Are Marijuana and Alcohol Substitutes? Another Faulty Analysis from the Washington Post

The Washington Post has been busy cheerleading yet again for legal marijuana. An August 7 article notes that marijuana is cheaper than beer in many places, and this is a good thing. The article is replete with analytical problems.

The authors review prices for marijuana leaf and bud in various outlets, and find that the price for a joint is falling. Hence, that price change may affect demand. But the first problem comes when the authors point to a possible benefit. If marijuana consumption increases, and if marijuana is a substitute for alcohol, and if, therefore, alcohol consumption were to decrease as a consequence of the substitution, and if marijuana were to be less harmful than alcohol, then cheap marijuana is a blessing for Americans. That’s a lot of “ifs,” but this is economic research, buttressed by published science. So we are told.

Here’s what the Post piece actually says: “A 2002 study in Australia found that lower marijuana prices increased marijuana consumption and decreased alcohol consumption, with users substituting the former for the latter. Considering that many experts say that marijuana is less harmful than alcohol on just about any measure, this may not be a bad thing.”

And sure enough, there is a link to a 2002 published academic paper from Australia examining prices for marijuana, which seem to be falling a bit. They further compare those prices to those for alcohol (beer, wine, spirits) and find them rising a bit. They then move from changes in price to an estimated volume of consumption of the products.

The research argues that “the nominal prices of the alcoholic beverages rose by 3-4 percent per annum on average, while that of marijuana fell by 2.5 percent.” Furthermore and at the same time, "per capita consumption of beer and wine declined at an average annual rate of 1.9 and .5 percent, respectively, while that of spirits and marijuana rose by .8 and 2.0 percent, respectively.”

It’s not clear what kind of “substitution” one is finding in these data; wine and beer presumably declined while marijuana consumption is said to have increased, but note that consumption of spirits appears to rise in concert with marijuana.

Regardless, the Post tells us that marijuana and alcohol are “substitutes” because this paper tells us that they are. Or does it? When one reads the actual study, it turns out that there is no demonstration that they are substitutes. What the academics say is this: “As consumption of alcoholic beverages and marijuana are likely to be substitutes, a fall in the price of the latter leads to less drinking.” While this statement is certainly plausible, it is nonetheless an assumption of the paper (what a professional in such matters tells me is "trying to estimate cross-price elasticity"), not an independent demonstration of the fact. (One is reminded of the hoary joke about the economist marooned with others on a desert island, with nothing to eat but cases of sealed canned goods. The economist, showing great ingenuity, saves the day: “Assume a can opener,” he declares to great approval.) There are ample anecdotes from law enforcement suggesting that marijuana and alcohol, rather than being substitutes, compliment each other in actual use patterns. Most habitual drug users are poly-drug users, and one finds routinely in blood or urine testing the presence of both intoxicants in the same individuals. Finally, Dr. Robert DuPont of the Institute for Behavioral Health points out that in both the Monitoring the Future study of youth and the National Household Survey of Drug Use and Health, those who do not use marijuana report drinking less alcohol than those who do use the drug.

Further, no one has any idea what volume of marijuana is actually being consumed in any research setting. Consumption can be assumed, often from proxies, but the only thing academics can actually measure is not product consumption but rather prevalence rates of use of the drug, those measurements sometimes refined by data obtained in response to questions about how much money users spend to buy their drugs. Moreover, even prevalence estimates depend upon self-reported sample survey data. The volume of consumption of marijuana is an unknown.

Alcohol consumption can be more closely determined from prices, because alcohol is sold in standard units with standard proofs; 12 ounces of beer, or 750 milliliters of wine, for instance, at a certain declared alcohol content. Further, the overall sales data are a matter of record. Neither of these facts is true for marijuana. How large is a standard joint, how much is being inhaled, and how much intoxication is produced are strikingly variable.

The Australian study referenced by the Post is based on recorded prices for marijuana plants. But this “unit” of organic material is not the pertinent measure affecting consumption. Rather, one needs to know the potency of the marijuana. That is, the content of the organic material contains a largely unknown measure of THC, the intoxicating compound, and in general THC potency has been rising over time, worldwide. Hence, the mass of organic material smoked could be rising or falling, but what one needs to know is the amount of THC represented by this leafy mass. In fact, it is quite reasonable to assume that as the potency of the plants increases, users could very well be “consuming” less rather than more of the organic material. Whatever the complicated facts, it is simply not likely that the researchers, measuring bulk plants, could infer consumption with anything like the precision of changes of 2 percent per annum, as claimed.

Concerning the “substitutability” of marijuana and alcohol, academic prudence says that it’s back to the drawing board. However, it still remains to be explained why the Post is so relentless in its presentation of positive, no matter how dubious, arguments on behalf of smoked dope.