In this post, I highlight five assumptions that the report makes to obtain estimates of tax revenue. To make sense of the policy implications, we need to understand what these assumptions imply the tax revenue that would be collected from a legalize-and-tax policy applied to marijuana.
Here's how the report (found here) estimates the tax revenue from a legalize-and-tax scheme:
Using year 2000 marijuana expenditure levels from the Office of National Drug Control Policy as a baseline, the report makes the following assumptions:
1. Marijuana's demand curve (relationship between price and quantity) is the same before and after legalization. Under this assumption, price and quantity change solely because marijuana becomes more or less costly to produce (i.e., supply conditions).
- The report then makes the strange claim that it might become more costly to sell marijuana in a legal scheme. This is strange because marijuana suppliers can always exercise the option of returning to the black market. If it really was more costly under a legalized system, producers would just revert back to the black market.
3. The demand for marijuana in the United States is the same as in 1972. Either that or use the demand from Australia. The world is a different place today than in 1972. The quality and potency of marijuana has also changed dramatically enough for some people to claim that it is a "different drug." I'm not claiming that it is a different drug, but if it has different quality attributes, we should expect price responsiveness to be different.
As an alternative estimate, Miron cites a study that used Australian data. Like the Netherlands comparison, using this study makes the implicit assumption that the United States and Australia are the same (i.e., have the same price-quantity relationship). They might be, but they might not.
4. The data represent what's really going on. As the previous three assumptions make clear, the report relies heavily on other marijuana studies. Most of these studies conducted a survey to obtain data. And, there's a big problem with surveys about illegal activity: people lie. Even purchase data are bound to have problems when the act of purchasing the drug is illegal. People hide that they're purchasing illegal drugs. Asking people about their activity and trying to observe their use is going to miss something, and that's going to affect the calculations.
5. Substituting from other goods (alcohol, tobacco, etc.) has modest, negligible effects on overall tax revenue. Miron recognizes the possibility of substitution from other taxed goods to marijuana, but he does not deal with this issue quantitatively. Instead, he dismisses this effect by referring to another conceptually unrelated effect that likely goes in the opposite direction -- that demand increases because punishments no longer apply. It might be that there are no good estimates out there on how people would substitute from alcohol to marijuana, but this is something we need to know (or at least think about more carefully).
As I am still sorting through these assumptions myself, I appreciate reading your thoughts in the comment box below. I have some thoughts, but it is a complicated problem. In particular, I would like to know: What effect do these assumptions have on the tax revenue estimates? Why? The report makes most of these assumptions on account of having inadequate data. Do you see any better ways to get around missing or misleading data? Also, if you have access to data sources that can help me sort out what's going on, I appreciate the input. Next week, I will have a more to say.
Note on Attribution
Since last week, I discovered the Jeffrey Miron has since extended the paper to cover the budgetary implications of legalizing all other currently illegal drugs. In particular, his revised version of the same paper is entitled "The Budgetary Implications of Drug Prohibition." His paper also updated his estimates for marijuana for 2006. After adjusting for inflation, the estimates in the new paper are $1.53 billion higher than the previous paper.
With a research question like this where the data are very noisy, an 18 percent blip in costs is not unheard of. What would be really useful is if Miron computed some measure of the variability in the data. For you statistics nerds, there are few standard errors in the report (though he talks a lot about uncertainty and variability). It would lend more credibility to the report if it quantified the data's variability in some way.
The Budgetary Implications of Drug Prohibition can be found here.