What is a Drug Crime? Part I

Before I continue on about the causes of prison growth, it is worth pausing to ask “what is a drug crime?” The official government statistics on which I’ve been relying use a simple taxonomy: an inmate is classified by his most serious offense. Violent crimes are classified as the most severe, followed by property crimes, then drug offenses, and finally "other" crimes.  So a “drug inmate” is someone convicted of one or more drug offenses, and possibly some “other” offenses as well.

This is perhaps a narrow definition of "drug offense," particularly if we are trying to estimate the effect of the war on drugs on incarceration. Consider a few other alternatives:

1. A murder resulting from a shootout between two drug-dealing gangs over territory or some other drug business issue. This will be classified as a violent offense, but it feels drug-related.
2. Pushing further, a murder resulting from a shootout between two drug-gang members as part of a power struggle following the incarceration of a high-ranking member of the gang which resulted in a power vacuum.
3. A robbery for drug money.
4. A robbery for drug money that is needed only because the war on drugs has likely increased the prices of drugs above what they otherwise would have been.
5. A robbery unmotivated by the need for drugs, but caused in part by the fact that prior low-level drug arrests, none of which led to incarceration themselves, led (or contributed) to a downward spiral—a harder time landing a job, maintaining a relationship, etc., etc.—that ultimately culminated in criminal behavior.

Should all these count as "drug incarcerations"?

A plausible argument can be made for each. But there are real costs to going down these routes as well.

First, if we are going to take a more comprehensive view of the drug war's costs, it is only proper to also account for its benefits as well—the war on drugs may be a net loss, but it still has some gross gains. So for every violent offender we reclassify as a "war on drugs" inmate, we'd need to somehow count the people who are not in prison because, say, the threat of arrest steered them to rehab and thus away from drug crimes, or the higher prices of drugs caused them to consume less (the demand for drugs is not unresponsive to price changes) and thus commit less violent crime, and so on. There may not be a lot of these people, but we need to estimate how many there are.

Second, we want to be careful not to retreat too far quickly to some sort of biological or sociological determinism, that the offender whose low-level marijuana convictions have made employment difficult is simply the victim of the drug war. Most unemployed people avoid lives of crime (I assume—I can't quickly find data on the percent of unemployed with criminal records), so at least for now¹ it seems reasonable to impute some of the responsibility for the crime onto the offender. So the person who commits robbery to fund his drug habit is... what? 1/3 drug offender, 2/3 violent offender? How would we even begin to determine such a ratio?

And third, these classifications are much more subjective than the coarse taxonomy used in official statitsics. This is not, per se, a problem, just a limitation that needs to be acknowledged. Using a more nuanced definition comes at the cost of more debates over definitions, and the need to better model our uncertainty about classification. How do we really know why Mr. G committed this robbery? How can we be sure that in a different world without a war on drugs he would not have done so? We can make estimates, but we need to note the imprecision they introduce.

This is not a problem unique to drug crimes. Consider “gang crimes.” The Chicago Police Department measures gang activity by looking at “gang-motivated” crimes, which are crimes committed by a gang member in furtherance of the gang’s interests. The LA County Sheriff’s Department, on the other hand, simply looks at “gang-related” violence, which is any crime committed by a member of a gang. The LACSD definition is more objective, since all it requires is information on the offender’s gang affiliation, not his motives, etc. The CPD definition seems to better capture what concerns us with gangs, but leaves much of that information to subjective assessments by law enforcement. How should the CPD classify a crime in which a member of Gang A kills the girlfriend of a member of Gang B due in part to a love triangle gone wrong, but also in part due to escalating tensions over a turf war. And does our analysis change if the victim was a member of Gang B as well?

I’ll stop here for now; I’ve likely crossed some sort of tl;dr line as it is. In one of my next posts, I’ll look at what the data can tell us about the prior histories of those convicted of index offenses—if we see evidence of lots of prior drug offenses, it would at least suggest that a definition that looks only at current offense is missing something important about the dynamics at play here.

 

¹I actually find the hard determinism literature quite provocative—this is one of my favorite articles on the topic—but the hard determinists certainly have not proven their case yet, so it seems perhaps best to design policies that still rely on the assumption of free will, but to also remain open to the limitations of such a perspective. But this is material for a much different post.