Policing Studies Measure Benefits To Crime Reduction—But Not Social Costs

Perhaps one of the more unexpected events this turbulent year has been the rate at which efforts to defund police departments gained national political traction. Local governments collectively spend roughly $100 billion per year on policing, and with big cities dedicating about 15 percent (if not more) of their budgets to police, a growing number of people are asking if it may make more sense to spend some of that money elsewhere, like on drug treatment, mental health, social work, or shelter. One of the many questions raised by the defund movement: Is spending on police justifiable from a policy perspective? 

Remarkably, we do not have a good answer. If nothing else, there are only a handful of studies that really undertake clear cost-benefit analyses of policing. But the bigger problem is that even these studies fail to answer the question of the costs of policing, and that’s because of how they’re framed, not technical or methodological issues. There are at least three significant ways that these studies go astray.

The first thing to note is that even taken on their own terms, most studies of the effectiveness of police are not really measuring the effect of police. In reality, studies of the impact of police on crime are more studies of the impact of what Jane Jacobs, in her book “The Death and Life of Great American Cities,” called “eyes on the street.” What really deters is the risk of being seen by someone, not necessarily by someone who can or will make an arrest. These studies show that police can be those critical eyes on the street, but they do not show that only the police can be those eyes.

This may seem counterintuitive. The key issue is that most of the deterrent effect from policing does not come from apprehension, which is quite low: Police make arrests in fewer than half of all reported violent crimes and one-fifth or less for reported property offenses, and people report fewer than half of all crimes.  As criminologist Daniel Nagin has pointed out, deterrence comes far more from the police’s “sentinel” function—simply their eyes on the street and willingness to intervene when crime occurs—than from their arrest powers.

But if observing what is happening is what matters most, showing that police effectively do so does not mean that only they can do so. Take private security. Few studies have measured the effect of (usually unarmed) private security on crime, even though the United States employs far more people as private security than as sworn police officers. The few studies that have, however, appear to show that private security also reduces crime, often quite substantially, and perhaps with less risk of lethal or serious violence.

The obvious counterargument is that private security only works because it can detain people until the sworn police make an arrest. Sure, to a point. First, even if true, it implies that we can rely far less on armed police officers than many current studies suggest. But second, and perhaps more important, it’s not entirely clear if that is true. As Nagin also points out, we tend to underemphasize the importance of informal punishment, which might actually be a far more significant deterrent threat than formal punishment, especially given how sporadically formal sanctions are actually imposed. Shame, stigma, that disappointed head shake from a person you respect: all these matter, too, and they can be imposed without arrest, much less any sort of trial or conviction. To the extent that these informal sanctions are key, then it’s the risk of any sort of credible detection of  a crime, not detection with the threat of formal arrest, that really drives deterrence.

In fact, given the often fraught relationship between police and communities of color, detection by non-police actors may be more effective a threat than detection by the police. In many ways, this is what motivates programs such as Cure Violence, which relies on people respected in the community to intervene among people who are at risk for violent behavior—not the police, but former gang members and local pastors.

Our policing studies have shown only that police can be sufficient, not that they are necessary—or at least not as necessary as studies that focus exclusively on policing seem to suggest. There may be cases where the threat of formal punishment is still necessary, but likely far less often than conventional wisdom holds.

The second thing about our current assessments of the effectiveness of police is that they are cost-benefit analyses that measure the wrong costs. (This is a problem that plagues research on incarceration as well.) Perhaps ironically, they measure the benefits correctly: These studies try, as best as possible, to estimate how much people value any reduction in crime that comes from increased policing. But when it comes to costs, these studies don’t try to measure the social costs of policing. In other words, these studies don’t count George Floyd’s death as a cost, or the costs of far less publicized uses of force. They don’t estimate the impact of reduced civic engagement that comes in the wake of police violence, or the emotional and physical toll of the constant fear of police violence that forces Black parents to have “The Talk” with their kids. They don’t account for any of the micro or macro costs of, say, the fact that as many as 79 percent of all young Black men in New York City were stopped by the police at the peak of Stop, Question, and Frisk: the shame, the fear, the emotional and other effects of so racially targeted a policy.

Perhaps this oversight shouldn’t surprise us, since the fiscal costs of policing are generally the costs experienced by those who write these studies. Years ago I called the NYPD after my wallet was stolen from my apartment. They recovered my wallet—that’s the benefit—and since I was treated politely and with respect, the only cost to me really was my taxes that paid for the police. But it is almost surely the case that accounting for the social costs of policing will significantly affect how costs and benefits balance.

There’s a separate problem with using fiscal costs as the cost. The bulk of police budgets go toward wages, benefits, and overtime, which means that police spending is a form of government stimulus spending. Not necessarily an efficient one, but not exactly a cost either. It also means that the only way to really defund police is to cut pay, which itself will carry with it real costs—unemployment and the dislocations that brings—that shouldn’t be ignored normatively or politically. (Failure to think about how much criminal justice spending goes to wages has led to problematic reform suggestions when it comes to prison closures as well.) 

Finally, studies of policing fail to confront the question of the opportunity costs of policing. Even if $1 spent on policing yields $1.63 in less crime—the findings of one well-designed and widely cited study—that doesn’t actually mean we should spend more money on policing. What else could that $1 get us? One study suggests that $1 spent on medical drug treatment cuts crime on average by about $4, and does so without the collateral costs of policing, and with a whole host of other physical, mental, medical, and social benefits. Other studies suggest that certain environmental changes such as improved lighting in neighborhoods could be as popular as hiring more police per dollar spent, and again without the collateral risks of police violence and with additional returns to people in those communities. In other words, showing that a $1 investment in policing has a positive social return does not automatically mean “invest more.” And that outcome can be completely consistent with the idea of “invest less,” especially when there are viable alternatives with higher documented returns, and when cities are already dedicating significant chunks of their budgets to policing.  

None of this is to say that police are irrelevant or consistently impose net social harms; calls for defunding are not inherently calls for abolition. But even those who strongly believe that police play an important and essential role in the criminal legal system need to understand that none of the studies we have clearly establish that police must play a central role in fighting crime, or that the amount we spend on policing is at all optimal. It seems quite likely that the studies we do have establish, at best, an exaggerated upper bound on the returns on policing investment. Given the overlooked social costs, the net return is most likely lower than what these studies suggest. So, the opportunity cost issue is all the more important to confront—especially since many of these studies suggest that much of the police’s “sentinel” efforts could be substantially performed by other actors.

John Pfaff is a professor at Fordham Law School and the author of “Locked In: The True Causes of Mass Incarceration and How to Achieve Real Reform.”