A justice system, by its very nature, provides benefits and imposes costs. On the one hand, the organized enforcement of universal rules prevents a Hobbesian “war of all against all,” and protects the conditions under which open societies and free commerce thrive. On the other hand, enforcement of the law, by definition, entails empowering the state to deprive individuals of their liberty, and any justice system will inevitably err in adjudicating people’s guilt and applying appropriate punishments. Enforcement is also expensive. The U.S. spends nearly 1% of its gross domestic product on the justice system – approximately $800 per person per year. 

Maximizing the public safety benefits of the laws, programs, institutions, and procedures that comprise the U.S. criminal justice system and minimizing their fiscal and social costs should be the principle that guides the design of that system and any efforts to improve it. At every decision point, policymakers should ensure that the next dollar spent reflects the best possible use of that dollar; that use should maximize social welfare at the lowest social cost, minimizing the most misery among the most people without making other people’s lives worse.

Certainly, that’s easier said than done. Like everything else, justice comes with opportunity costs. Investing more resources in a justice system means investing less in competing social priorities. However, a justice system that produces benefits that exceed its marginal costs can free up more money for long-term social investment. 

An exceptional justice system ultimately has the potential to deliver a double dividend to society by lowering both crime and punishment. By promoting policies that generate deterrence and by incapacitating the highest-risk and most prolific offenders, society can have both greater safety and greater freedom, outcomes which have long been considered to be in tension with one another but which need not be at the margin. This essay provides a high-level overview of the most consistent empirical findings in crime research and, leveraging that knowledge, suggests several ways in which resource allocation within the criminal justice system can be shifted to better fit the evidence.

Deterrence is relatively cheap

There are two broad ways through which a justice system can improve public safety. The first is by arresting and incarcerating people who commit crimes. Doing so abates the crimes that those individuals would have committed had they been on the street, free to offend. The second is deterrence, i.e., sufficiently raising the perceived costs of crime so that would-be offenders choose to abstain. To the extent that people who might commit crimes are deterred by the prospect of being arrested and punished, we’ll see less crime and fewer crime victims.

While both of these mechanisms have the potential to get us to the same level of public safety, deterrence is almost always the more efficient option. If we can convince someone not to offend in the first place, victims don’t suffer the harm of crime, no one loses their liberty, and the public avoids paying for a term in jail or prison. For this reason, for a given level of effectiveness, it makes sense to prioritize policy options that generate significant deterrence effects over options that rely on incarcerating large numbers of people after they’ve committed crimes.

People are myopic

Most of us discount unlikely events that might happen in the distant future relative to certain events that will occur in the near future. If I drive home from the bar while intoxicated, I won’t have to sleep in my car tonight, a welcome benefit that I will experience in the present. And while it’s risky to drive drunk, the odds are good that I won’t cause an accident or be caught by a police officer during my short drive home. Of course, if I am caught, I’ll be arrested and adjudicated, an experience that, for many people, would be life altering. But that probably won’t happen – the National Highway Traffic Safety Administration estimated that there are between 500 and 2,000 DUI violations committed for every one DUI violator arrested – and, even if it does, my penalty won’t be meted out for many months. For people who commit crimes, many of whom are young and male, and some of whom have already found themselves involved in a high-risk lifestyle, a tendency to discount the future is probably even more ingrained than it is for the average person. Punishment must grapple with this seemingly fundamental pillar of human nature. 

Our low certainty-high severity policy regime suggests that we are using a mix of tools that almost certainly does not maximize the effectiveness of the marginal public safety dollar.

The research literature now reflects a broad consensus that offenders are deterred to a greater extent by the certainty of punishment than by the severity of punishment. In other words, deterrence is more likely to be achieved by investing in detection than by severe but unlikely punishments. America’s justice system is ripe for improvement along this margin: Most serious crimes – including about 70% of robberies and 50% of serious assaults – lead to no punishment at all because the police are unable to gather enough information to make an arrest in the first place. The clearance rates are even lower for less severe crimes like burglary and theft, but punishments can still be harsh. 

To summarize the available evidence, our low certainty-high severity policy regime suggests that we are using a mix of tools that almost certainly does not maximize the effectiveness of the marginal public safety dollar. While the United States spends more than double what other industrialized nations spend on prisons, its police spending lies somewhere in the middle. We have thus chosen to double down on severity, investing less intensively in certainty. The result is unnecessarily high levels of both crime and punishment.

Serious crime drives the social costs

Each year, approximately 9 million index crimes are committed in the United States. These include violent crimes such as murder, rape, robbery and serious assaults, and property crimes such as burglary, motor vehicle theft, and larceny (a category that includes shoplifting). Just 21,000 of these 9 million crimes are murders but estimates of the cost of offending generated by economists suggest that the murders – which account for just 0.2% of all crimes – by one estimate, account for 25% of the social harms of crime. 

From a human perspective, this concentration of harm is tragic. But from a policy perspective, it is a tremendous opportunity. While reducing shoplifting by 20% would do little to change the social burdens of crime, reducing murders by 20% will have an appreciable impact. Law enforcement can greatly reduce the social burdens of crime by focusing narrowly on lethal violence. 

