Will the Current Crisis in Police Legitimacy Increase Crime? Research Offers a Way Forward

Police legitimacy is all too timely a topic. As I am writing this commentary, instances of police use of lethal force against unarmed civilians are prominent. In Cincinnati, North Charleston, Cleveland, Staten Island, and elsewhere, video recordings have documented what appears to be unprovoked or at least unwarranted violence by a uniformed officer against someone stopped for a minor violation. The fact that the victims in these publicized encounters are minorities, and that the officers’ version of events is sometimes contradicted by the video evidence, suggests that such attacks may be quite common, as has long been asserted by the African American community. Starting with the death of Michael Brown in Ferguson, Missouri, the “Black Lives Matter” campaign has provided a slogan for a growing perception that young black men are targeted by the police in an all-too-literal sense. In June, Gallup reported a drop of 5 percentage points in the fraction of the adult American public who have “a great deal” or “quite a lot” of confidence in the police, as compared to the previous poll on this topic in 2013 (http://www.gallup.com/poll/183704/confidence-police-lowest-years.aspx). While the police still enjoy more trust than most public institutions, it appears that the publicity given to unwarranted police violence has undercut their legitimacy, at least for the time being. What are the implications of this loss?

Tom Tyler, Phillip Goff, and Robert MacCoun argue that police legitimacy is an end in itself but also an important asset in gaining the cooperation and compliance of the public, both with police authority and with the law itself. As the authors suggest, these ideas are becoming well established. In fact, they are supported by a 2004 report of an expert panel convened by the National Research Council (NRC), for which Tyler was a consultant and source (Skogan & Frydl, 2004). The NRC panel contrasted legitimacy with lawfulness, noting that “by legitimacy we mean the judgments that ordinary citizens make about the rightfulness of police conduct and the organizations that employ and supervise them” (p. 291). By this definition, legitimacy is subjective but influenced by the actions of police: “When they adhere to the rules, maintain their neutrality, and treat people with dignity and respect, police legitimacy increases” (p. 292). The NRC panel went on to observe that legitimacy is important not only in influencing the behavior of individual citizens but also in the political process: “If citizens trust the police, they will be willing to invest more authority in the police and spend more taxpayer dollars on them” (p. 291). Thus, the NRC panel’s analysis is the basis for a set of predictions about the consequences of the loss in public trust, such as has occurred over the last year or two:

Note that the reduction in police budgets (3) would tend to amplify the hypothesized increase in crime (1) by reducing police capacity to surveil high-crime areas and “clear” crimes through arrest and conviction—that is, to deter crimes and incapacitate some criminals. The same could be true for the second mechanism, if reduced cooperation undercuts police capacity to investigate crimes and build strong cases against perpetrators. What does the evidence suggest?

In their assessment of the effect of legitimacy on crime rates, Tyler, Goff, and MacCoun stress the direct effect of police legitimacy on voluntary compliance with the law. However, they suggest that the deterrent effect of police activities, while not irrelevant, tends to be weak and uncertain. Below, I provide a sampling of the evidence that I believe supports a stronger conclusion: that police resources—the department’s budget and capacity to enforce the law—have a demonstrably direct effect on crime rates (Durlauf & Nagin, 2011). If the legitimacy “crisis” has the effect of reducing police budgets and law-enforcement mission, the predictable result will be to further increase crime rates. The sad irony is that crime, and especially violent crime, tends to be concentrated in poor minority communities, which would suffer disproportionately from a crime increase.

The evidence on the criminogenic effects of a reduction in police capacity to enforce the law starts with historical incidents in which police capacity dropped sharply (Andenaes, 1974; Sherman, 1992): In North America, the list includes the police strike in Boston in 1919, the 16-hour Montreal police strike of 1969, and the Baltimore police strike of 1974, each of which was followed by a breakdown in public order. It seems plausible that a suspension of police activity in any large city today would have similarly dramatic consequences. The total deterrent effect of the police on crime (i.e., the hypothetical difference between the crime rate at current police capacity and what it would be with zero capacity) is very large. That conclusion sets the stage for an inquiry about the possible crime effects of much more typical changes in police capacity of a few percentage points in either direction. Fortunately, there is considerable experimental and quasi-experimental evidence on just that subject, evidence that helps establish a causal effect of police capacity by identifying instances in which changes were exogenous, rather than being influenced by the trend in crime.

In 2003, state budget cutbacks mandated by a voter referendum in Oregon forced an abrupt cut in Oregon State Police capacity by 36%, resulting in a similar-sized reduction in citations issued. A careful analysis documented a 12% to 14% increase in highway traffic fatalities as a direct result (DeAngelo & Hansen, 2014).

The Office of Community Oriented Policing Services, a federal program known by the acronym COPS, awards funds to localities to hire additional police. Two economists, working with data from the period 1990 to 2001, found that this program was successful in increasing police capacity in jurisdictions that secured a grant, and the result was substantial reductions in auto thefts, burglaries, robberies, and aggravated assaults (Evans & Owens, 2007). Similarly, a program in England and Wales called the Street Crime Initiative awarded resources to 10 of 43 police districts amounting to £24 million per year in fiscal years 2003 and 2004; the estimated effect of this quasi-experimental program was to reduce robbery (the target crime) by enough to pass a cost-benefit test (Machin & Marie, 2011).

Several articles have analyzed the effects of police redeployments caused by terrorist alerts. Klick and Tabarrok (2005), for example, observed that changes in the terror alert level set by the Department of Homeland Security mandated changes in police deployment in Washington, D.C., that were unrelated to crime trends. Using daily crime data during a period in which this system was in place, they showed that the level of crime decreased sharply during high-alert periods. The decrease in the level of crime was concentrated in the National Mall, the focal point of the terror alert and hence the boost in police presence. A similar type of evidence was uncovered by Di Tella and Schargrodsky (2004), who analyzed the effect of the redeployment of Buenos Aires police following the bombing of a Jewish synagogue. The analysts demonstrated that the temporary concentration of police near potential Jewish and Muslim targets had the side effect of sharply reducing the local auto theft rate.

