Drunk Driving--Posner's Comment

This topic could be thought a continuation of our last week's topic, the trans fat ban in New York City. We are again dealing with safety regulation. The case for punishing drunk drivers may seem clearer than the case for banning trans fats in restaurant meals because the externality is more pronounced and consumer competence is not in issue, but the appearance is deceptive. Becker's proposal for heavier penalties for drunk driving could be criticized as paternalistic, because it regulates an input rather than an output. If there are 1.4 million annual arrests for drunk driving, and if we assume realistically that this is only a fraction of the actual incidents of drunk driving, yet only 2,000 innocent people are killed by drunk drivers, then it follows that most drunk driving is harmless. Why then punish it with arrests and severe penalties? Why not just punish those drunk drivers who cause deaths or injuries to nonpassengers? In fact we do punish such drivers, under such rubrics as reckless homicide (if the victim dies) or reckless infliction of bodily injury. And the punishments are severe. Why punish the 99+ percent of drunk driving that is harmless? Indeed, if the penalties for reckless homicide are optimal, the implication is that the number of deaths from drunk driving, 17,000 a year, is also optimal. This is actually a plausible inference. If there are only 2,000 nonpassenger deaths (other than that of the drunk driver himself) caused by drunk driving every year (and how many of the accidents in which a drunk driver is involved are actually caused by the drinking?), then the probability of being killed by a drunk driver is very small, and the value of life estimate that I used in my post on the trans-fat ban should be usable here as well to conduct a cost-benefit analysis of drunk driving. The probability of a drunk driver's killing someone must also be small, given the number of drunk drivers implied by the arrest statistics. Suppose the annual probability that a drunk driver will kill a nonpassenger is .001 (as it would be, given the 2,000 victim figure, if there are 2 million drunk drivers, which is a very modest extrapolation from the arrest figure, since many drunk drivers are not caught). Then the expected injury cost from drunk driving is $7,000 (.001 x $7 million). (This corresponds to Becker's $10,000 figure, which seems to me too high, as it disregards the drunk drivers who are not arrested. Notice that if only a third of drunk drivers are arrested each year, the expected-cost figure drops to $3,333.) This implies that a driver who derives at least $7,000 in utility per year from drinking while driving (more commonly, shortly before driving) is behaving optimally and should not be punished at all. The larger issue that the drunk-driving question raises is the choice between ex ante and ex post regulation. Health inspections of restaurants and, yes, a ban on trans fats are examples of ex ante regulation. Such regulation prevents dangerous activity rather than waiting for the danger to materialize and using the legal system to punish the injurer. The tort system is an example of ex post regulation. If you drive recklessly but don't injure anybody, you have not committed a tort. Tort law comes into play only when an injury occurs. The theory is that an optimal tort penalty for the injury deters tortious conduct, not perfectly--or there would be no tort cases--but well enough. Criminal law is a mixed bag. Crimes that result in injury are punished, usually quite severely, but much preparatory conduct--attempts and conspiracies--is punished as well, even when no harm results (the failed attempt, the abandoned conspiracy). Arrests for speeding--and for drunk driving--are examples of ex ante regulation. The economic argument ex ante regulation is that ex post regulation is often inadequate. This is obvious in the trans-fat situation--it would be impossible to figure out which victims of heart disease owed the disease, and to what extent, to which restaurants. In the case of reckless homicide, the answer is less clear. Suppose drunk driving is inefficient--the drunk driver derives less utility than the expected accident cost--and so we want to deter it by punishing the drunk driver who kills or injures a nonpassenger. Suppose the value of life is $7 million and 10 percent of the drivers are not apprehended. Then the optimal penalty would be a fine of $7,780,000 ($7 million √∑ .9). Few drivers could pay that, so the trick would be to impose an equivalent disutility on them by nonpecuniary means, such as imprisonment. This is not to suggest that punishing drunk drivers who are arrested, the method that Becker endorses, can't achieve the correct deterrence. But punishing just the ones who kill might be more efficient--there wouldn't be as much need for policemen, there would be fewer trials and prison terms, and probably many drunk drivers are quite harmless, for it is unlikely that everyone who drives while drunk has an equal probability of causing an accident. In general, heavy punishment of fewer people is chaper than light punishment of more people. Thus, only if ex post punishment failed to deter optimally would there be a strong case for punishing drunk drivers who are not involved in accidents with nonpassengers (assuming drunk driving is inefficient, the assumption questioned earlier in this comment)--or at least a strong case based on the simple model of rational choice that underlies my analysis. Maybe drunk drivers systematically underestimate the effect of drinking on the likelihood of an accident, or believe that they can fully compensate for the danger by driving more slowly, or exaggerate the degree to which they can hold their liquor, or exaggerate their driving skill. Maybe their drunk-driving behavior is addictive, and they do not realize, before starting to drink, that they won't be able to avoid drinking before they drive. These might be grounds for ex ante regulation--even for regulations anterior to arrest, such as stiff alcohol taxes.

