A liberal theory of social welfare: Fairness, utility, and the Pareto principle

Yale Law Journal

Volumn 110, Issue 2, 2000, Pages 173-235

  Chang, Howard F ^a  

^a UNIVERSITY OF PENNSYLVANIA LAW SCHOOL   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 0041886887     PISSN: 00440094     EISSN: None     Source Type: Journal    
DOI: 10.2307/797571     Document Type: Article

References (241)

1. Amartya Sen, The Impossibility of a Paretian Liberal, 78 J. POL. ECON. 152 (1970). The Royal Swedish Academy of Sciences cited Sen's work on this issue as one of the contributions justifying its award of the 1998 Nobel Prize in Economics to Sen. See Royal Swedish Acad. of Scis., http://www.kva.se/eng/pg/prizes/economics/1998/ecoback98.pdf (last visited Sept. 24, 2000).
2. Amartya Sen, Liberty, Unanimity and Rights, 43 ECONOMICA 217, 235 (1976).
3. E.g., Louis Kaplow & Steven Shavell, The Conflict Between Notions of Fairness and the Pareto Principle, 1 AM. L. & ECON. REV. 63, 65-66 & n.5 (1999).
4. LOUIS KAPLOW & STEVEN SHAVELL, ANY NON-INDIVIDUALISTIC SOCIAL WELFARE FUNCTION VIOLATES THE PARETO PRINCIPLE 1 (Nat'l Bureau of Econ. Research, Working Paper No. 7051, 1999),
5. forthcoming as Louis Kaplow & Steven Shavell, Any Non-Welfarist Method of Policy Assessment Violates the Pareto Principle, 109 J. POL. ECON. (2001).
6. Kaplow & Shavell, supra note 3, at 76.
7. E.g., Amartya Sen, Utilitarianism and Welfarism, 76 J. PHIL. 463, 468 (1979) (defining "welfarism" as the view that "[t]he judgment of the relative goodness of alternative states of affairs must be based exclusively on, and taken as an increasing function of, the respective collections of individual utilities in these states"). Thus, welfarism rejects the relevance of nonutility information. Infra note 22 and accompanying text.
8. Kaplow & Shavell, supra note 3, at 65-66 n.5.
9. Id at 65 n.3; Sen, supra note 1, at 153.
10. Amartya Sen, Social Choice Theory, in 3 HANDBOOK OF MATHEMATICAL ECONOMICS 1073, 1075 (Kenneth J. Arrow & Michael D. Intriligator eds., 1986).
11. Amartya Sen, Personal Utilities and Public Judgements: Or What's Wrong with Welfare Economics?, 89 ECON. J. 537, 549 (1979).
12. Sen, supra note 2, at 235.
13. Kaplow & Shavell, supra note 3, at 64.
14. Id. at 73.
15. J.J.C. Smart, An Outline of a System of Utilitarian Ethics, in J.J.C. SMART & BERNARD WILLIAMS, UTILITARIANISM: FOR AND AGAINST 3, 6 (1973).
16. LOUIS KAPLOW & STEVEN SHAVELL, PRINCIPLES OF FAIRNESS VERSUS HUMAN WELFARE: ON THE EVALUATION OF LEGAL POLICY 47-48 (John M. Olin Ctr. for Law, Econ., and Bus., Harvard Law Sch., Discussion Paper No. 277, 2000),
17. forthcoming in 114 HARV. L. REV. (2001).
18. Id. at 1, 454.
19. This analysis should be of interest even to those who do not believe that the Pareto principle should apply to all social choices, as long as we can agree that there are at least some choices to which we should apply the Pareto principle. This Article reveals that we can apply fairness principles to those matters to which the Pareto principle is also applicable; that is, a belief in the Pareto principle with respect to those questions does not commit us to a welfarist theory regarding those issues.
20. Bernard Williams, A Critique of Utilitarianism, in SMART & WILLIAMS, supra note 14, at 75, 105.
21. Bernard Williams, A Critique of Utilitarianism, 75, 105. Id.
22. Bernard Williams, A Critique of Utilitarianism, 75, 105. Id.
23. Sen, supra note 6, at 478.
24. Sen, supra note 10, at 548. As Robert Goodin has noted, Sen's use of the term "non-utility information" here may be misleading, because "information about why individuals want what they want" is "still very much information about individual utilities." Robert E. Goodin, Laundering Preferences, in FOUNDATIONS OF SOCIAL CHOICE THEORY 75, 76-77 (Jon Elster & Aanund Hylland eds., 1986). Nevertheless, I use this term as Sen uses it.
