The Spike in Homicide in 2020
The murder spike was caused by the COVID-19 pandemic. It was not about policing or protests.
On Wednesday, September 22nd, the New York Times reported on the 2020 Uniform Crime Report, noting that the FBI data had been uploaded to the FBI’s Data Explorer. As I write this on September 26th, I cannot access those data, but I assume that the Times reporting is accurate and that next week the FBI will release the 2020 numbers and they will be essentially similar to the Times story.
What I have in mind here is not really a hot take, but rather a prebuttal. For various reasons, crime data takes a long time to process, so by this point in September 2021, a lot of 2020 crime data has been gathered from other sources—mainly from big-city law enforcement agencies. So, even before the release, the picture is very clear.
In 2020, crime in America fell for the 19th straight year, the longest decline ever recorded using modern crime statistics.
Also in 2020, America experienced the largest one-year increase in homicide ever recorded using modern crime statistics.
Two record events, going in opposite directions.
There’s a lot to unpack here. The reason I feel motivated to offer a prebuttal, however, is that along with the 2020 crime data that has been circulating in advance of the FBI’s data release, a lot of preemptive explanations have been circulating as well. A lot of these explanations feel like what I would call policy-based evidence-making, rather than evidence-based policymaking. The picture is confused, things are great, things are terrible, and it is relatively easy to pick out the data that fits an existing narrative and run with that. Hence the need for a prebuttal.
So let me start by saying that I think the explanation here is very straightforward. But let me tease the answer a little and first talk about the key to breaking the code.
Obviously, the headline here is the spike in murders, but you cannot understand what caused the spike in murders without also understanding why crime fell for the 19th straight year. Or, if you think one thing caused crimes to go down and a different thing caused homicide to go up, you would need to explain that. And you would need to explain how Thing 1 did not affect Thing 2 and vice versa.
So that leads to my first point in my prebuttal, which is that the most of explanations for why murder spiked in 2020 simply ignore the overall decline in crime. Folks in this camp just want to talk about Thing 1 (murder) and ignore Thing 2 (crime), and as you know from your Suess, Thing 1 and Thing 2 are a team. They cannot be understood by themselves.
As I hope to show you down the page, the problem with most of the arguments getting a lot of air time these days is that they only explain why murder spiked and they leave you scratching your head why overall crime declined (not to spoil my newsletter, but perhaps just to remove the cork to give it some air, I would point out that most of the explanations centered on policing fail this test).
Why ‘Crime’ and Homicide are Almost Always Correlated
The first idea to confront when exploring reasons for the murder spike and the crime decline is that in most years, in most places, homicide and other types of violence are intimately related. A lot of homicides are nonfatal violence that went sideways. It could be a drug deal that has been completed peacefully in the past that turns into a shootout. It could be a fight outside a bar where the victim who is punched doesn’t simply fall to the ground but instead hits their head on the curb. It could be a stickup with a nervous robber who points the gun without intent but ends up shooting. It could be a drug deal that has been completed peacefully in the past that turns into a shootout.
The point is that typically there is a pretty stable rate of homicide—called a case fatality rate—that accompanies other types of violence. Each robbery has a chance to become a homicide. As does an assault or a rape or even a burglary. Normally, you can’t neatly untangle these crimes. As each type of violence increases, homicide goes up when some of those crimes inevitably become murder. And conversely, usuals as other types of violence go down, so does homicide.
That brings us to homicide in 2020.
Looking at the data visualized in this Times graphic, there can be no doubt that the homicide increase in 2020 was real, far out of the ordinary, and seismic. Cataclysmic even. These facts are not up for debate. There were 16,425 homicides in 2019. There were 21,500 in 2020. Five thousand more people were killed in 2020 than in 2019.
Why Homicides in 2020 were Different
A clue in the data, however, is that not only did murder trends and crime trends diverge, violence trends and murder trends diverged as well. Murder increased by 29% and violence increased by 5% in 2020 over 2019. There is clearly a divergence between the murder spike and the small violence increase.
While there is a possibility that there was some major change that affected case fatality rates in 2020—and suddenly nonfatal assaults and robberies were more likely to become fatal than in the past—there is a much simpler explanation. That is, that many of the homicides in 2020 were not robberies or assaults that started as something else and ended in a homicide.