Comments

[Bjartur]

a side note. don’t know if it has been made before but on posner's comment "[drivers] believe that they can fully compensate for the danger by driving more slowly"

driving slowly although seemingly rational to avoid accidents when driving under the influence is also a signal to law enforcement that not all is well with the driver. In other words the officer might have greater reason to pull the driver over if he/she is going 30 on a road that allows 55 mph. as a result the intoxicated driver might be tempted to fake a normal driving style which is beyond his skills at that particular point in time -- increasing the likelihood of an accident. Although I can see some problems with such clauses the driving under the influence law could perhaps have escalating clauses based on for example speed. i.e. you are allowed more leeway in terms alcohol in the blood if you are caught at 10mph below the speed limit as opposed to 10mph above the speed limit.

A final note. Warmest season greetings and thanks for an entertaining blog.

BS/SKH

[Matt Rognlie]

Although I admire your contrarian style, I'm afraid that this post is almost a parody of rational choice economics.

Consider the central assertion: that punishing those who cause a fatal accident can be more efficient than punishing drunk drivers in general. It rests on the premise that potential drunk drivers will rationally calculate the probability of a deadly crash, match it against the expected "utility" from their actions and come to a logical decision. Yes, you qualify this assumption near the end of the post -- but in the process include a howler:

"Maybe drunk drivers systematically underestimate the effect of drinking on the likelihood of an accident, or believe that they can fully compensate for the danger by driving more slowly, or exaggerate the degree to which they can hold their liquor, or exaggerate their driving skill."

Maybe?

Really, if you go and ask anyone involved in drunk driving about the mathematical likelihood of an accident, you're likely to get an answer (if you get an answer at all) that is orders of magnitude off. Not just a poor estimation, but a fundamentally wrong one that obliterates the cost/benefit calculus you suggest. (This is a testable hypothesis -- I suggest that curious souls enter a few drinking establishments and try it!)

Now, if you ask about the likelihood of being caught by police, you're not likely to get an accurate answer either -- but it will probably be closer to correct. And regardless, the key isn't the precise number, but the fact that there's a much higher chance: people usually know others who have been nailed with a DUI, making the idea of risk infinitely easier to grasp. Ex ante punishment isn't superior here because it more effectively taps into some abstract model of rational choice, but because it better remedies widespread poor decision-making.

[Matt Rognlie]

I should add that this question mystified me:

"And how many of the accidents in which a drunk driver is involved are actually caused by the drinking?"

Um... almost all of them, as can be immediately inferred from the numbers. Instances of drunk driving comprise only a small fraction of total vehicle miles-driven, but cause an enormous percentage of total fatal accidents. This implies that the conditional probability of a fatal accident while drunk is far, far higher than the correponding probability for driving in general. And this means that, indeed, most accidents involving a drunk driver were caused by drinking.

This seems clear enough that I wonder whether I'm misinterpreting your question. Does it mean something else?

[robert]

The use of a technology whereby the driver cannot start the vehicle until he blows into a breathalyzer-type device and the device ascertains that he/she is not drunk, may be the soundest way by which to deal with this problem. The criminal justice system could order its use for non-fatal drunk driving offenders, and have the offenders pick up the cost for its purchase and maintenance.

[Lawrence Indyk]

It's too bad there was no discussion of the latest issue in drunk driving which is the proposal for optional or compulsory installation of engine-lock "breathalyzers". Some communities have begun trials of requiring convicted drunk drivers to place such devices in their cars, and I believe New Mexico, which has an intense drunk driving problem, has investigated universal compulsory installation somewhat in the spirit of mandatory safety device laws requiring air bags and electronic tire pressure monitoring.

