The sharp rise in murders across much of the country in 2015 has posed a new puzzle among criminal justice researchers. But at least one criminologist is coming around to a provocative new theory.
In 2015, there was a 17 percent increase in homicides across 56 large US cities, according to research by criminologist Richard Rosenfeld for the US Department of Justice. Much of the increase came from 10 cities, which saw an average increase of 33 percent. This is alarming: For the past few decades, crime and homicide rates have plummeted in the US. So for there to be a fairly significant increase in US cities, which could signal an end to the crime drop, is worrisome.
Researchers don't yet have hard answers for why murder rates rose in 2015, or if the numbers indicate a reversal of the long-term trend.
But one theory that's come up — and analyzed in Rosenfeld's study — is the "Ferguson effect": the idea that the Black Lives Matter protests against police brutality have scared police, chilling proactive policing practices that stop and prevent crime, and emboldened criminals, leading them to commit more violent crimes. The effect gets its name from Ferguson, Missouri, where the first major Black Lives Matter protests over police brutality took place after the police shooting of Michael Brown.
Rosenfeld, a criminologist at the University of Missouri in St. Louis, was once a leading skeptic of the theory, publishing a paper that suggested crime rates had been rising in St. Louis before the protests in Ferguson, which is near St. Louis.
But after studying the issue more closely for the Department of Justice, and finding the rise in murders in 56 large US cities, Rosenfeld has come around to the possibility of the Ferguson effect.
Rosenfeld considered three theories for the rise in murders he found: the rise in heroin use potentially driving violent black market activity, recent drops in incarceration, and the Ferguson effect. Rosenfeld ruled out heroin and de-incarceration, he said, because they've been trends for years, so they can't really explain a recent rise in murders. But the Ferguson effect really is a recent phenomenon that would explain a sudden, recent rise.
Now, Rosenfeld isn't completely convinced that the Ferguson effect is the definitive answer for the rise in murders. For one, there isn't that much good data on 2015 yet — the fact he could only look through data for 56 large cities, but no smaller cities or suburban or rural areas, is a testament to this. And only murder rates are increasing sharply, not all crime or violent crime — something that there's not yet a very good explanation for.
But Rosenfeld's shifting view is notable nonetheless: He now sees a theory he once tried to debunk as more credible. I called him before the official release of his new report to understand how the Ferguson effect may work, and what its implications could be for criminal justice policy. What follows is our conversation, edited for length and clarity.
The two theories behind the Ferguson effect
German Lopez: There are two theories for the Ferguson effect. One is that police have been scared by criticism and are pulling back, and that's leading to more crime. The other is that Black Lives Matter protests have reinforced community distrust, which has boiled over, leading to community members taking the law into their own hands since they can't trust police. Which one of these do you think has more credibility, in light of your research?
Richard Rosenfeld: I find the second somewhat more plausible. But I have to say that we don't have the data yet to fully evaluate them. Well, let me qualify that a bit: We do have the data — the FBI has the data, but it won't release year-end data until sometime next fall.
So we can evaluate the depolicing argument, right? If depolicing is occurring and it comes with pulling back, we should expect arrest rates to be declining in those cities where we've seen very sharp increases in homicide.
Now, there is a paper out of Johns Hopkins [University] that looks at that in the case of Baltimore, and finds a "Freddie Gray effect" in Baltimore — that is, he finds declining arrest rates associated with increases in homicides in Baltimore [after Freddie Gray died in police custody, leading to protests and riots]. In one other city, Chicago, I'm prepared — although we don't have systematic evidence that I'm aware of — to believe that something like depolicing is contributing to more homicide there.
But I don't know that that's the case in any other city.
"I do find it somewhat hard to believe that the entire increase could be explained simply by the withdrawal of the police"
I am familiar with the city of St. Louis; I work fairly closely with the police department there. And the chief there, Sam Dotson, has good evidence that the arrest rates in St. Louis did not drop through 2015. They went down slightly immediately after the Ferguson incident, because police were being redeployed from normal patrol operations to address protest activity around the city of St. Louis and in Ferguson and elsewhere. But relatively soon, they returned to more normal levels. So I don't think that depolicing a terribly good explanation for what was going on in St. Louis.
We don't have evidence yet to determine if depolicing contributed to the homicide increases we've seen in other large cities.
I do find it somewhat hard to believe that the entire increase could be explained simply by the withdrawal of the police. That assumes that the police have an enormously powerful influence on crime rates. And while the research shows that police do influence crime rates — in particular, police actions [known as] "proactive policing" can reduce crime — it's limited.
