The possibility of a fair Paretian.

• Author: Chang, Howard F.
• Position: Response to article by Steven Shavell and Louis Kaplow in this issue, p.237

In my preceding article, I respond to the claim that the weak Pareto principle implies welfarism.(1) Louis Kaplow and Steven Shavell make this claim in a series of recent papers.(2) In my article, I have consistently adopted the definition of welfarism that is standard among social choice theorists.(3) Under this standard definition, welfarism "restricts the information that can be utilized in ranking social states to utility information corresponding to those social states."(4) In their prior writings, Kaplow and Shavell have indicated that they intend to refer to this same concept.(5) They contrast welfarist theories with notions of "fairness," which "are based, at least in part, on factors unrelated to individuals' well-being" such that "knowledge of the effects of a legal rule on each individual's well-being will not generally be sufficient to assess the fairness of the rule."(6) Kaplow and Shavell use the terms "well-being" and "utility" as synonyms.(7) They claim that "any conceivable notion of social welfare that does not depend solely on individuals' utilities" will sometimes violate the Pareto principle.(8)

In my article, I show that this claim, expressed without qualification, as Kaplow and Shavell state it, is false. I present a "conceivable notion of social welfare that does not depend solely on individual utilities" that complies with the Pareto principle. Under either F* or F**, the rankings of alternatives will depend on information other than individual utilities. This fairness information may include, for example, the source of the utility produced by each alternative, which would be relevant if our fairness criterion F is liberal consequentialism.(9) This information is necessary under F* or F**, because the rankings under each theory require the identification of the fairness optimum to partition the set of alternatives. If we change the fairness information corresponding to each alternative, then even if the utility information corresponding to each alternative remains unchanged, the fairness optimum changes, and thus so do our rankings.(10) Therefore, the rankings are a function of information other than individuals' utilities.

Thus, contrary to the claim in the title of one of their forthcoming pieces, Kaplow and Shavell do not show that "Any Non-Welfarist Method of Policy Assessment Violates the Pareto Principle."(11) They do not (and indeed, as I have shown, cannot) show that "any" such theory violates the Pareto principle. They show only that some non-welfarist theories violate the Pareto principle. In particular, they can show only that non-welfarist theories that comply with independence and continuity conditions over an unrestricted domain must violate the Pareto principle.(12)

In their reply to my article, Kaplow and Shavell deny that either F* or F** is "a Pareto-conflict-free notion that is not solely based on individuals' well-being."(13) The basis for their claim is obscure, but Kaplow and Shavell apparently seek to expand their definition of welfarism to include F* and F** and accordingly add new restrictions to their definition of fairness. In their reply, Kaplow and Shavell suggest for the first time that their notion of welfarism would allow rankings to depend on more information than just individual utility levels. First, they assert that rankings under their framework could depend on information about individuals, so that social choices "could give more to Joe because he is tall and less to Jill because her preferences are objectionable."(14) Even assuming that the definition of welfarism allows such information about the characteristics of individuals to be morally relevant, this concession is still insufficient to bring F* and F** within the definition. If F is liberal consequentialism, for example, then F* and F** require not information about individuals but rather information about the effect of each alternative on the type of utility enjoyed by each individual. Under either F* or F**, it is insufficient to know that Jill has objectionable preferences. We must also know the degree to which each alternative satisfies her personal preferences.

Perhaps Kaplow and Shavell now intend to include such information about alternatives within their framework. They suggest, for example, that they do not intend to include any fairness regarding distribution in their definition of fairness.(15) In their prior writing, however, they exclude only "some" distributive concerns, namely those based "solely on individuals' well-being."(16) They contrast these welfarist concerns with "theories bearing on the just distribution of income ... not based solely on individuals' well-being," which "can be shown sometimes to ... conflict with the Pareto principle."(17) Indeed, they claim specifically that their formal proof of a conflict between the Pareto principle and fairness "encompasses distributive theories that are not based only on individuals' utilities" as subject to their claim."(18) Thus, because the distributive fairness in F* and F** is based not only on individuals' utilities but also on other information, they are among the fairness theories supposedly subject to their claim regarding a conflict with the Pareto principle.

Kaplow and Shavell can bring F* and F** within their definition of welfarism only by modifying that definition to allow in any fairness information that can affect the distribution of utility. After all, the fairness criterion F could give weight to any fairness notion criticized by Kaplow and Shavell, including corrective justice and retributive justice.(19) I gave liberal consequentialism as only one example of a fairness theory F. If F instead is a theory that gives weight to notions like retribution and corrective justice, then so will F* and F**, which make the fairness optimum under F relevant for ranking alternatives. Thus, F* and F** would allow in considerations that Kaplow and Shavell specifically seek to exclude from our analysis of legal rules. If they seek to define welfarism this broadly, then they have departed radically from the standard definition and left themselves with a notion of welfarism...