Assuming that most potential drunk drivers would or could not circumvent the system by using compressed air or getting "sober help", the potential life-saving consequences could be enormous at a small (relative to the cost of an automobile) expense for the device and a greatly reduced need for the enormous dedication of law enforcement resources or for inefficient taxes.

Drunk drivers would either be directly deterred by the inability to start their vehicles or indirectly deterred from getting heavily inebriated by the prospect of a long wait in the cold to sober up. They need not be deterred by the punishments of fines and incarceration, and would not need to compensate the system for injuries which never occur.

The question remains, however, if the total cost of all the devices would exceed the total benefit. I have not come across any economic analysis - and I would imagine the law enforcement benefits would be difficult to estimate. Furthermore, the addition of the device to the cars of the many reliably good drivers could operate similarly to taxes on alcohol or driving that catch innocents.

Then again, my experience has been that even reliably good drivers will occassionaly drink too much at a bar or party, or lend their car to their children who might do the same and cause a rare, but tragic, accident. And despite the fact that they are statistically more likely to cause an accident, chronic drunk drivers may be too few in number to cause the bulk of fatalties - so perhaps even an "innocent driver" could benefit more from a lock than its cost if it prevents that one tragic mistake from happening.

[KipEsquire]

It seems to me that Judge Posner is arguing against the very propriety of "reckless endangerment" as a broad category of crime.

Whom do I harm if I let my large dog run around public spaces off leash? So why have leash laws at all and not just punish post facto dog bites instead?

Whom do I harm if I frolic down the street brandishing a loaded firearm? So why make that an offense unless and until the weapon accidentally goes off and harms person or property?

Whom do I harm if I drive drunk? Same analysis, same absurd result.

Becker is right: it's all about the externalities.

[David]

I think that drunk driving is a very important issue that receives too little attention. The harm caused by drunk drivers consists not just of the fatalities (something like 50% of all traffic deaths), but also injuries and property damage. If we eliminated (or nearly eliminated) drunk driving, we could probably reduce the toll of traffic accidents by half. That would be a great benefit to society.

The problems in fighting drunk driving are several. First, drunk driving (if it doesn't result in serious injury) is too sporadically and too lightly punished. Moreover, there is too little social stigma. If a DWI conviction carried the same social stigma as, for instance, a shoplifting conviction, I think we would see far fewer drunk drivers.

Enhanced penalities and stepped-up enforcement are good ideas, but they only go so far. The public should be educated (more so than they already are) about the dangers, and more efforts should be made to provide alternatives to drinking and driving. Unfortunately, taxis and public transit are virtually non-existent, or unused, outside a handful of the nation's largest cities -- probably because they are either inconvenient or unduly expensive.

Investment in good, public transportation systems would reduce drunk driving and also help in other ways, by reducing pollution and street congestion, and by conserving gasoline. In other parts of the world, nearly every medium-sized city has an effective subway system. Bringing public transit to more of America would be a good start to solving this and other problems.

[MaxWilly]

I think that drunken drivers is a very big problem in a world today. And its very important to explain peoples that they should not drunk, or they should not drive.

[guy in the veal calf office]

The United States has almost 40,000 deaths per year from automobile accidents, and about 40 percent are due to drunk driving.Not according to the NHSTA study, which included as "alcohol related fatality" every incident in which a driver or motorcyclist had a BAC above .08. Note first that person who caused the fatality was not considered. Second the study says that 40% (16919 of 42,836 accidents) were "alcohol related." At the next Congressional hearing on speeding the special interest analog to MADD will be touting the fact that one third of accidents are "related to speeding." (See e.g., http://safety.fhwa.dot.gov/speed_manage/facts.htm).All the posters suggestions may be valid, but the inaccurate and unquestioned recital of a self-interested bureaucracy's facts should always be avoided because the consequence is terrible "public health" based legislation.

[Jake]

The idea of automobile ignition locks keyed to a breathalyzer device has promise. But consider the following scenario. Jones, a responsible citizen who enjoys a social cocktail, but would never drive while drunk, goes to a restaurant and orders a martini while waiting for his meal. Right after finishing the martini, but before getting his meal, Jones gets a call on his cell phone from a family member who is in an emergency situation. Unfortunately for Jones, when he leaps into his car, the compulsory breathalyzer concludes that he should not drive. A member of Jones' family suffers greatly, or even dies, due to his inability to come quickly to the rescue.

Is this reasonable and effective regulation?