25. Sen, supra note 10, at 547-48.
26. Sen, supra note 6, at 476.
27. R.M. Hare, Ethical Theory and Utilitarianism, in UTILITARIANISM AND BEYOND 23, 31 (Amartya Sen & Bernard Williams eds., 1982).
28. Id. at 30.
29. Id. Similarly, Hare observes that "we are most unlikely, even if we give sadistic desires weight in accordance with their intensity, to encounter a case in which utility will be maximised by letting the sadist have his way." Id. Kaplow and Shavell give a similar response to the problem of objectionable preferences. KAPLOW & SHAVELL, supra note 15, at 416 (suggesting that "the gain to the sadist will be exceeded by the aggregate utility loss to others").
30. John C. Harsanyi, Problems with Act-Utilitarianism and with Malevolent Preferences, in HARE AND CRITICS 89, 96 (Douglas Seanor & N. Fotion eds., 1988) (noting that if the ratio is sufficiently large, then "we will have to conclude that the social-utility maximizing policy will be to kill all Jews").
31. R.M. Hare, Comments, in HARE AND CRITICS, supra note 28, at 199, 245. Hare also suggests that "in all actual cases" it is better "to push our institutions in the direction of the abandonment of harmful pleasures and desires, and hope that those who now indulge in them will soon change their ways." Id. at 246. Similarly, Kaplow and Shavell argue that "adopting policies that are contrary to the current preferences of some could produce a change in their preferences, and, over the long run, social welfare may rise as a result." KAPLOW & SHAVELL, supra note 15, at 416. Making the immorality of genocide contingent on the mutability of the intolerant preferences of the majority, however, seems no more plausible than making it contingent on the ratio of proponents to victims. Even if we were persuaded that these preferences are immutable, such that liberal policies would not change the intolerant preferences of the majority, we would still be inclined to oppose genocidal policies that would satisfy these preferences.
32. Hare, supra note 25, at 30.
33. Id. at 31.
34. Id. Similarly, J.J.C. Smart defends the utilitarian position that "there are no pleasures which are intrinsically bad," including sadistic pleasures. Smart, supra note 14, at 26. Smart explains that "[o]ur repugnance to the sadist arises, naturally enough, because in our universe sadists invariably do harm." Id. at 25. Thus, our distaste for sadism merely reflects "our distaste for the consequences of sadism," which the utilitarian has already taken into account. Id. at 26.
35. Sen, supra note 6, at 476.
36. Hare, supra note 25, at 26.
37. In Palmore v. Sidoti, 466 U.S. 429 (1984), the Supreme Court held that a state cannot deny a divorced mother custody of her child on account of her interracial remarriage. The Court rejected the legal relevance of racial prejudice in society: "Private biases may be outside the reach of the law, but the law cannot, directly or indirectly, give them effect." Id. at 433; see Loving v. Virginia, 388 U.S. 1 (1967) (holding that the Fourteenth Amendment prohibits antimiscegenation laws); Shelley v. Kraemer, 334 U.S. 1 (1948) (holding that the Fourteenth Amendment prohibits states from enforcing racially restrictive covenants); Buchanan v. Warley, 245 U.S. 60 (1917) (holding that the Fourteenth Amendment prohibits laws that forbid blacks to reside in white neighborhoods).
38. See Howard F. Chang, Liberalized Immigration as Free Trade: Economic Welfare and the Optimal Immigration Policy, 145 U. PA. L. REV. 1147, 1210-21 (1997) (arguing that racist and xenophobic preferences should not count as reasons to adopt immigration restrictions).
39. Goodin, supra note 22, at 78.
40. Id. at 81.
41. Id. at 77.
42. RONALD DWORKIN, TAKING RIGHTS SERIOUSLY 234 (1977). Dworkin does not, however, specify what counts as a "good." If we define "good" broadly enough to include the satisfaction of external preferences, then any external preference can also be described as a personal preference. Therefore, to maintain the distinction between external and personal preferences, Dworkin must have a narrower definition of "good" in mind, perhaps a definition restricted to the sensory or physical experiences of the holder of the preference.
43. Id. at 234-35.
44. Id. at 235, 275-76.
45. Id. at 235.
46. Id. at 276. To give a satisfactory result in Hare's genocide hypothetical, a liberal theory of social utility would exclude the satisfaction of external preferences entirely rather than merely discount this satisfaction by some factor. Otherwise, there would still exist some ratio of Nazis to Jews that would imply that genocide would increase social utility. See supra note 28 and accompanying text.
