Pretty much the most interesting blog on the Internet.— Prof. Steven Landsburg

Once you get past the title, and the subtitle, and the equations, and the foreign quotes, and the computer code, and the various hapax legomena, a solid 50% English content!—The Proprietor

Monday, May 17, 2010

Drunk Driving Licenses

Apart from the occasional self-indulgences like gloating (prematurely) or playing gotcha, I like to write about questions which I find interesting in the hopes that so will my band of readers. A question is interesting usually because it is hard; a question is hard usually because there are non-obviously-wrong arguments for both sides. One such case is whether it should be against the law to drive drunk.1

Let's start with a few assumptions (all which are, to the best of my knowledge, undisputed facts):2 The influence of alcohol degrades driving performance. There may not be any threshold beneath which there is no effect, but even if there is a threshold, it is quite low. Even under the current state of legal prohibition against drunk driving, tens of thousands of people die in the U.S. every year because of traffic accidents which would not have happened (or been fatal) if all involved drivers had been completely sober. Many of those who die (or are otherwise injured, often severely) were not themselves under the influence, or even passengers of the affected drivers who could arguably have said to assume the risk.

So, case closed; so much for that "hard" question, right? Not quite.

After all, the question for debate was not whether it should be against the law to negligently or recklessly injure or kill innocent bystanders. That is the one with the easy "yes." The question was whether driving drunk per sei.e., regardless of whether anybody has been hurt—should be subject to legal punishment.

Answering "yes" to that question seems straightforward, but runs headlong into one of the most basic and soundest principles of good law-making: That the state should only impose legal punishment, civil or criminal, where there is a specific, involuntary victim.3 Without such a victim, the state is just meddling: The driver after all chose to drink and presumptively for good reason: enjoyment is a good reason. Without any offsetting loss somewhere else—and in cases where there was no victim there appears to be none—legal prohibition of drunk driving seems sub-optimal: In so far as it deters the driver, it makes him or her worse off by the opportunity cost of drinking. In so far it does not, it makes him worse off by the legal punishment (and the rest of us by paying the cost of imposing that punishment). So laws against drunk driving just seem to be contrary to sound principle and economically inefficient.

The obvious response to that is that while any individual act of drunk driving may be victim-less, each creates a small, but non-trivial, risk of serious harm. Therefore drunk driving should be prohibited regardless of whether any particular instance actually causes any harm.

Except that is not the way we regulate risk generally and for good reason. Generally, individuals are free to do all sorts of risky things: Hang-gliding, working as a miner or lumber jack, operating an off-shore oil-rig, performing (or having) high-risk surgery, or even just plain old driving a car completely sober. The risky behavior itself is not prohibited; only if the harm occurs is there any legal sanction: the risk taker just has to bear his own damages and, depending on circumstances and legal standards, the damages suffered by others.

That is a good rule, both from the standpoint of principle for a free society and, not coincidentally, from the standpoint of economic efficiency. If you internalize all the benefits and risks to one person and then allow that one person to make the decision whether to engage in that risky behavior,4 you will generally get just the right level of risk taking: Those activities which overall are beneficial will be undertaken and those which are not, will not.

The problem with applying this rule to drunk driving is that the rest of our legal system is insufficient to deal with cases in which a serious accident actually does occur. Damages to the victims are easily evaded through bankruptcy (or even the threat thereof). Even if these damages were non-dischargable, it seems doubtful in many cases that the perpetrator would ever be able to offer full financial restitution. Even criminal penalties for drunk driving causing severe injury or death to innocent bystanders (while much harsher than they once were) are still extremely lenient compared to those meted out to criminals who cause the same amount of harm intentionally.

So, when the risk actually occurs, we will not adequately deter the perpetrators and often cannot sufficiently compensate the victims of drunk driving. This under-deterrence will, by standard economics, lead to an overindulgence in the risky behavior. So we have a law which places additional deterrence against the very act of taking the risk, regardless of whether it actually occurred: the law against drunk driving. This is probably wise and creates more economically efficient incentives than a regime which permits drunk driving in combination with the limits on deterrence and compensation mentioned above.

Still, there is something about this which rankles this author's principle explained above. And perhaps the combination over-deters drunk driving too? Certainly, given physiological fact, it seems hard to believe that it does not under-deter drivers just beneath the legal limit or over-deters drivers just above it (or both). And is it not a dangerous precedent towards all sorts of slippery slopes?

So I would suggest a solution for at least a part of the problem: An optional license to drive drunk available only to those who can post sufficient bond or insurance and assume sufficient criminal indemnity that they will not be under-deterred. Note that such a license would not protect against legal liability for damage caused to others: Drivers with such a license would still face full civil and criminal sanction for that. The only difference is that the mere fact of driving under the influence would not itself be a crime for such drivers.

Needless to say this is not an entirely satisfactory solution. The collateral or bond posted would have to be large enough to cover judgment even in very severe cases (like multiple deaths of innocent bystanders) and would hence likely would be in the tens of millions—in other words, anybody who could afford a drunk driving license already can afford a chauffeured limousine, a currently available alternative. As for the insurance option, it seems likely that the adverse selection problem would be so severe that the premium would also be more expensive than currently available legal options.

So are we stuck in the current second best (at best) solution?

1 That I consider this to be a hard case may convince some of my readers that this author is a "case" of quite another kind. But please hear me out.

2 Should any of them not be, kindly correct me in comments, preferably with citations—but please avoid claiming that I made any unstated assumptions or jumped to conclusions.

3 Yes, this author is aware that there are other, indeed many, laws which violate this principle. But in these other cases, the author is against those other laws because they violate this principle (or at least that is a shorthand for the author's opposition). Whether and why laws against "drunk driving" should be any different to this author (and those similarly inclined) is what makes this question interesting.

4 This assumes that the risk taker is rational. And, yes, the author is aware that there is evidence that some people in some situations do not act rationally; indeed, most people most of the time do not consciously reason the way the caricature of homo economicus does. Yet, this author is willing to defend in another post the proposition that within a liberal legal system and society, most mentally competent adults make most important decisions in their lives as if they were homo economicus. In short the rationality assumption only be true by introspection for small fringes of society (including many economists and fellow travelers), but nevertheless true as a matter of positive economics. And, specifically to the subject at hand, while it may be disputable whether an inebriated individual is fully rational, the person who decided to get inebriated generally would be.