This paper “What the Numbers Say about How to Reduce Imprisonment: Offenses, Returns, and Turnover” is somewhat out of date as it is based on 2016 data and was published in2020, but a conversation brought it to mind. The paper asks: what would it take to get US imprisonment back to 1970s levels, before the mass incarceration era? Through most of the 20th Century, imprisonment rates were fairly constant. Getting back to the 1970s level would require reducing the 2016 imprisonment rate by 75%. The main punchline is that reducing mass incarceration is not simple, but involves different processes and that the patterns vary greatly between states, so what worked in one state is not necessarily relevant for another state.
States vary greatly in their imprisonment rates and in whether they increased or decreased imprisonment during the 2000s. They vary greatly in the mix of offenses of the people they imprison, in the average sentence lengths for particular offenses, and in the rate at which they return people to prison for violations of the conditions of parole. While a majority (54%) of those in prison in 2016 were violent offenders, only 29% of those who spent time in prison between 2000 and 2016 were violent offenders. Non-violent offenders go to prison for shorter sentences and are the large majority of people who have experienced some time in prison. Most of the decline in imprisonment happened for nonviolent offenses. Most people who were released from prison for either nonviolent or violent offenses stayed out of prison. Returns to prison were higher for those released to supervision than those released unconditionally, with most returns to prison for the supervised being due to revocations with no new criminal sentence; when there was a new crime for either the supervised or the unsupervised, it was more often nonviolent than violent, even for those whose first offense was violent. Released nonviolent offenders were re-imprisoned for new crimes (overwhelmingly nonviolent) at higher rates than released violent offenders. Most returns to prison happen within the first few years. Long periods of supervision are not appropriate.
I looked at how various kinds of potential reforms would affect imprisonment rates. Because more than half of all prisoners were scheduled to be released within five years, the biggest way to reduce imprisonment would be to send fewer people into prison by not sentencing people to prison at all for lesser offenses. This includes drug sentences, but also various nonviolent offenses and the lower levels of violent offenses, including some assaults, some robberies, and some sexual assaults. Other potential reductions can come from releasing people who have already served a long time in prison and reducing re-incarceration of people after imprisonment for technical violations or lower-level offenses that would not normally receive a prison sentence.
Reformers across the political spectrum are calling for a rollback of mass incarceration. The U.S. rate of incarceration in state prisons would have to decline by 75% to return to its 1970s level. How might this be accomplished? This Article provides descriptive statistics about the mix of offenses, sentence lengths, and admission types and shows that no single approach can undo mass incarceration. Those classified as violent offenders are a majority of those in prison, but nonviolent offenders are a majority of those entering, leaving, or having been in prison. A majority of those in prison are scheduled to be released within five years, meaning that steep reductions in prison admissions can have a large impact on imprisonment rates. Revisiting the sentences and parole options for those who have already been in prison ten years or more could have some impact. An examination of the rate of returns to prison after a first release from prison suggests that the rate of committing a new crime is low and that reductions in revocations for violations of the conditions of supervision are an important avenue for reducing incarceration. The U.S. states vary greatly in their mixes of prisoners by offense, sentence length, and returns to prison for parole violations with no new crime as well as in their histories of trends over time. States will vary markedly in which reforms will affect their prison populations, and assumptions based on old data may not hold true as conditions change.