Why The War On Drugs Matters In Mass Incarceration, Part 1: Who Goes To Prison.


There’s been a lot of academic and popular work trying to understand the causes of “mass incarceration” in the United States. For a while a common perception was something like “The War On Drugs Caused Mass Incarceration”. This perception has recently been accurately challenged by academics and writers to the point where a counter-narrative has developed, something like “The War On Drugs Had Nothing To Do With Mass Incarceration”. (I’ve even found myself associated with that position.)

To be clear, I don’t think that was the intention of those academics and writers, and I think it’s good that the simplistic narrative of “The War On Drugs Caused Mass Incarceration” has been challenged with quantitative evidence. If all you want to analyze is “the incarceration rate”— the number of people who are in prison at any given time — then drug offenders really don’t seem to be the most important group.

However, the pendulum may have swung too far in that direction, and the counter-narrative isn’t much better. To understand mass incarceration, you should want to analyze more than just “the incarceration rate”. One metric can never adequately describe the full human and social realities of something as big and complicated as the American criminal justice system. After all, people aren’t hallucinating that “The War On Drugs” happened, and if you look at other metrics as well, it seems to have had a clear and significant impact.

In this series of relatively short articles, I hope to look “The War On Drugs” using several additional perspectives like prison admissions, felony convictions, arrest rates, and racial bias. My goal is for these articles to be useful as examples of looking at a topic in criminal justice from different angles.

Mass Prison Terms, Not Just Mass Incarceration:

One statistic that seems to show clear effects of “The War On Drugs” is the number of prison sentences per year. Again, I think this describes real consequences that “the incarceration rate” overlooks. Imagine that one policy adds a fifth year to a prisoner’s four-year term, while another policy turns a probationary sentence into a one-year prison sentence. Those two policies would have increase “the incarceration rate” by the same amount, but they would mean very different things for the individual involved.

My guess is that issuing more prison sentences is more punitive and disruptive than adding the same amount of time onto existing prison sentences — and in fact a lot more prison sentences are being issued than there used to be.

The number of new prison sentences per capita, or the new court commitment rate, grew from 58 new prison terms per 100,000 people in 1977 to 165 in 2006 before declining over the past decade to 134 in 2015. Even the 1975–1977 rates were over the apparent baselines of the previous fifty years (see sources and notes for some caveats about that comparison).

To put those numbers into perspective: If the new court commitment rate had stayed at 1975 levels, then there would have been about 6.6 million prison sentences from 1975–2015. If the rate had stayed around 1960 levels (to whatever extent that’s even comparable) then there might only have been about 5.4 million.

But in reality, there were about 13.4 million prison sentences from 1975 to 2015. So the “mass incarceration” era has seen millions and millions of “extra” prison sentences, perhaps as many as seven or eight million in total, issued on top of what previous policies would seem to have led to, and entirely apart from longer prison sentences or changes in parole. The cumulative effects of those sentences go above and beyond “the incarceration rate” as such.

The “Extra” Prison Terms By Offense:

What explains these millions of “extra” prison sentences? I’ve never seen a satisfying answer, to be honest. From 1980 to 1990 the new court commitment rate at least doubled from in 24 states from New Hampshire to Iowa to Oklahoma to Arizona and increased by at least 50% in 18 more. That’s a change was so great, so fast, and so national that it seems to demand a single clear mechanism, but to my knowledge no one’s found one.

One approach to finding such a mechanism might be to try to identify and examine the seven million “extra” prison sentences specifically. Who were they issued to, and on what charges? Unfortunately I haven’t seen full, consistent historical offense data for both state and federal prison terms. But I have seen breakdowns of state prison terms by offense category from 1977 to 1995 and for 2001, 2006, and 2011. By interpolating the missing years this can roughly cover state prison terms from 1977 to 2015, which includes 12 million prison terms, nearly 90% of the 1975–2015 total.

The following graph/gif shows the offenses of the 12 million state prison terms from 1977 to 2015, both over time and cumulatively. The cumulative totals are presented along with a hypothetical where 1977 new court commitment rates by offense repeated every year, and the difference between them. It’s hard to deny the War on Drugs here:

Of the 12 million new state prison terms from 1977 to 2015, about 3.1 million — 25.8% — were for drug offenses. 3.1 million is a lot, but drug offenses were also a much smaller share of state prison commitments prior to the mid-to-late 1980s:

Under the 1977 baseline hypothetical, there would have been about 5.5 million new state prison terms from 1977 to 2015, with only about 600,000 for drug offenses (11.5%). So by this standard there were 6.6 million “extra” state prison terms from 1977 to 2015, and perhaps 37.8% — 2.5 million out of 6.6 million — were for drug offenses. (An additional 20.6%, or 1.4 million, were for “other offenses”, and even within the “violent” and “property” categories growth seems to have been concentrated in lower-level offenses.)


