Liberal Blogger Matt Yglesias Wants to Tax Your Beer! I Want to Tax Drunk Drivers.

Matt Yglesias writes:

[E]xcessive alcohol consumption is associated with a lot of health problems, and not only problems for excessive drinkers—drunk driving and the linkage between alcohol and violence impose significant costs on other people. Mark Kleiman estimates that “Doubling the tax on beer (from a dime to twenty cents a can) would reduce the assault rate by at least 5%, and maybe as much as 20%.”

Meanwhile, you could get a healthy chunk of the revenue needed to pay for health care reform. [..] I don’t want my beer to get more expensive. But at the same time, I do want to see comprehensive health care reform. So someone will have to pay something. And this is a pretty good option.

There's pretty good economic evidence to suggest that alcohol consumption is fairly sensitive to price. And there's lots of good economic (and commonsencial) evidence that alcohol consumption is associated with a wide array of undesirable outcomes, from increased highway fatalities, to decreased productivity at work, to increased violence against women, to worsened health outcomes. Increasing alcohol taxes would, therefore, almost certainly save lives.

One problem, however, which is somewhat unique to liquor consumption, is that these behaviors aren't particularly strongly associated with drinking unto itself. They are associated, rather, with drinking to excess and/or engaging in other, particularly stupid sorts of behaviors while doing so. I know the evidence on this is mixed, but many studies have suggested that moderate alcohol consumption is in fact associated with improved health outcomes -- especially if you're drinking red wine and especially if you're a guy. A person who has a couple of drinks a couple of days a week, and who never drives or has the compulsion to engage in violence while doing so, imposes virtually no negative consequences either on himself or on society. Drinking doesn't cause negative externalities in the same way that, say, driving (traffic congestion and pollution) or smoking (second-hand smoke) intrinsically do.

People say it's difficult to tax the undesirable behaviors associated with drinking without taxing drinking itself. But is that necessarily so?

In 2006, there were 1.1 million arrests for drunk driving in the United States (source), not counting Florida which didn't report its statistics. Fine each of those people $8,000, and you'd have almost about $9 billion more to pay for health care every year. Why $8,000? Because that's the figure, according to a 2001 paper (.pdf) by Steve Levitt (the Freakonomics guy) and Jack Porter, that would be required to internalize the negative externalities associated with driving drunk.* By the way, if you're concerned that this tax might be regressive, you could scale it according to a person's income, as they do for traffic fines in Finland.

Of course, if you were actually to fine people $8K every time they got a drunk driving conviction, you wouldn't raise quite as much as $9 billion. Faced with a choice between an $8,000 fine or a $20 taxi fare, a lot more people would have Yellow Cab on speed dial, and you'd have fewer revenue-producing arrests.** But this is a feature of the policy rather than a bug -- you'd be stopping drunk driving. Moreover, it's exactly the same feature/bug problem you'd run into by raising alcohol taxes in general, or any time you were trying to use tax policy to disincentivize an undesirable behavior.

The drunk driver tax would also produce at least as much revenue as the liquor tax. The CBO estimated (.pdf) in December that raising alcohol taxes to $16 per proof-gallon (by about 10 cents a drink over current levels) would raise about $6 billion per year. Even if our $9 billion total were decreased somewhat by deterred drunk driving -- and again, that would be a Good Thing -- we could probably beat that number.

Of course, there are probably some jurisdictional or federalism issues with this. And like Yglesias, I'd be willing to chip in a few extra cents with every beer to help pay for universal health care. (Although as a self-employed person who could probably expect some cost savings from alternatives to employer-based insurance, this is not necessarily a selfless act for me -- though I do drink a lot of beer!).

But a drunk driver tax would be fairer, more efficient, and more Pigovian than a beer tax -- and good luck to the liquor industry in trying to oppose it.



* Levitt and Porter don't account for fines and other sorts of explicit and implicit financial penalties (e.g. raised insurance premiums) that drunk drivers already face; on the other hand, they also don't account for externalities associated with merely maiming, rather than killing, innocent victims. Nor do they blame drunk drivers for the deaths of the other passengers in their cars, figuring that those people had knowingly accepted the risks of driving with someone who had been drinking to excess.

** Then again, there is probably some low-hanging fruit. The number of drunk driving arrests varies a lot from state to state -- from a high of 861 per billion vehicle miles traveled in Alaska to a low of 26 in Delaware. These differentials have virtually no correlation with the amount of drinking in each state -- slightly more Delawareans than Alaskans are binge drinkers, for instance. While I'm enough of a libertarian to be opposed to things like random traffic checks, clearly some states can afford to be more vigorous in their pursuit of drunk drivers.