47. DWORKIN, supra note 40, at 275.
48. Id. at 273.
49. Id. at 235. Bruce Ackerman takes a similar concept of neutrality to be fundamental in the liberal state. This neutrality principle precludes anyone from justifying a legal regime by claiming that "his conception of the good is better than that asserted by any of his fellow citizens" or that "he is intrinsically superior to one or more of his fellow citizens." BRUCE A. ACKERMAN, SOCIAL JUSTICE IN THE LIBERAL STATE 11 (1980).
50. Dworkin has explained that his claim of "double-counting" was "meant to summarize the argument, not to make it." RONALD DWORKIN, A MATTER OF PRINCIPLE 366 (1985).
51. H.L.A. Hart, Between Utility and Rights, 79 COLUM. L. REV. 828, 842 (1979); see also
52. C. Edwin Baker, Counting Preferences in Collective Choice Situations, 25 UCLA L. REV. 381, 386 (1978) (suggesting that "the egalitarian nature of utilitarian analysis would seem to be corrupted not by counting all of people's preferences but by favoring those with only a certain type of preferences, i.e., personal preferences" and that "ignoring [an] 'external preference' would appear to be 'half counting'").
53. DWORKIN, supra note 48, at 362. In a similar spirit, Paul Brest declares that "decisions are irrational insofar as they reflect the assumption that members of one race are less worthy than other people." Paul Brest, The Supreme Court, 1975 Term - Foreword: In Defense of the Antidiscrimination Principle, 90 HARV. L. REV. 1, 6 (1976). This assumption would represent a defect in "the process by which race-dependent decisions are made" providing a rationale for the "antidiscrimination principle" that is distinct from "harmful results of race-dependent decisions."
54. Paul Brest, The Supreme Court, 1975 Term - Foreword: In Defense of the Antidiscrimination Principle, 90 HARV. L. REV. 1, (1976) Id.
55. Similarly, Larry Alexander classifies biases that "rest on erroneous judgments of others' inferior moral worth" as "intrinsically immoral" preferences. Larry Alexander, What Makes Wrongful Discrimination Wrong? Biases, Preferences, Stereotypes, and Proxies, 141 U. PA. L. REV. 149, 192 (1992). He distinguishes "intrinsically immoral" preferences from those that are "extrinsically - contingently - immoral because of the effects on others of acting on those preferences." Id. at 194. Thus, like Dworkin, both Brest and Alexander would reject such intrinsically immoral preferences as justifications for decisions even before we weigh the costs that the satisfaction of these preferences imposes upon others.
56. DWORKIN, supra note 48, at 363.
57. Id. at 361.
58. JOHN RAWLS, A THEORY OF JUSTICE (1971).
59. Some have proposed that we exclude a narrower class of preferences. For example, Baker suggests that "taking into account" external preferences "is not always inappropriate" but that "counting these preferences when they deny the inherent equality or worth of persons is improper." Baker, supra note 49, at 403. Eric Rakowski considers the suggestion that we exclude "preferences attributable to the belief that some persons are not entitled to equal moral consideration" but still count "desires that are not antiegalitarian," but he concludes that this solution is "unavailing" because other external preferences are also problematic. ERIC RAKOWSKI, EQUAL JUSTICE 29 n.13 (1991). For example, Alexander notes that other "preferences based on moral ideals," such as the belief that heterosexuality is "the only morally proper form of sexuality," that "the role of sex partner is gender-specific as a matter of morality," or that "it is immoral for women to perform certain roles and for men to perform certain roles," need not entail a belief that "men and women have differential moral worth." Alexander, supra note 50, at 164. Nevertheless, Alexander considers these preferences to be "intrinsically immoral." Id. at 192.
60. DWORKIN, supra note 48, at 361, 363.
61. Kaplow & Shavell, supra note 3, at 75.
62. DWORKIN, supra note 40, at 235.
63. Brian Barry suggests that such a utilitarian "would seem rather eccentric to take account of my judgment as an ingredient in his judgment about social welfare." Brian Barry, Lady Chatterley's Lover and Doctor Fischer's Bomb Party: Uberalism, Pareto Optimality, and the Problem of Objectionable Preferences, in FOUNDATIONS OF SOCIAL CHOICE THEORY, supra note 22, at 11, 37.
