If you can’t change a law you don’t like, just stop enforcing it. That’s the philosophy behind the “progressive prosecutor” movement—which, with the help of big left-wing campaign donations, has elected anti-incarceration district attorneys in numerous big cities. Last week, Alvin Bragg took his position as Manhattan DA and promptly instructed his prosecutors, in the words of the New York Times, to “ask judges for jail or prison time only for the most serious offenses—including murder, sexual assault and economic crimes involving vast sums of money—unless the law requires them to do otherwise.”
Bragg insists that making incarceration a “last resort” will actually make society safer and peppers his memo with citations to friendly studies. Given the electoral success of progressive prosecutors in big cities, the movement isn’t going anywhere anytime soon. So it’s worth asking exactly how much punishment we can forgo, and how much leniency we can afford, before we start paying the price in increased violence and other crime. Naturally, a great deal of social science aims to answer that question.
The short answer: evidence exists—and indeed, it’s common sense—that leniency can be the best approach to certain offenders, especially first-time offenders who might be able to avoid a criminal record entirely if given a second chance. But that doesn’t mean that we can play nice with a much broader swath of criminals, or make categorical promises the way Bragg has that many harmful crimes won’t result in prosecution, without consequences. How long will urban voters tolerate those consequences when violence is already on the rise?
The biggest-picture studies ask what happens to crime when overall incarceration levels change, as happened dramatically nationwide in the 1980s and 1990s. The national imprisonment rate kept rising more moderately through most of the 2000s and has fallen slowly since. A fairly broad consensus holds, first, that rising incarceration played at least some role in controlling crime during the major buildup, contributing to the incredible 1990s crime drop; and, second, that incarceration can have diminishing returns as more marginal, less-worrisome offenders are locked up. The left-wing Vera Institute of Justice noted in a 2017 review that “overall, the increased use of incarceration through the 1990s accounted for between 6 and 25 percent of the total reduction in crime rates.” In terms of diminishing returns, however, one of the more methodologically complex studies in the literature, from 2012, found that “each additional prison-year served prevented approximately 30 index crimes” for the period 1978 through 1990, but “for the period 1991–2004, the comparable value is eight.”
Of course, if the returns to rising incarceration diminished, the process will presumably play out in reverse for decarceration: little effect on crime for small, sensible steps, but growing problems as decarceration becomes more aggressive and increasingly dangerous offenders are treated with kid gloves. In 2007, justice reformers lauded tough-on-crime Texas for backing off on low-level, nonviolent offenses; now we have a Manhattan DA reducing charges for felons caught carrying guns.
Other studies try to discern what happens to individuals’ future offending behavior, rather than to overall crime rates, as a result of incarceration. Of course, prison is effective at incapacitating offenders while they’re actually in prison, though most offenders are not locked up for extremely long periods. (Misdemeanors, by their usual definition, come with jail sentences of less than a year; those released from state prisons in 2018 had served a median of 1.3 years.) Fear of punishment also deters people from committing crimes to begin with, though there are limits to how much we can increase that fear by making punishments more and more severe.
Some have theorized, however, that the experience of prison itself might increase crime afterward by making offenders harder to employ and giving them criminal social connections. An alternative hypothesis is that prison reduces future offending by chastening criminals, rehabilitating them, or scaring them straight.
Both theories seem mostly overblown. Rigorous studies focusing on such effects were recently summarized in a helpful review by Charles E. Loeffler and Daniel S. Nagin. This is a hard thing to study: when one individual is incarcerated and another is not, there’s often a reason, so researchers can’t just blindly compare offenders who do time with those who don’t. The trick is to find situations where incarceration is effectively random, and thus similar people get treated differently—a “natural experiment.” A common approach, for example, is to look at places where cases are assigned to judges more or less randomly, but the judges have varying tendencies toward leniency, so certain marginal offenders are incarcerated by some judges but not others. Another approach is to look at “discontinuities” in sentencing rules, where a slightly more serious offense bumps up the offender’s punishment to include incarceration.
As Loeffler and Nagin summarized, “most such studies find that the experience of postconviction imprisonment has little impact on the probability of recidivism,” though there are stronger signs that pre-trial incarceration might have bad effects on future offending. Generally, while being locked up may keep offenders from committing crimes while behind bars, it does not greatly affect their chances of rearrest, new charges, or reconviction over a longer time period—sometimes as long as five or ten years in these studies.
The major caveat to these papers is that they pertain to marginal offenders whose incarceration came down to random factors. As Loeffler and Nagin themselves stressed, these are people “who have been convicted of less serious crimes and have less extensive prior records.” Even relatively lenient judges will send the scariest offenders to prison, after all, so there’s no randomness as to whether a convicted murderer is incarcerated.
The same caveat applies to a much-discussed paper from last year by Amanda Y. Agan and two coauthors. Like studies involving judge assignment, this paper exploited the fact that cases are more or less randomly assigned to assistant district attorneys in Suffolk County, Massachusetts. The paper finds that for the offenders whom some ADAs would prosecute but others would not, being prosecuted for a nonviolent misdemeanor increased offending over the next two years—an effect biggest for first-time offenders. And when the head district attorney created a “presumption of nonprosecution” for certain nonviolent misdemeanors, it seemed to have a consistent effect, at least in the short term.
If we do a better job of predicting which people will reoffend if not incapacitated and which will not—a task that we don’t seem particularly skilled at, but that risk-assessment tools have improved on recently—it’s possible to reduce incarceration and crime simultaneously, simply by incarcerating more accurately. We can also improve the swiftness and certainty of punishment by giving offenders speedier trials and improving policing, rather than relying on the sheer severity of punishment for deterrence.
As a whole, the results of this literature are a call for careful consideration of which offenders might not need prison. They’re not a call to decarcerate aggressively and thoughtlessly—as Bragg and other progressive prosecutors want to do—unless we all want to relearn, and live through, the lessons of the 1990s.