« On Corporations and Pine Cones | Main | What Happened to Occupy? »

Thursday, April 25, 2013

Some More Evidence Against the War on Drugs Hypothesis

I started my previous post with a graph showing the steep rise in the total prison population. So I want to start this one with a graph showing the steep rise—and then the sudden plateau, and then the decline—in the share of prisoners in prison on drug charges.

Two things jump out on this graph. First, after a rapid rise the percent of drug offenders peaks at 21.8%, in 1990. Second, from 1990 forward the fraction steadily declines, with only a few upticks here and there, to 18.4% in 2008; by 2009, it was down to 17.8%. In other words, in 1990, nearly 80% of all prisoners were non-drug offenders, and by 2009 that percent had risen to more than 84%. And almost all of these other inmates are serving time for violent or property offenses.


Adding in the federal system, which is much more drug-focused—about half of all federal prisoners are serving time for drug crimes—does not change numbers or trends much: 24.1% in 1990, 22.1% in 2009. This is unsurprising: despite the extensive (in fact, quite excessive) attention it receives from legal academics, the federal system held only 13.5% of all prisoners in 2011, and until the 2000s it wasn’t even the largest prison system in the country, lagging behind California.

In short, as I’ve noted before, there simply aren’t “enough” drug offenders in prison for the drug-incarcerations-drive-prison-growth story to hold. Yes, in both relative and absolute numbers there are a lot more than there were before the incarceration boom, but not so many that they can be seen as the primary engine of prison growth.

But I can make the case for the relative unimportance of drug incarcerations even more precisely. The table below reports the number of violent, property, drug, and other offenders in 1980 and 2009 in state prisons.




% Change

% Contribution


























Between 1980 and 2009, state prisons added 1,120,200 prisoners, and of these 551,000—or 52%—were convicted of violent offenses. Increase drug incarcerations explain only 21% of the growth.

Of course, there are multiple ways to spin this table—counting is hard. One version is that 68% of the growth in prison came from incarcerating people for index offenses. Another version is that 48% of the growth came from locking up non-violent (i.e., property, drug, and other) offenders. Assuming, of course, that none of the property and drug offenders were actually arrested or even charged with violent crimes but struck plea bargains that dropped the violent offense. 

But either way, the table deals a solid blow to the drug-incarcerations-drive-prison-growth thesis.

This is a graphic example of why it is always essential to pay attention to baserates: the percent increase in drug commitments has been huge because the initial level was so low. Thus the overall contribution of increased drug commitments to the increase in total prison population (21%) is not nearly as large as the percent change in drug incarcerations (1,175%) alone might suggest. It seems counterintuitve that drug incarcerations could grow almost four times as fast as violent incarcerations yet be less than half as important, but such is the nature of baserates.

Drug commitments were the fastest growing segment, but not the most important segment. 

We see this elsewhere in the prison debates as well. Consider, for example, female prisoners. We used to hear that women had become the fastest growing population in prison. This is not surprising at all—in fact, the opposite effect would be quite unexpected. In 1990, there were 44,065 women prisoners in a system with 743,382 prisoners; in 2010, there were 105,200 women in a system with 1,544,200 prisoners. Thus the number of women prisoners grew by 139%, compared to 106% for men. Yet of the 800,818 prisoners added between those years, 739,683—92.4%!—were men. Rates of growth can be deceptive if the populations start from different bases.

Again, this is not to say that drug arrests and incarcerations have played no role in prison growth, or that rolling back the War on Drugs will have no effect on incarceration. But it does suggest that whatever effects there are will be indirect and complicated. A simple policy of drug decarceration will not yield the returns that many think.


Posted by John Pfaff on April 25, 2013 at 10:36 AM | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Some More Evidence Against the War on Drugs Hypothesis:


Feed You can follow this conversation by subscribing to the comment feed for this post.

This is really interesting stuff, John. Here's another question. How many people are incarcerated in state facilities that are not prisons? (I assume this includes all city-level and county-level facilities?) Do you have a sense of the shifts in numbers and offense types in this cohort?

Posted by: Dan Filler | Apr 25, 2013 11:56:04 AM

The label on the vertical axis should be "share" or the numbers should be multiplied by 100.

Posted by: anon | Apr 25, 2013 1:22:26 PM

I think the picture is more complicated than is presented here. a big impact the war on drugs has is not straight-up convictions for drug offenses, but for the huge number of parole violations that are related to drug use. For example, I currently have a client who was previously convicted of a non-drug offense. While on parole, he was caught with marijuana on him. His parole for his original offense was revoked and he's now serving four years ostensibly for the original offense. He has no drug conviction, but he's now serving four years because of drugs.

In my experience, a dirty drug test is one of the most common ways in which defendants violate their parole. Perhaps my experience is atypical -- or perhaps even when you take into account the parole violators who are in essence serving time for a drug offense, the numbers are not meaningfully different. But I'll be a bit skeptical of these numbers until I see that impact.

Posted by: LODK | Apr 25, 2013 2:53:16 PM

The parole violation question is a good one, and I'll have an entire post about the role of parole violations on prison growth later on. My general sense is that the effect of parole violations--for whatever reasons they happen--is generally overstated. Non-trivial, but not central, but also hard to fully disentangle.

Posted by: John Pfaff | Apr 25, 2013 3:00:51 PM

Fascinating. I'm sorry if I overlooked this here or in the previous post, but what is "other"? E.g., is securities fraud a "property crime" on this theory, or "other"?

I look forward to hearing more.

Posted by: Q | Apr 25, 2013 11:21:59 PM

The problem with your thesis is that it is the initial contact with the criminal justice system that often leads to a life of crime. Approximately 50% of all drug arrests are of people under 21. Most of these arrests are for simple possession of marijuana. Once found guilty a young person can legally be discriminated against in both housing and employment for the rest of their life. With no hope of ever getting a decent job, either a life of selling drugs or petty theft is the life most of these youths caught with marijuana find themselves trapped in. A single arrest for marijuana may over the course of that person’s life, lead to a dozen or more arrests that never should have happened. Certainly the using of jails as mental health facilities with little to no mental health care professionals has exacerbated the problem, but it is unlikely the cause. One only needs to look at the ethnic makeup of any person system to see that.

Jails and prisons are places where people learn the trade of taking from other people to survive. They are good for little else. Society puts these young people that are simply possessing or consuming a plant in these institutions and saddle them with criminal records so they are not allowed to be productive members of society and then wonders why we have a crime problem. The answer is we manufacture criminals where none previously existed.

Posted by: Tom Butler | Apr 27, 2013 11:54:28 AM

Post a comment