64. Williams, supra note 18, at 106.
65. Id.
66. Brest, supra note 50, at 8.
67. Amartya K. Sen, Rational Fools: A Critique of the Behavioral Foundations of Economic Theory, 6 PHIL. & PUB. AFF. 317, 326-29 (1977). Williams uses the term "commitments" in a similar sense. Williams, supra note 18, at 112-13.
68. The preference for public policies that satisfy one's own altruistic preferences might be deemed political, but then personal preferences would be political in precisely the same sense.
69. John C. Harsanyi, Morality and the Theory of Rational Behaviour, in UTILITARIANISM AND BEYOND, supra note 25, at 39, 47, 54.
70. Id. at 47.
71. More recently, Harsanyi has endorsed Dworkin's claim that "the very nature of utilitarian ethics suggests the exclusion of all external preferences," including altruistic preferences. Harsanyi, supra note 28, at 98.
72. Thus, Yew-Kwang Ng argues that in order to avoid "double or rather multiple counting," we should ignore "concern for the welfare of others" in "social evaluation." Yew-Kwang Ng, Utility, Informed Preference, or Happiness: Following Harsanyi's Argument to Its Logical Conclusion, 16 SOC. CHOICE & WELFARE 197, 199 (1999). Other economists have argued that cost-benefit analysis should not include ethical or altruistic values.
73. E.g., Paul Milgrom, Is Sympathy an Economic Value? Philosophy, Economics, and the Contingent Valuation Method, in COINCENT VALUATION: A CRITICAL ASSESSMENT 417, 418-22 (Jerry A. Hausman ed., 1993).
74. Brest, for example, concludes that decisions that reflect "racially selective sympathy and indifference," that is, "the unconscious failure to extend to a minority the same recognition of humanity, and hence the same sympathy and care, given as a matter of course to one's own group," are "unfair," like "those reflecting overt racial hostility." Brest, supra note 50, at 7-8. Brest explains that "such treatment violates the cardinal rule of fairness - the Golden Rule." Id. at 8. We can distinguish these inherently discriminatory preferences from purely self-regarding preferences, which may advantage some individuals over others merely as an incidental consequence of satisfying these preferences in a market economy. For example, "a preference for music advantages individuals born with musical talent," and "a preference for reading novels advantages individuals born with literary talent." KAPLOW & SHAVELL, supra note 15, at 414 n.865 (noting that "virtually any taste may have third-party effects that result in relative advantages and disadvantages to other people"). We can give weight to these purely self-regarding preferences, which are not "intrinsically immoral." Alexander, supra note 50, at 194.
75. Hare, supra note 25, at 26.
76. Rakowski has suggested that a utilitarianism that excludes external preferences "forsakes its motivating idea." RAKOWSKI, supra note 53, at 26. The truth of this claim depends on what one takes to be the "motivating idea" of utilitarianism. The claim would be correct if this idea is that "whatever moral principles would be chosen by self-interested, risk-neutral individuals ignorant of their own desires and abilities are justified." Id. at 24; see John C. Harsanyi, Cardinal Utility in Welfare Economics and in the Theory of Risk-Taking, 61 J. POL. ECON. 434 (1953) (deriving utilitarianism from such a premise).
77. Kaplow and Shavell similarly "consider contexts in which all individuals are symmetrically situated before events arise that call for the application of legal rules" to justify welfarism. Kaplow & Shavell, supra note 3, at 68. This idea, however, "begs the question" of whether we should consider this particular hypothetical to be the appropriate test for a moral theory. Barbara Fried, Can We Really Deduce Utilitarianism from the Pareto Principle? 41 (Dec. 6, 1999) (unpublished manuscript, on file with The Yale Law Journal). We can reject this idea as the basis for moral principles and substitute a different motivating idea. If the motivating idea is instead an ideal observer committed to equal concern and respect for all persons, then the exclusion of external preferences would be a means to implement, not abandon, that idea.
78. Harsanyi, supra note 63, at 56; see RICHARD A. POSNER, THE ECONOMICS OF JUSTICE 83 (1981) (criticizing utilitarians who "ascribe value" to "envy and cruelty, because these are common sources of personal satisfaction and hence of utility").
79. Shavell also suggests in his earlier work that "illicit utility (such as would arise when a person batters another for pleasure)" may not enter "social welfare." STEVEN SHAVELL, ECONOMIC ANALYSIS OF ACCIDENT LAW 160-61 (1987).
