Drugs of Choice

Doubling down on explaining drug use with price theory

In my previous newsletter, I asked whether Narcan saves lives. The basic price theoretic point is that Narcan reduces the cost of using opioids. Thus, some of the reduction in cost accrues to drug users who will get more enjoyment from increased consumption of opioids. The question is how much of the benefits accrue to drug users in the form of enjoyment in comparison to the benefits from the reduction of opioid-related overdose deaths. What I pointed out is that the magnitude of these benefits are related. The more lives saved, the less of the benefit that accrues to drug users as a result of additional consumption. The fewer lives saved, the greater the benefit to drug users from additional consumption. How these benefits are distributed depends on the price elasticity of demand for opioids.

This ruffled some feathers. Some people contacted me to tell me that the post was silly. We cannot use price theory to explain drug use, they said, because drug users aren’t rational. These are not people moving along a demand curve. Their behavior is erratic. Some criticized me for ignoring actual price elasticities of demand to downplay the success of Narcan. Others told me that I was ignoring the psychology of drug users. Still others noted that addiction is a biological condition, detached from economic incentives.

I reject all these criticisms.

First, I never assumed that drug users are rational. They need not be rational for the theory to apply. Price theory is about applying rational frameworks (not rational people) to explain human behavior. It does not require understanding what is going on in the decision-maker’s brain.

Second, the typical negative relationship between price and quantity demanded doesn’t even rely on rationality. As Gary Becker explained, this type of relationship between price and quantity demanded only requires that the person has a budget constraint. This is something that Brian wrote about previously.

In addition, while I didn’t mention actual elasticities, I did describe several scenarios. I described what happens in the inelastic case and what happens in the elastic case. As a benchmark, if the price elasticity of demand for opioids is -1, then there would be no effect on deaths. There is a relatively recent paper by Olmstead, et al. in the Journal of Health Economics that estimates the price elasticity of demand for heroin. Their baseline estimates suggest a price elasticity of demand for heroin of -0.8 with a standard error of 0.23. The point estimate itself suggests that the presence of Narcan would reduce opioid-related overdose deaths modestly. However, from these estimates, we cannot rule out that there is no effect on deaths. (Note that the relevant elasticity is the Hicksian elasticity. Since they control for income, the estimates are best interpreted as Marshallian elasticities. Whether this biases the implications toward or away from “no effect” depends on whether heroin is a normal or inferior good and the share of the average consumer’s income that goes to drug consumption. The authors own estimates of the income elasticity are quite small, 0.1, but positive.)

Finally, I don’t deny that there are psychological and biological aspects to drug use. However, that isn’t relevant to the analysis. What the price theoretic analysis does is it asks, holding other things constant, if price theory can inform our understanding of the market for opioids.

This gets to a broader point here at Economic Forces. We think price theory is broadly applicable. Yes, it is applicable to the market for oranges or grapes, but it is also applicable to other types of goods, like illicit drugs.

With all that being said, this week I thought I would double-down on the discussion of drugs in the context of economic decision-making. Let’s push the boundary of what price theory can explain by trying to explain what types of drugs that people use based on economic incentives. To this, I’m going to rely on the work of my colleague, Henry Thompson, and his coauthor, Justin Callais, and their work on “Doing Drugs.”

Economic Incentives and Drug Use

There are many different types of drugs. One might wonder whether price theory has anything to say about the types of drugs that people decide to use. This isn’t to say that there are not other factors that determine which drugs people use. Nonetheless, price theory might be able to explain an aspect of these choices.

There is certainly reason to believe that economic incentives play a role in the decisions that people make about what types of drugs to use. For example, not long ago, Major League Baseball went through quite a scandal about steroid use in baseball. The scandal was due to the fact that steroids are considered performance-enhancing drugs. Steroids help athletes to develop larger muscles and improve recovery times. As a result, a player who uses steroids has an advantage over a player who isn’t using steroids. Given that contracts for professional baseball players are quite lucrative, it is not surprising that a number of players were alleged to have used steroids.

But what if that principle applies more generally? What if labor market contracts in a variety of settings explain the types of drugs that people decide to do? Thompson and Callais present a theory of drug use consistent with this idea.

Think about two different types of labor contracts. Sometimes compensation is based on output. The worker is paid depending on what they produce. Other times, compensation is based on inputs. This is a standard wage-based job in which the worker is paid based on the number of hours worked. All else equal, if someone is doing drugs, one would expect that the decision about what drugs to use will depend on their labor market contract.

For example, suppose that I have an output-based contract. This might be something like a sales job in which the employee earns a commission on every sale. All else equal, to the extent to which a person in an output-based contract does drugs at all, one would expect them to be more likely to use stimulants than depressants or psychedelics. Stimulants might help the worker to stay alert longer, work longer hours, and have the energy necessary to make a hard sale at the end of the day.

