America took a bite out of crime
And rejected the explanation for it..
This month, I’m giving two talks on the ongoing decline in crime. The first is Wednesday, April 15th at 2 PM EST, hosted by the National Prevention Science Coalition, on “The Promise of Prevention to Address Addiction.” You can register here. The other is not a public event, so I’ll have to see what the rules of engagement are, but I hope to share some thoughts afterward because it’s a great panel.
In other news, I was delighted to be elected to the Council on Criminal Justice (CCJ) this month. CCJ plays a crucial role in the field of crime and justice, as the nonpartisan center of gravity for policymakers and the public. CCJ has just released a final report on the Implications of AI for Criminal Justice, the Women’s Justice Commission is going to release reports soon, and the Crime Trends workgroup, which I chaired, after we lost Rick Rosenfeld, had important insights. Check them out.
On to the show. This Substack has grown quite rapidly, and I think we’ve been in the Rising Top 100 for News off and on for the last six months. To welcome those new readers, I have compiled all the crime-decline essays of the last three years into a single post that organizes the argument. As we await the final 2025 numbers from the FBI, which may be out later this month, and some indication of the 2026 trend, it’s timely to revisit where we are and where we have come from.
Because where we are is astonishing. If the prognostications hold when all the data are in, it is likely that the US had the largest one-year crime decline in seven of the nine categories we use to assess crime. And it just missed being the largest decline in the other two. It was a year unlike any in modern record-keeping, and one we are unlikely to see again. And we probably do know why it happened, we just don’t much like the answer.
Understanding the Rapid Decline in Violence in the US (2023-2026)
Gun homicide soared during the COVID-19 pandemic, quickly peaked, and rapidly reversed. But rather than returning to pre-pandemic levels, homicide rates have fallen well below their pre-pandemic levels to historically low rates. Other types of violent crime, as well as property crime, have also fallen since the pandemic, and that decline is accelerating. When crime in 2025 is fully counted, we are likely to find that most types of crime are near or below their lowest levels since 1960.
The thesis here is that the unifying factor that explains all of these declines is the unprecedented COVID-era investment in local government.
Through the American Recovery Plan (ARPA), local governments have invested directly in fighting crime with a mix of evidence-based investments. And they have hired new police and invested in new technologies. But perhaps more importantly, they have invested heavily in a workforce of teachers, counselors, clinicians, and grants managers who work directly with the young people most at-risk of violence and victimization. The scale of the investment, over $300 billion, is unprecedented, as have been the returns on that investment.
Crime is way down because we have made substantial investments not just in intervening with people already in or on the cusp of our justice systems, but also through universal prevention, which has removed obstacles and promoted opportunities for millions of young people.
The crime spike and the crime decline
Looking at the very long term, crime in the US is currently near record lows. But the path here first climbed alpine peaks. The graph below shows that property crimes peaked in the late 1970s and in 1990 began a rapid and steady descent. Overall violence peaks a little later, in the early 1990s. Those crime peaks are almost unimaginable now. In a single year in New York City, there were over 2,200 homicides and more than 100,000 cars stolen. As I write this in mid-April 2026, about one-third of the way through the year, there have been 65 homicides and 3,000 stolen cars in New York City.
Homicide in the 2020s
Long-term trends are helpful for understanding broad eras, but they hide a lot of critical variation. The first half of the 2020s has been a time of tremendous volatility, at least by historical standards. Nowhere is that clearer than in homicides. Homicides were generally rising in the US between 2015 and 2019, but spiked with the onset of the pandemic. In a typical year, homicide changes +/- 4.5%, but between 2019 and 2020, homicide increased by 31%. This was by far the largest one-year increase in homicide since modern recordkeeping began in 1960—the largest prior one-year increase was 11%.
After 2022, homicides began to fall rapidly, by 11% in 2023 and 15% in 2024. All the early indications, from the Real-Time Crime Index, the Major Cities Chiefs Association, and the Council on Criminal Justice, show even larger 2025 declines of 18-20%. These effects compound, so an 18% decline in 2025 is a drop that starts from well below the COVID-19 peaks. This would put the homicide rate around 4.0 per 100,000, which is lower than at any point in the last 65 years.
The Violence Decline
Over the last three years, the decline in homicide broadened into a general decline in violence. Between 2022 and 2024, rape fell by about 14%, robbery by about 10%, and aggravated assault by about 6%. In other words, the retreat from the COVID-era surge has been broad, not isolated. Homicide has fallen the fastest and drawn the most attention, but the larger story is that violence as a whole has been trending down sharply. And as with homicide, the preliminary 2025 numbers show even larger declines, with rates of robbery falling another 18%, rape down almost 9%, and aggravated assault more than seven percent.
Since 1960, crime has declined in all nine categories only 11 times, the last time in 2014. The decline from 2023 to 2024 included the largest single-year decline (to that point) in both homicide and property crime.
But the 2025 decline is unique. As noted earlier, we are still waiting for the official 2025 national crime statistics from the FBI. The data above showing the 2025 expected declines are from Jeff Asher’s Real Time Crime Index, which has been extremely accurate in portraying national crime trends. Those data show the largest one-year decline in seven of the nine categories. And where it was not the single largest one-year decline, it was the second-largest (larceny) and third-largest (rape).
At this point, we have firmly established that the decline in crime is broad-based. While it began as a decline in gun violence and gun homicide, by 2025, the crime drop was steep and all-encompassing.