Ex ante regulation of drunk driving will fail for the same reasons that Prohibition and the Soviet Union did.

We would do far better to limit the issuance of drivers licenses to persons who can prove that they passed, at a minimum, a high school physics course that might have taught them the potential consequences of piloting a 5,000 chunk of steel around their neighbors. Such a regulatory policy would also have the salutary effect of freeing up congested roads, for drivers who actually understand such matters are a small minority. And limiting driving privileges to better educated persons, who are better equipped to logically foresee the consequences of their actions, should reduce the number of drunk drivers.

Of course, ending the widespread policy of handing out drivers licenses like water from public fountains is hopelessly unpopular. Thus it will never come to pass.

If we want to hand out a drivers license to any idiot who wants one, we cannot complain very much about idiots who drive drunk.

[dWj]

The $3,333 figure is relevant only if you are going to fine those drunk drivers that you don't catch. A $10,000 fine is appropriate -- if the probability of a drunk driver being caught is P, the deterrent effect will be ($10,000)*P, and the costs we infer (knowing that we only catch drunk drivers P of the time) works out to ($10,000)*P.

This assumes the marginal cost is the same as the average cost, but you both seem happy with that assumption in your posts, and I'm not particularly averse to it. (If the density of drunk drivers gets high enough, innocent people start modifying their behavior so as not to become victims; I would guess costs associated with that aren't included in the $10,000, but, more to the point, they are more likely to increase faster than linearly with the amount of drunk driving, suggesting a higher marginal cost than average cost.)

[Jack]

Looks like America's rapidly increasing inequality plays a hefty role here as well.

OK, so we use some abstract construct to devise the "perfect" monetary fine of $10,000. (Is it an accident or an oversight that the penalty is primarily monetary and not time or blow machines in vehicles?)

Getting gigged for $10,000 is hardly more than cause for gentle ribbing over the next round of drinks at the yacht club, (I've seen it.) a tough swat for those at median household income of $45,000 who are saving for kid's college, paying a medical bill, or hoping against the odds to retire someday, and a crushing blow to those 38 million living at or below the $10,000 "poverty line" that's unlikely to be paid.

Likewise the solution of taking a cab home is a no-brainer of little cost to the yachtie, a decision to be made in a foggy state of mind for the median guy and out of the question for the poverty level and most below median. (Yes Phyllis, we know that drinking has been added to the list of unaffordable luxuries and the poor shoudn't drink unless they can afford a cab home and be "accountable for their actions" but then there exists the real world where they do drink and have not the resources to be "accountable".

Still, we arrived at this societal "cost" of perhaps $10,000 so if we're going to try to run America at these and even higher levels of inequality, perhaps we'd copy Denmark? or Sweden? with a percentage of income approach. The $10,000 would be 20% of the median household income, but, at that level discretionary income is next to zero so perhaps $5,000 and ramping up progressively so the $250,000 household would feel the pinch of $50,000? Wait! Upper income policy wonks where are you going?

In the current environment monetary penalties scaled to income aren't very likely. How's that for an understatement? So, perhaps we need to know more about those who DUI, where they drink, why they end up driving after, etc. and then try to design our disincentives, educational programs and pro-active solutions accordingly. Jack

[Bruce M]

Matt Rognlie: You ponder the following: "And how many of the accidents in which a drunk driver is involved are actually caused by the drinking?" You assume that just because someone is drunk and causes an accident that the intoxication is always the cause of the accident. In other words, you assume that but for the alcohol intoxication, the accident would not have taken place. That is a fallacious assumption, even though people are convicted on it every day in courts across the country. Perfectly sober people cause accidents through negligence all the time. Maybe the person was drunk, talking on a cellphone, and yelling at 3 kids in the back seat who are throwing balls and screaming. Why do you assume the alcohol, and only the alcohol, contributed to the accident?

What's even more frustrating is when a drunk person and a sober person have an accident, and it is presumed that it is the intoxicated person's fault merely because the other person was not drunk. Sober people negligently crash into drunk people all the time. Why should the person with ethanol in his blood be deemed responsible? In such intoxication manslaughter/assault prosecutions, the statute (at least here in Texas) requires the state to prove that the defendant was intoxicated and caused serious bodily injury/death while driving a vehicle which was CAUSED by the intoxication. They NEVER prove that causation element, courts of appeals never require it with respect to legal sufficiency appeals ("he was drunk, there was an accident, therefore he caused it due to his intoxication").

