Shortly after George Floyd was killed by a Minneapolis police officer, America began debating the idea that the best way to rectify the historic racial inequities in the criminal justice system is to “defund the police.” Opponents of the idea retort that if we police less aggressively, crime will increase, a burden that historically disadvantaged communities will bear disproportionately.

Empirically, we have a great deal of evidence that putting more police on the street reduces crime. We know that Black and Latino people are more likely to be victims of a serious crime than Whites or Asians. We also know that recent data suggests that homicides spiked in the United States’ largest cities last year by an average of 30 percent.

Yet none of that answers the most important question: For communities suffering under high burdens of both policing and crime, does the benefit of more police outweigh the costs of the policing to both individuals and communities?

Even if you concede that policing does reduce crime considerably — which many “defund the police” advocates don’t — it’s possible that more policing does reduce crime for mostly whiter and more affluent neighborhoods, while historically disadvantaged groups mostly bear the substantial costs of being policed. Most of the debate over the past year has implicitly assumed such a tradeoff.

Yet a new paper distributed by the National Bureau of Economic Research suggests this assumption isn’t quite right. Policing really does disproportionately burden Black communities (especially when it comes to low-level offenses), but it also disproportionately benefits them — not just by protecting victims from the people police arrest, but by protecting people who might otherwise have turned to crime.

The authors looked at 38 years of police employment data for America’s 242 largest cities, not just to see the policing’s effect on crime, but specifically, what its differential impacts were on different racial groups. In tune with earlier research, they found that bigger police forces reduced “index” crime — serious offenses such as violence, burglary or robbery. Depending on the exact model, they found that hiring somewhere between 10 and 17 new officers averted one homicide every year. And as the opponents of the “defund police” movement have suggested, this indeed disproportionately benefited Black communities: “In per capita terms,” they write, “the effects are approximately twice as large for Black victims.”

Financially, that means investments in policing would save one life for every $1.6 million to $2.7 million spent. That’s actually a pretty good cost-benefit ratio for a government program, as grim as it may seem to put a dollar value on any single life. And that’s before we consider any other benefits of public safety. A broad decrease in crime doesn’t just benefit homicide victims. It reduces rape and the associated trauma. It lowers the number of nonfatal injuries. It saves people from losing property they can’t necessarily afford to replace. And it reduces the pervasive fear of crime, which is bad in itself — as are the things people do to avoid it, from staying home at night to paying for expensive alarm systems to barring all their windows and doors. When you add all that in, $2.7 million looks pretty cheap.

Still, that’s just the financial cost. What about the personal cost of all those arrests?

Here’s the surprising part: More policing, they find, reduces not just victimization, but also arrests for index crimes, which can carry long prison sentences. It was reasonable to fear that adding police “solves” our crime problem through mass incarceration of Black men, but that doesn’t seem to be the case. In an associated note published through the Niskanen Center, the authors write, “The decline in index crime arrests is four to six times larger for Black civilians than whites, which suggests that investments in policing are unlikely to have contributed to the massive and racially disparate growth in the scale of incarceration in the United States during the last four decades.”

This is not to say that this is all benefit, no cost. While arrests for index crimes went down with extra police, they found that each additional officer hired made between seven and 22 additional arrests for lower-level “quality of life” offenses such as liquor-law violations and drug possession. And arrests for those crimes, specifically, seemed to be as much as three times higher among Black civilians. That’s not good for the community either; arrests are unpleasant and expensive, and frequently being hassled by police is a source of fear for a lot of people, not to mention an infringement on their dignity.

The question, then, is whether we can find some way to beef up the police presence in vulnerable communities without turbocharging this toxic and racially skewed dynamic. I think there are ways we can do that through better training and accountability and higher salaries to attract and retain top-quality officers. But we probably won’t get there unless we can move right and left off the unhelpful polls of “less” or “more,” and toward “more policing, but also better.”

Read more: