The last five years have been a remarkable time in America for crime and justice.

First, in 2020 and 2021, there was the largest one-year increase in homicides ever recorded, and then a spike in broader measures of violence. By 2022, the pendulum had begun to swing, and violence peaked. In 2023, crime declined, gaining momentum in 2024, and is now accelerating into 2025, with historic falls in crime and violence not just mirroring the 2020 drop but exceeding it. Violence is now approaching levels not seen since the 1960s, or perhaps ever. This year may have the lowest homicide rate in modern records.

Why has violence dropped this much, at this time?

Below, I will make the case that four factors explain both the crime spike and the crime drop. Here are the Cliffs Notes.

1. Local governments (townships, cities, and counties) experienced massive layoffs in the spring of 2020 as governments reeled from lost tax revenue. These jobs include many of the people who work most directly, hands-on, with young men who are at most risk of committing violence and being victimized. And crime spiked. When those jobs came back (there are more people in those jobs now than at any point in the 70 years crime and violence plummeted).

2. Drug markets demand spiked in the pandemic when other avenues for illicit drugs were closed (bars, restaurants, work, school), and trauma led to more customers. But after the pandemic, overdoses slowed, then declined, then declined rapidly. And violence declined.

3. Routine activities changed. During the pandemic, people stayed close to home, which protected many, but also put young people with historic disputes close to one another, in deeply traumatized places. And shootings spiked. When people returned, en masse, to business as usual routines and eyes on the street returned to normal, crime declined. But along the way, a lot of those most likely to shoot were cut down in the frenzy of COVID-era violence, and violence declined.

4. Policing was decimated during the pandemic. Already facing staffing shortages, law enforcement experienced disproportionate waves of illness, along with directives to limit contact with the public, and possibly, some effects of the George Floyd protests and the police legitimacy crisis. The result was that there were simply fewer police on the streets, and thus less deterrence, and more crime. As police deployments returned to more standard levels, crime declined.

These are the national forces that affected all places. The national factors varied across places, playing a larger role in some places more than others, where local concerns had a greater impact. However, while individual cities, towns, suburbs, exurbs, and unincorporated areas may have responded differently to the COVID-19 crime crisis, there is little doubt that the rising tide swamped all boats. There is equally little doubt that the ongoing crime drop is exerting a gravitational pull on crime rates today, across the country.

It is troubling that most reports about the crime spike and the crime drop focus on a single place, as if it were isolated from the larger crime context. It was not. I am not arguing that what happens in Philadelphia somehow affects crime in Paducah. But it is certainly the case that big, national forces can simultaneously affect both through the same mechanism.

This is a golden opportunity to reset crime and violence in America at a new, lower level. If we are to fully take advantage of the ongoing decline in crime and violence, then we should begin by asking what explains the entirety of the crime drop, rather than immediately investigating its parts.

The COVID-19 tsunami washed through all of America. As it has receded, it pulled back with it at least some of the problems that predated it and uncovered some unexpected opportunities. Those are the places to look for the next solutions.

Share

According to new data from Jeff Asher and AH Datalytics, and from the Major Cities Chiefs Association, crime has fallen substantially in 2025, building on the historic decline in 2024. I have been arguing for some time now that the crime decline is not just real and substantial, but accelerating. The velocity, which has assumed an almost ballistic trajectory, separates this decline from previous declines. And, the decline is expansive: crime is down in virtually every category—personal and property, stranger and domestic.

But nowhere is the picture clearer than it is with the homicide rate.

Addressing the Regression to the Mean Argument

Before I present the supporting arguments for the four crime explanations, I would like to offer a rebuttal to the argument that crime is simply returning to normal.

In the figure below, I chart a decade of crime trends (2010-2019) and project those trends through the 2020s. The trend is up throughout the decade, and a reasonable expectation for the 2020s would have been for a slight increase in homicide over time. But even if we assume that crime in the 2020s would remain at the previous decade’s average, the conclusion is clear: crime in 2025 is well below expectations. We have burst through our expectations. Something much more substantial than regression to the mean is at work.

Leave a comment

Crime Prevention Through Local Government

Local (county/city) governments employ five times the number of employees as the federal government and provide or facilitate most prevention services to the highest risk population (young men). The last point is critical. One challenge to a lot of explanations for crime spikes and declines is that they don’t focus on whether those mechanisms directly impact the people most at risk of violence and victimization. For example, the police legitimacy crisis is often linked to the 2020 violence spike, but perceptions of police legitimacy seem to have changed more among people who are not particularly at risk of violence and victimization, and much less among people who are at high risk.

