Beer and Taxes

Worth noting, Washington, Oregon, and California are amazing states for locally crafted beer… which means the popular stuff has higher alcohol content. Measuring by the gallon favors states that drink piss.

Well, I was going to be very sad until I read Sam's comment. I hope the theory is fact. Because I do love my beer.

Camden, our news intern, is creating an excel spreadsheet to see what all these stats mean.

Spreadsheets and beer don't mix.

The 0.76 tax is ~43 cents/six-pack. Applied to volume, I'm thinking it favors quality over quantity. Local brews might benefit.

Local micro-brews are exempt (as they are already exempt from the majority of the state's beer tax), so shouldn't hurt either jobs or the availability of good local brew. This is for those mass-market beers that Eric Idle calls “like making love in a canoe — f#@king close to water”

Bob – are you sure local brews are exempt? if not we should push for them to be.

Meinert,
They are exempt. At least that's what the Senate proposed.

Does anyone know what the state defines as a local micro-brew – like which breweries are just to either side of the break-point of 60,000 barrels/year?

As a microbrewery owner, that's good, but for bars that sell industrial-manufactured brews for the masses (myself included) in addition to microbrews capturing more refined palettes, the tax is not so great for their bottom-line — to say nothing of 7 Eleven shoppers.

With our already regressive sales-tax-heavy tax structure, this is yet another drag for folks of modest means. I'm just weighing in with a little cross-class/beer industry solidarity and concerned about the hit on bars.

With respect to your hard-working intern, that bar graph is meaningless. I'll walk you through what you need to do to get the answer you're looking for.

The question you posed is whether “a higher tax rate on beer would impact consumption.” Obviously that answer is yes. A higher tax rate on beer means a higher price for beer, and unless beer is a Giffen good (it's not), then the law of demand says that the higher price means consumption will decrease. If that's what you really wanted to know, you don't need that chart, I just gave you the answer.

I think the question you're really interested in is HOW MUCH consumption is impacted by a given increase in the price of beer. In econo-speak, you want to know the “elasticity of demand” for beer. If the price of beer increases by 1%, then the elasticity of demand for beer tells you by what percentage the quantity of beer consumed changes.

To determine the elasticity of demand, you want to regress the log of the quantity of beer consumed against the log of the price of beer. You'll want to control for state effects (a fact that you actually recognized, even if you didn't know the technical term for it), which can be done by running your regression with panel data. (Essentially, you run 50 different regressions, one for each state.) However, you can't just regress quantity against price, because there is dual causality. If the price of beer changes, then the quantity consumed changes.

But if the quantity available changes, then the price changes. You can control for this using instrument variables. A good instrument would be sales tax. A better instrument might be the beer-specific tax. Regression using instrument variables is essentially a two-stage regression: first, you regress the price of beer against the log of the instrument variable (use the beer tax, I'm almost positive it's a going to be a better instrument than general sales tax or beer tax + sales tax); second, you regress the log of the quantity of beer consumed against the log of the estimated price of beer obtained from the first regression.

The result of this two-stage regression using instrument variables and controlling for state effects will almost get you your answer. The coefficient on the log of the price of beer from this regression will tell you the elasticity of demand for beer. It's going to be negative something. If the coefficient is -0.8, then a 1% increase in the price of beer will lead to a 0.8% decrease in consumption of beer. If the coefficient is -1.2, then a 1% increase in the price of beer will lead to a 1.2% decrease in consumption of beer.

The elasticity of demand looks at the change in the total price of beer, you're wanting to know about a change in the tax on beer. So your last step would just be to solve an elementary equation to figure out how big of an increase in the beer tax would lead to a 1% increase in the total price of beer. The solution of this variable is then the percent change in the beer tax associated with the percent change in quantity you obtained a moment ago.

For example, suppose the coefficient from the regression above was -1.08, and you determined that it takes an 8% increase in the beer tax to result in a 1% increase in the total price of beer. Then it could said that an 8% increase in the beer tax results in a 1.2% decrease in beer consumption. You can scale these number to see how much you'd have to increase the beer tax if you were hoping to decrease consumption by a certain amount. Or you could see how much consumption would decrease if you were thinking about implementing a particular increase in the beer tax.

I totally need to get Maddux'd at your brewery one of these days!

As for the “impact on consumption”…

Using Josh's calculation above, we're talking about less than 29 cents per six pack.

I've sat back and watched as the liquor taxes have increased, and have not seen a decline in consumption because of the tax increase (rather, the down economy). Realistically, if beer prices go up by $0.30 per six pack because of this, I don't really think many people would notice the difference, unless they are total lushes. And in that case…hey, more money for the State!

Granted, this is completely anecdotal, but that's just how I see it. And, IMO, it's better than raising the overall sales tax.

My apologies Josh, I missed this when I read the article the first time:

“Do the high taxes inhibit beer consumption? Or are they a sign of states with high beer consumption (meaning the states with the highest beer taxes know how to make some money off their derelict citizens)?”

I would say yes. If you follow the methodology I suggested above, you will find that an increase in price leads to a decrease in consumption. The analysis described there separates out demand effects from supply effects. That is, it isolates how a change in price causes a change in quantity from how a change in quantity causes a change in price.

The result of that analysis will show that an increase in price leads to a decrease in consumption. In fact, a quick Google search found a university study using a nearly identical method to what I described above. The only non-trivial difference from their study and what I outlined is that I forgot to say to include income as a regressor, an unfortunate oversight by me. Their study shows that a 1% increase in the price of beer will lower the quantity of beer consumed by 0.53%. You can read their study here here. I'm sure you can find many similar studies attempting to calculate the elasticity of demand for beer.

Since an increase in the price of beer causes a decrease in the consumption of beer, and a tax on beer increases the price of beer, then logically a tax on beer decreases the consumption of beer.

If you see a flaw in my reasoning (I don't, but I'm open to critiques), then perhaps another route would be this:

Forget about the price of beer. Regress beer consumption against beer taxes and income (actually, the log of each of those things). Use panel data to control for state-specific fixed effects. And use an instrumental variable to estimate beer taxes. I think income+property+gas+cigarette tax would be a good IV. The lower they are, the higher a state would need to set its beer tax to make up the revenue gap. And the taxes on those other things don't affect beer consumption, at least not when you're also controlling for income. I'll bet you a beer that you'd find a coefficient on the ln(beer tax) regressor that is less than zero at a statistically significant level. Regardless of who wins the bet, this regression would definitely answer your question.