Debunking the Ferguson Effect

The Brennan Center for Justice released a report on Thursday that explores crime trends in 30 cities in 2015. In it, the report’s authors sought to test a hypothesis put forth by some scholars and journalists about a purported crime wave this year.

First, some brief background. In a widely shared May 29 op-ed titled “The New Nationwide Crime Wave” in The Wall Street Journal, the Manhattan Institute scholar Heather Mac Donald wrote about an apparent rise in homicides and other major crimes in some American cities in recent months. She called this rise the Ferguson effect, a term she borrowed from St. Louis police chief Sam Dotson, and described it as a “current surge in lawlessness” caused by “the intense agitation against American police departments over the past nine months.” In other words, the heightened scrutiny of American law enforcement led them to hesitate more on the job, thereby driving up crime.

Mac Donald’s thesis provoked a significant amount of debate, both in defense of and in opposition to it. My colleague David Graham described the Ferguson effect as “the Bigfoot of American criminal justice: Fervently believed to be real by some, doubted by many others, reportedly glimpsed here and there, but never yet attested to by any hard evidence.” My other colleague Ta-Nehisi Coates called it an “utterly baseless suggestion.” But a number of major news outlets, including The New York Times, explored the idea and thereby gave it some legitimacy; FBI Director James Comey and DEA chief Chuck Rosenberg also cited it.

The Brennan Center’s report is an attempt to shed some methodological light on the debate. “These stories have been based on a patchwork of data, typically from a very small sample of cities,” the report notes. “Without geographically complete and historically comparable data, it is difficult to discern whether the increases these articles report are purely local anomalies, or are instead part of a larger national trend.”

To fill these gaps, the researchers looked at homicide rates and crime rates from January 1 to October 1 from 25 of the 30 largest American cities. (Five of the cities did not provide data, the authors note.) The entire report is worth reading, but two graphs in particular stood out to me. The first one is this overview of the murder rate for those 25 largest American cities.

Based on their data, the Brennan Center projects that homicides will rise slightly overall from 2014 to 2015. Any rise in the homicide rate is troubling, especially for the handful of cities largely responsible for the increase, but the relative uptick is still small compared to the massive two-decade drop that preceded it. Mac Donald warned in her May op-ed that “unless the demonization of law enforcement ends, the liberating gains in urban safety over the past 20 years will be lost.” That fear is unfounded so far.

The second graph drives home this point on a city-by-city basis and helps explain how a nationwide spike in homicides seemed plausible. The Brennan Center compared this year’s projected homicide rate in the cities it examined to the last year each of those cities had a similar rate.

In three cities, the murder rate returned to 1990s levels and remained high in some others. But in many more, there are only minor variations from the current lows. Homicide rates in cities with numerical upticks this year like New York, Los Angeles, and Chicago still resemble the historically low homicide rates those cities’ residents have enjoyed for almost a decade.

The researchers also noted that some cities fared worse.

There is no question a few cities have seen troubling increases in murders. Murder rates in Baltimore are now at 1990s levels. And in Milwaukee and St. Louis—where murder rates were already relatively high—murder rates have risen sharply, and are now near 1993 levels. Rather than a national pandemic, it appears that the increases in murder rates are localized, suggesting that community conditions are a major factor.

To offer a possible explanation, the researchers noted these five cities also share lower incomes, higher poverty rates, falling populations, and high unemployment compared to other cities. “The relationship between economics and crime is hotly debated," the report states, “but it is possible that the weak economies of these cities are a contributing factor to their high murder rates.”

From all of this, two things stand out. First, the fears of a “a new nationwide crime wave” are premature at best and wildly misleading at worst. The numbers make clear that violent crime is up in some major U.S. cities and down in others. “However, there is not yet sufficient evidence to conclude that these levels will persist in the future or are part of a national trend,” the report concludes. “Although headlines suggesting a coming crime wave make good copy, a look at the available data shows there is no evidence to support that claim.” And with the crime wave’s existence unsupported by evidence, so too is the claim that heightened scrutiny of law enforcement caused it.

Second, and perhaps most importantly, this debate may have been avoided if the United States had a better nationwide system for reporting crime data and provided monthly city, state, and national updates instead of annual ones. I’ve written before on the gaps and delays in U.S. criminal-justice data and the misunderstandings they can foster in public policy discourse surrounding criminal justice. Even the Brennan Center report notes that recent crime and homicide data “could not be secured” from Columbus, El Paso, Indianapolis, Nashville, or San Francisco, five of the largest U.S. cities. The turn towards data-driven policing since the CompStat revolution of the 1990s makes the difficulty of obtaining regular, reliable, and public data even more galling.

Perhaps the best illustration of the problem is the voluminous amounts of statistical reports churned out by the rest of the federal government on a regular basis. NASA releases monthly reports on planetary heat temperatures tracked by orbiting satellites, for example, and the U.S. Labor Department provides weekly reports on new jobless claims and monthly reports on the national unemployment rate. This task is not impossible for law enforcement agencies around the country, and the benefits it could offer to public discourse and academic study is a compelling enough reason to try.