When national trends clash with on-the-ground realities
Murder is down nationally, but what drives murder is local trends. Ten years ago, in 2014, the number of murders in America dropped below 14,300, the lowest level since 1968 (when America’s population was far smaller). In 2019, there were 16,425 people murdered. In 2020, this jumped to 21,570, an unprecedented 31% increase. Though official numbers aren’t released until late summer, I estimate that 2023 will show approximately 18,800 murders. If this estimate is accurate, it would represent an 11% annual decline in murder nationwide.
One take-home message may be that our criminal justice systems, imperfect as ever, are functioning better than they have since 2020, but still not as effectively as they once did.
National murder numbers are important for obvious reasons. But thinking of violent crime in terms of a national “trend” doesn’t help our understanding of what actually causes and prevents crime. Changes in crime are not preordained, nor are they inevitable or cyclical. What’s getting better can continue to improve. What’s bad can always get worse. Crime isn’t TikTok; crime doesn’t actually “trend.” At least not on a national level. To understand the meaning behind crime data, we need to focus on the local level: the city, the neighborhood, the block and even the individual. And here, the results are much less “trendy.”
From publicly available data, I’ve compiled 2023 murder numbers from cities, metro areas, large counties and two states (69 cities and 84 jurisdictions in all) that constitute half of all murders nationally and a quarter of America’s population. Fifty-six of these polities saw a 2023 decrease in murders while 25 saw an increase (and four were unchanged).
A trend, statistically, is simply an average changing in one direction over time. But conceptually, a trend implies something else, something greater, as if the changes are the result of an outside force, like a rising tide that lifts all boats.
That’s where the concept of national crime trends breaks down. There is no tide or magnetic field that pulls gun triggers in the same direction. Even macro-level factors, such as COVID-19 or post-Floyd urban unrest of 2020, affect localities in very different ways.
Contrast changes in crime with national economic trends. When interest rates go up, borrowing money costs more. Decreasing national home sales over time is indeed a trend. A tight labor market in one town affects hiring trends in a neighboring village. There’s common cause and some universal general effect. There may be exceptions, but they are outliers and not just the tails of a normal statistical distribution.
Imagine if inflation were like murder. One could never see prices increase 34% in New York while at the same time decreasing 30% in New Jersey. But between 2018 and 2023, that’s exactly how murder changed in New York City and Newark.
A trend, statistically, is simply an average changing in one direction over time. But conceptually, a trend implies something else, something greater, as if the changes are the result of an outside force, like a rising tide that lifts all boats.
Crime prevention is more like politics, in that all crime is local. The problem of thinking of crime is “trending” up or down is that it infers some superhuman agency which discounts the people and organizations who work hard to prevent and deter lawbreaking and violence.
This is not to say that national crime trends are impossible. Changes in, say, federal gun regulation could indeed spark a national trend. But policing, prosecution and incarceration (plus any alternatives to the traditional criminal justice system) are, by and large, left to states and localities, and they play out at neighborhood and individual levels. As a result, the $64,000 question isn’t “What’s the national trend?” but “What works here and why?”
At the local and state levels, there have been major changes in law, policy and prosecution in recent years, and policymakers, researchers and advocates have debated their effect ad nauseam. Perhaps the onus should be on skeptics to prove why these changes wouldn’t matter. A common fallacy in crime analysis is the idea that the absence of definitive proof is proof of absence. Of course it’s true that correlation doesn’t prove causation, but it often remains a most excellent clue.
For example, nobody can say with confidence precisely how much successful prosecution impacts crime. But the answer certainly isn’t “not at all.” At the very least, a prosecutor affects policing, including who is arrested, and incarcerating repeat violent offenders prevents violence.
The $64,000 question isn’t “What’s the national trend?” but “What works here and why?”
It is surely significant, then, that Baltimore saw a 21% decline in murder in 2023. Over the same period, there was a 10% increase in arrests, more gun convictions and one-third more murder convictions. A new prosecutor — competent, less critical of policing, presumably law-abiding — replaced a now twice-criminally-convicted former prosecutor who was, to be kind, less competent. Better prosecution and a decline in murder could just be coincidence, but why would they be?
Not far from Baltimore is the nation’s capital, Washington, D.C. Last year in Washington, murders increased 35% and carjackings doubled. (The increase in carjackings is indeed part of a local trend. Carjackings have increased more than sixfold, 530%, since 2019.) Police arrests in Washington are half pre-2020 levels, and, unlike what we see in many other cities, have not begun to return to pre-COVID-19 levels. The United States prosecutor for Washington chose not to prosecute 56% of arrests in 2023, an astoundingly large number — albeit an improvement from 67% in 2022. Before 2014, fewer than 30% of Washington arrests were not prosecuted, which is still higher than in most cities.
The prosecutor recently said, “If we want to be safer in the long run, we cannot prosecute and arrest our way out of it [the crime problem].” Really? No matter what he believes, should not the prosecutor (or the mayor, or the police chief) at least pretend their job matters? If not, quit and turn it over to somebody who believes in the crime-fighting potential of the criminal justice system. Washington’s 2023 rise in violence is certainly not entirely the fault of the federal prosecutor, but some of it is. The question, again, isn’t “if” but “how” and “how much.”
Changes in crime and violence have different causes in different cities. Often, one doesn’t need a particularly well-fitting thinking cap to figure out what works and what doesn’t. Competent leadership, a focus on proximate causes of violence, and the removal of a small number of repeat violent offenders matter more than some may want to acknowledge. Crime will continue to trend down if, at the local level, leaders do more of what works and less of what doesn’t.