clock menu more-arrow no yes mobile

Filed under:

The End of Policing left me convinced we still need policing

One of the most prominent books on police abolition doesn’t have a good answer on violent crime.

Minneapolis police officers riding bicycles
Bicycle police officers in Minneapolis.
Tony Webster

As calls to defund or even abolish the police get more mainstream attention, many people have a question: Won’t this create a huge surge in violent crime?

The police abolition movement has its intellectual roots in the work of African American prison abolitionists Angela Davis and Ruth Wilson Gilmore and their group Critical Resistance, which in turn took inspiration from the earlier work of the Norwegian sociologist Thomas Mathiesen. But when the conversation turns specifically to the police, many activists and advocates point to the work of Brooklyn College sociologist Alex Vitale, whose 2017 book The End of Policing earned a blurb from Gilmore; its publisher, Verso, has made it available for free in ebook form during the current surge of interest in radical ideas on policing.

As someone who’s reported on and off on criminal justice issues from the standpoint of quantitative social science for years (my first print article after graduating college was a 2003 research-based argument for rolling back mass incarceration), all of these interviews left me confused. Vitale argues that police reform as conventionally understood by liberals is doomed because “policing is fundamentally a tool of social control to facilitate our exploitation” and we need to scrap it. He says we can get by with fewer police by dropping efforts to criminalize drug use, sex work, and unauthorized border crossings and use the savings to finance a generous welfare state that would tackle “the entrenched, radicalized poverty concentrated in highly segregated neighborhoods which are the main source of violent crime.”

But there’s a substantial literature in economics and sociology arguing that more police on the beat equals less violent crime. One effort to quantify this precisely is a 2018 Review of Economics and Statistics article by Aaron Chalfin and Justin McCrary. It estimates, based on a big set of police and crime data from large and midsize cities between 1960 and 2010, that every $1 spent on extra police generates about $1.63 in social benefits, primarily by reducing murders. One needn’t take this literature as gospel truth, but one of the go-to scholars on the abolitionist position should be able to — and want to — counter the prevailing academic claim that investments in policing pay off in reduced violent crime.

Figuring that perhaps Vitale’s interlocutors simply didn’t ask the question squarely, I picked up his book and read it. I wasn’t particularly expecting to be convinced, but I did try to be open-minded and hoped that at a minimum, I would understand the contours of the debate. The End of Policing is a good book that contains some good ideas about the potential to use housing and mental health policy to address certain classes of problems that are now largely dumped on the criminal justice system. I learned things about the history of policing in America and globally. But I found the book didn’t contain an answer to the question about what a huge reduction in the number of police would mean for violent crime. Vitale doesn’t address the research that worries me or present alternative research that might reassure me.

American policing needs to change. And there’s at least some reason to think that reducing the scope of policing can and should be a big part of that change. Fairly mild policy changes undertaken over the past few years have delivered results in terms of fewer police killings of unarmed people, and there’s reason to believe that plenty of opportunity exists for further reform.

But policing is important. There’s evidence that the number of police has an effect on crime, especially violent crime. And when crime soars, not only do the direct victims suffer but we run the risk that economically diverse cities will unravel as people with means flee to the suburbs. The people brushing past these worries with a casual nod to Vitale are relying on unearned authority, both about the impact on crime and about the possibilities of reform.

Vitale’s argument on crime is weak

Vitale’s main thesis is that, historically and causally speaking, the main origins of police forces were as instruments of social control more than public safety, reflecting various kinds of elite fears about urban working classes, immigrants, and runaway enslaved people. And he sees the problems with policing as too deeply ingrained to be combated with reform.

“The origins and function of the police are intimately tied to the management of inequalities of race and class,” he writes, and “a kinder, gentler, and more diverse war on the poor is still a war on the poor.” He says that “any real agenda for police reform must replace police with empowered communities working to solve their own problems.”

And while the book sketches out some convincing areas in which we could get by with less policing and some promising ideas for social programs that might be more useful instead, you will not find a detailed vision of what those empowered communities are going to do about violent crime. This somewhat cavalier attitude on how we would deal with violent crime in a post-police world is driven by Vitale’s intense skepticism that status quo policing is doing anything to reduce crime. But he doesn’t provide evidence for this proposition.

Instead, he remarks, with a citation of a 1996 book by the late political scientist David Bayley, that “there is no correlation between the number of police and crime rates.”

That simplistic correlation argument (maybe high-crime jurisdictions hire more cops) is all he offers to allay fears that there might be a big risk to scrapping or massively scaling back America’s police departments — there’s no counterargument presented to the contrary literature, or even an acknowledgment that it exists. But the existing research is much more quantitatively sophisticated than this.