Serious crime is concentrated among a small number of people and places

Violence is rarely random. A small number of neighborhoods in a small number of cities are responsible for the lion’s share of the serious violence – including murder. Within these communities, gun violence is concentrated among a small number of social networks, and the trigger pullers are often well-known to law enforcement. Even in a city of millions, violence is driven by perhaps a few thousand people on a few hundred city blocks. The problem is, therefore, manageable. To really reduce rates of serious violence, it’s critical to engage with several thousand people where they live, rather than several million people citywide. 

Optimizing deterrence

Empirical evidence underscores the need for policymakers to focus on scalable policies that deter crime and strategically incapacitate high-risk offenders. This evidence base yields several key insights:

  1. In thinking about the two most fundamental policy inputs that social planners have at their disposal to address immediate public safety challenges – police and prisons – the available literature suggests that police are likely to be the better investment at the margin. Indeed, research shows that, across many cities and many years, when cities put more police officers on the street, crime and violence have declined. Interestingly, those crime declines have not led to an increase in arrests that are most likely to lead to a prison sentence, a sign that police are generating important deterrence effects. 

    Regarding investments in prison beds, the research suggests that the public safety gains are smaller than for police. While the crime reductions from imprisonment appear to have been fairly large in the 1980s – when incarceration rates were comparatively low – recent research suggests that the marginal benefit of imprisonment has since fallen considerably. A particularly influential paper published in the Journal of Law & Economics in 2012 found that a 10% increase in the number of prisoners in a given state leads to a 1-2% reduction in serious crime. In contrast, recent research on the effects of police hiring suggests that a 10% increase in the number of police officers hired leads, on average, to a 5-10% reduction in serious crimes. California’s recent experience with “justice re-alignment,” the state’s response to a federal court order which reduced the use of incarceration considerably, is also instructive. Despite the fact that criminal sentences became shorter – with the number of state prisoners declining by 17% in the first year of the reform – violent crimes, those that drive the social costs to a considerable degree, did not rise in California as a result of the policy shift. The research did find that re-alignment led to a small increase in property crimes, driven almost entirely by motor vehicle theft. The evidence thus suggests that while imprisonment does mechanically reduce crime, at present levels of incarceration, investments in law enforcement are likely to be a better bet. 

  2. Regarding how to invest in the police, it stands to reason that the best investments will deter serious offending and increase the certainty of apprehension for the most serious crimes. Research shows that when police departments invest more resources in investigating shootings, these crimes are more likely to be cleared by an arrest. Research likewise shows that when law enforcement, including both police and prosecutors, focus intensively on a small number of individuals and groups, building major cases and engaging in tactical deterrence, significant declines in violence can follow. Often, these declines in violence are not accompanied by increased arrests at the city level. Both crime and imprisonment can thus decline simultaneously, as occurred in NYC during the ten years leading up to the COVID-19 pandemic.
  1. Tailored strategies are best. Because the “power few” drive so much of the crime problem, the greatest gains can be achieved when attention is concentrated on the most proximate drivers of crime. Concerning shoplifting, this means building major cases against the retail theft rings that drive millions of dollars in losses, rather than a broad-based increase in criminal penalties for shoplifting, the burdens of which will be shared by a large number of unfortunate people – those with mental illnesses, those who are homeless, etc. – and who tend to account for little of the problem. Regarding burglaries, this means investing in technology (e.g., GPS trackers, cameras, license plate readers) that allows law enforcement to identify high-volume burglary crews rather than simply increasing criminal penalties for everyone. This is critical because only 14% of burglaries are cleared by arrest. The problem isn’t that we’re too “soft on burglary;” we’re simply not investing enough in detection. 
  1. Low-level “quality of life arrests” can compromise police legitimacy and drive racial disparities but sometimes do have value. Still, available evidence suggests that large numbers of low level arrests have little effect on serious crimes. While researchers do not yet understand all of the mechanisms involved, one of the reasons may be that the loss of 3-4 hours of a police officer’s time processing an arrest for drug possession, takes her away from one of the core ways that police deter crime in the first place, which is maintaining visibility and engaging with genuinely risky situations – e.g., a house party that spills out into the street or a verbal argument that could escalate into a shooting. Research in Dallas, Texas shows that when police are called away from their regular beat to respond to a severe traffic accident, crimes rise in areas they would have been patrolling. The bottom line: visibility matters.
  1. When it comes to crime control, research does not provide all the answers, but it does point us toward the most likely solutions. While every community is different, experience suggests that investments in law enforcement’s capacity to build major cases and target the most criminally active people are likely to pay far greater dividends per dollar spent than either low-quality “quick rip” arrests or ratcheting up the harshness of criminal sanctions. 

This essay does not aim to detract from critical investments that policymakers have made and continue to make in social service supports and community capital, which have also improved public safety. Indeed, the lessons of crime control are applicable to crime prevention: the most significant gains can likely be achieved by targeting social services towards the people and places that drive violence. In both cases, the broader takeaway is that smarter allocation of resources offers the possibility of reducing both crime and imprisonment. 

Aaron Chalfin is Associate Professor of Criminology at the University of Pennsylvania and a Senior Fellow at the Niskanen Center.