Instead, in 2020, they were more homicides where the goal of the crime was homicide.
This is more of an epidemiological argument than a causal inference, but the data points toward this conclusion.
Explaining the Crime Decline and the Murder Spike: The COVID-19 Pandemic and Response
When I first wrote about this a year ago, I gave the following explanation for the homicide spike, and I would offer the same explanation today. The only change I would make is that the closures I wrote about have persisted far longer than I anticipated. Schools, government offices and businesses, social services and community centers, churches and pools, remained closed throughout 2020 and some remain closed today—my office, and perhaps yours, has yet to reopen.
I think the summer of 2020 is mainly a routine activities story. Young men are stuck at home instead of working or going to school. They are experiencing the same anxiety the rest of us are experiencing, but in neighborhoods with a long history of violence they have the added burden of accumulated traumas, including unresolved disputes. And critically, the people they have serious beefs with are also stuck at home and close by. Imagine someone you loved was killed or shot or beaten and you know (or you think you know) who did it (that is the motivation). You are stuck at home without access to supports, a caring teacher, work for a sense of purpose, or professional supports (that is the lack of guardianship). And the guy who did it is just a few streets away (that’s the target). It’s just a toxic situation.
That’s the explanation for the homicide spike. Dense clusters of young men stuck at home with little to do carrying the burden of past trauma, knowing those with whom they hold deep grudges are close by and home too.
For an explanation for the homicide spike to hold, it must be true everywhere, for almost everyone—homicide spiked in cities, suburbs, and exurbs, in cities with red mayors and blue mayors. This explanation does.
A little thought experiment helps to clarify. Suppose that in normal times, these young men were home half the time and out and working some, or in school some, or just out, the other half. If this is also true of their enemies, then the chance that both are home at once is small, maybe 1 in 4. With a 25% chance, you are both around at the same time, the search costs for finding your target are high. Now suppose the pandemic forces you both home three-fourths of the time. Now the chance you are both home about doubles. Suppose during the most intense period of closures, you are home 90% of the time and your enemy is too. Now the chance that they are home when you are home and looking for them has more than tripled. It’s easy to see why there would be more deadly encounters.
But, what about the Crime Decline?
I said upfront that any explanation for the murder spike must also explain the crime decline. I think COVID-19 as a causal mechanism does this as well. It’s important to note that the ‘crime’ measures used to calculate the FBI crime rate are tilted toward property crime. In the FBI statistics, Part I property crime is defined as burglary, larceny-theft, and motor vehicle theft. Because the volume of those property crimes is so much higher than the number of violent crimes, there are six or seven property crimes per violent crime in the total number. The 2020 numbers will show big enough declines in property crime to offset the 5% increase in violence which leads to an overall decline in crime.
Thus the critical question: what would cause property crimes to decline while violent crimes increased? The COVID-19 pandemic and response. It is straightforward to show how the pandemic could lead to declines in burglary, larceny-theft, and motor vehicle theft.
Motor vehicle theft would decline in 2020 due to COVID-19 because more people’s vehicles would be in their driveway than a parking lot or on the street, more of the time. It is a lot easier to steal a car from the Walmart parking lot than it is out of your driveway. Larceny-theft would decline for any number of reasons due to people being home more—one easy example is shoplifting which would decline when stores were closed. Residential burglary would decline as more people were at home, and in a nation of 300 million+ guns, burglars are as careful as they can be to target unoccupied residences (this is why residential burglaries in the US are mainly daytime affairs, and burglaries in the UK (where civilian gun ownership is rare) are more evenly spread between day and night). Commercial burglaries of unattended and closed businesses would be expected to increase, though not nearly enough to offset the other negative effects on crime.
Alternative Explanations for the Murder Spike
That the COVID-19 pandemic and response explains the murder spike will strike some as too obvious to even write about and others as far-fetched. For those who see it as far-fetched another explanation—changes in policing—is probably their preferred explanation. So, let me offer this prebuttal.