Police officers reduce crime

Excessive use of force has been well-documented in American policing for years. So though I’ve been shocked by images of police running riot on protesters this spring, I’m not exactly surprised. And when I first got wind of the modern policing and crime empirical literature, I didn’t really want to believe that more police can improve crime outcomes.

But this is a taste of what researchers who have been trying hard to come up with plausible causal estimates for the impact of police on crime have found:

  • Jonathan Klick and Alex Tabarrok took advantage of changes to the terrorism “alert level” that happened for several years in post-9/11 Washington. When the federal government deemed the terror alert level to be elevated, DC police would mobilize extra officers, especially in and around the capital’s core, centered on the National Mall. Using daily crime data, they found that the level of crime — in this case largely car thefts rather than violent crime — decreased significantly on high-alert days, and the decrease was especially concentrated on the National Mall.
  • Klick, John MacDonald, and Ben Grunwald looked at an episode when the University of Pennsylvania had its campus police increase patrols within its defined zone of Philadelphia, and used a regression discontinuity design to discover that crime fell about 60 percent (this time with a larger decline for violent crime) where the extra officers went.
  • Steven Mello looked at a huge surge in federal funding for local police staffing associated with the 2009 stimulus bill. Exploiting quasi-random variation in which cities got grants, Mello showed that compared to cities that missed out, those that made the cut ended up with police staffing levels that were 3.2 percent higher and crime levels that were 3.5 percent lower — again with a larger drop in violent crime.
  • John MacDonald, Jeffrey Fagan, and Amanda Geller looked at a program in New York called Operation Impact that would surge additional officers into high-crime neighborhoods and found that a wide range of crime — assaults, robberies, burglaries, violent felonies, violent property crimes, and misdemeanor offenses — fell in response to the surge.
  • Richard Rosenfeld’s field experiments show that “hot spot” policing, where extra officers go to specific high-crime locations, not only reduces crime in the hot spots but reduces crime (in this case, specifically gun assaults) citywide.

Patrick Sharkey, a Princeton sociologist who is clearly sympathetic to the goals of the defunding movement, writes in a Washington Post piece arguing for a greater role for local leaders and communities in containing violence that “those who argue that the police have no role in maintaining safe streets are arguing against lots of strong evidence. One of the most robust, most uncomfortable findings in criminology is that putting more officers on the street leads to less violent crime.”

Note that this finding doesn’t have to come from a place of utopian thinking about police officers. Indeed, one particularly striking thread of evidence about the importance of policing to crime control comes from instances of police misconduct. Tanaya Devi and Roland Fryer find that when big-city police departments are hit with civil rights investigations following high-profile brutality incidents, the officers respond with work slowdowns that lead to soaring crime. Stephen Morgan and Rhiannon Miller document this dynamic specifically in Baltimore and Rosenfeld documents it in St. Louis.

Bocar Ba and Roman Rivera, looking at Chicago, agree there was a crime spike during this period, but they suggest the direct impact of the brutality on civilians may be the real issue. However, both research teams, in different ways, find that increased oversight outside the context of a major scandal leads to beneficial results. Further from the limelight, Alexandre Mas found in 2006 that when police unions in New Jersey lose a salary arbitration fight, arrest rates decline and reports of crime go up.

None of this reflects well on the professionalism or ethics of America’s big-city police officers. But that in turn underscores that the importance and efficacy of what police officers do doesn’t hinge on believing they’re angels. The number of officers patrolling the street has an effect on the murder rate.

And unfortunately, Vitale’s failure to grapple with this literature is part of a broader trend toward handling causal inference poorly.

Vitale doesn’t talk about trade-offs

In the course of a larger argument in favor of disarming police officers (something that on some margin is almost certainly a good idea), Vitale makes the striking claim that “traffic stops would be less deadly for officers and the public if the police carried no weapons.”

This would be a huge policy win for the police and for communities, given how frequent traffic stops occur. But Vitale offers little evidence or argument for the claim beyond a footnote to a 2015 article by Greg Smithsimon that appeared in Metropolitics. The 2015 article is not an extended argument on behalf of the claim about unarmed traffic stops being safer. Instead, it presents as evidence a single paragraph that I want to quote in full. The idea is that we know unarmed police would be safer because English police — who do not carry guns and work in a country where gun laws are extremely strict — die at a lower rate than American police:

What’s more surprising is what we forget when people say that the police need guns because they do a dangerous job: it’s more dangerous because of their guns. Surveys of police who are unarmed find that their concerns include not only danger to civilians, but the psychological harm done to police who fire weapons, and a belief that arming police makes officers’ jobs more dangerous (Squires and Kennison 2010). Thirty police were killed in the US in 2014, while a police officer was last killed in Great Britain in 2012. Even accounting for the UK’s smaller size, a dozen cops would have died on the job in that time if they faced the rates of American police “protected” by their weapons.

