Communities across the country are reconsidering how they handle nonviolent misdemeanors such as disorderly conduct, disturbing the peace and possession of small amounts of illicit drugs. In many cities — including Philadelphia, Los Angeles and Baltimore — reform-minded district attorneys have been elected after promising to scale back the prosecution of such offenses. But they face pushback from police and community members who worry that not punishing people for low-level infractions will simply encourage more crime.
It's an important debate, partly because misdemeanors account for 80% of all prosecutions nationally. Hoping to bring evidence to bear on the question, we examine in a new study the effects of prosecuting nonviolent misdemeanors on future criminal involvement in Suffolk County, Mass. — where Boston is located — using data from 2004 to 2018. Specifically, we looked at the effects of prosecuting nonviolent misdemeanors on defendants' future criminal legal involvement. Across the board, we find that being more lenient on the defendants — that is, erring toward non-prosecution — has big benefits. People who are not prosecuted for misdemeanors are much less likely to find themselves in a courtroom again within two years. Entanglement with the legal system itself seems to be a risk factor for future criminal prosecution.
Figuring this out was not straightforward. Simply comparing people who are prosecuted with those who are not would be misleading, because prosecutors carefully decide which cases to pursue and which to dismiss. (You could imagine that they might focus on the more serious misdemeanors, for example.)
To estimate the causal effect of prosecution, we need something that sorts people into prosecution or non-prosecution in an essentially random way. It turns out that the way nonviolent misdemeanor cases are handled in Suffolk County, Mass. opened the door to just such a natural experiment. Overall, 79% of misdemeanors tend to be prosecuted, while 21% are not. But the percentages vary greatly by prosecutor: Some are simply more lenient than others. What's more, the prosecutors rotate through arraignment hearing rooms in a way that's unrelated to which cases will be arraigned on particular days. Therefore, once we control for things such as court, month and day of the week, it is random which prosecutors handle each case. That randomness let us home in on people who committed similar crimes but received different treatment.
We examined 67,553 cases in all. Not being prosecuted for a nonviolent misdemeanor, we found, decreases the likelihood of any new criminal complaint — an arrest, basically — within two years by 58% and decreases the number of new criminal complaints by 69%. These effects are statistically significant and they grow over time; effects after three years are even larger. We find reductions in new misdemeanor complaints as well as in new felony complaints, and in new violent offenses as well as in new nonviolent offenses.
The effects of non-prosecution are biggest for first-time defendants. That supports the hypothesis that helping people avoid their first contact with the criminal legal system helps them avoid a path where criminal activity is more likely. Consider what happens to a prosecuted defendant, even if they are not convicted: A criminal record of the arrest will still be added to the state database. That record is then visible to other law enforcement agencies and potentially to employers, who may choose not to hire the person. Having a criminal record can have collateral consequences in many domains (reducing access to public benefits or housing, for instance) that also increase the likelihood of future criminal activity.
The sheer time that it takes to defend against prosecution exacts a toll, too. People who are not prosecuted walk away after the arraignment hearing, but prosecutions typically take six months or so. That means taking days off work for hearings and meetings with lawyers, not to mention the stress of worrying about the case. These disruptions could cause defendants to lose their jobs if they have them, or increase alcohol or drug use. And these effects could, in turn, increase the chances of future criminal activity.
Then there's conviction: 26% of cases that are prosecuted lead to a misdemeanor conviction. Such convictions could make it even more difficult to find a job, which again could lead to more criminal activity. ("Ban the Box" policies that prevent employers from asking about past convictions on job applications do not appear to change this; employers can still avoid hiring people with criminal records when they do a background check at the end of the interview process.) And the punishments — typically probation or a fine — create other hardships.
Our results imply that future criminal legal involvement would fall in Suffolk County, Mass. if all prosecutors behaved more like the most lenient ones. But what would happen if the county implemented a presumption of non-prosecution for low-level offenses, as is advocated by many reform-minded district attorneys? This could push all prosecutors to be more lenient than even the most lenient prosecutor was during the period we studied.
It turns out that Suffolk County Mass. implemented exactly this sort of policy in 2019, when it elected one of these district attorneys — Rachael Rollins, who took office early that year. Using data on new complaints filed in 2019 and early 2020, we considered the effects of this policy change on non-prosecution rates and subsequent criminal legal involvement. As intended, non-prosecution increased by 15 to 20% relative to before Rollins took office. And — in line with our previous results — we again find that non-prosecution caused a reduction in new complaints; the effect was about the same size or larger for this group.
We might worry that even if future criminal legal involvement falls for current defendants, a policy of non-prosecution might increase crime rates more broadly if local residents start to perceive that there are no consequences for bad behavior. But when we look at reported crime rates in Boston, we don't see any evidence that crime increased in the year after Rollins took office; if anything, it seems to have fallen.
All of this suggests that district attorneys such as Rollins are on the right track. Pushing minor cases — trespassing, shoplifting, driving infractions and so on — through the criminal legal system appears to do more harm than good. There probably are other policies that would deter future crime, such as easing access to treatment for mental illness and substance-use disorders. But many of the people we studied might have simply made a mistake that they will never make again, if they are shown mercy and allowed to resume their normal lives without prosecution or punishment.
There is still a lot we do not know about where to draw the line — which cases to prosecute and what the punishments should be. As communities become more lenient, they need to keep a close watch that public safety is maintained. But this research is encouraging for policies that show leniency toward nonviolent low-level offenders.
Amanda Agan is an assistant professor of economics at Rutgers University.
Jennifer Doleac is an associate professor of economics at Texas A&M University and director of the Justice Tech Lab.
Anna Harvey is a professor of politics at New York University and director of the Public Safety Lab.
This piece was written for The Washington Post.