A 90 percent increase in homicide in Cleveland, for example, or 75 percent increase in Milwaukee in 2015 or doubling the homicide rate in Orlando, Florida — those changes are so large it's hard for me to believe that they're brought about simply by withdrawal of the police. Perhaps if the police simply stayed home everyday [and] went on strike, maybe. But that's not what's been happening.
GL: What about the other possible Ferguson effect theory?
RR: I'm at this point — with the caveat we don't have the evidence we need to draw strong conclusions — more partial to the second explanation: that what may have happened is longstanding grievances, specifically among black communities and in some cases Latino communities, with the police are more activated by the heavily publicized incidents in police use of force, especially those involving unarmed, young black men in the latter part of 2014 and 2015.
There is good sociological and historical evidence that if people lose confidence in the police to protect them, or if they lack trust in the police because they believe the police are harassing them or behaving unfairly, they do tend to take matters into their own hands. So one tends to see preemptive killings and retaliatory shootings go up.
So that's the hypothesis that I came away with having looked at now 56 cities, not simply St. Louis, for what may have happened in 2015 and, in some cities, lasting to the current year.
Depending on which theory one ascribes to, there are starkly different policy implications
GL: The two Ferguson effect theories seem to have completely different implications. I know some of the conversations have suggested that if police are pulling back, we should stop being so critical of police. But if it's lack of trust in police, it seems that it would have completely different implications.
RR: That's the Heather Mac Donald takeaway: to stop criticizing police and let them go back to do their jobs fully and safely, and crime rates will go down. If depolicing is the primary reason for why crime rates have gone up, that argument would have some support.
If it's community lack of confidence in police, I think there are two policy implications: One, the police — and maybe this is related to the depolicing argument — have to do more than they have to get out of their patrol cars, meet with community residents, and work on problem-solving issues with the local community residents. They have to have more impact. They have to knock on people's doors. They have to hold community meetings. They're going to take a lot of criticism at those meetings, and that's part of the job. It's going to be in a slow, somewhat arduous process, but it's the only way to reestablish trust.
The other policy consequence, however, is the police have to do a much better job in protecting minority communities from the very high levels of homicide and other serious violent crime that's present in so many of these communities.
"I read the survey evidence of African-American attitudes toward the police as, 'You don't protect us from violence to the degree that you should. And when you are here, you harass us.'"
Now we're talking about something that predates Ferguson and other use-of-force incidents. If one boils it down, [disadvantaged minority communities] have two major gripes about the police.
One is that [police] harass [them] over petty issues — small amounts of drug, profiling, and so forth.
The other is [police] don't adequately protect [them] from the threat of violence. And that one translates to, "They don't treat homicides that occur in our community, especially when the victim or the offender has a criminal record, as seriously as they treat homicides in other communities. They don't investigate them as extensively. They don't throw the resources at them as they do for homicides that occur elsewhere."
It's another twist on the issue of whether black lives matter. I read the survey evidence of African-American attitudes toward the police as, "You don't protect us from violence to the degree that you should. And when you are here, you harass us." That's obviously simplifying things, but there are two prongs.
Even if the Ferguson effect is real, it may not be happening all across the US
GL: Based on your analysis, do you think that the Ferguson effect is some sort of national trend, or is it something that's happening in some cities but not others? Because that's something that seems to be getting lost in the broader conversation.
RR: Again, we don't know. Anyone who is interested in this issue — and that would include me, other criminologists, Heather Mac Donald, the Brennan Center, the Washington Post, Vox — all of us have to rely on these relatively small and in some ways unrepresentative samples of large cities. And we have to be very careful about drawing broad conclusions from that kind of data.
I'm reasonably confident that homicide rates did go up and rose sharply in the year 2015 in most large cities. Of the 56 I looked at, homicide rates were up in just a little over 40 of them. And in 10 of those cities, the 10 that accounted for two-thirds of the total rise in those cities, they were up 33 percent.
And then you've got cities of the sort I mentioned — Cleveland, Milwaukee, Orlando — where you've got a real, true rise in homicide increases. These are percentage increases based on a relatively large base. So we're not looking at doubling of homicides from two to four; we're looking at doubling of homicides from 50 to 100 or 75 to 150. Those are worrisome increases, and they need to be explained.
