Should Hosts Be Liable for Serving Liquor to Guests Who Cause Accidents While Driving under the Influence?--And Other Issues of Third-Party Liability--Posner
Some 40 percent of the roughly 40,000 fatal traffic accidents a year in the United States involve a drunk driver, but this figure is down from almost 60 percent 25 years ago, although the number of fatalities has not declined (unsurprisingly, since there are more drivers). I do not know what contribution if any "social host liability" has made to the decline. The term refers to laws (statutes or, less commonly, common-law--that is, judge-made--doctrines), now in force in a majority of states, that make an employer, a bar or restaurant or liquor store, or a purely social host, liable in tort for a guest's (employee's, customer's, etc.) injuring someone while driving under the influence of alcohol (or it might be drugs), if the host knew or should have known that the guest was drunk and would drive, yet served or continued serving him liquor. Actually, the liability of commercial establishments that sell liquor to obviously drunk patrons is nothing new; it has long been imposed by state "dram shop" laws. The extension to other hosts is novel.
Social host liability raises the interesting general question of when someone who has not inflicted an injury should be punished for having failed to prevent it. In general, such liability is not imposed. For example, it is not a tort to fail to rescue a stranger in distress even if you could do so without incurring any danger, or other significant cost, to yourself. I am not my brother's keeper. Or, as the courts say, there is no "good Samaritan" duty enforced by the law (it is different of course if one has agreed by contract to undertake a duty of rescue). One reason not to impose such liability is that it might induce potential rescuers to steer clear of any situation in which they might be punished for failing to attempt a rescue; so strong swimmers might avoid beaches at which poor swimmers congregate. In addition, adjudicating cases of third-party liability for failing to prevent an injury would often be plagued by uncertainty--who was actually in a position to attempt a rescue, what would the risk to him (or to them) have been, and how likely is it that the attempt if made would have succeeded?
These do not seem serious problems with regard to social host liability, even when the host is an individual rather than a bar. Of course, there may be tricky questions regarding whether the host should have known that the guest was drunk and would drive, but these are not much more difficult than other factual questions in tort cases. Still, it might seem superfluous to impose liability on a third party when there is a clearly liable second party (as often there will not be in a rescue case)--the drunk driver. (The victim is the first party.) If he is punished whether civilly or criminally (or both) should he cause an injury because of his drinking, why should a third party be liable too? It is only a partial answer that the injurer may be judgment proof (even if he carries liability insurance, as he may be required by law to do, his insurance may not cover the full extent of the harm that he causes), because criminal law enables severe punishment to be inflicted on judgment-proof wrongdoers. It is true that criminal punishment does not compensate the wrongdoer's victim, but people can protect themselves by first-party insurance from the financial consequences of being injured in a traffic accident. The focus of policy should be on deterrence and other means of prevention.
A possible explanation for social host liability is that the combination of tort and criminal sanctions for dangerous behavior is not thought an adequate deterrent. More than a million drivers are arrested every year for driving under the influence, and there are those 16,000 or so annual deaths in accidents involving a driver who is driving under the influence. These figures in themselves do not prove anything, because it is necessary to weigh any benefits of an activity that causes potential harms, even potential fatalities, against those costs. But there is skepticism that people who drink so heavily and uncontrollably as to become a menace to other drivers and court arrest and prosecution for drunk driving are making utility-maximizing judgments. Suppose we think they are more like children than competent adults, and agree that parents should be liable for their children's vandalism if knowing their children's propensity to vandalize they fail to take reasonable steps to control them. Then there would be a strong argument for social host liability, provided the costs are not excessive. It would be akin to accomplice liability for selling a gun to a person who one had reason to think would use it to commit a crime. One would not want in a case such as that to rely on the existence of heavy criminal penalties to deter the buyer from using the gun to commit a crime.
Another way to think about social host liability is to imagine that an off-duty policeman is at a bar and sees an obviously drunk person leave and get it into a car and start to drive off. The policeman could arrest him. In effect, social host liability makes bars and other hosts a kind of auxiliary police force, though conscripted, rather than hired, to prevent criminal activity.
Third-party liability is so common that it is hard to believe it has no economizing features. The doctrine of respondeat superior, which makes an employer liable for the torts committed by his employees in the course of their employment for him, is the most common example. It is more extreme than social host liability, because it does not depend on the employer's knowing or having reason to know that the employee was careless or irresponsible. The economic rationale is that we want the employer not only to be careful in selecting, training, and monitoring his employees, but also to consider substituting capital for labor inputs and making other adjustments that might reduce the incidence of employee torts; in other words, we want him to consider making changes in activity (his use of inputs, perhaps the scale of his production) and not just changes in his care.
All these are examples of collective punishment, a term that simply means threatening to punish those who fail to prevent a harm that cannot be as efficiently prevented directly. It is a question of fact rather than (I think) of principle whether in particular circumstances collective punishment is an efficient method of minimizing harm.