Local government, by contrast, engages directly with young men through various initiatives. Local government employs police and other first responders. But also, K-12 teachers and community college professors. And coaches, counselors, and social workers. And clinicians who deliver mental, behavioral, and physical health services. And employees who manage the grants that fund nonprofit, community-based organizations that directly deliver services to high-risk populations.

In 2020, local government employment plummeted. Between March and May 2020, more than 1.25 million local government jobs were lost, according to data from the Bureau of Labor Statistics. These do not appear to be furloughs, but rather staff separations. Suddenly, many of the most critical people in the lives of high-risk young people had disappeared. This is a strong explanation for the spike in crime.

This is also a strong explanation for the crime decline. Beginning in 2021, there was a slow return to pre-pandemic levels in local government employment. By 2023, when the crime decline was accelerating, these jobs had nearly reached pre-pandemic levels and were increasing steadily and predictably. By 2024, local government employment had surpassed pre-pandemic levels.

Disrupting Drug Markets

Research is clear that any change in drug markets, particularly disruptions in supply or demand, causes violence. There are no property rights in drug markets; one of the only recourse available in a drug market dispute is violence. There were clearly disruptions in supply and demand during the pandemic. During the pandemic, alternative illicit markets were closed (work, bar, school, etc., and yes, lots of people buy lots of drugs in those places). This likely substantially increased demand within open-air drug markets. Plus, the trauma of the pandemic drove other users to these markets. There also appears to have been some substitution of a cheaper drug (Tranq), which could also cause market disruption.

Nicole Johnson and Caterina Roman have an important take in PLOS ONE, where they show the increase in shootings in Philadelphia and how the location of shootings aligns with the known location of open-air drug markets. In the map, redder areas indicate larger increases in shootings during COVID, and the stars mark the open-air markets. Virtually every star is in a dark red spot, and dark red areas are disproportionately close to the stars.

This also fits the trajectory of the crime decline. As routine activities resumed a more normal pattern, the pressure on drug markets declined, and shootings slowed. In my discussions with prosecutors and law enforcement, they point out another factor, one that is hard to show in the data. That explanation is that while there are a lot of guys in drug networks who are willing to shoot back, there are relatively few who are willing to shoot first. They argue that the toll of 2020 and 2021 was so deep, and the actors involved so concentrated in place, that large numbers of the people who were willing to shoot first were killed, injured, or captured. And that also explains the decline in shooting.

And finally, drug overdoses seem to have peaked and begun their own decline, matching the general pattern of violence. While not all overdoses are from street drugs obtained in open drug markets, the general pattern suggests a decline in demand, which has been shown to reduce violence.

Routine Activities

Routine activities theory states that crime occurs when three elements converge: a motivated offender, a suitable target, and the absence of guardianship. When there is less convergence, there is less crime.

This theory was spectacularly successful in explaining crime in 2020. During the height of the pandemic, very few people were vulnerable because routine activities shifted from public to private settings. People stayed closer to home, including both potential victims and potential offenders. Additionally, as the next section details, there was a decrease in guardianship. During the pandemic, almost every category of public crime declined, while private crime (such as domestic violence) increased.

The exception was homicide. But this too fits the theory. Violence is concentrated in a few places, and those places have sustained violence—violence is concentrated in the same places year after year.

During COVID, when homicide spiked and other violence declined, emerging research suggests that homicide increased within places that had already high concentrations of crime. Homicide did not spread to new places, and there was simply more of it in places where it was already unusually common. This also fits routine activities. Young people in these neighborhoods were cut off from prosocial resources—they did not go to school, or work, or church, or see their mentors and advisors. They were stuck in places with historical trauma and in close proximity to people with whom they had long-standing disputes. And they, like all of us, experienced the added trauma of living in a pandemic. It should be no surprise that shootings increased. And shootings beget shootings, and the vicious cycle spiraled.

The return to normal was slow. While mandated closures ended relatively quickly, remote and virtual work and school were far more prevalent in late 2020 and 2021 than in 2019, and the number of vulnerable people in public spaces increased slowly. This caused an increase in other types of violence, which appeared to have increased in 2021 (exact figures for 2021 are not available, as the FBI made a significant change in data collection that year, resulting in limited data).

The routine activities story in the COVID-19 recovery also aligns with the timing of the crime decline. While vulnerabilities increased as people resumed normal activities, guardianship also slowly increased, as more police returned to routine policing after disruptions due to reasons of officer health and safety. And normal routines also included unusually high rates of employment. Unemployment skyrocketed in 2020, but it rapidly declined. For many groups, including teenagers, unemployment was lower in 2023 than it had been since the 1960s.