GL: One place that still makes me a bit skeptical of the Ferguson effect is New York City. If you'd expect to see a Ferguson effect, it seems like it would be there. It has a mayor who's perceived as critical of police and a large population of aggrieved minority residents, yet it doesn't seem to have had the same increase in crime and homicides as these other places. Do you have any good ideas for why?
RR: I don't.
New York is anomalous in so many ways. New York has had this truly spectacular crime drop — sharper than the national crime drop — over the last two decades or more. Now why that may be remains a matter of some debate. But New York is something of an anomaly with respect to the crime increases that occurred in the more recent past.
One might expect that New York, as the nation's largest city, should see comparable changes or absence of changes as Los Angeles or Chicago. But LA is up right now, and Chicago has been up for many, many months. And New York really stands as an outlier.
It's not clear if the Ferguson effect will hold up in the long term
GL: One thing I'm trying to gauge is how temporary the Ferguson effect might be. Do you have any good idea, based on your research?
RR: We can't know yet whether it's a reversal in the long-term decline in homicide and other crime we've seen in the past couple decades. That's going to require data over the next few years. We should not — and I certainly do not — use a single year of an abrupt increase and indicate that the long-term downward trend is now over.
But I do think it's important to understand — as best we can — why homicides rates increased as precipitously as they have over the last year or so.
We just talked about policy implications. If we use this as a reason to redouble efforts to fix the rift between the police and disadvantaged minority communities, and redouble efforts to investigate or respond more effectively to serious crime in those communities, then I would expect this to be relatively temporary. But as I said, those efforts are going to take some time.
We've been here before. During the 1960s, there were major urban insurrections throughout the United States in the big cities, [and] crime rates increased after having declined through the 1950s. But in the early '60s, they began going up. In the mid-'60s, many cities were on fire.
"If we use this as a reason to redouble efforts to fix the rift between the police and disadvantaged minority communities … I would expect this to be relatively temporary"
If you think back to those incidents that were the catalyst for urban insurrections, they were almost always incidents involving tense contact and conflict between the police and typically young black men. They tended to be the catalyst for those disorders. And it activated longstanding grievances, and that's one of the reasons we saw the big rise [in crime] we did in the '60s and on through the '70s. But crime rates peaked, and then began to come down.
But even with the uptick in 2015, we're still running with homicide rates in most cities that aren't back to where they were in the '60s and '70s.
If the use-of-force incidents diminish, that won't necessarily address the underlying grievances. But it may make them somewhat less consequential in terms of crime increases.
All of this is highly speculative. If we're seeing a reversal from the long-term trend, we have to wait at least another year or so to see how it's looking.
This issue exposes the lack of good, current crime data
GL: One of the consistent issues we've raised is the lack of current crime data. What makes this such a big problem?
RR: We could know more than we do. Our crime monitoring system — if that's what one wants to call it — is based on the Uniform Crime Reports from the FBI. It is woefully inadequate, and it need not be. I've been calling for a long, long time for a Uniform Crime Reports program to release monthly crime statistics in a very timely way.
So we should be seeing crime data, at least for the large cities in the United States, one or two months after the month the data was collected — just as we do for unemployment; just as we do for inflation; just as we do when influenza breaks out, and the CDC is able to give us weekly case counts.
I'm not suggesting that we're going to have crime data of the sophistication or sheer magnitude or timeliness of the economic indicators, but we can do a lot better.
I pointed out in the paper I wrote that back in the 1930s the US did release monthly crime data. It's striking. There were fewer police departments then, right? But those crime reports were filled out in pen and ink or a manual typewriter, and then sent by the local post office to Washington.
All the big city departments and many other departments now digitize their crime data, and have it available for their own use in almost real time.
Indeed, local police departments don't send their crime data directly to the FBI in Washington. They send their crime data to their state [Uniform Crime Reports] program; there are 49 of them. The state program then vets the data for completeness and accuracy, and then some of those state programs send it on month-by-month to the FBI, where it sits until nine, 10 months after the end of the collection year. That just baffles me.
GL: One example that comes to mind is the Philadelphia Police Department, which right now has homicide data from up to two days ago on its website. It's bizarre.
RR: One could argue that if these cities have all their own data, what do they need the FBI for? The problem is that if you want to know that what's going on in your community is being driven by purely local idiosyncratic factors or is part of a broader trend, you need to be able to compare your community with others. And that's what the Uniform Crime Reports were supposed to enable police departments to do. That's why they're called uniform.
The local police can only do so much with their own data — to figure out what's going on in their communities.