D. Welfarism and Neoclassicism: Some Differences
4. The Possibility of Paternalism?
Let me distinguish between weak and strong paternalism.219 Strong paternalism holds that persons can be made better off in the teeth of their preferences. That is, (1) P prefers S1 to S2, but nonethe- less (2) S2 can be better for P’s welfare than S1. Weak paternalism holds that persons can be no better off notwithstanding the satisfac- tion of their preferences. That is, (1) P prefers S1 to S2, but nonethe- less (2) S1 can be no better for P’s welfare than S2.220
Neoclassicism denies both strong paternalism and weak paternal- ism. This is easy to see. Neoclassicism incorporates an unrestricted preference-based view of well-being, which says that P is better off with S1, as compared to S2, just in case he prefers S1 to S2. If P does prefer S1 to S2, then—the neoclassicist concludes—he is better off with S1, not S2 (pace strong paternalism). The neoclassicist also con- cludes that this is a real welfare improvement, rather than a case where S1 is not better than S2 for P (pace weak paternalism).
Sen, The Discipline of Cost-Benefit Analysis, supra note 40, at 940.
218. This is overstated: an outcome could be better than another with respect to effi- ciency, while the second outcome is better with respect to X1, X2 . . . Xn, such that the out-
comes are NBNW with respect to the balance of X*, X1 . . . Xn and the agency has moral
discretion on balance. Nonetheless, it seems correct to say that if choices are rarely NBNW with respect to X*, it is less likely that they will be NBNW with respect to X*, X1. . . Xn.
219. For discussions of paternalism, see GERALD DWORKIN, THE THEORY AND PRACTICE OF AUTONOMY (1988); JOHN KULTGEN, AUTONOMY AND INTERVENTION: PARENTALISM IN THE CARING LIFE (1995); DONALD VANDEVEER, PATERNALISTIC INTERVENTION (1986); and
Joel Feinberg, Legal Paternalism, 1 CAN. J. PHIL. 105 (1971).
220. See Feinberg, supra note 219 (distinguishing between “weak” and “strong” pater- nalism). My distinction is quite different.
Welfarism denies strong paternalism, but it admits weak pater- nalism. This is pretty easy to see, too. Welfarism incorporates a so- phisticated preferentialist view of well-being. This says (placing to one side, for now, the issue of restriction) that P is better off with S1, as compared to S2, just in case (1) P prefers S1, and (2) P would pre- fer S1 under suitably idealized conditions. If P does prefer S1 to S2, then—even if he would prefer S2 under suitably idealized condi- tions—he does not actually prefer S2 to S1, and he therefore cannot be made better off with S2. This shows the inconsistency between so- phisticated preferentialism and strong paternalism. On the other hand, if P prefers S1 to S2 but would prefer S2 under idealized condi- tions, then P is not made better off by S1. This shows how weak pa- ternalism flows from sophisticated preferentialism.
An example may make the point clearer. P prefers sitcoms to op- era but would prefer opera under suitably idealized conditions. The public television station’s decision to show opera instead of sitcoms does not make P better off, because he actually prefers sitcoms. So the strong paternalist view that P is a beneficiary of the station’s choice of opera is a mistake. Yet the fact that P actually prefers sit- coms does not mean that he benefits if, instead of opera, the station chooses to broadcast sitcoms. After all, P’s idealized preference is in favor of opera. Sophisticated preferentialism gives equal weight to this idealized preference, along with P’s actual preference, in deter- mining what improves his welfare. For the sophisticated preferentialist, neither programming choice is better for P’s welfare.
P is not better off with the sitcom, nor is he better off with the opera, since the dual and conjunctive conditions for a welfare benefit— actual preference satisfaction plus idealized preference satisfaction— are not true of either option. Among other things, this dual and conjunctive account of welfare bears out weak paternalism.
What does this mean for regulatory agency policy? It means that the agency can (weakly) decide that some persons who believe them- selves affected by an agency’s choice are actually unaffected. Persons who are counted as “Winners” or “Losers” under neoclassicism may become “Neutrals,” neither benefited nor harmed by the choice, and therefore ignored for purposes of evaluating that choice in light of overall well-being. Consider, as a plausible example, the laws prohib- iting recreational use of drugs.221 Persons who prefer drug use must be counted as Losers from an anti-drug policy within a neoclassical framework,222 but they may well be seen as Neutrals by the welfarist 221. See generally DOUGLAS N. HUSAK, DRUGS AND RIGHTS (1992) (presenting detailed philosophical argument against prohibition of recreational drug use).
222. More precisely, certain persons who prefer drug use must be counted as Losers from an antidrug policy, within a neoclassical framework. It is arguably consistent with that framework to say that drug addicts (persons whose preference for drug use is effec-
(if she thinks they would disprefer drug use under idealized condi- tions).
It also may mean, a bit more robustly, that where actual prefer- ences are changing, agencies can rely upon idealized preferences as the sole criterion for determining an option’s welfare effect. Imagine that P now prefers S1 to S2, and that he will continue to prefer S1 if it is chosen by the agency; but he would change his mind and come to prefer S2 if that were chosen. Then, neoclassicism has no clear basis for determining which option benefits P. Here, it is neoclassicism which may well end up classifying P as a Neutral, while the welfarist could plausibly say that whatever option P would ideally prefer is the one that benefits him.223 In such a case (on a plausible specification of sophisticated preferentialism), P is a Winner with S1 if he ideally prefers S1, and P is a Winner with S2 if he ideally prefers S2.