partial reduction in underage drinking may confer benefit in proportion to the total. Thus, the study suggests that a 10 percent reduction would be worth about $53 per household, or 10 percent of the total cost per household. Of course, a complete analysis would require an assessment of the costs of achieving the reduction as well as the benefit; if the 10 percent reduction is achieved through a set of programs that cost $10 billion, then the net gain per household would be just $43.

There are two problems in doing this sort of accounting exercise well. First is the epidemiologist’s problem of discerning the actual consequences of eliminating (or reducing) underage drinking. What reduction would there be in highway crashes, crime, and school dropouts and in all the long-term effects of these events? All such consequences are the result of complex multicausal processes; knowing that there is alcohol involvement in some percentage of such cases leaves one far short of knowing the causal importance of drinking. A further complication is introduced by the realization that the actual effects of a reduction in underage drinking will depend not just on how much of a reduction is accomplished, but also on what sort of collateral consequences will occur. What effect will that intervention have on routine activities, such as weekend driving with friends, the use of other illicit drugs, or dating? The answers may be important in influencing the net consequences, which may well depend on the nature of an intervention to prevent or reduce underage drinking.

The second problem is to develop and implement a sound accounting system for translating outcomes into a measure of social burden. The choice of accounting rules in this context necessarily reflects decisions about deep issues in understanding the public good. Two specific issues are particularly thorny: whose preference should count in defining relevant consequences? Should the social cost computation include subjective losses or only production losses?

The presumption in our society is that the public good is the sum of individual preferences. A reasonable exception may be the preferences of teenagers, who tend to place too little emphasis on their long-term well-being and too much emphasis on pleasing their peers. That commonsense view of adolescent human nature, coupled with the fact that underage drinking is illegal, provides some justification for ignoring the pleasures of drinking as perceived by teens and accounting only for the harmful consequences.

Although few people think that the value of the life of someone killed or permanently disabled by a drunken teenager is limited to his or her lost earnings, that in fact has been the accounting rule used in the traditional