Variation across cities
There are almost 19,000 law enforcement agencies in the United States. About half have 10 or fewer officers, and they include every type of policing from big cities to college campuses to tribal police. Not surprisingly, there is often tremendous variation in their experiences with crime. In 2004, Steve Levitt wrote a seminal article about the crime decline in the 1990s and showed that crime had declined in almost every major city. From this, he established that the decline was a national phenomenon.
Summarizing the Crime Decline and Explaining its Cause
What we have established here is that the crime decline occurred everywhere, across all types of crime, and all at once. This suggests that the crime-decline generating process can only be national in scope. A lot has been written explaining the crime decline. New technology employed by local law enforcement. A new crime-fighting strategy. New evidence-based programs, perhaps a community-police collaborative. I am not diminishing any of these explanations, only noting that they all seem united by a single, larger explanation.
And that explanation is the American Recovery Plan Act, usually referred to by its acronym ARPA. ARPA was approved by Congress in 2021, and funds started flowing shortly thereafter. The gears of government grind slowly, so it is reasonable to think it took a year or two for the money to find its way from the federal coffers to its recipients locally, whether that was in the form of a paycheck to a newly hired staffer or funds distributed to a local non-profit. And some of that money is still flowing, though there is a cliff around the bend as it must be spent by the end of 2026.
And what could funds be used for? From the Treasury Department:
The Coronavirus State and Local Fiscal Recovery Funds (SLFRF) program authorized by the American Rescue Plan Act, delivers$350 billion to state, territorial, local, and Tribal governments across the country to support their response to and recovery from the COVID-19 public health emergency.
Through SLFRF, over 30,000 recipient governments across the country are investing these funds to address the unique needs of their local communities and create a stronger national economy by using these essential funds to:
Fight the pandemic and support families and businesses struggling with its public health and economic impacts
Maintain vital public services, even amid declines in revenue resulting from the crisis
Build a strong, resilient, and equitable recovery by making investments that support long-term growth and opportunity
There’s a lot of flexibility in those directions, especially the last bullet. One thing ARPA funds were definitely used for was to hire new staff after so many had been let go during the pandemic. As I’ve shown several times before, between March and May 2020, more than 1.25 million local government employees lost their jobs. To provide a sense of scale, this is approximately equal to the total number of active-duty military personnel across all six branches. And ARPA funds explain why the number of local government employees returned to pre-pandemic levels by 2024. And why the number of local government employees today far exceeds pre-pandemic levels.
It is hard to overstate how massive this investment was. For context, consider that the $350 billion investment in state and local government is about 67% of the $550 billion state and local governments spent on police and courts during this same period. One way to think about this is that ARPA increased state capacity around public safety by up to two-thirds. That is a lot.
Government Employment and Homicide
What I haven’t shown before is how that trend compares to the crime decline. Here, I’ve put both series on the same scale, the homicide rate and the number of local government employees. The idea here is simply to show how the two trends compare over a relatively long period, about six years.
The blue line shows the number of local government employees, and the red line shows the homicide rate. They are mirror images of each other. I am not saying that the sudden loss of a million-plus local government employees caused the homicide spike, but it sure didn’t help. And I’m not saying the return to full local government employment caused homicide to decline, but it definitely helped.
The dollars replaced state and local revenue lost to the pandemic and were used to maintain vital public services. That sounds ponderous, as public finance discourse often does. But what those dollars purchased in large part were more teachers, counselors and clinicians at a moment when they were scarce. These are the people who interact directly with young people at the greatest risk of violence and victimization to develop opportunities for them and to reduce the risks they face. The term for this work is “prevention”.
Why we resist this answer
It is the complexity of the crime-generating process that is our largest stumbling block to living in a world with less crime. America clings stubbornly to its thematic roots of rugged individualism. We look to a new sheriff to clean up corrupt towns. We want justice to be administered individually, not factory-farmed in huge, anonymous justice systems. We agonize over missing people and unsolved crimes. After work, we relax with a true crime story and its montage of senseless violence. We wait patiently for a steady hand to sort right from wrong. In this telling, America is a loose coalition of people, gently networked in small communities that should rightfully be isolated from society’s dangers. These communities exist everywhere, in rural crossroads, in gated suburbs, in our bustling cities.
We gaze dreamily into the heavens and say over a campfire that our neighborhoods could be protected if only someone who could separate good from evil came to town with the gumption to take these problems head-on. It’s a romantic vision but detached from reality.
Because the truth is that national forces move everything, everywhere. Do you live in a rural town or a big city? What kind of work do you do? Are guns plentiful or scarce? Is your neighborhood homogeneous or diverse? Do you have a lot of formal education or not so much? Are you safe as you go about your day, or are you in danger? Are you sick or healthy? Are your kids and parents nearby or far away? Do you see your friends all the time or only once in a while? The answers to all of these questions are mainly determined by what is happening collectively across the country for all of us. Cars replace horses, cities replace farms, suburbs pull people from cities, exurbs pull people from suburbs, the city recovers, and people return. The gravity of a nation imposes its will on all of us.
This is not true all of the time with crime. In fact, it is the exception, not the rule. Most years, the federal government does very little, and crime goes up or down in your town because of something the sheriff does or doesn’t do, or something that comes to town, or something that leaves.
But once in a while, a crack epidemic grips the nation. Or mass incarceration seizes the narrative and is imposed everywhere. Or the federal government, in trying to solve one gigantic problem, accidentally solves another. That appears to be what happened here.