Pointing to the strong safety record of unarmed English police as “evidence” for the proposition that arming American cops doesn’t improve officer safety fails to account for the reality that the United States has 120 guns per 100 civilians while England and Wales have fewer than 5. Obviously, if 96 percent of America’s civilian-owned small arms vanished, you would also want to scale back police gun possession.

By the same token, Vitale proclaims that “there is no evidence that our country’s drug problems have been improved by driving millions into prison. Since 1982, drugs have become cheaper, higher quality, and more widely available than ever before.” In other words, he doesn’t just think that legalizing all drugs would be beneficial on net, he thinks there is no trade-off in terms of higher rates of drug abuse.

This doesn’t really make sense. And based on what we’ve seen from states that have legalized marijuana, it doesn’t seem to be true. Advocates are very proud of the fact that youth pot smoking (which remains illegal) has not risen in Colorado. But adults are getting high more, including a rise in problem usage. Of course, that doesn’t mean marijuana should be illegal — the prohibition cure can be worse than the vice — just as alcohol is legal in America despite the public health downsides.

Vitale himself discusses the alcohol analogy, urging us to “look at the alcohol abuse rates and problem behavior in places like Italy and France,” where legal restrictions on alcohol are fewer. But he does not actually cite any numbers. In fact, European countries have higher rates of problematic teen drinking, as well as a higher rate of cirrhosis deaths. What’s true is that drunkenness is less problematic in Europe because there is more mass transit and fewer guns, so the range of possible alcohol-related harms is narrower.

But laxer pot rules mean more pot consumption. Laxer alcohol rules mean more alcohol consumption. And laxer heroin rules would likely mean more heroin consumption. One can certainly make a case for this (criminalization has not been a rousing success story), but it requires some real argument.

Writing about how we should legalize all sex work to reduce policing, Vitale is similarly blithe, arguing that “despite decades of police enforcement, commercial sexual services remain easily available.” That’s true. But studies of prostitution legalization — including ones that generally find positive effects like fewer rapes and sexually transmitted infections — have found that legalizing sex work (an idea that is, not coincidentally, much more popular with men than women) leads to more consumption.

This kind of thing, where Vitale identifies a form of policing he doesn't approve of and then says we should do away with it, takes up a lot of the book. Alongside the chapters on drugs and sex work, there’s one that considers immigration enforcement (he votes to scrap it) and another about the policing of political dissenters, which includes an interesting look at the history of political police in France that is longer than his discussion of police officers’ impact on violent crime. Your mileage may vary with these ideas, but a consistent theme is that the causal reasoning is a little shaky and the willingness to consider trade-offs nonexistent.

Reform is possible

By the same token, Vitale is dismissive of promising reform ideas to reduce police misconduct.

“Much of the public debate has focused on new and enhanced training, diversifying the police, and embracing community policing as strategies for reform, along with enhanced accountability measures,” he writes. “However, most of these reforms fail to deal with the fundamental problems inherent to policing.”

Many faddish implicit bias trainings don’t really seem to work. But there are promising results from several different procedural justice trainings. More to the point, Vitale himself says that “in some ways training is actually part of the problem” because “in recent decades, the emphasis has shifted heavily toward officer safety training.” Instead of receiving training that creates an exaggerated sense of threat (police work is dangerous, but officers’ death rates are lower than for fishers or roofers), police should be provided with deescalation training (which has been found to be at least somewhat effective) and, more importantly, required to use it with real consequences for officers who don’t.

Even the relatively superficial reforms enacted between the killings of Michael Brown in 2014 and George Floyd have led to a reduction in police killings in big cities and fewer killings of unarmed people.

But we’ve barely scratched the surface of potential reforms that would really get tough on misconduct without compromising the basic concept that police are useful.

Right now, collective bargaining agreements make it extremely difficult to fire police with records of misconduct. Those who are dismissed are often ordered to be rehired. And police officers who are permanently fired — which, to be clear, means they have passed a high bar for badness — often get hired at other jurisdictions. Meanwhile, the “qualified immunity” doctrine immunizes police for civil penalties for misconduct.

Per what records are available, a relatively small number of officers are committing most of the misconduct, but studies show that bad behavior can spread like a virus to peer officers. Getting rid of the worst 5 percent of officers could eliminate an enormous share of the misconduct, halt the spread of bad norms throughout departments, and open up new hiring opportunities to create more diverse forces.

