[Philip Cook, guest-blogging, January 7, 2009 at 8:15am] Trackbacks
A Free Lunch?

In the pre-Prohibition era, the saloons would advertise a "free lunch" for the working man. The catch was the expectation that the working man would have a beer with that lunch. The beer was definitely not free. Thus the most famous saying associated with economic science -- "there's no such thing as a free lunch" -- originates with alcohol retailing.

Given that historical reference, I enjoy the irony that alcohol taxation is a "free lunch," or at least close to it. An increase in the federal alcohol tax could benefit most people directly (in the sense that they would come out ahead financially), while increasing economic efficiency for all.

Let me fill in the background on this not uncontroversial claim. While taxes usually impair economic efficiency by distorting incentives, that is not necessarily the case. If there are negative externalities associated with a particular activity, then it is "underpriced" and will lead to over-indulgence. Compelling examples include carbon emission and driving during rush hour.

Needless to say, drinking, while a pleasurable activity to many (including me), generates negative externalities. If you add up the deleterious effects on crime and highway safety and STD transmission and the quality of parenting, it is easy to surpass a dollar per drink on average. The public is inadvertently forced to share that cost, only a small part of which is covered by alcohol tax revenues. (The current federal excise tax amounts to 5 cents per beer, and even the highest state tax rates are lower.)

It would be more efficient if drinkers were confronted with the full price of their decision of how much to consume. Raising the tax rate can help accomplish that purpose. And it would be fairer to have the drinkers pay the social costs (in proportion to their drinking), rather than have those costs borne willy-nilly in everyone's insurance premiums, income tax rates, uncompensated risks, and so forth.

However, while it the "drinker pays" approach seems fairer than the current system and certainly more efficient, it is far from perfect. The problem is that that average social cost conceals a great deal of variation.

A drinker who enjoys one glass of wine or beer with dinner every night is unlikely to impose any cost on others as a result. If that same person drank all her weekly ration on Saturday night before driving home, the total tax bill would be the same, but the social-cost calculus would be far different. The public consequences of private drinking choices depend on who, when, and where, not just how much.

Thus the alcohol tax appears to be a crude remedy at best.

Instead of taxing alcohol, it seems that it would be preferable to "tax" the socially costly consequences of abuse directly: Impose still heftier penalties on drivers convicted of driving while intoxicated. Have social workers finger negligent and abusive parents for sanctions. Step up enforcement against alcohol-related violent crime (and maybe those drunks who put themselves at foolish risk of being victimized). Subsidize condom distribution. And so forth.

Of course these remedies, while seemingly better targeted, are costly in terms of use of public authority and revenues. So let's look again at whether alcohol taxes are really all that "crude." It is relevant and revealing to look at the distribution of alcohol consumption.

Per capita consumption in the United States runs about 500 drinks per year, where a "drink" is a 12-ounce beer, a 5-ounce glass of wine, or a 1.5 ounce shot of 80-proof spirits (all of which have about the same amount of ethanol) But that average also conceals a great deal of variation: about 35 % of adults abstain, and drinking is very concentrated within the larger group who do drink.

The famous 20-80 rule of marketing applies -- 20% of the consumers of most any commodity account for 80% of the total purchased. Removing the abstainers, that means that 13% of adults consume 80% of the ethanol, and thus pay 80% of the tax. (I've checked this estimate against actual self-reported drinking, and it works pretty well.)

What's more, only about 7% of adults drink more than that 500-drink per capita average. That means that 93% of the American public contribute less than average to the alcohol tax.

As a thought experiment, consider increasing the alcohol tax by 10 cents per drink and then distributing the proceeds annually to every adult, $50 each. All but 7% would come out ahead on this deal. Given the preventive effect of higher alcohol prices, even that group would benefit from lower auto insurance rates and in other ways.

This thought experiment reminds us of the nice feature of alcohol taxes -- unlike other prevention measures, this one generates revenue. And taxes no longer seem quite so crude or unfair, being nicely concentrated on the heaviest drinkers where we also find most of the abuse and social costs.

Of course in practice the distributional consequences of an alcohol-tax increase depend on how the money will be used. This year the extra revenues from any tax increase will be put to closing state budget deficits, which is to say that they will take the place of an increase in some other tax rate or a cut in expenditure. The ultimate question when it comes to distributional and efficiency consequences is how the alcohol tax stacks up against the alternatives.

The discussion of distributional consequences also must include mention of jobs. The alcohol industry has fought against nominal tax increases with great success on this basis. Not content to see tax rates erode with inflation, they are campaigning to have Congress roll back the 1991 increases, arguing that alcohol taxes reduce national employment. Of course this claim is not to be taken seriously.

The excise tax rate affects the size of the alcohol industry (and ancillary industries, such as funeral homes and trauma surgery). But aggregate employment is not affected by the beer tax, only the portion of the economy devoted to beer.

I'm sure at this point you are eager to hear just how high I would go when it comes to alcohol taxes in an ideal scenario. One standard would be to return the tax to the level that prevailed in some previous period, such as 1951 or 1975 -- whenever the "good old days" occurred in your life or mine. I think that historical standard is an interesting reminder of trends, but provides little real guidance.

Another possibility is the public health standard, namely to save as many lives as possible. The problem is this -- The higher the tax, the more lives are saved, up to some level so high that we are back to Prohibition. The public health standard takes no account of the pleasures of drinking, and thus provides no basis for balancing pleasure against cost.

The most defensible approach in my mind is to set the tax equal to the average marginal social cost of a drink, perhaps with some distinctions between beer and spirits, or between on-premise and off-premise service. Estimating the precise levels would require careful up-to-date analysis. But we don't have to do that precise analysis to know for sure that the social costs are much higher than the current tax rates. In particular, the increases that are being proposed by various governors this year are just a small step in the right direction, far less than the full social costs.

Any complete account of the public interest in alcohol taxation must deal with a few other issues -- with the alleged health benefits of drinking (which I'll turn to in a subsequent blog), and with possible substitutions. Would higher taxes lead consumers to substitute illicit drugs or dangerous moonshine?

The concern about an upsurge of moonshining or home manufacture in the face of higher prices is misplaced. Even if the 1950s, when federal taxes were a multiple of what they are today, illicit production filled only a miniscule piece of the market.

Some folks may try their hand at home beer- or wine-making, but unless it's an enjoyable hobby it would not be a productive use of your time even if commercial prices increased by 50%.

The possibility of substitution to other intoxicants is a real one, but the evidence again points the other way. As it turns out, drinking and illicit drug use are not alternatives -- rather they tend to go together. Almost every illicit drug user also drinks, and a high percentage of those who seek treatment for drug abuse also have problems with alcohol. One analysis of alcohol tax changes found that marijuana smoking declined with higher alcohol taxes -- the two intoxicants are complementary.

In any event, the evidence on the virtuous effects of tax increases that I cited in my previous blog is based on data analysis of what happens in states that raise their tax in comparison with states that don't. The effects on mortality and all the other outcomes incorporate and reflect all the substitutions that consumers make in response to those higher alcohol prices.