It's now well known that crime rates have decreased in the United States and in other developed nations over the past two decades. The reasons offered have included rising incarceration, improved economic conditions, better policing, aging populations, decreased crack use, abortion policy changes, reductions in childhood exposure to lead, or some combination of these and other factors.
None of these evaluations is considered definitive. Many assessments point to an array of possible causal factors without much supporting evidence. And those that identify a single explanation for the decline, such as changes in abortion policy or reductions of lead in the environment, have been criticized for attributing an implausibly large causal role to that one factor without accounting for the influence of others. No study has incorporated a broad range of possible explanations of recent crime trends in a single analysis—until now.
The Brennan Center of Justice at the NYU School of Law this month published an ambitious empirical investigation of 14 factors identified by criminologists and media accounts as having contributed to the U.S. crime decline over the past two decades.
Like all studies, this one has its limitations. The authors, Oliver Roeder, Lauren-Brooke Eisen and Julia Bowling, acknowledge that they were unable to obtain the data they needed to evaluate the role of several factors, including abortion policy, lead levels, inflation rates, and consumer confidence. In these cases, they relied on past research to estimate their probable influence. That resulted in some problems. For example, the authors find that the growth in income contributed to the crime drop. But they did not adjust the income data for inflation, and other research has found no significant influence of inflation-adjusted income on crime rates.
Omitted variables aside, however, the Brennan study is important for what it did—and did not—reveal as causes of the crime decline.
Despite the inclusion of multiple explanatory factors in its analysis of the crime drop, the study directs major attention to a single ingredient: the growth of imprisonment. Unlike other single-factor explanations of crime trends, however, the authors conclude that prison expansion has had only a modest influence, perhaps explaining 5%-10% of the crime drop. And nearly all of that effect had disappeared by the turn of the current century.
They maintain that imprisonment growth has diminishing returns on crime reduction:
“. . the current exorbitant level of incarceration has reached a point where diminishing returns have rendered the crime reduction effect of incarceration so small, it has become nil” (p. 7).
The effect of imprisonment on crime rates, the authors claim, is greater at lower levels of incarceration because only the most serious offenders are imprisoned. As imprisonment levels increase and less serious offenders are sentenced to prison, the effects of imprisonment on crime diminish.
This is not a new argument; other students of mass incarceration have made the same point. But the Brennan study extends prior research with more recent data and their analysis is persuasive. There remains little question that prison expansion contributed more to past crime reductions than it does now.
But the “diminishing returns” argument remains open to debate. Growing rates of imprisonment do not necessarily result in the incarceration of less serious offenders. Prison expansion may simply capture a larger fraction of serious offenders. Besides, imprisonment rates began to level off after the turn of the century. That, too, may have resulted in smaller incarceration effects on crime. The Brennan study is but an initial step in understanding the complex role of imprisonment in the crime decline.
According to the study, prison expansion is not the only factor that has had little or no impact on recent crime trends.
The authors find no evidence, for example, that “right-to-carry” laws that permit non-felons to carry guns in public have led to decreases in crime. Nor do they find any effect of the death penalty on crime trends. What factors, then, did play a part in the crime drop?
The authors' empirical analysis discloses three contributing factors: the aging of the population, decreased alcohol consumption, and income growth. Based on past research, they also speculate that declining inflation rates and growing consumer confidence may have resulted in crime reductions. The authors' results are based on an analysis of yearly crime trends across the U.S. states. In a separate analysis of large cities, they also find that the introduction of Compstat programs in local police departments, which hold police managers accountable for crime rates on the basis of real-time crime data, contributed to monthly crime declines in the 2000s.
None of these factors, by itself, had a large effect. Taken together, all of them explain no more than half of the crime drop, based on the authors' estimates. That leaves the other half of the U.S. crime decline unexplained.
Explaining the dramatic decline in crime rates over the past 20 years has proved to be one of the most difficult research challenges facing social scientists in the U.S. and elsewhere. But the difficulty of the task is no reason to give up the effort.
Knowing why crime rates go down is just as important from a research and policy standpoint as knowing why they go up. If we understand the causes of crime reductions, we can work to sustain them over time. The Brennan study has not solved the riddle of the crime drop, but it has shown that recent crime trends are amenable to reasoned and evidence-based inquiry and it has identified at least some of the contributing causes.
Social scientists, with constructive prodding by the press and policymakers, need to build on this important work and continue searching for answers.
Richard Rosenfeld is the Founder's Professor of Criminology and Criminal Justice at the University of Missouri – St. Louis. He chairs the National Academy of Science's Roundtable on Understanding Crime Trends. He welcomes comments from readers.