What many people neglect to realize is that DWI is a political crime. MADD has people sit in courtrooms across the country to make sure prosecutors/judges are not lenient on DWI defendants (they cause a hissy fit when they are with mass mailings, etc). Many other activities just as, if not more, dangerous than DWI are not only perfectly legal, but not even advised against. DWTOC (driving while talking on cellphone), DWF (driving while female), DWYM (driving while young male), DWE (driving while elderly), DWR (driving while religious... the more jesus stickers on a car the more banged up it is, which is good evidence that religious people are bad drivers, particularly those who are more busy praying than watching the road), DWS (driving while stressed-out), DWTM (driving while text messaging), DWWDVD (driving while watching DVD), and DWHS (driving while having sex). Why is none of these activities illegal? I'd rather be driving next to a 35 year-old guy who had 4 beers with lunch 6 hours ago than next to a 22 year-old sorority girl talking on her cellphone. (I'd rather be driving next to the 22 year old sorority girl on her cellphone than next to a 95 year old geriatric, and I'd rather be driving next to the a car with a "Darwin fish" on the bumper than next to a car with a "Jesus fish" on the bumper).

Out of all the dangerous and unnecessary activities people are currently allowed to partake in while driving, why is only DWI (certainly not the most dangerous) unlawful? Politics.

And I do not accept for one minute the argument that society has determined the social utility of having the right to do all of these non-DWI activities outweighs the danger caused by them, but only the danger of DWI is outweighed by the social utility in having the right to DWI.

[john]

What is the object of the sanction: deterrence, retribution?
If retribution is the object - stiff sentences of incarceration are appropriate, if also quite expensive (given direct costs of incarceration, and the indirect costs associated with interrupting people's lives in such a drastic way).
If deterrence is the object - rehabilitative sentences, including those already suggested like ignition interlock devices, auto forfeiture, electronic monitors, and administrative incentives such as conditional licensing under stated conditions, are preferrable.
Drunken driving, like shoplifting and assaultive conduct, seems to be pretty compulsive behavior that is resistant to progressive sanctions. The prisons are filling up with people who are unable to maintain sobriety, and who insist upon driving. Given that 4,000,000 US citizens are now in jail, prison, or under court supervision, incarceration may be a sanction that is too costly to consider for this type of behavior - until a point of high recidivism is reached.

Although some drunk drivers are first timers, the large majority of offenders have an established dependence. Perhaps sanctions ought to be tied to the level of dependence, as established 'post ante'. Accurate assessment tools exist. Why spend the money to incarcerate someone celebrating a wedding or graduation, who does drink for the first time, drive, and do harm? This might be a more cost effective way to address the problem, rather than by grading offences upon blood alcohol levels, numbers of convictions, the presence of property damage or personal injury or death, etc...

[Bernard Yomtov]

Bruce M.,

You misunderstand Matt's argument. He is not assuming anything. He is making a simple deduction. If, say, 1% of drivers are drunk, but 10% of accidents involve drunk drivers, it's perfectly logical to argue that the vast bulk of accidents involving drunks are caused by the fact that they are drunk.

Maybe you could disprove this, by showing that drinkers are especially accident-prone even when sober, but it's not easy. Just citing a few posible exceptions is meaningless.

[Bernard Yomtov]

I also agree with Matt and Alex that much of this reads like a parody of rational choice. Surely, as Kip implies, there are behaviors that are so unjustified that the utility derived ought not carry any weight. Yes. This is paternalistic, or puritanical, or judgmental, or something. So what?

But even leaving this aside, I question the usefulness of such off-the-wall numbers as a drunk driver's utility as a guide to policy-making. What is this “utility?” You can take a cab home, and sometimes you will need to take another back to get your car the next day. What’s that? Fifty dollars per occasion, if that, and a big part of that is just transferred to the cab driver. (And we should be talking about “net utility” anyway- with a reduction for the risk assumed) So, no more nonsense about utility, please.

As for the question of ex ante punishment, I think Posner overlooks an important point. We want to reduce the number and duration of drunk driving incidents. An arrest gets a drunk driver off the road. If we assume that the 1.4 million arrests came in the middle of the drivers' trips on average we have cut the duration of 1.4 million trips in half, with some consequent reduction in accidents.

