« Drugs and Prisons: A Recap, and a Complication | Main | On "Moving On" »

Wednesday, May 29, 2013

The Surprisingly Minor Impact of Drug Incarcerations on Racial Imbalances in Incarceration

I want to turn my attention here from the impact of drug incarcerations on prison’s size to their impact on the distributions of prison populations, specifically their racial distributions. Given that blacks are over-represented in prison relative to their share of the general population, and given that blacks are over-represented among drug incarcerations relative to their share of the general population (more on this in a moment), it is easy to see how many can argue that drug incarcerations explain the racial disparity in prison populations. But is this true?

The easier answer is “clearly no.” The harder answer is “still no, but with a bit more nuance.”

First, the easy answer. In 2009, non-Hispanic blacks1 made up just under 43% of all prison inmates and just over 50% of all drug inmates. Yet had we released all drug offenders of all races and ethnicities from prison in 2009, blacks would have comprised a hair under 41% of all prisoners. So the disproportionate number of drug incarcerations explains about 2 percentage points of the racial gap in prison populations.

This really shouldn’t be all that surprising. In 2009, only about 18% of all prisoners were in prison on drug charges. That fact, combined with the fact that the racial gap for drug offenses was only slightly larger than that in general, implies that racial disparities in drug sentencing simply can’t explain much of the disparity in overall prison populations.

But there is a trickier question to consider: is the disparity really a disparity? After all, the proper comparison isn’t the percent of blacks in prison to the percent of blacks in the population as a whole: it’s the percent of blacks in prison to the percent of blacks who commit offenses. And blacks commit crimes at noticeably higher rates than whites.

Consider murder, for which we have perhaps the most reliable offender data. Blacks committed 52.5% of all murders between 1980 and 2008; in 2009, blacks comprised 46% of all inmates locked up for murder and manslaughter. Compared to their 13.1% share of the US population, a 46% share for murder seems excessive, but it seems about right when compared to their 52.5% share for murders committed.

Some academics have argued that racial disparities will be lower for more serious offenses, since arrests are less discretionary for these crimes and scrutiny is higher; as we work our way down to less serious offenses, the argument goes, discretion rises and so too do disparities.

Does this explain the observed disparity—when compared to the population base-rate—for drug incarcerations? Unfortunately, the data are woefully insufficient.

There simply isn’t data on drug offending. For obvious reasons, we can’t use official drug arrests as a proxy for underlying drug offending. If we are concerned that blacks are disproportionately arrested for drug offenses, drug arrest data is clearly incapable of identifying that effect.

Some scholars, like Michelle Alexander and Michael Tonry, turn their attention to surveys of households and students on drug use, and argue that similar self-reports on drug use across races suggest that the imbalance in drug arrests and incarcerations reflect racial biases in criminal justice system.

I am quite confident that racial biases permeate the criminal justice system—although at least when it comes to sentencing the magnitude they can be hard to estimate—but there are at least three reasons to be skeptical of household and school surveys:

  1. Most obviously, use isn’t sale, and almost everyone in prison is in prison for distribution. Why should we assume that racial distributions are the same for consumption and distribution?
  2. There is evidence of racial differences in self-reporting: when it comes to arrests, for example, blacks tend to under-report more than whites. If similar patterns hold for drug use, then similar self-reports actually reflect higher levels of use among blacks. (Though again, we need to know the correlation between use and sale for this to be relevant.)
  3. There are selection effect problems with household and school surveys. We should be particularly concerned with school surveys, since those who sell drugs (or use drug heavily) are least likely to be in school when the survey is administered. Moreover, blacks graduate from high school at substantially lower rates than whites: the black male graduation rate, for example, is nearly ten percentage points lower than the white male rate. If we assume that drug use declines with academic achievement—which seems to be the case—then if drug use is equal across race/achievement groups (i.e., black A-students use at the same rate as white A-students) we should expect the average use level for blacks in high-school to be lower than that for whites, if the disproportionately-higher number of dropouts for black students are concentrated among the lowest levels of achievers. Similar rates thus again suggest higher use for blacks. (But once more, this only matters if use is at all a useful proxy for sale.)

That said, there is some evidence that police focus disproportionately on crack sales in open-air drug markets—at least in Seattle, where one leading study was conducted, crack appears to be the drug in question during 33% of all exchanges but 75% of all police actions—and that this disproportionate focus2 explains some of the racial disparity in arrests. Which in turn likely contribute to the racial imbalance, such as it is, in drug incarcerations.

And this imbalance in drug arrests could exert an important but indirect influence on the overall racial distribution of the prison population. These arrests, even if they do not result in incarceration, could destabilize future lifetime outcomes (employment, marriage, etc.) that are thought to contribute to the desistance from crime. So the racial disparities in offending for more-serious crimes could be in part the product of disparities in enforcement that do not necessarily lead to incarceration.

In other words, the racial distribution of offending is not some exogenous given, but is an endogenous rate that can be shaped by enforcement choices made earlier in time. We should not simply look at the racial distributions of offending and arrests and incarceration and, if they are similar enough, assume that there is no work to do when it comes to race and criminal justice.

That said, the evidence on the relationship between arrests and, say, lower employment is ambiguous. Ex-convicts, for example, inarguably earn less than non-convicts, but there could be a strong selection effect at play here. And quasi-experimental evidence suggests that this selection effect could be quite strong.

I think that is enough for this post. The tl;dr version: racial disparities in drug incarcerations cannot explain racial disparities in total prison populations, and those disparities (with respect to the population baseline) are going to be heavily influenced by more-serious offenders, for whom the racial composition of inmates remains fairly close to the racial composition of offenses. But just because offense-incarceration disparities are not as great as one might think does not imply that the underlying disparities in offending are independent of choices made by various criminal justice agencies and actors.


1The BJS defines its racial groups as non-Hispanic white, non-Hispanic black, and Hispanic. This is technically incorrect, since “Hispanic” is an ethnicity, not a race, and thus whites and blacks can both be Hispanic (i.e., Hispanic should be broken into Hispanic white and Hispanic black). However, my own analysis of the National Corrections Reporting Data revealed that well over 90% of Hispanics self-identified as white, so treating non-Hispanic blacks as blacks in total is a fairly accurate approximation.

2Whether it is truly disproportionate is tough to say. Although much of the panic surround crack was just that—a panic—there is evidence that crack was in fact a distinctly harmful drug. As the markets have matured, much of the violence may have dissipated, but some of the public health concerns remain.

Posted by John Pfaff on May 29, 2013 at 10:20 AM | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference The Surprisingly Minor Impact of Drug Incarcerations on Racial Imbalances in Incarceration:


The comments to this entry are closed.