Editor’s note: The following is adapted from the author’s April 22, 2021, testimony to a Senate judiciary subcommittee hearing on behavioral health and policing.

A key misperception seems to be driving much of the conversation about how to reform policing in the United States: that police violence is a likely outcome of an investigative or enforcement interaction—particularly when those interactions involve people of color. For example, in a recent report for the Manhattan Institute, Eric Kaufmann found that eight in ten African-American survey respondents “believe that young black men are more likely to be shot to death by the police than to die in a traffic accident”; and “among a highly educated sample of liberal whites, more than six in ten agreed.” A 2016 Morning Consult poll found that twice as many black respondents reported worrying about those they know becoming victims of police brutality than of gun violence, generally.

These beliefs are completely at odds with the data on police use of force. A recent study published in the Proceedings of the National Academy of Sciences (PNAS) shows that police kill young black men (between the ages of 25-29) “at a rate between 2.8 and 4.1 per 100,000,” which is far lower than the rate at which they die in traffic accidents. That same PNAS study put the odds of dying at the hands of police at one in 1,000 for black men. Contrast that with the odds for all Americans, of any race, of being killed by gun assault, which, according to the National Safety Council, are dramatically higher, at one in 298. Given that black men are more than ten times more likely than their white counterparts to be the victim of a homicide, it’s quite clear that their risk of death at the hands of police is far lower than their risk of being killed by gun violence, generally. Furthermore, 83 percent of respondents to a Pew Research survey guessed that the typical police officer has fired his gun at least once on the job; in fact, only about one in four (27 percent) actually do.

The data are clear that police very rarely use force, and when they do, it very rarely results in serious injury. In 2018, police officers in the U.S. discharged their firearms an estimated 3,043 times. That year, they made more than 10.3 million criminal arrests. This means that, at most, 0.4 percent of officers purposely discharged a firearm in 2018 and that, at most, police applied deadly force with a firearm in just 0.003 percent of arrests.

As to non-deadly force, in 2018, a research team of doctors and a criminologist published a thorough study of police use of force in the Journal of Trauma and Acute Care Surgery. The study analyzed more than a million calls for service to three mid-sized police departments in Arizona, Louisiana, and North Carolina over a two-year period. Those calls resulted in more than 114,000 arrests. Police used physical force in one of every 128 of them, meaning that more than 99 percent of arrests involved no use of force. Using expert medical examinations of suspects’ medical records, the study found that 98 percent of suspects on whom police used physical force “sustained no or mild injury,” and 1.8 percent of suspects sustained moderate or severe injuries. Only one suspect was fatally wounded by police gunfire during the study period.

Nor are uses of force likely when you look at particularly dangerous police encounters. In 2020, the NYPD made 20,935 arrests of suspects who possessed weapons and responded to 161,278 calls to 911 for persons in crisis. Yet the department recorded just 42 firearm discharges in 2020, including off-duty shootings. By and large, police are pretty skilled at dealing with dangerous or unstable subjects without relying on physical force.

Of course, there’s room for improvement, but we must explore opportunities for reform carefully. Pulling the wrong policy lever can have disastrous effects, particularly on crime. That’s a risk we should be especially wary of now, given the sharp uptick in shootings and homicides across the country.

Last year, for the first time since 1995, the U.S. saw at least 20,000 criminal homicides, according to estimates based on preliminary figures. That’s an increase of about 4,000 additional homicides compared to 2019. While that’s still a ways off from the nearly 25,000 homicides the country experienced in 1991, some cities have seen their homicide numbers approach and even surpass their 1990s peaks. In 2020, Cleveland, Ohio, had its highest murder tally since 1982. Minneapolis, Minnesota, had more murders than every year since 1995. Cincinnati, Ohio, and Louisville, Kentucky, both set new homicide records.

Moreover, those upticks were not evenly distributed; they were heavily concentrated in minority neighborhoods already struggling with elevated levels of crime. Conversations about policing in America tend to focus heavily on racial disparities in various enforcement statistics, but it’s also worth noting that the burden of rising violent crime falls disproportionately on the shoulders of people of color. In New York City, going back at least to 2008, at least 95 percent of all shooting victims were either black or Hispanic; similar proportions hold for 2020, when shootings were nearly double 2019’s total.