This may seem small, but Posner makes the mistake of assuming that arrest and accident risks are uniformly distributed across drunk drivers. Not so. it is likely that the most dangerous drivers are the ones most likely to be arrested. The drunker the driver, the more dangerous, and the more likely to drive erratically enough to attract police attention, so arrests probably prevent a disproportionate number of accidents.

Also, arresting drunk drivers creates a class of drivers - those with previous arrests - who face harsher penalties for drunk driving than others, and are therefore more heavily deterred. Again, it is likely that these are, on average, the more dangerous drivers. The chance of being arrested rises, obviously, with the frequency with which one drives drunk, as well as the degree of drunkenness. It is thus fair to say that those with previous arrests, facing harsher penalties, are more dangerous on average than others, and ought to be more heavily deterred.

So to me the ex ante policy looks pretty sensible.

[Ken Arromdee]

"If, say, 1% of drivers are drunk, but 10% of accidents involve drunk drivers, it's perfectly logical to argue that the vast bulk of accidents involving drunks are caused by the fact that they are drunk."

That's true for your very precisely worded hypothetical, but not for several close variants that seem more like what people are really claiming.

For instance, consider this similar argument: "If, say, 5% of drivers are drunk, but in 10% of accidents a drunk driver is at fault, then..." Your conclusion would not be correct in that case, since there are laws which automatically consider the drunk driver to be at fault, and this could inflate the percentage of at-fault accidents from 5 to 10 percent.

[Bruce M]

Thank you Ken, you saved me a paragraph's worth of typing. Also, not all the 9% of drunk drivers involved in accidents are the cause of the accident, nor can it be said that 100% of the 9% of drunk drivers who cause accidents caused the accident solely by virtue of their being drunk (i.e. some percentage of them would have caused an accident or been involved in an accident even if they had been sober).

Implicit in my logic is giving a drunk person involved in a (possibly fatal) car accident the benefit of the doubt. This does not come naturally.

[Bernard Yomtov]

Ken,

I'm not sure I understand your objection. If 5% of drivers are drunk men, and 10% of accidents involve drunks, isn't fair to conclude that drunks get into more accidents than average, even if they are not really 100% their fault?

Bruce M.

The argument is a statistical one. Just because there are exceptions doesn't invalidate it.

[Bruce M]

Bernard: It sure does if you're a prosecutor using that statistical argument to convict someone for intoxication manslaughter beyond a reasonable doubt (which means proving the intoxication caused the accident beyond a reasonable doubt). You might be able to prove it by a preponderance of the evidence with such statistics (maybe) but not BARD.

[Bernard Yomtov]

Bruce M.,

I don't disagree with you as regards specific cases. My comments, and I gather the subject of the post, has to do with setting policy so as to prevent or deter drunk driving.

If it were shown that drunk drivers were no more likely than sober ones to cause accidents this would not be an issue at all. But they are, as any sort of reasonable look at the statistics shows. So for policy purposes we need to take them seriously. That does not mean that the statistics are, or ought to be, sufficient for a conviction.

[Jon]

Drunk-driving and other alcohol-related fatalities are preventable, but so too are many non-alcohol related driving fatalities. A cost-effective prevention policy would target both types of fatalities, but the popular press seems focused mainly on alcohol-related fatalities.

The following NHTSA statistics are from the web site: www.alcoholstats.com.

1982: 43,946 crash fatalities, including pedestrians (17,773 non-alcohol related, 26,173 alcohol-related with a BAC of .01 or above) – 61% are alcohol-related

2005: 43,443 crash fatalities, including pedestrians (26,558 non-alcohol related, 16,885 alcohol-related with a BAC of .01 or above) – 39% are alcohol-related.

The number of alcohol crashes has fallen substantially in the past 24 years (as has per capita alcohol consumption), but you would never know that if you read the newspapers.

I do not know how many non-alcohol crashes involve deaths of other persons besides the driver and occupants of the vehicle at blame, but it seems likely that it is at least equal to that number killed by drinking drivers. Alcohol-related crashes have fallen substantially due in part to a higher minimum legal drinking age and better enforcement of existing laws. Targeting repeat offenders and those with very high BAC levels are likely to produce some additional reduction in fatalities. Indeed, the average BAC level among drunk drivers in 2005 was 0.18.

