Policing and Crime: What the Evidence Shows

How economists measure whether more police reduce crime, and why the answer has been hard to find.

Do more police officers reduce crime? It seems like an easy question. In fact it is a classic example of why measuring cause and effect is hard, and how economists have found clever ways around the problem.

The chicken-and-egg problem

If you compare cities, those with more police often have more crime. That does not mean police cause crime. Cities with more crime tend to hire more police. This is a problem economists call endogeneity: the thing you are studying and the outcome affect each other, so a simple comparison misleads.

Natural experiments

To get around this, researchers look for natural experiments, situations where police numbers change for reasons unrelated to crime.

One well-known study by Jonathan Klick and Alexander Tabarrok used terror alert levels in Washington, D.C. When the alert level rose, extra police were deployed around the National Mall for reasons that had nothing to do with local street crime. Crime in those areas fell during high alerts. Other studies have used the timing of police hiring grants and election cycles. Much of this research finds that more police do reduce crime, especially violent crime, though estimates vary.

The weather is not a cause

Imagine noticing that ice cream sales and swimming accidents rise together. Ice cream does not cause drowning; hot weather drives both. Comparing police numbers and crime across cities has the same trap, because a third factor, the underlying crime problem, drives both. Natural experiments are like finding days when ice cream sales changed for reasons unrelated to weather.

Beyond numbers

Economists also study how police are used. Evidence suggests focused approaches, like patrolling crime hot spots, can reduce crime without simply pushing it elsewhere, a concern called displacement. At the same time, researchers note that policing has costs beyond budgets, including arrests for minor offences that can damage people’s job prospects and trust in institutions, especially in communities that already feel over-policed. A full accounting weighs the crime prevented against these harms.

Assuming correlation settles the question

It is tempting to point to any chart showing police and crime rising or falling together. Without a way to separate cause from effect, such charts can support almost any conclusion. The most reliable evidence comes from studies that find changes in policing unrelated to crime itself.

Key takeaways

Simple comparisons of police and crime mislead because cities with more crime hire more police.
Natural experiments, like terror alert deployments, help separate cause from effect.
Much of this research finds that more police reduce crime, especially violent crime.
Focused policing can work, but policing also carries social costs that belong in the accounting.