Rethinking Racial Profiling: A Critique of the Economics, Civil Liberties, and Constitutional Literature, and of Criminal Profiling More Generally

Bernard E. Harcourt, Columbia Law School


New data on highway stops and searches from across the country have spawned renewed debate over racial profiling on the roads. The new data reveal consistently disproportionate searches of minority motorists, but, very often, an equal or lower general success rate – or "hit rate" – associated with those searches. Economists are developing new models of racial profiling to test whether the data are consistent with policing efficiency or racial prejudice, and argue that equal hit rates reflect that the police are maximizing the success rate of their searches. Civil liberties advocates are scrutinizing the same data and, in most cases, reaching opposite conclusions. They argue that equal hit rates merely reflect similar offending patterns by race and thus that the disproportionate searches are racially biased. Meanwhile many constitutional commentators decry racial profiling on the highways as "plainly unconstitutional," while courts draw technical legal distinctions to easily dispose of civil suits alleging racial profiling on the roads.

The debate over racial profiling on the highways is becoming increasingly empirical, technical, and engaged. It is also focusing increasingly on the issue of policing efficiency. The problem is, the debate is asking the wrong question and tracking the wrong statistic. The key question is not whether racial profiling maximizes the success rate of searches, and the key statistic is not the comparative hit rate by race associated with those searches. Instead, the key question for purposes of the empirical, policy, and constitutional analyses is: What are the conditions under which it is justifiable to use race in policing? The key statistics, it turns out, are the comparative elasticities of offending to policing and the relative offending rates of the different racial groups.

When we pose the right question properly, it becomes clear that both sides of the debate have it wrong: the use of race in police searches is neither plainly unconstitutional nor simply efficient. Racial profiling on the highways can be justified as an effective law enforcement tool or, from a constitutional perspective, as a narrowly tailored policing technique that promotes a compelling governmental interest in law enforcement if the following three conditions are satisfied: (1) racial profiling has a long-term negative effect on the profiled crime, (2) while increasing the efficient allocation of police resources, (3) without producing a ratchet effect on the profiled population. These three specific conditions will only be satisfied in certain identifiable situations of comparative elasticity and offending as between racial groups.

Under these narrow conditions, race can constitutionally be used in policing to advance a traditional law enforcement interest in combating crime. There may be other non-law enforcement interests that warrant using race in policing as well. There may be a compelling interest in having a prison population that reflects more accurately the demographic distribution of the offending population, or even of the population as a whole. There may be a compelling interest in combating crimes committed against historically disadvantaged groups. But with regard to the traditional law enforcement interest of fighting crime, race can only properly be used under these three narrow conditions.

Race in this criminal justice context should not be treated differently than in other contexts, such as education or employment. If racial profiling satisfies the three narrow conditions, then opposition to racial profiling should be based on the grounds of affirmative action: because of this country's institutional history of racism, or in order to achieve a more balanced carceral population, or for other compelling reasons, the police should endeavor to minimize the minority representation in prison by profiling white offenders. Conversely, if racial profiling does not satisfy any one of the three conditions, then racial profiling should be conditioned on compensating innocent minority motorists who are searched for wasting their time, for diminishing their dignity, and for inflicting emotional harm. If one of the conditions is not satisfied, innocent minority motorists are being used for other purposes – for example, to increase search success rates regardless of a ratchet effect – and they should be compensated for the taking.

As an empirical matter, it is fair to speculate, drawing reasonable inferences from other data sources, that minority motorists may have slightly lower elasticity of offending to policing because of diminished job opportunities and other market alternatives, and slightly higher offending rates because of drug trafficking patterns. As a result, racial profiling on the roadways may increase the overall costs to society, including the amount of profiled crime, and likely produces a ratchet effect on the profiled population, resulting in a greater disproportion of minority arrests or negative contacts with the police over and above any possibly higher offending rate. This is going to have significant repercussions on minority motorists: it is likely to more unevenly distribute criminal records, supervision, and post-punitive collateral consequences, and to significantly boost the public perception that minorities are drug users and drug dealers. Racial profiling on the highways, accordingly, is poor crime policy and, because of the ratchet effect, is not narrowly tailored to the governmental interest in law enforcement. Given that no federal or state agency has attempted to establish the three narrow conditions under which race could properly be considered in policing, the practice of racial profiling on the highways should be considered presently unconstitutional.

The important point is, however, that racial profiling on the highways is presently unacceptable not because of any per se constitutional bar on using race in police searches, but rather because of the mathematics of criminal profiling. The central problems with racial profiling – possible adverse long-term effects on the profiled crime and probable ratchet effect – are problems about criminal profiling in general, not about race. The same problems may infect any type of criminal profiling, whether of minorities for drug possession, of the wealthy for tax evasion, of single mothers for welfare fraud, or of white males for domestic terrorism or serial murder. To be sure, the ratchet effect is most disturbing when it plays on race, as well as gender, social or family status, class, or wealth. The ratchet effect violates a core principle of punishment theory, namely that anyone who is committing the same crime should face the same likelihood of being caught, and that race, gender, class, or status should not affect that equation. But both the ratchet effect and the long-term effect on the profiled crime are phenomena that may undermine any scheme of criminal profiling. In this sense, the debates over racial profiling on the highways should make us reexamine our views on the larger question of criminal profiling more generally.