However, the rise in the number of non-alcohol related deaths is striking, especially given the improvements in vehicles and highways since 1982. The rise in driving speeds must be partly to blame as well as the sheer amount of driving that Americans do. I have driven in Germany and Australia in the past two years. My general observation was that drivers in these countries were much better at obeying the posted speed limits, maintaining safe driving distances, etc. This seems like another example (like obesity) where many Americans fail to take low-cost responsible actions, and rather blame problems on someone else (such as the fast food industry or the alcohol industry). A proper role of economics is to determine the optimal mix of alcohol- and non-alcohol related crashes. The data suggest that the actual balance has shifted.

[guy in the veal calf office]

Bernard:

The internal logic of your comment is sound, but it doesn't really shed any light on the discussion for me, anyway. To get a real assessment of externalities, I think we need a better grasp of, for example, what is a good proxy for "drunkeness" (a .08 BAC is not, it represents a powerfully lobbying groups version of "zero tolerance" for drinking and driving, not for drunken driving). Secondly, the collection of statistics are questionable and driven by obvious bias.

Also, people, including Professor Becker, are using "cause" while the NHSTA uses "related" and those are definitely different concepts.

As previsouly pointed out, when an accident occurs with one person having at least a .08 BAC, then it is categorized as alcohol related. If speeding was involved, it is also categorized as speeding, and if weather involved, its is also weather related. And so on. You have "relation" inflation. Now each of those might be a contributing cause, but then it should be parsed: 40% to drink, 20% to leering, 40% to being in a drag race.

Also, the policeman providing the statistics might exclude speed, being stupid, gawking at fetching physical specimens, but he never will exclude alcohol (for third parties), because that is a legal decider. A .08 is an overrepresented relation.

In addition, why exclude some relations? Driving slow, stupid, or on a cell phone, are "relations" that are not collected. They don't have an interest group yet.

If only a single cause was assigned (or many causes apportioned into a single number), and the eligible causes were more numerous, the difference in your 5% and 10% numbers would reduce.

For example, a person going 30mph (or 150mph) on an interstate highway is probably going to cause an accident, regardless of his having a single cocktail. That should not count in the drinker numerator.

This brings up the fact that a .08 BAC is only randomnly associated with being drunk in a motor skills sense. Better drivers can stomach more before having their skills reduce to the level of the truly terrible drivers.

So, granted your logic is impeccable, but not very useful to an accurate discussion of drunk drivings externalities.

Incidentally, I could knock back a bottle of absinthe and out-drive any non-taxi driver from New York. They are a menace.

[Jack]

Whew! Any study of statistics and human behavior can get complex and confusing very quickly!

Guy in Veal: Interesting numbers and conclusions, and yes what appears to be a rise in non-alky related fatalities is due to more people and more driving. Our fatality rate per million miles has been dropping for 40 years but seems stalled out recently. At a cost of $800 per capita we'd be well advised to add cutting it in half in ten years to our lengthy list of needed "Apollo programs".

Interestingly while further reductions in alky related still seems to be the low hanging fruit, perhaps increasing seat belt usage is most beneficial. Hard to know the numbers today but perhaps?? 70% always buckle up? and more some of the time? Just for comparison with stats below.

"According to NHTSA, for every percentage point increase in seat belt usage, 280 lives can be saved. MADD knows that the best defense against a drunk driver is a seat belt. The fact is, of those killed in alcohol-related traffic crashes, 76 percent were not wearing their seat belt. Had they been, a significant portion of them would be alive today."

"Drunk drivers typically do not buckle up, nor do they make sure their passengers are properly restrained. The sad fact is that two-thirds of children killed in alcohol-related crashes are passengers driven by an impaired driver. We also know that seat belt use for children generally decreases the more impaired a driver becomes."

"While higher-risk drivers are a small portion of the population, they pose a significant threat to innocent motorists. On a typical weekend night, only one percent of drivers have a BAC of .15 or higher, but high BAC drivers were involved in over one-half of all alcohol-related traffic deaths in 2000."

Guy; it is as you say that the higher BAC's are most closely related to fatalities but the .08 and below create a lot of less deadly accidents as well.

"And, about one-third of all drivers arrested or convicted of DUI are repeat offenders."

J: So we do need to find ways of behavior modification: From below, it looks as though ignition interlocks and electronic home arrest offer the most bang for the buck and avoids a lot of the problems of higher monetary penalties.