Although this argument might sound somewhat controversial, the argument is really no different than the Major League Baseball example. Players are compensated with salaries, but those salaries are based on performance. If you hit more home runs, then you will receive a more lucrative contract. If there is a drug that contributes positively to one’s performance on the job, people are more likely to use that drug than a drug that doesn’t help with performance. (And less likely to use drugs that would harm their performance.)

You even see evidence of the price theoretic prediction in popular media. Movies about out-of-control guys on Wall Street, known for its long hours and commission-based income, almost always includes scenes in which the drug of choice is something like cocaine.

But let’s not get ahead of ourselves. Although I’m using the word “drugs,” I’m doing so quite liberally. I’m not necessarily talking about illegal drugs. Sure, stimulants include illegal drugs, like cocaine, but nicotine is also a stimulant and is sold in every gas station in America, in many forms.

The theory also doesn’t say that labor contracts make someone more or less likely to consume drugs, illegal or otherwise. The theory simply predicts that if the person does drugs and if the person is compensated in a particular way, we should be able to predict which types of drugs the person chooses.

This isn’t even necessarily a causal argument. Suppose a non-drug user becomes a drug user. If that person has an output-based labor contract, all else equal one would expect that the drug of choice for the new user would be a stimulant. However, one could just as easily argue that someone who is already using stimulants might self-select into a job with output-based compensation.

It is more difficult than one might think to test the prediction. If output-based compensation is associated with more stimulant use and input-based compensation is associated with less stimulant use, it is possible that these people in these different contracts are also systematically different in other, observable ways.

For example, maybe output-based compensation is higher on average. We couldn’t rule out that it is actually the level of income itself that determines the stimulant use. Or maybe there are systematically different patterns of drug use and different patterns of compensation among different age groups. Then the use of stimulants might simply be driven by age, even though it varies by compensation type. Furthermore, if those with output-based compensation use more stimulants, it could also be the case that people in these type of labor contracts are just more likely to be drug users independent of the type of drug.

To test this prediction about labor market compensation and the type of drug use, Thompson and Callais use data from the National Survey on Drug Use and Health, a comprehensive and representative, national survey that relies on anonymous responses. Using the data available from the survey, they match people in the survey with others who are most similar in terms of observable variables like income, education, and age. They do this through both distance matching and propensity score matching.

As their baseline approach, they match people who are self-employed (to proxy for output-based compensation) with those who work at not-for-profit firms (to proxy for input-based compensation). [Note: They also test the robustness of the results, but I won’t go into that here. You can read the paper.]

They have an extensive list of observable controls related to both characteristics of the individual as well as the individual’s risk attitudes and access to different types of drugs. By matching people in these two groups using these observable characteristics, it makes it more likely that observed variation (to the extent it exists) in the type of drug use in these pairs is primarily explained by the remaining difference: the difference in their labor market contracts.

Using a nearest neighbor approach (matching an individual with 1-3 of his or her closest neighbors in terms of observables), they find that the self-employed are 0.3 - 0.5 percentage points more likely to use prescription stimulants. They also find some evidence that the self-employed are more likely to use cocaine. By contrast, the self-employed are no more likely to use prescription sedatives, hallucinogens, or ecstasy. Each of these results are confirmed by their propensity score matching approach as well.

What these results seem to indicate is that the terms of labor market contracts do seem to have some effect on the choice of which drugs to use. Those with output-based compensation, like the self-employed, are more likely to use stimulants, but not more likely to use other types of drugs.

One could argue that the same type of person who uses stimulants is also more likely to self-select in self-employment. That’s fine. Again, they claim no causation. And remember, they are controlling for risk-taking attitudes and the self-employed are no more likely to use other types of drugs than the control group, so this isn’t just a preference for drug use generally or a difference in risk-taking attitudes more generally.

Price theory, it seems, has something to say here.

Conclusion

What this post hopefully illustrates is that economic incentives matter for decision-making. There might be a number of other factors that matter. When talking about drugs, there are entire academic literatures outside of economics that have studied drug use and addiction. An analysis of drug use within the context of price theory does not imply that this other work is wrong. It also doesn’t imply that the behavior explained by price theory is more important or more informative than those other approaches.

The basic lesson is that price theory can allow us to think through different types of decision-making, even if they are things that people might consider outside the typical purview of economics. Nonetheless, any decision that people make that involves costs can be understood through the lens of price theory. This is true even when the people making the decisions are erratic or otherwise unpredictable. Real-world constraints discipline decision-makers because those real-world constraints are the source of costs. As long as those constraints exist, price theory has a role to play in understanding human behavior.