On Balance: Bringing Benefit-Cost Analysis to Policing Practices

February 12, 2017

By Barry Friedman 

The Journal of Benefit Cost Analysis and the Policing Project at New York University School of Law teamed up to host a symposium on the use of benefit-cost analysis in a domain in which it is all too absent: policing. (Policing tends to have many definitions, but generally we mean it here to refer to any use of force or surveillance of the populace for reasons of achieving public safety.) The goal of this Symposium on Benefit-Cost Analysis of Policing Practices, and the conference that preceded it, is to interest more scholars in working in this vital field, and to identify and begin to tackle some of the methodological challenges the field faces.

The past few years have seen constant turmoil in the country around policing. This is true of street policing, like stop-and-frisk or the use of force, particularly against brown and black men and boys. It is equally true of policing surveillance: tactics such as location tracking and Stingray cell phone location devices, facial recognition and predictive analytics.

Listening to the rhetoric surrounding these controversies, it may seem as though the issue is primarily political, when in fact much of the problem is a lack of information needed to make healthy regulatory decisions. According to the Bureau of Justice Statistics within the US Department of Justice, we spend over $100 billion dollars annually on public safety—yet we have remarkably little understanding of what works and what does not.

Over the past several decades, policing has become more regulatory. Rather than simply trying to apprehend bad guys, which is what we commonly imagine as part of law enforcement, police practices today increasingly are aimed at the population at large, with the goal of regulating conduct broadly. The idea is to use means of policing that keep anyone from even thinking of violating the law. This is true of airport security, drunk driving road blocks, programmatic stop and frisk, and widespread surveillance. But do the regulatory practices policing agencies are adopting make sense?

Benefit-cost analysis (BCA), so prevalent around other regulatory domains, is comparatively thin around policing. There is some good work, but too little of it. And, very little of the scholarship deals with the social costs of policing. How can we evaluate policing practices without paying attention to these social costs, be they privacy interests around surveillance, or racial impacts, or the psychological costs of police stops concentrated in particular neighborhoods?

In the lead article in this symposium (Ponomarenko and Friedman, 2017), my co-author Maria Ponomarenko and I discuss the lessons learned at the convening held to discuss policing BCA, and some of the challenges scholars in this field may face. For the conference and the symposium, we gathered scholars from diverse fields, some of whom work on BCA regularly, and some of whom work in policing but not BCA. We had criminologists, as well as economists, legal scholars, and quantitative methodologists from the fields of environment, public health, and other areas.

One of the more interesting questions tackled at the conference, and reflected in the articles in this Symposium, was how to define the frame for a BCA around policing, which in turn will point towards the types of benefits and costs that should be included. Assume a police chief wants help with a BCA of some tactic or technology, like use of a license plate reader to locate people with outstanding warrants. Some of the costs are evident, like the cost of the equipment. So too are some of the benefits, like locating a fugitive from the judicial system. Yet, should the chief also consider the downstream costs of incarceration and the possible impact on the offender’s family? Should the chief take into account that the warrant may be for a very low-level offense? A social planner would want to consider all of this. But decision making in the justice system is fragmented, and so it is difficult to consider broad systemic effects.

One of the greatest challenges in the policing field is how to monetize intangible social costs and benefits. If, for example, widespread use of stop-and-frisk hurts community trust, or a community policing program enhances trust, how do we value those changes? Similarly, how do we value the psychological costs of police tactics, such as SWAT raids or stop-and-frisk? Contingent valuation is one method, but then the question is how to design and implement the special surveys or experiments, which can be costly, in a way likely to contribute to our understanding of these policing programs? Moreover, are there other ways to assess these costs and benefits, such as clever unearthing of revealed preferences?

Finally, those at the conference discussed the difficulties of assessing the effectiveness of policing practices, given that randomized controlled trials often are not practical when policing practices are at issue, in part because of ethical concerns with randomizing treatments that may have very real consequences for communities. We talked about natural experiments, and making use of rollouts and rollbacks, to learn what works and what does not.

The first job of government is assuring public safety. Without it, much else government seeks to do can be for naught. Yet, too few scholars are at work on the benefits and costs of policing, and we know far too little. We welcome scholars to the task.


•    Ponomarenko, Maria and Friedman, Barry. (2017) “Benefit-Cost Analysis of Public Safety: Facing the Methodological Challenges” Journal of Benefit-Cost Analysis 8(3), 305-329.