Back in 2013, the Economist created a version of a chart originally presented in Bernard Harcourt’s 2011 article “An Institutionalization Effect: The Impact of Mental Hospitalization and Imprisonment on Homicide in the United States 1934- 2001.”
That Economist chart, which is now over a decade old and whose data series stops nearly a quarter of a century ago, recently went viral on Bluesky, where it was given a sort of progressive spin — the moral of the story there was that America’s prisons are full of mentally ill people in need of treatment, not punishment.
It was also covered by Conn Carroll in the Washington Examiner, who offered a more right-wing spin, arguing that a large share of America’s homeless population ought to be coercively institutionalized in mental hospitals or drug treatment facilities.
I think that people are quick to draw these inferences from the chart because both of the policy claims are pretty reasonable. Specifically, I believe that:
Providing mental health and substance abuse treatment services to convicts rather than having them cycle in and out of prison is a good idea.
While housing scarcity is the main statistical driver of homelessness (a large fraction is cyclical and involves employed people without serious mental illness sleeping in their cars), there is a non-trivial population of chronically homeless people who suffer from addiction and other illnesses who probably should be coerced into treatment.
Meanwhile, the chart is eerie, but I think it’s pretty misleading.
Despite the symmetry of the lines, it’s just not true that the currently incarcerated population consists primarily of people who would be institutionalized in mental hospitals absent the de-institutionalization trends of the 1960s.
It’s hard to know exactly what would have happened, counterfactually, had we not shuttered those institutions. But the demographics of the present-day prison population are very different — much younger and more male — than the demographics of the historical mental hospital population. What’s more, while it’s hard to blame Harcourt too much for ending his series in 2001 when he published back in 2011, there’s no reason for people like Carroll to be doing that in 2025. Notably, the incarceration rate peaked in 2008 and, as Keith Humphreys pointed out over the summer, is poised to plunge in future years.
Fifteen years of sharply declining incarceration have not generated a boom in institutionalization of the mentally ill (in part because the institutions themselves mostly don’t exist anymore) or a surge in crime.
Sources: Harcourt, “An Institutionalization Effect,” Bureau of Justice Statistics
Again, without disputing that there is reason to believe there are people both on the street and in prison who could use treatment, I just don’t think any of the predictions a reader of Harcourt’s original chart would have made have panned out.
In particular, the steadily declining aggregate institutionalization rate since 2008 has not generated a steady increase in homicide. The actual relationship between incarceration and crime is bidirectional: locking people up can reduce crime, but high rates of crime also lead to lots of people being locked up. And the real relationship to mental health issues is much more tenuous than the eerie symmetry of the original chart suggests.
But getting this all straight starts with trying to better understand what Harcourt was arguing in the first place.