Vitale scorns the notion that trying to make police forces more racially diverse should be a priority — citing evidence that this has little impact on police shootings (other studies, I should say, do show that black officers make fewer discretionary “suspicious behavior” stops). But Anna Harvey and Taylor Mattia show that litigation resulting in court-mandated affirmative action programs “not only increased black officer shares, but also substantially reduce racial disparities in crime victimization.”

Of course, since Vitale doesn’t think police reduce crime, he doesn’t look into the idea of whether hiring black cops helps black people avoid victimization by criminals. But the toll of violent crime in black neighborhoods is staggering. In one 2013 study of census data, black people were nearly five times more likely than white people to be murdered.

As the journalist Wesley Lowery wrote recently, the mainstream black critique of the cops has long been that there is too much harassment and too little effective policing, not that they want the police to vanish.

What we need to do is take the full range of harms stemming from what Jenée Desmond-Harris has labeled the simultaneous overpolicing and underpolicing of black neighborhoods. Cases where police officers kill unarmed black and Latino men are incredibly traumatic, but they’re just the tip of the iceberg of traumatic interactions that, as my colleague Sean Collins writes, “leave many black Americans frustrated with and fearful of police.”

So it’s important to say clearly that while there is a strong research base for believing that having police on the beat reduces crime, these same studies find that the aggressive “suspicious behavior” stops and stop-and-frisk tactics that have poisoned police-community relations have no real crime-fighting value. By contrast, Duke University professor Philip J. Cook finds, in a clever study that compares the level of police effort that goes into murders versus non-fatal shootings, that police departments could solve more violent crimes if they put more effort into investigating them. The winning formula is more accountability, less harassment, and more police work.

Of course, as Vox’s Aaron Ross Coleman explains, even as black communities express relatively little enthusiasm for ending policing (a June 14-16 Economist/YouGov poll found 22 percent of African Americans favor abolishing the police), they do want real investment in their communities and not just cops. Given the vicious round of state and local budget crises that Covid-19 is poised to inflict, painful trade-offs between different forms of local spending are necessary. But that’s a downstream consequence of bad federal policy, not a fact of life. The right answer to demands for more investment in effective public services is to make more investment in effective public services.

America needs more public spending

Vitale is very persuasive on two points. Police officers spend a lot of time adjudicating conflicts between people experiencing homelessness and people annoyed by those people’s presence. They also spend a lot of time as de facto untrained community mental health providers. He is surely correct that it would be better to actually solve America’s housing and mental health treatment problems rather than use police as Band-Aids.

As the defund debate has played out in the public sphere, an idea has taken hold that not only should America spend more on social services but that the police are the reason we can’t or won’t do that. A recent viral tweet depicting a teapot labeled “defund the police” filling the cups of education, universal health care, housing, etc. very much captures the spirit of the times.

It does not, however, capture the budget realities. As the experts at the Urban Institute point out, police spending is a relative drop in the bucket of state and local government budgets; at the federal level, it’s even smaller.

It’s true that the government should be spending more on housing and mental health programs and that doing so would probably reduce crime. But it would probably reduce crime by freeing up officers to do more police work. And there’s no particular reason the money for it has to come out of police departments. If you compare the United States to Europe, the reason Europe has a more generous welfare state is a much higher overall level of spending — not that the US has overfunded the police.

Indeed, as President Obama’s Council of Economic Advisers pointed out, while the United States has more prison guards and far more prisoners than the average country, it actually has 35 percent fewer police officers.

The problem with America’s police officers is that they’re too unaccountable and lawless, operating with too much job security and a sense of impunity, not that there are too many of them.

We can take away the armored personnel carriers and other military gear. We can (and should) stop the “stop and frisks” with no probable cause and “no knock” warrants that elevate the virtues of drug seizures out of proportion to the risks. We can train officers in deescalation tactics and require them to use them. But if we proceed with the assumption that police do not matter and reform is too hard, what could likely result is rising crime and unraveling city communities.

Affluent residents will retreat behind armed guards. Nobody on the left is currently advocating for a strategy of “privatize the police,” but once you discover that the link between policing and crime is real, it’s clear that’s what lies down this path — replacing an insufficiently democratic, insufficiently accountable police force for a much less democratic and much less accountable patchwork of private guards who’ll safeguard those with money and ignore the marginalized people that defunding was supposed to help.

Correction: An earlier version of this article misstated the finding of the Bocar Ba and Roman Rivera paper, saying they confirmed that depolicing was responsible for the Chicago crime spike when in fact they were raising doubts.

Sign up for the newsletter Today, Explained

Understand the world with a daily explainer plus the most compelling stories of the day.