Studies conducted throughout the country over many years have shown that additional policing has significant crime-reduction effects; in other words, generally speaking, more police means less crime. And just as increasing crime disproportionately impacts people of color—particularly black men—decreasing crime disproportionately benefits them. From 1991-2014, the U.S. experienced a sharp decline in homicides. A 2019 analysis of that decline showed that, during that period, “the decline of homicide-specific mortality led to increases in [life expectancy] of . . . 0.14 years . . . for white males,” versus “1.00 years . . . for African American males.” As the lead author of that study wrote in his 2018 book, “the impact of the decline in homicide on the life expectancy of black men is roughly equivalent to the impact of eliminating obesity altogether.”

Police executives, rank and file officers, and other defenders of the institution often cite falling crime as an example of policing’s success. One question that might be worth considering, given the current state of debate about American policing, is whether we ought to characterize an institution whose stated purpose is to produce outcomes that disproportionately benefit minorities as systemically racist towards those very minority groups. The answer to the question isn’t simple. Black and Hispanic communities do disproportionately bear the costs of policing, including false-positive police interventions, which is partly but probably not wholly attributable to underlying variations in violent crime levels. But at a minimum, if we’re assessing the racial impact of policing, we should look at both sides of the ledger—not just the costs but the benefits, too.

I suspect that sincerely held misperceptions about police have shaped the overarching goal of the reform movement, which, at the moment, seems to be to minimize in any way possible the footprint of police and the criminal justice system more broadly. We have heard calls to “defund” the police that, in some cities, officials have heeded. We have also heard calls to divert more responsibilities—particularly functions like traffic enforcement and responding to mental health calls—away from police to unarmed civilians. The assumption is that doing these things will reduce the scope of interactions that might result in tragedy. But even if that were true—in the case of traffic enforcement, for example—it would also reduce the scope of interactions that might lead to the discovery of more serious criminal activity. The Kansas City Gun Experiment, a 1995 National Institute of Justice study that analyzed enforcement and deployment decisions targeted in hot spots for gun violence, found that “Traffic stops were the most productive method of finding guns, with an average of 1 gun found in every 28 traffic stops.” Many “lower-level” police interactions lead to the discovery of contraband, the closure of open warrants, and the recovery of illegal firearms. Zeroing out those encounters, therefore, would not be cost-free.

On the mental health front, there is some evidence to suggest that we should continue to invest in efforts to augment the police by deploying civilian crisis-intervention teams to calls involving people in crisis. Charles Fain Lehman reviews some of that evidence in a Manhattan Institute study. Among the approaches he evaluates is a popular, oft-cited, and promising effort launched in Eugene, Oregon, called CAHOOTS (Crisis Assistance Helping Out On The Streets). While effective, CAHOOTS is also a case study in the limits of programs loosely referred to as “alternatives to policing.” As Lehman notes:

CAHOOTS responders are highly specialized. In 2019, they covered just 17% of Eugene 911 calls, with 75% of those calls being a welfare check, providing transportation to someone (usually homeless or in need), or assisting the police already on the beat. Even in those relatively limited circumstances, CAHOOTS responders still called for backup in roughly one in every 67 calls for service in 2019.

It is hard to see how one might scale up that model to cover the other 83 percent of 911 calls to the Eugene Police Department. Doing so would involve not only dramatically increasing the program’s $2.1 million budget but also identifying a large population of trained crisis-intervention professionals and, most significantly, asking those professionals to handle situations that grow increasingly risky as responsibility expands. In other words, though such programs provide useful services now, they are not good models for replacing the police. As a complement to policing, however, they may be a useful model for other cities to adopt.

When you consider the sheer volume of mental-health calls received by police (not to mention the fact that these calls often come in the late-night or early-morning hours), it becomes clear that we simply don’t have the capacity to shift this particular responsibility in total. Another complicating factor: it is often difficult to determine, based on 911 calls or dispatcher information, whether a call is safe for civilian responders trained in mental health crises or needs a police response. A study in Philadelphia found “that some medical or public health activity initially masquerades as crime or other policing work and some events eventually determined to be police/crime activity can initially appear to be public health related.” The study went on to note that “[a]bout 20% of activity in this area does not appear predictable from the initial call type as handled by police dispatch.”

None of this is to say that we should not work vigorously to improve policing outcomes. Indeed, I have outlined a number of reforms that would be worth pursuing in the short and long-run. However, as we undertake reforms, we must not lose sight of the government’s first duty, which is to provide for the public’s safety. As serious violent crime across American cities continues to rise this year, that mission is more critical than ever.

Photo by Scott Olson/Getty Images


City Journal is a publication of the Manhattan Institute for Policy Research (MI), a leading free-market think tank. Are you interested in supporting the magazine? As a 501(c)(3) nonprofit, donations in support of MI and City Journal are fully tax-deductible as provided by law (EIN #13-2912529).

Further Reading

Up Next