Economists tend to think of public policy in terms of inputs and outputs: a policy is undertaken and there is a response. One can then evaluate how well the policy worked in terms of costs and benefits. When it comes to police work, the usual assumption has been that the function of police is to reduce crime, and that a higher quantity of police will tend to lead to lower levels of crime.

This approach isn’t exactly wrong, but it is incomplete. If the social goal is to have lower levels of crime, there may be multiple ways of accomplishing that goal along with hiring additional police officers. Policing is more than just the number of officers, but also has to do with how the policing is carried out: for example, are police often on foot or perhaps bicycles, being seen in the community, or are they arriving in cars with sirens blaring only after a crime has occurred? How do police manage their interactions with community members, including the continual need for aback-and-forth transition between being protectors and aggressors?

In the most recent issue of the Journal of Economic Perspectives, two papers address these issues of thinking through the functions of policing–and the input and output functions that researchers should use when studying policing. Emily Owens and Bocar Ba write “The Economics of Policing and Public Safety” (Fall 2021, 35:4, pp 3-28). Monica Bell contributes “Next-Generation Policing Research: Three Propositions” (pp. 29-48). I won’t try to summarize the articles here, but instead will pass along some thoughts worth considering.

(Full disclosure: I’ve been the Managing Editor of JEP since the first issue in 1987, so I am perhaps predisposed to find its articles to be of interest. These articles and indeed the entire corpus of JEP, from the most recent issue back to the first issue 35 years ago, is freely available online compliments of the publisher, the American Economic Association.)

There is considerable evidence, as Owens and Ba point out, that additional police do reduce the crime rate. The challenge to this research, as in so many cases, is to find a plausible way of separating cause and effect: for example, if one observes that cities with high crime rates had more police while cities with lower crime rates had fewer police, there would appear to be a positive correlation between police and crime. But of course, correlation isn’t causation, and responding to higher crime rates with more police doesn’t mean that the police caused the crime. The empirical challenge has been to find situations in which the number of police increased or decreased for reasons that had nothing to do with crime, and then to track what happened. One classic study, for example, note that politicians often boost the number of police before elections for political reasons–whether or not crime had actually been increasing–and that these increases in police do typically reduce crime. There is also considerable evidence that alternative strategies of policing, like the policies enacted in New York City in the early 1990s, helped to reduce the crime rate.

The benefits of lower crime rates can be divided into two categories: one is people not victimized; the other is that those who are not victims of crime can worry less and take fewer actions to protect themselves. On a per person basis, being a crime victim is much worse than just worrying about it. But there are many, many more nonvictims of crime than victims, and when you add up all the benefits, the total benefits of crime reduction are probably much higher for the nonvictims of crime than for the would-have-been victims.

Being the subject of police engagement is costly, and Owens and Ba report that about 24% of civilians have some engagement with police in a given year. These costs are not evenly distributed across society. Indeed, as Owens and Ba point out, those who fear crime and support more aggressive policing will also sometimes be those who do not bear the costs of more aggressive policing. Owens and Ba write:

External pressure on police departments tends to encourage crime reduction, with less attention to legitimacy costs. Even if a department was able to identify socially optimal police policies that carefully consider the distribution of direct and indirect benefits, it is not obvious those policies would be implemented by individual police officers. Current tools available to police managers to encourage individual officers to behave in the interest of the department are, from a personnel perspective, limited, and have been only rarely shown to alter police behavior in the field. Instead, many compensation and oversight strategies tend to encourage individual officers to make arrests and to emphasize officer’s personal safety. While these goals are both important and laudable, they are different from engaging with the public in a socially optimal way.

Owens and Ba take a granular look at the incentives of the various actors in the police structure. For example, politicians have an incentive to appeal to the median voter and to raise revenue. Thus, their policies toward policing will either vocalizing the views of those who fear crime or those who feel the costs of aggressive policing, depending on who they see as the median voter in the community. Politicians will also tend to favor actions by the police that lead to more revenue. These incentives may not coincide with encouragement of socially optimal policing behavior.

At a day-to-day level, police officers have an incentive to get home safely, clear offenses, and avoid complaints. Police departments can try to affect their behavior in the ways that officers are hired, trained, and monitored. But police unions typically enforce a salary structure where raises are based on experience and promotions–and there is little or no room for financial incentives aimed at other aspects of police behavior. In addition, police unions care about protecting their members from any and accusations. Many police departments have community outreach events, but the community members who participate in these events are not a randomly chosen cross-section. Some police departments do not reflective the ethnic mix of their community.

Putting all this together, Owens and Ba write:

While police departments and officers are tasked with solving a complicated social welfare problem, the structure of institutional incentives is relatively simple. The dominant incentives faced by police departments are to develop policies which provide indirect benefits—to make civilians feel safe and to see the police “doing something” about crime. As long as only a small fraction of the population is directly affected by criminal victimization and only a small fraction of the population bears the cost of achieving the direct and indirect benefits of crime reduction, we would expect that rational, vote-maximizing politicians might not object to policies that either under-provided or over-provided police engagement in specific areas of concentrated disadvantage. …

On top of truly optimal social policies being difficult to identify, individual officers
within a department have substantial discretion in how they engage with the
public. Standard strategies that organizations use to provide incentives for workers
are limited by structured wage and promotion mechanisms, high monitoring costs,
and limits on the ability to sanction employees. There is currently little evidence
base with which one might identify screening, training, or monitoring strategies
that support a department of officers who are able to make welfare enhancing decisions
about civilian engagement.

Monica Bell raises many similar issues from a slightly different perspective. She stresses the importance of spelling out the costs that arise from more aggressive police interactions: for example, there are studies that aggressive policing in certain areas is associated with lower educational performance for teenager, for lower stress-related health outcomes ranging from high blood pressure to anxiety/trauma, and even to increased residential segregation or lower voting rates. In short, the costs of aggressive policing are not to be brushed as a minor inconvenience.

Bell points out that alternative methods of reducing crime are often under-researched. Here’s one example:

But there are a number of other community-based programs or alternatives to traditional policing that remain largely unstudied, even though some of them are becoming models for other jurisdictions across the nation. For example, CAHOOTS (Crisis Assistance Helping Out on the Street) started in Eugene, Oregon, to send two-person clinical response teams to aid people in mental health crisis, without relying on armed police officers. Although the program has existed for more than three decades, in summer 2020 it gained national attention and became the model for numerous pilot programs—in San Francisco, Denver, Rochester, Toronto, and more. Eugene’s CAHOOTS program is funded and overseen by the police department, but some other emerging programs are funded and managed separately from police. Despite its long duration … CAHOOTS has never been rigorously evaluated.

Another largely unevaluated program seeks to develop quantitative metrics, based on local community involvement, for steps that might reduce crime. Examples might include improved outdoor lighting in certain areas, or an exercise program, or a youth jobs program.

My point here is not to be starry-eyed about possible alternatives to police, but instead to take seriously the idea that if more police are the only item in your policy toolkit for reducing crime, you aren’t serious enough about actually reducing crime–or about building the kinds of communities where young men in particular are less prone to crime. I would also say that if police and appropriate policing strategies aren’t a substantial part of your overall policy toolkit for addressing crime, you also aren’t serious enough about reducing crime.