30 months of great news on falling crime

Things are getting way better, but some big unanswered questions remain

Homicides and non-fatal shootings soared in 2020. The level was still below the worst period of American crime, but the uptick represented the fastest increase in violence on record. The trend continued in 2021, albeit at a reduced pace, and I was worried during the winter of 2021-2022 that it might take a really long time to get things back under control. But as the summer of 2025 gets underway, the news seems good: Violence fell a little in 2022, and then a lot in 2023 and 2024. And while trends vary somewhat from place to place, the overall trend this year continues to be positive.

Jeff Asher wrote back in May that 2025 was on pace to be perhaps the lowest murder rate year on record. But even if we fall short of that milestone, we’ll be within spitting distance.

What’s more, when it comes to violent crime, progress begets progress. Every time there’s a murder, uniformed officers need to secure the crime scene, which means they’re not available for routine patrol work. Detectives can’t work on non-fatal shootings or burglaries if they’re busy chasing murderers. This is why I worried the crime spike would be hard to reverse in the first place. But since it has reversed, it seems like it should be possible to keep the momentum going without dramatic new policy changes.

Of course, it would be nice to know why crime has fallen so much over the past 30 months.

Unfortunately, America’s practices around collecting crime statistics are really bad. The most recent official national crime stats came out last September and reflected crime data from 2023, and while the set of participating law enforcement agencies contains almost 95 percent of America’s population, that’s not 100 percent. The best source for more up-to-date data is the Real Time Crime Index compiled by independent researchers supported by Arnold Ventures. This is an invaluable project that involves artisanal collection of data from law enforcement agencies, focusing on the largest jurisdictions. We know from experience that this is a large enough sample — 377 agencies covering between a third and a quarter of the population — that the national trend is almost certainly going down.

But one shortcoming of this method is that it doesn’t let us test causal theories. In Georgia, for example, the only reporting agencies are Atlanta (which contains less than 10 percent of the Atlanta metro area’s population), Athens, and Dunwoody. This means we can’t test how rapidly crime is falling in Atlanta relative to its suburbs, or whether crime is falling faster or slower in Greater Atlanta than in rural parts of the state. A sample that’s dominated by big cities plus a random smattering of large suburban counties is fine for assessing broad national trends, but getting a clearer portrait of what’s happening elsewhere would allow insight into the details. Without the details, it’s hard to know why the trend is what it is.

So in terms of what exactly is driving the drop in crime, we need to remain somewhat agnostic. But we can still draw some conclusions, including the disappointing conclusion that we still haven’t solved the problems that led to the crime spike five years ago.