There have been some reports in recent weeks which suggest that marijuana legalization is creating some drag on beer sales in states like Colorado, Washington, and Oregon (and there have also been reports arguing the opposite). Although I don’t purport to know what the long-term effects of marijuana legalization will be, I can say that I see no evidence that legalization has had an effect on beer sales in the short term.

Let’s delve into the (scant) existing evidence and why I differ on this issue from some other analysts.

Dissecting the Data

First, I think there are data issues with some of the analyses. For example, a recent analysis by Cowen and Company used Nielsen data to note that in Denver, “total beer volumes in that market have fallen 6.4 percent year-to-date and craft beer volumes have dipped five percent.” Now, I think Nielsen data is great, and I don’t doubt that in the channels they measure those numbers have some validity, but the problem is that Denver is probably one of the markets where scan/POS misses the most volume given the incredibly strong craft on-premise scene.

In 2015, premises use (brite tank sales) were equal to 3.3 percent of total shipments in Colorado, the highest percentage in the nation. That doesn’t even count to-go sales and other ways brewers are selling beer that isn’t being measured. Given that premises use TTB data is up 57 percent this year, that could explain a large portion of any declines being seen. And that’s exactly what you see if you look at excise tax data/shipment data. In fact, the Colorado Department of Revenue shows that consumption of beer in Colorado is up relative to 2015.

More generally, even pointing to a change in consumption isn’t enough to prove anything, since there are a ton of confounds in trying to figure out whether beer consumption is increasing or decreasing and why. The first would be to look at population growth, since per capita consumption is going to tell us more about long term trends than absolute consumption in any given year. Next would be to look at other factors that affect the beer industry, such as unemployment, other parts of beverage alcohol, etc. Many of the analyses I’ve seen ignore these other variables and simply look at sales in a place before and after a change in marijuana regulation. Although that may tell part of the story, doing these types of comparisons without controls makes it very easy to confuse noise for signals.

I’m not convinced that anyone has clearly demonstrated what the causal mechanism would be for marijuana legalization decreasing beer sales.

Finally, even if an analysis were to control for all of these variables and find a short term relationship, I’m not convinced that anyone has clearly demonstrated what the causal mechanism would be for marijuana legalization decreasing beer sales. The usual explanation provided is called a “substitution effect,” where consumers substitute one good for another, in this case marijuana for beer. I haven’t seen any data that proves this one way or the other, but the mechanism here could further be broken into two possibilities: consumers have less money to spend on beer since they are spending it on marijuana or consumers are simply substituting the effects of marijuana for the effects of alcohol.

On the first, there may be some evidence that legalization is actually creating more disposable income for consumers. A recent paper by the Marijuana Policy Group (MPG) looking at the economic impact of marijuana on Colorado suggests that the average 21+ resident in Colorado may actually have spent less on marijuana in 2015 than they did in 2014. How is that possible? For one, the MPG didn’t see a huge surge in demand, with increased sales partially driven by tourism (one estimate is that 4 percent of Colorado’s record 2015 tourist numbers were driven by marijuana) and much of the rest coming from the shift from the black to the legal market. Much of the remaining small increase in internal demand can be explained by population growth–Colorado adds nearly 100,000 21+ adults a year. Finally, any remaining growth in demand is offset by the drop in prices brought on by legalization. Put it all together and when you’re just considering marijuana spending, MPG’s data show that the average 21+ Coloradoan has more money in their pockets now than before.

On the second possibility, that marijuana is replacing alcohol occasions, that would again require controlling for other beverage alcohol sales as well, since total beverage alcohol varies less than the shifts within that consumption between beer, wine, and spirits. Maybe I’m in the beer echo chamber, but why are so many of these studies focusing on beer rather than simply beverage alcohol? Arguing that marijuana affects any of the three (beer, wine, or spirits) without looking at the other two misses the clear relationship between an increase in one and a decrease in one of the other two.

To close, let me underline that none of the above should suggest that marijuana won’t have an effect on the beer industry in the longer run, rather that I don’t see anything definitive one way or the other so far. Although I critiqued the Cowen and Company conclusion that marijuana has had an effect on beer sales in legalized states, I find their overall research agenda highly informative. Saying that there is no effect in the aggregate is very different than saying there is no effect in different portions of the market, and given widely ranging incidence rates by gender, ethnicity, income, and geography, marijuana has the potential to affect different segments of the beer market in quite different ways.