Actually, the concentration of police capacity in particular places is not just happenstance but has become part of a widely practiced strategy known as “hot spots” policing (Weisburd & Telep, 2014). Twenty years ago, the Minneapolis Hot Spots Patrol Experiment demonstrated that a concentrated police presence in high-crime blocks reduced crime in those areas, even though the police were given no specific instructions except to show up (Sherman & Weisburd, 1995). Subsequent field experiments and other evidence have further strengthened the case for focusing police resources in high-crime areas, finding a direct effect in the target areas and little or no displacement while providing some indication of which tactics seem most effective in cooling the “hot spots” (Braga & Bond, 2008; Braga, Papachristos, & Hureau, 2012).

Also relevant is the quasi-experimental evidence that urban expenditures on private security have large crime-prevention effects. Cook and MacDonald (2011a) found a direct dose-response effect of private security in the context of business improvement districts created in Los Angeles during the 1990s, with each dollar spent on private security preventing $20 worth of crime and no evidence of displacement to other areas. Presumably, legitimacy is of little consequence when it comes to the extra surveillance generated by private security. Rather, the presence of closed-circuit cameras and private guards enhances the likelihood of apprehension—and the perception of that reality—for would-be criminals. The same can be said for LoJack, the vehicle-tracking system that greatly increases the likelihood that auto thieves and chop shop operators will be apprehended, with dramatic deterrent effects on theft (Ayres & Levitt, 1998).

Consistent conclusions emerge from this and other evidence: that both property and violent-crime rates are directly influenced by the capacity of the police to surveil public places and respond to crime when it occurs. Police budgets and civilian cooperation with the police are both important inputs into that capacity. The secular increase in both budgets and some measures of cooperation since the 1970s may deserve part of the credit for the extraordinary drop in crime rates (Cook & MacDonald, 2011b). If fewer dollars go to the police as a result of bad publicity, the result will be higher crime rates than would otherwise be the case, undermining this progress.

What about the direct effect of legitimacy loss on crime? As noted by Tyler, Goff, and MacCoun, laboratory experiments provide an empirical base for supposing that ordinary people are more inclined to comply with rules if they perceive the rules and the enforcement regime as fair. For example, Friedland, Thibaut, and Walker (1973) ran an experiment with 96 undergraduates in which they were assigned to a worker’s role in a complex management game. The amount of “cheating” in the game was influenced by whether “bribery” was allowed (thus undercutting the legitimacy of rule enforcement) and whether workers’ performance benefitted only the “managers” or both “managers” and “workers.” The laboratory evidence is intriguing and persuasive in establishing a causal mechanism, and it should provide the impetus for new field research demonstrating that this mechanism is important in influencing street crime. The evidence from the field on this matter is not as compelling as the findings cited above concerning police capacity.

Lorraine Mazerolle and her colleagues (Mazerolle, Bennett, Davis, Sargeant, & Manning, 2013) conducted a Campbell Systematic Review of research on police legitimacy through 2009. They reported strong effects for “compliance and cooperation,” especially of offenders, but varied results for crime depending on how it was measured:

Finally, police-led legitimacy interventions showed a marginal effect on reoffending as an indirect outcome measure. When reoffending was broken down by measurement method, studies that measured reoffending using official police data and self-reported reoffending showed no effect of legitimacy interventions; however, studies that measured self-reported victimization showed a large decrease in revictimization as a result of the interventions. (p. 73)

Further research is needed on how legitimacy interventions affect the tendency to self-report victimization and offending and to report crimes to the police (Blattman, Jamison, Koroknay-Palicz, Rodrigues, & Sheridan, 2015).

Subsequently, Mazerolle, working with Angela Higginson, extended the review through 2012 and focused specifically on the crime effects of “legitimacy policing of places” (i.e., of geographic areas). They found 33 studies of police-led interventions that included at least one element of procedural justice. “Overall, we find that spatially focused policing interventions that incorporate a procedurally just dialogue result in a measurable decrease in crime and disorder in the intervention areas” (Higginson & Mazerolle, 2014, p. 429). While some of these studies included strong evaluation designs, the interventions tended to be quite complex. The relevant mechanism or mechanisms that account for the crime drop are not well identified in interventions like “Weed and Seed,” which was used in 14 of the 33 studies. “The Weed and Seed programs involve a wide variety of strategies including active law enforcement, prevention, intervention and treatment, neighborhood restoration, and community policing” (p. 441). That community engagement (an aspect of procedural justice and hence legitimacy) is one aspect of the Weed and Seed programs does not settle the question of the importance of that particular design feature. I would conclude that while legitimacy may have a direct effect on real-world crime rates, the importance of that specific mechanism has not yet been well established in research. The tragedies of current events make the need for such research more pressing.

Tyler, Goff, and MacCoun make a strong case for taking police legitimacy seriously, and they point out that many police officials and other public leaders have recognized its importance. That recent events have undercut the public’s trust in the police is a setback to this effort; increased training, the expanded use of body cameras, and other measures may ultimately repair the damage to public perception, but that remains to be seen. The stakes are high. Whether crime increases as the direct effect of lost legitimacy or as the indirect effect of reduced enforcement capacity, one result will be to further burden poor minority communities. Those communities would benefit from more policing, not less, but of the right sort. Research offers a way forward; it cannot determine society’s goals, but it can provide the means to achieve the goals of a just society, and to do so more efficiently and effectively.