COVID-19 and Crime: How Responses to the Pandemic Affected Homicide Rates

Why did you decide to study the effect of COVID-19 on homicide rates?

I had already begun work on a much longer-term project studying weekly homicide counts in American cities. Since the beginning of the pandemic, there has been anecdotal evidence about homicides decreasing in certain cities. Arnold Ventures was interested in a quick turn-around project to get some more systematic evidence into the debate, so they asked us to do this report. 

I was interested to learn that homicides were actually up in the first three months of the year. 

Overall they’re up about 6 percent for the year so far. If you look at the change in January and February, compared to the same months of 2019, there were very sizable increases. I had not anticipated those increases and don’t know how to fully account for them. We saw about a 28 percent increase in January, a 17 percent increase in February and March, and then a 22-point decline in April. Typically as weather warms up, we see an increase in street crime, including homicides. So the decline seems to be associated with the first phase of the stay-in-place orders and business closures. In May we continued to see a decline, but it was only half as large. That, and other factors, led us to predict that we’re going to see homicides increase over the next few months.

Nobody can explain why crime goes up or down in a discrete period of time with 100 percent confidence. However, it is very hard to explain such a pronounced decline in homicides at a time you would expect them to go up. There really is no other plausible explanation for such a strong decline other than the pandemic. 

Is there any precedent for this kind of precipitous decline in homicides? 

Possibly, but on a month-to-month basis, not to my knowledge. We had a big decrease in homicides during the ​’90s. But that unfolded over a number of years, not from one month to the next. 

Why would a pandemic reduce homicides to such an extent?

It all boils down to the level of public activity. Homicides depend on relatively vibrant activity patterns — lots of people on the street. When the streets are empty, the opportunities for all kinds of street crime go down. The notable exception is domestic homicide, but overall we should expect homicide rates to decline as the streets are emptied and businesses close. High levels of unemployment also tend to coincide with lower levels of crime. We saw that during the Great Depression and the Great Recession. But as activity patterns go back to normal, we should expect these types of crime to begin going back up. That’s what we think we’re seeing in the smaller decline of homicides in May compared to April.

Crime is multifactorial — lots of things impact crime rates. So you wouldn’t expect crime in every city to respond to the pandemic in the same way. In fact, there was a significant number of cities where violence did not go down. Overall, homicides declined, but not everywhere. 

Thirty-nine of our 64 cities experienced homicide declines in April and May. Twenty-five did not. 

Why do homicides tend to go down during economic downturns? That seems counterintuitive.

Recessions mean people are out of work, and people who are out of work tend to stay home rather than be out and about. So activity is reduced. I’ve also found in my own work that when inflation rates decline, as they did during the Great Depression and Great Recession, we tend to see fewer violent crimes and property crimes. The reason appears to be that during periods of rampant inflation, like during the ​’70s, there are greater incentives for thieves to ply their trade.

Thomas, what does your research (as discussed in your book ​“Bleeding Out”) say about the best way to reduce homicides? 

There are three core anti-violence principles that I uncovered in my research. First, the most successful strategies are very focused on the highest-risk people and places. Second, the strategies are balanced between police- and non-police approaches. Third, they are fair — they are legitimacy-enhancing rather than legitimacy-reducing. There seems to be a relationship between the legitimacy of the state, specifically law enforcement, and the level of violence in the community. As legitimacy drops, the people in these communities simply don’t use the criminal justice system to mediate conflicts and instead turn to violence. We saw a significant increase in homicides after the unrest in Ferguson, Baltimore, Chicago, and other places. We may be in for the same thing in the wake of the George Floyd protests. 

Can you talk about the example of Oakland, which cut its homicides in half using the kind of balanced strategy you mentioned?

In Oakland they used a strategy called ​“focused deterrence.” It was a partnership between police, service providers, and community members. There had been a history of very difficult community-police relationships in Oakland, yet this strategy was successful. Basically, they identified the people at the highest risk of violence and then a team of people sat down with them and tried to intervene. They presented a message of empathy and accountability. And that has also been effective in other places.

In the report you mention how a loss in trust between people and police can lead to an increase in violent crime. Do you think that will happen now?

It is something I’m worried about. After Ferguson, there was talk among conservatives of a Ferguson Effect — a national slowdown in policing that was causing a crime spike. Researchers did not find a national slowdown, but there were isolated slowdowns that may have driven up homicide rates in Chicago and Baltimore. If a so-called Blue Flu happens, it can definitely affect homicide rates in a city. That doesn’t mean some cities aren’t overpoliced, but it’s a matter of the drawdown being arbitrary versus planned.

After Ferguson I studied arrest rates for low-level offenses, and what I found is that police had been drawing back for years before the unrest without homicides going up. What the data appear to show is that if police ease up on these lower-level offenses, it doesn’t necessarily portend increases in more serious crimes. 

Can you talk a little more about this linkage between crime rates and perceptions of police legitimacy?

As Tom said, there’s emerging research suggesting that if people think they’re treated fairly and justly by police, they’re less likely to take matters into their own hands and more likely to rely on the traditional functions of law enforcement. We don’t just see this in low-income communities of color. One of the driving forces in the arming of the American citizenry is the belief that when you need them the police won’t be there to protect you, so you have to protect yourself. When you think about it, that’s exactly the same logic that drives the concern and fear in communities of color — that, ultimately, you’re on your own. 

Based on your research, do you anticipate that the decline in homicides over the past two months will lead to a lower year-over-year homicide rate for 2020?

No, I don’t think it will. We saw a big uptick in homicides for the first three months of the year, a sizable decline in April, a less sizable decline in May. I don’t think we’ll see much of a decline in June, and I think homicide rates will go up for the rest of the year. So overall I think we’ll probably end 2020 with a higher homicide rate than 2019. I hope we’re wrong, but that’s how we’re reading the evidence.