"Economic losses due to motor vehicle crashes cost the nation approximately $230.6 billion each year, an average of $820 for every person living in the United States."

Point .08 OK? I checked this calculator for my 180 pound weight and found that (in theory) I could slam down four gin and tonics and head for the highway, and would have to grab a fifth "one for the road" to ensure I'd be over .08 by the time I was busted. The calculator doesn't scale for age but in middle age I'm fairly sure that after the five drinks in an hour I'd need help getting to my car much less being roadworthy. Give it a try:

http://www.ou.edu/oupd/bac.htm

Driving while Republican??

This one is fatalities per state and a bit tongue in cheek, but barely. While the lowest 4 blue states of the N-E have the advantages of a lot of public transportation, why would OK have twice the highway deaths/million as WA? Or SC have more than twice the fatality rate as motor-mad CA? And what would MS an WY have in common to top the list with three times the fatalities as WA or OH?

http://www.driveandstayalive.com/info%20section/statistics/stats-usa_indiv-states_per-capita_2003.htm#table-1

Best bang for the buck?

Ignition Interlock: Breathtesting ignition interlocks are designed to prevent anyone with a positive BAC from starting or driving a car. Attaching an interlock to a car for a year after its operator is convicted of driving while intoxicated would reduce recidivism by an estimated 75% and alcohol-related fatalities by 7%. It would save almost $7,900 per vehicle equipped. Including equipment and case management costs, interlock costs would total approximately $950 per vehicle.

Electronically Monitored House Arrest
Implementation of this program decreases recidivism by an estimated 31%, causing DUI crashes to decrease by about 3%. Per person arrested, the program would cost nearly $1,400 and could avoid an estimated $3,400 in crash costs and almost $1,800 in incarceration costs.

Intensive Probation Supervision with Treatment: Intensive probation supervision with treatment is an alternative to incarcerating repeat offenders. This early intervention program seeks to reduce alcohol-impaired driving by addressing repeat offenders’ drinking habits and provides intensive individual counseling and monitoring. Implementation of this program decreases recidivism by an estimated 48%, causing DUI crashes to decrease by 4%. Typically, per person arrested, this program costs approximately $1,200 and can avoid an estimated $5,300 in crash costs and $500 in incarceration costs.

Child Safety Seat Law: Infants and children who are seated in places other than the back seat account for nearly 34% of child fatalities in the United States and those seated in the back seat without proper restraints account for an additional 32% of fatalities. Drinking drivers are more likely than other drivers to transport children improperly. Traveling in a child seat reduces the chance of a crash death by an estimated 71% for infants and 54% for children age 1–4.

J: How 'bout it? A well publicized, automatic 10 days in the clinker for first offense DUI with minors aboard on top of all other penalties?

The CDC found that sobriety checkpoints can reduce impaired driving crashes by 18 to 24 percent. These checkpoints are especially effective when coupled with media campaigns that raise the visibility and awareness of drunk driving enforcement efforts in the community with the bottom line goal of deterring impaired driving before it happens.

J: I hate the idea of checkpoints, but 20%???

My guess it that selecting some or all of these measures would do much more than increasing the severity of penalties without increasing the odds of getting caught. Not sure what to do about the Red state problem. Any ideas? Jack

[Matt Rognlie]

Bruce M, I understand that you'd need an entirely different level of proof to show, positively, that an individual case of drunkness caused a particular accident. But that's not what I'm doing. As Bernard noted, this is an entirely statistical argument: it works because when we're dealing in aggegrate numbers, there's no need to prove causation beyond a reasonable doubt for any particular case.

And it's pretty simple. The numbers betray an extremely strong relationship between the state of being drunk and the occurence of automobile accidents. Now, you could theoretically quibble with this by saying that it's only a "correlation." But absent a causative mechanism, I can't see how you could possibly explain the bulk of this correlation.

Sure, as Bernard also mentioned, you could say that the kinds of people likely to be drunk are simply more accident-prone in general. But this would be have to be an extremely strong relationship to even compare to the size of the statistical "alcohol effect." You'd have to make heroic -- and frankly implausible -- assumptions that drunk driver types are hundreds of times more likely to get in accidents even when sober. And this is just to make a dent in the statistical death toll of drinking.

For the vast bulk of the statistical correspondence between drinking and accidences, there simply isn't any non-causative explanation. And this implies, mathematically, that almost all accidents involving alcohol were indeed caused by drinking.