Policing and the Murder Spike
When people saying changes in policing explain the homicide spike, I think they mean several different things, and each is worth mentioning very briefly.
The Protests. One common alternative explanation is that the protests caused the homicide spike. One way that is interpreted is extremely literally—that most of the additional homicides were at the hands of the protestors. There is little evidence to support this notion. It is difficult to get a clear number, but most estimates are that around 25 people died concurrently to a protest. With almost 5,000 additional homicides in 2020, this is a small part of the explanation.
The Protests (Part II). The other protest-related explanation is that so many policing resources were diverted to respond to the protests that the deterrent power of the police was so reduced and the costs of violence were therefore so reduced that violence flourished. There are two rebuttals to this argument.
First, if this was true, the reductions in deterrence would certainly have affected all types of crime. Most particularly, aggravated assaults and robberies would have risen at the same rate as homicide. Why would those motivated solely by murder feel suddenly unshackled but those motivated by pecuniary gains (robbers) respond differently?
Second, the volume and duration of protests were insufficient to explain the increase in homicide. Here’s the thought experiment. If there are 750,000 sworn officers in the US and they work a typical year of 2,000 hours that sums to 1.5 billion policing hours annually. If there are constant rates of return to policing, as is often assumed in economic studies, then a 29% increase in homicide equates to a 29% reduction in deterrence. For the protest explanation to be true, it would have to be true that police spent 435 million hours policing protests, all at a cost of a total loss of any conventional deterrence from those policing hours. That’s a lot of hours.
Police Legitimacy Crisis. It became widely apparent in 2020 that policing faces a legitimacy crisis in the United States. That crisis was, of course, the impetus for the protests. As an explanation for the homicide spike, it would manifest in two ways.
Police work to the rule. The idea here is that police, in response to an unsympathetic public, engaged in what used to be known as a work to the rule strike—where they only engaged in that portion of their duty explicitly prescribed. In practice, the argument is that police stopped engaging in proactive policing, particularly stops that involved questioning individuals to develop probative evidence instead of calls for service. Putting aside the value of these kinds of activities, an analysis by the New York Times suggests that officers only spend about 10% of their shift engaged in this kind of ‘proactive’ policing. Even if every single police officer stopped proactive policing after May 25th, this would not explain more than a small part of the 29% homicide spike.
The Public Disengages. The flip side of the legitimacy crisis is that it signals an unwillingness of the public to interact with the police, so, for instance, fewer witnesses would step forward after a shooting. However, my discussions with people who work in high violence communities (both in the public and the police, and this is an anecdote, not data) suggest that the dynamic is not as straightforward as it sounds. In places with the highest violence and not coincidentally historic mistrust of police, police illegitimacy was already high. In these neighborhoods, the George Floyd murder and the response to the subsequent protests were more of a confirmation of an existing set of beliefs. In places less exposed to violence, the murder was more revelatory and had broader impacts on the public’s sentiment. That would suggest that the role of the police legitimacy crisis was relatively minor.
Officer Health and Safety. The fifth category of changes in policing as an explanation for the murder spike has gotten less attention than the other four. That is, many police departments implemented COVID protocols that required officers to play a more hands-off role in maintaining public safety, for their safety, particularly when transmission rates were high. Here, for instance, are the Philadelphia police department’s protocols, which would undoubtedly, and reasonably, lead to less officer engagement. This change certainly had some effect on policing, but it is again unclear why this change would explain the homicide spike and not the broader reduction in crime.
Overall, it is clear that the US experienced a tragic increase in homicide in 2020. It is also evident that the pandemic was a substantial explanation for that murder spike. The rise in murder appears unrelated to other changes in criminality, suggesting that the change in routine activity increased the number of people who sought to kill, and sadly increased the availability of their targets.
As the COVID-19 epidemic, hopefully, and mercifully wanes, this pressure will be reduced, and homicides should begin to fall. So far in 2021, that appears to be the case—homicide is no longer rising and there are signs at the end of the summer of 2021, that homicide rates have begun to decline. But the effects of the pandemic will linger. Hurt people, hurt people, and more people and more communities were harmed by homicide than at any time in a generation. Those wounds will not heal without an outpouring of resources and supports.