National overview
Progressive people respond to a surprising crime study
Every year, something like 13 million crimes are filed in the United States. These charges, ranging from traffic violations to serious assaults, may be less flashy than crimes, but they are the most important way Americans experience the criminal justice system. We prosecute crimes because, among other things, we want there to be fewer of them, and we believe that prosecution prevents the offense. But in a recent blockbuster article, the opposite is made striking: prosecution of crimes increases the likelihood that they will offend again. The newspaper was announced by supporters of progressive district attorneys who used their position to unilaterally apply reforms to the criminal justice system, including the refusal to prosecute many crimes. Boston DA Rachael Rollins, who provided the data for the study, claims it confirms the wisdom of her approach. So do other reformers like the Chicago District Attorney, Kim Foxx, and San Francisco District Attorney Chesa Boudin. However, policymakers need to be careful before reaching such broad conclusions. The newspaper can just as easily be read to endorse more modest reforms – especially considering the all-encompassing principles of criminal law on which it is silent. The paper is the work of three researchers: Amanda Agan of Rutgers, Jennifer Doleac of Texas A&M and Anna Harvey of NYU. Both Doleac and Agan have previously published research challenging progressive policy preferences, and their new findings are unlikely to be driven by a desire to achieve a politically comfortable outcome. To conduct their study, the three obtained data on each criminal case filed in Suffolk County (Boston’s home) between 2004 and September 2018. They analyzed the relationship between whether a crime is prosecuted and whether he was subsequently reappointed, indicating that he had transgressed again. The consequences are frightening: if you are not prosecuted for a crime, the likelihood of a future crime charge is reduced by 60 percent, and of a future crime charge by 47 percent. It also significantly reduces the likelihood of future violent, motor vehicle and disorder / theft offenses, although it is not drug abuse. In other words, prosecution not only keeps subsequent crimes at bay, but also increases the chances of re-offending. This is according to the authors of the newspaper because any deterrent effect is exceeded by the effect on the prospects on the labor market. Unemployment can lead to crime, and being prosecuted can increase one’s chances of unemployment. It also creates a criminal record, which employs the offender less and thus makes them more susceptible to crime. How did the researchers come to these conclusions? To understand how prosecution affects the risk of retaliation, we can not just compare prosecuted and unpunished offenders to see what more crimes are committed. A person may not be prosecuted just because she does not consider a risk. There are confusing variables that both determine what we need to control for. To work around this issue, the paper uses an “instrumental variable” that is not related to the confusion. All offenses charged in Suffolk County are charged by an Assistant District Attorney (ADA). Based on the other cases of ADA, the authors construct some of their ‘indulgence’, i.e. their tendency to prosecute a given offender. The ADAs mostly seem to agree on prosecutions – they will usually prosecute about 70 percent of the cases and usually drop about 20 percent. But about ten percent of the time, it varies in their derivative indulgence. The allocation of these ten percent of offenders to ADAs of varying flexibility becomes the source of randomness. What this means is that most of the papers apply to the “marginal” offenders. If you do not prosecute all offenses, it will not reduce everyone’s risk of offending, but if you do not want to prosecute the marginal offense – the one on the line between prosecution and non-prosecution, he reduces the chance to transgress again a lot. By the end of the article, the authors generalize these marginal offenders. They find that non-prosecution has a major impact on all offenders, and that is, on average, a 15 percent reduction in re-offensive risk. The effect is most concentrated among those who are usually persecuted: if they are persecuted, the authors write, they are much more likely to transgress them. Oddly enough, the effect on those who are not usually prosecuted is distinguishable from zero. If ADAs prosecuted those they would not normally do, there would be no average effect on their future propensity to offend. When I asked Doleac about this finding, she suggested that it might reflect the ADA’s judgment on guilt: The people who are most likely to transgress may also be the ones who are most lenient with ADAs – the young, mentally ill, etc. . – well-adjusted adults who have made a mistake – are those for whom ADAs have the least sympathy, but for whom prosecution can have a major, negative impact. According to Doleac, this indicates that there is a significant difference between debt and risk. To me, it also indicates that ADAs are not big judges about the consequences of their prosecution choices. There is a second important detail that the authors pay attention to, but which is missed in some commentary on the newspaper: Most of the non-prosecution effect they measure is the result of first offenders, who are much more likely to commit crime as prosecuted. In contrast, the prosecution of repeat offenders of any kind has little noticeable effect on the likelihood that they will be offended again in the future. This is not surprising, since most offenses are committed by a handful of offenders – criminological research consistently finds that a small, offensive population commits the most crime. For those outside the population – including many first-time offenders – prosecution will probably not deter them from doing something they would not do anyway, but it could have detrimental consequences that would drive them to crime. So should we prosecute fewer offenses? We can not draw too dramatic a conclusion from one study of one country, no matter how well designed. And while this one relies on the latest statistical techniques, we should always be wary of findings that can only be reached through extensive statistical interrogation. The sheer complexity of the instrument used in the paper, combined with the huge effects it finds, should temper enthusiasm – there are simply too many research degrees not to do so. That said, we can cautiously conclude that the best evidence says that the marginal offense should be prosecuted less frequently. But if ADAs are bad at judging the consequences of their prosecution, we should not assume that they are good at telling the marginal offense of the future serial offender. Whether the results of the study are wrong or ADAs are poor judges on how prosecution relates to future transgression, we need to be careful not to give them too much leeway to decide who is a marginal case and who is not. We can rather offer a rule of thumb: wander if you do not doubt not to prosecute crimes in the first place. Diverting these offenders, with the threat of worse punishment if they transgress again, can help clean up dossiers while keeping crime to a minimum. It will also free ADAs to concentrate on repeated violations. The target of repeated offenders will reduce the risk of abuse of the first diversion by making it clear that a ‘second chance’ will not be followed by a third, a fourth, a fifth, and so on. Research on California’s “three-strike law”, for example, indicates that increasing punishment for repeat offenders can be a powerful deterrent. The above approach differs from the idea that we should generally prosecute offenses much less – a valid interpretation of the article’s findings, but not necessarily the rights, for two reasons. First, deterrence is not the only reason to prosecute an offender. Proponents of non-prosecution tend to take up a victimless crime such as drug possession and prostitution. But offenses can also include offenses such as simple assault and car theft – crimes that harm others. Such crimes reasonably elicit a claim for retaliation. It makes us morally offended to think that someone who commits a serious but not criminal assault can get away with it. Second, systematic reduction of indulgence can affect the decision-making of all criminals and thus increase their propensity to offend in the long run. The newspaper shows that Rollins’ attempt to prosecute crimes did not increase the crimes as a whole, but the data it uses is only responsible for the period between her election in January 2019 and March 2020, when the coronavirus crisis has begun. It is entirely possible that criminals will adapt in the long run, and the number of offenses will increase. Changes in the blanket policy may cause increased resentment. The increase in California in 2014 to the threshold for crime theft, for example, has predictably led to an increase in theft at the city level, suggesting that offenders change their behavior in response to such shifts. It can be time consuming and exhausting to come face to face with the justice system, and can increase on the side of the case rather than reduce the tendency to offend. Even those among us who are very concerned with public safety should be interested in creative solutions that keep crime and disorder to a minimum. At the same time, policymakers should not pre-empt themselves, as some are in a hurry to defend police departments and reduce the use of more serious charges. Good research is the basis of good policy and this research makes a valuable contribution to public safety policy. But we have to be careful about how far we go with it – careful changes on the sides are always safer than the blankets.