Each year, police kill more than 1,000 people in the United States. Very few of these incidents, less than 2 percent, result in any criminal charges. To be sure, some deaths involve police saving others or themselves from danger — but just as surely some incidents involve police misconduct. Many observers argue that the close working relationship between prosecutors and police contributes to the low rate of indictments.

How do police officers react when a new prosecutor takes over? Are police more cautious about using force if they feel more uncertain about whether they might be investigated?

In new research, Haritz Garro and I find that when a new district attorney takes office, police kill fewer people. Deaths decline no matter what the new prosecutor’s political party or policy platform might be. During that time, we find no changes in arrest rates or assaults on police officers. That suggests that officers can de-escalate situations without either refraining from policing or risking greater harm themselves.

Prosecutor turnover leads to fewer deaths by police officers

We studied recent elections of local prosecutors, commonly called district attorneys or DAs. We observed election outcomes for all 2,315 districts that elect this position in the United States, from 2012 to 2017.

We then compared the number of killings by police in all districts where the incumbent district attorney won reelection with those in districts where a new candidate won office. To do so, we compiled data from the Prosecutors and Politics Project, Mapping Police Violence, the FBI and the Census Bureau.

We found that a new district attorney taking office is associated with a 17 percent reduction in police killings, equivalent to 0.08 deaths per district per year.

But this doesn’t tell the whole story. Many districts have little competition for the prosecutor position, especially in rural areas. Law enforcement in districts where incumbents are sure to win could be very different from law enforcement in districts where the incumbent routinely faces new challengers who have real chances of winning.

To address this, we focused on 157 districts with close primary or general elections. We compare districts where a challenger narrowly defeated an incumbent with districts where an incumbent district attorney narrowly defeated the challenger. These districts were demographically very similar, being generally more urban, and had very similar policing results (such as arrests, deaths, convictions and so on) before the election. As a result, any subsequent changes in policing are likely to result from the election’s outcome.

In districts where elections are close, the differences are even more dramatic. After a new candidate ousts an incumbent, the number of deaths by police falls by 40 percent, equivalent to 0.3 deaths per year per district or at least one death in a four-year term.

The effects are similar, no matter the civilian’s race

Black, Latino and American Indian adults in the United States face a higher risk of being killed by police than White adults. The risk is exceptionally high for Black men, who face an estimated 1 in 1,000 chance of being killed by police in a lifetime.

But when a new prosecutor takes office, the drop in police killings doesn’t vary significantly by race or ethnicity. White deaths decrease by 40 percent and Black deaths decrease by 38 percent. In other words, while prosecutor turnover reduces police killings, it does not close race and ethnicity gaps in police violence.

What’s more, after a new district attorney takes office, we found no changes in property crime, violent crime or arrests. This suggests that when officers want to, they can de-escalate situations that might have otherwise turned deadly, and aren’t simply avoiding those situations altogether. Nor do we see an increase in assaults on or deaths of police officers.

Why do police behave differently when there’s a new prosecutor in town?

Others have found this link between police accountability and use of force. Still, some might be surprised by our findings. How might changes in the district attorney’s office influence high-stakes policing decisions?

Certainly, we do not expect officers to calculate the odds of an indictment while making split-second decisions. Instead, we imagine at least three reasons to expect police officers to respond to a new candidate’s election.

First, district attorneys work closely with police officers every day. If officers felt known and respected by an incumbent, they may become cautious about using force after the incumbent is ousted.

Second, a new district attorney introduces uncertainty. Officers may behave cautiously if they are unsure how a prosecutor will handle future cases involving police. Prosecutors behave differently in such things as where they focus, how intensely they investigate facts that come their way, and how they use grand juries. For these reasons, officers might change their behavior even if they don’t expect to be indicted.

Third, a new district attorney may change policies. Progressive prosecutors often campaign on promises of holding the police accountable. They’ve won more races in the past several years, mainly by defeating incumbents in Democratic primaries. We do find some suggestive evidence that police are particularly cautious when Democrats win. However, even when the new district attorney is Republican, independent or nonpartisan, police killings drop significantly. The policy position of the newly elected district attorney might be important but cannot fully explain the decline in deaths that we observe in the data.

Prosecutors and police overhauls

Very few police officers are indicted on charges related to their behavior on the job. Focusing solely on the number of indictments, however, can obscure how the district attorney can rein in police violence. Since district attorney turnover alone results in fewer police killings, communities can benefit from more competitive elections. District attorney elections in more populous jurisdictions have become more competitive in the past decade, in part because more progressives have challenged incumbents. Still, from 2012 to 2017, 70 percent of district attorney elections were uncontested.

The fact that police violence falls after election of a new district attorney hints that police violence can also drop if a community assigns outside prosecutors to handle cases involving police officers. Much as happens when a new district attorney takes office, knowing that an unfamiliar prosecutor will come in each time an officer is under investigation might shift police behavior, resulting in fewer deaths to investigate in the first place.

Allison Stashko is an assistant professor at the University of Utah’s David Eccles School of Business.