by the federal government, the civilian-collected database fatalencounters.org estimates that, since 2015, Black and Hispanic civilians comprise 29% and 17% of instances of fatal use of force. The overall picture provided by aggregate statistics can differ from particular conditions in specific cities. In Minneapolis, the site of George Floyd’s death, nearly 60% of all police shootings since 2008 involved a Black civilian (see here).
• Race disparities in policing reflect multiple potential sources of inequities and discrimination. Some of the disparities could be coming from differences in crime rates across neighborhoods. Black and Latino civilians have lower levels of income and wealth, live in neighborhoods with higher levels of pollution, have access to lower quality public schools and obtain lower levels of education, on average. These factors tend to decrease economic opportunity and are associated with higher levels of crime. These unequal circumstances can also contribute to police officers engaging in racial profiling or statistical discrimination (the practice of using race as a signal of criminality). Police officers can also exhibit what economists call "taste-based racial discrimination", whereby officers are prejudiced against Black and/or Latino civilians and behave with particular animus towards these groups. These biases do not need to be active or conscious; researchers have compiled a wealth of evidence on the ubiquity of subconscious or implicit racial biases, and these biases likely affect the ways that police interact with civilians. Beyond the actions and attitudes of police officers and differences between communities, there are also higher level policies that direct policy focus and resources in ways that can have large and unequal impacts on minority communities. The United States has a history of criminal justice institutions that have formalized discrimination in practice, including through vagrancy laws, legal enforcement of segregation, “broken windows” policing and the “war on drugs”; the legacy and echoes of these systems contribute to race disparities we see today. 
• How much of the racial disparity in police treatment is driven by police officer prejudice or bias? A goal of economics research into racial disparities in policing has been to attempt to identify the role of taste-based racial bias or police officer racial animus as a contributor. In general, most studies find evidence that is consistent with racial profiling. Studies that look at vehicle searches or stop and frisk find that when you compare individuals who look the same in the data — by individual characteristics, neighborhoods, or crime conditions — Black individuals, and to some extent Hispanic individuals, are much more likely to be searched. Similarly, studies find that police tend to stop fewer Black drivers in traffic stops at night when it is more difficult to observe race (see here). But the evidence that racial animus — which is narrowly defined in the literature — plays a large role in driving disparities is less consistent. Researchers have constructed empirical tests that try to identify whether differences across races that persist after taking into account other characteristics can be attributed to taste-based racial discrimination. For instance, the “hit rate” test asks whether police searches of individuals are equally likely to find contraband across the race of civilians searched (if it is found that minorities are less likely to be carrying contraband, then a higher rate of searching minorities would be taken as an indication that minorities are being searched at higher rates out of prejudice). Several studies have applied this approach to data on police searches of vehicles and have found mixed evidence for this form of racial bias, that depends on both context and research methods used (see here). For example, a study of the New York Police Department (NYPD) practice of “Stop, Question, and Frisk” found that officers applied a uniform standard of guilt when deciding whether to stop individuals of different races, and failed to find evidence of racial animus among officers. A separate approach to testing for racial bias compares the behavior of officers of different races and asks whether the treatment of opposite-race civilians is worse than treatment of same-race civilians, where a difference in outcomes is taken as evidence of bias. These methods have found mixed evidence of taste-based racial discrimination in vehicle searches (see here and here) but have led to findings of this form of discrimination in traffic accident investigations, and traffic fines.
• Scholars have only recently directly evaluated disparities in police use-of-force and more specifically fatal incidents. Because records on use-of-force are not consistently maintained across departments and are not reported to the federal government, research has been hampered by the lack of reliable data on this question. However, emerging research is finding that disparities in the use of force against minorities persist after taking into account a range of characteristics. One of the first studies on this topic used nationwide data on the number of police shootings in each county and found that minorities are more likely to be shot by the police and that this disparity could not be explained by differences in county-level arrest rates across race groups. A 2019 study found that relative to white civilians, Black civilians are more likely to have low-level force used on them by officers, and that this disparity could not be fully explained by differences in incident or offense characteristics; however, when focusing on lethal use of force, the study failed to find evidence of racial discrimination. A related study found that while all officers are more likely to use force when dispatched to minority neighborhoods, white officers have a disproportionate increase in their force propensity when policing minority neighborhoods. The evidence on disparities in lethal use-of-force is more difficult to establish because this occurrence is comparatively rare. Much more research is needed to determine the extent of race disparities and discrimination in use-of-force, especially in lethal force.
• New research is evaluating interventions to understand what works to reduce the incidence of police misconduct and race disparities in policing. A first strand of research, including our own, looks at the extent to which findings of disparity in treatment correspond to widespread police behavior versus the actions of particular police officers. Studies document that there are large differences in individual officer behavior, including through officer differences in racial bias in assigning fines for speeding, in arrest behavior and indicators of arrest quality, and in pedestrian stops. This matters for proposing interventions. The fact that individual officers behave in very different ways suggests that there is scope for impacting racial disparities in policing through additional monitoring and oversight of officer behavior. For example, an analysis of detailed data from the Philadelphia Police Department found that the use of predictive analytics in hiring could substantially reduce the incidence of future police misconduct of officers once they are on the job. Making greater use of civilian oversight, through focusing on civilian complaints, also seems to have the capacity to improve policing outcomes. Research has found that civilian complaints are strongly linked to other negative behavioral outcomes for police officers, including use-of-force. New work finds that increasing police oversight and transparency can substantially reduce civilian complaints without negatively impacting police productivity or increasing crime rates. Predictive Early Intervention Systems (EIS) can leverage civilian complaints and other experiential data to flag which officers are most at risk of an adverse use-of-force event, and assist in administering tailored interventions for these officers. There is some evidence that police training in procedural justice and de-escalation can positively affect police actions: A recent experimental evaluation of this kind of training reduced the likelihood that trained officers resolved incidents with an arrest or were involved in use-of-force incidents. Scholars and practitioners have also considered the merits of other interventions such as explicit and restrictive department protocols on use-of-force, body-worn cameras, implicit bias training, diversity in hiring, targeted firing of officers and various forms of constitutional policing. But the current evidence on the efficacy of these interventions is limited or mixed, and additional research is needed to establish whether they may effectively reduce police use-of-force as well as racial disparities in policing.