More About Institutionalization vs. Homicide Rates

I mentioned yesterday a Texas Law Review paper by Bernard Harcourt about how adding mental hospital and prison incarceration rates together did a better job of predicting murder rates than other models--and that as the total of the two rates went up, there was a strong negative correlation to murder rates.

Today, I heard from Professor Harcourt, who pointed me to a paper in process here which uses state level data and a broader range of social variables--and finds that while there is variation from state to state, the same essential results pop out: as states increase the percentage of population who are locked up in either mental hospitals or prisons, the murder rate declines.

There are signs of elasticity on this. When the institutionalization rates drops to very low levels, murder rates don't continue upward at the same rate because nearly all the dangerous people are out on the street. When the institutionalization rates rise to very high levels, murder rates don't get any lower because those who are locked up increasingly include people who might be a bit eccentric, or mentally ill, but not dangerous to others.