Policing (Deterrence)

As a society, the first place we look for an explanation for a change in crime tends to be the police. The thing about policing, though, is that it tends to change very little over time. D'Alessio, Flexon and Stolzenberg have an interesting recent article that includes a table of the total number of sworn law enforcement officers per capita. In 2013, there were 2.4 police per 1,000. In 2016, there were 2.55 per 1,000. And in 2020, there were 2.44 per 1,000. These are very small changes. Police staff levels are quite stable.

What is not stable is the number of officers actually deployed each day who are available to respond to calls for service. During COVID, many police departments sent instructions to officers to curtail their interactions with the public to reduce their likelihood of exposure to the virus. This included reducing in-person contacts and traffic stops, changing patrol practices, and even limiting arrests. It is not difficult to find memos from police leadership to officers with these instructions. Here is some guidance from the International Association of Chiefs of Police. And concerns when they were not instructed to change their activities. And despite these precautions, many police officers became ill, and there was a 250% increase in mortality among male officers.

The net effect is that routine policing was substantially curtailed. Justin Nix, Jessica Huff, Scott E. Wolfe, David C. Pyrooz, and Scott M. Mourtgos tested the effect of the COVID-19 pandemic and the George Floyd protests on changes in violence in Denver. They show that in the immediate aftermath of the emergency declarations in March 2020, pedestrian and vehicle stops and drug and disorder arrests declined precipitously. And the effect on crime? One set of models finds, “large-scale reductions in stops and drug-related arrests were associated with significant increases in violent and property crimes.”

That there is an effect of reductions in policing on crime during the COVID-19 emergency is increasingly difficult to dispute. However, the mechanism by which reductions in policing lead to an increase in crime is subject to much more debate. My argument is that the key idea is the simplest: police deter crime, and with fewer police around, fewer crimes are deterred.

I may have a little more to say in the coming weeks. It is tempting to go through competing explanations for the crime spike and the decline. there are certainly other theories that explain either the spike or the decline, but these are the only four that seem to have symmetrical explanatory power: they explain both.

Note

Mother’s Day is behind us, now it is on to Father’s Day (let’s go!!). Fortunately, Dad is easy to shop for. Max Read maintains a list of Dad thrillers that you can rent for Dad, probably for nothing. And he’ll be thrilled. But wait, there’s more. If you scroll down a little further, down below the Dad Thriller list, Max has a list of what he calls Halogencore movies, which he describes as “stories of beaten-down people acquiescing to or negotiating compromise with power.” Sounds awful, right? No! Moneyball, which is the greatest movie ever made, falls into this category. Margin Call, which is the greatest movie you’ve never heard of, does too. Anyway, it’s a cheap date, check it out.

Musical Interlude

When I was a kid, we’d rent the second floor of a ramshackle house at the beach every summer, and the family would come together from all over. The world was more boring then I suppose, and the beach wasn’t just another place to be while you looked at your phone. There was magic in that week, every time, especially by the waves. At the end of the week, I’d make my way to the surf and sit in the dunes and wonder if later that day and the next, and every day after that, the surf would crash just as it did while I watched, and this would go on and on, even when I wasn’t there to see it. I was struck that my observing of the waves was not instrumental in their being, when it had seemed so fundamental over the last few days.

We lost Mike Peters last week, founder and leader of The Alarm. I don’t think I knew Mike’s name until he passed. I know very little about the musicians who make my music, I prefer not to. If I do, then the music is theirs and I am just borrowing it. This way, it’s just mine. A couple of weeks ago we spent some time with dear friends who we do not see enough, and Mike’s death brings me another wave of sentiment. Here, too, it is not a splashing the keyboard sentiment, but happy memories.

The Alarm existed in the hard-ish, punk-ish rock world between the polish of U2 and the screeching self-reverence of the Sex Pistols and I loved them. In a high school segregated into Led Heads, Dead Heads and Talking Heads, they were in my corner. This was my first concert and it was music that made you feel like you could do anything. But my listening was not instrumental in their being either, and when I heard Mike was gone and looked in my playlist, The Alarm wasn’t even there.

I have learned a lot about Mike Peters in the last couple of days, and there’s a well-produced bio. He was a super-talented guy, with a good heart and all the loops and bends that life brings. One happy surprise was that Mike was a contemporaneous fan of Big Country, as was I. I hazily recall thinking The Alarm was a New York band, but they were self-evidently Welsh, and Big Country was a Scottish punk band, so it all fits. A nice confluence, when two people you like enormously turn out to be connected themselves. For a while, Mike’s after-work job was fronting Big Country after their iconic lead singer, Stuart Adamson, passed.

So, in that spirit, I’ll close with this tribute, which feels on the nose.

I’m not expecting to grow flowers in the desert,

But I can live and breathe and see the sun in wintertime