I don’t know how reasonable a baseline that is, and 38% isn’t even a majority of the “extra” prison terms, let alone all of them. Even considering that particular percentage, you might get different results depending on your choice of baseline year or what systems you include. Drug offenses seem to explain rather little of prison sentence increases prior to the mid-to-late-1980s, so choosing a baseline year earlier than 1977 would probably slightly lower the estimate. On the other hand, adding federal prison terms would probably slightly raise the estimate.

(I have a very rough guess, using even more interpolations than presented here, that using a 1970 baseline but including the federal system would mean that 34% of “extra” prison terms from 1970–2015 were for drug offenses.)

I don’t think we should get too hung up on the details of that percentage either way. Some of the “extra” prison terms on drug offenses might have been pretextual. That is, some were probably prison terms for people who probably would have gone to prison anyway on other charges, like going after Al Capone on taxes. And some of the “extra” prison terms violent, property, or public order/other offenses were surely related to “The War On Drugs” as well. Either way, drugs have to explain a significant fraction.

Lifetime Prison Status:

A distinct, but closely related, metric is the number of unique individuals who have ever been to prison. Again, this could describe real consequences that the raw number of prison sentences doesn’t. A first-time prison sentence is presumably more punitive and disruptive than a second-time prison sentence, since the mere fact of a prison record matters.

Unsurprisingly, it is estimated that a much higher percentage of Americans have a prison record today than before “mass incarceration”. Here I’ll follow a recent paper in Demography by Shannon, Uggen, Schnittker, Thompson, Wakefield, and Massoglia. According to them, the percentage of American adults who had been to prison slowly dropped from around 1.35% in 1948 to around 0.93% in 1979, which would be 2.2 to 3.2 million people at 2010 population levels. But the percentage then increased rapidly and by 2010, 3.11% of American adults had been to prison, or 7.3 million people. Crudely, that might suggest that about 4–5 million people have prison records today who wouldn’t have prison records under earlier policies.

Any analysis of how much of that can be attributed to “The War On Drugs” is very speculative — without more longitudinal individual-level data, such as the changing distribution of “first-time prison terms” over time, it’s very hard to know anything for sure. But if “The War On Drugs” accounts for around 38% of extra prison sentences, I think it’s possible that it accounts for around 38% of extra prison records. That might mean 1.5 to 2 million adults have prison records today because of “The War On Drugs”. (Perhaps a bit less — I think drug offenses were 33.5% of felony defendants in large urban counties in 2009 and 29.4% of first-time felony defendants, but I don’t have those numbers for a good baseline year.)


People have rightly pointed out that drug offenses account for a relatively small fraction of “the prison population” at any given time. While correct, that shouldn’t be used to dismiss “The War On Drugs”. The number of prison sentences issued on drug offenses exploded around 1985 and stayed elevated for at least a quarter-century, perhaps declining a bit in the last few years. That adds up to millions of “extra” drug terms and at least a large fraction of the growth in prison sentences and prison records over the “mass incarceration” period. (I think it’s at least plausible it’d be a majority of that growth if we could account for related charges, but even if it isn’t it’s nothing to dismiss.)

Notes And Sources:

SVG-to-GIF code was taken liberally and gratefully from Noah Veltman’s block on the subject: https://bl.ocks.org/veltman/1071413ad6b5b542a1a3

1929–2015 national population from the BEA API. Pre-1929 population from: https://www.census.gov/population/estimates/nation/popclockest.txt

1923–1982 prison sentence counts are from: https://www.bjs.gov/content/pub/pdf/hcsus5084.pdf, Table 3–8 on page 36, except 1974 which is from: https://www.ncjrs.gov/pdffiles1/Digitization/34707NCJRS.pdf, Table 5 on page 22. 1978–2015 new court commitment counts from: https://www.bjs.gov/index.cfm?ty=nps I distinguish between “persons received from court” for 1923–1970 and “new court commitments” for 1974–2015 because I believe there was a change in definitions around then, with the latter being offenders sentenced to at least one year, and the former with some states including that. See here (page 440):

Also some years are missing states and so on. I didn’t do any interpolation of missing states or anything. You’ll just have to blame the BJS! 1977 state new court commitments are from: https://www.bjs.gov/content/pub/pdf/p92.pdf, Appendix Table 2 on page 10.

1977–1990 offense percentages are from: https://www.bjs.gov/content/pub/pdf/p92.pdf, Appendix Table 1 on page 10. 1991–1995 offense percentages are from:
https://books.google.com/books?id=0CgnfzWO8HEC&lpg=PA14&ots=wSAiqOmik2&dq=%22new%20court%20commitments%20to%20federal%22&pg=PA16#v=onepage&q=%22new%20court%20commitments%20to%20federal%22&f=false, Table 1.20 on page 16. 1995 data was then-preliminary. 2001, 2006, 2011 offense percentages are from: https://www.bjs.gov/content/pub/pdf/p12tar9112.pdf, Table 5 on page 7.

Felony conviction numbers referenced at the end (more on that in a subsequent article) are from https://www.bjs.gov/content/pub/pdf/fdluc09.pdf, Table 10 on Page 12, and Table 14 on page 18.