80. Other economists make similar suggestions. E.g., Robert D. Cooter, Economic Analysis of Punitive Damages, 56 S. CAL. L. REV. 79, 87, app. at 99 (1982) (suggesting that an economic analysis may exclude an "illicit benefit" that an injurer enjoys from wrongful behavior);
81. Alvin K. Klevorick, On the Economic Theory of Crime, in NOMOS XXVII: CRIMINAL JUSTICE 289, 299 (1985) (suggesting that an economic analysis of the social value of criminal acts may discount "the criminal's gain" from the crime);
82. George J. Stigler, The Optimum Enforcement of Laws, 78 J. POL. ECON. 526, 527 (1970) (suggesting that when "society has branded the utility derived from [crimes] as illicit," the "gain to the offender" is not "a gain to society").
83. Like Dworkin, others have suggested the exclusion of all other-regarding preferences from consideration. E.g., Jean Hampton, Rational Choice and the Law, 15 HARV. J.L. & PUB. POL'Y 649, 671 (1992) (suggesting that to avoid "unfairness," the parties to an ideal social contract would consider only "self-concerned" preferences and exclude "other-regarding interests that could be either malicious (thus skewing the contract 'unfairly' against those who were hated) or beneficent (thus skewing the contract 'unfairly' in favor of those who where loved)");
84. Peter J. Hammond, The Economics of Justice and the Criterion of Wealth Maximization, 91 YALE L.J. 1493, 1499, 1501 (1982) (reviewing POSNER, supra note 70) (arguing that counting antisocial preferences is "ethically unacceptable" and suggesting that only "good" and "self-interested" preferences should count, where "good" is defined using "ethical criteria" to determine "moral worthiness").
85. Harsanyi, supra note 63, at 56.
86. Id. Thus, Harsanyi endorses the use of a "censored utility function" that excludes antisocial preferences, because an ideal observer would have no "moral sympathy" for such preferences. JOHN C. HARSANYI, RATIONAL BEHAVIOR AND BARGAINING EQUILIBRIUM IN GAMES AND SOCIAL SITUATIONS 62 (1977).
87. Goodin, supra note 22, at 85.
88. Id.
89. Id. at 86.
90. DWORKIN, supra note 40, at 277.
91. Kaplow and Shavell criticize Dworkin's theory using similar examples, pointing to "the difference between an opera singer who performs for the money and one who would not be induced to perform by the money alone but does choose to perform because of the pleasure of pleasing an audience" and noting that "under the posited theory, the latter singer should not be permitted to perform." KAPLOW & SHAVELL, supra note 15, at 414 n.864.
92. W. Michael Hanemann, Valuing the Environment Through Contingent Valuation, J. ECON. PERSP, Fall 1994, at 19, 33 ("When estimating demand functions for fish prior to Vatican II, no economist ever proposed removing Catholics because they were eating fish out of a sense of duty.").
93. Harsanyi, supra note 63, at 55. Harsanyi also calls this idea "[t]he principle of individual self-determination," HARSANYI, supra note 73, at 61, or "the principle of acceptance, because it requires us to accept each individual's own personal preferences as the basic criterion for assessing the utility . . . that he will derive from any given situation," id. at 52.
94. RAWLS, supra note 53, at 448.
95. JOHN STUART MILL, ON LIBERTY 6 (Longmans, Green & Co. 1921) (1859).
96. Harsanyi, supra note 63, at 56.
97. Id. at 55.
98. Id.
99. Id.
100. Id. For example, Harsanyi suggests that we are justified in using a "corrected utility function" for an individual if we think that the individual "would approve of this" correction if "made aware of the possibility that his actual preferences were based on factual mistakes and of the likely costs of these mistakes to him." HARSANYI, supra note 73, at 61-62.
101. Similarly, Kaplow and Shavell assume that "individuals understand fully how various situations affect their well-being," so that their arguments apply to "individuals' actual well-being rather than to individuals' well-being as reflected in their mistaken preferences." Kaplow & Shavell, supra note 3, at 65 & n.2. I discuss the implications of the distinction between manifest and true preferences when this issue becomes relevant. Infra note 153.
102. RAWLS, supra note 53, at 450. Similarly, Richard Epstein argues that "no one person is entitled to elevate his or her beliefs about how others should act above those of anyone else." RICHARD A. EPSTEIN, FORBIDDEN GROUNDS: THE CASE AGAINST EMPLOYMENT DISCRIMINATION LAWS 415 (1992) (emphasis added). Yet he otherwise seems to favor a calculation of costs and benefits that does not require us "to ask whose preferences are legitimate and whose are not." Id. at 43.
103. DWORKIN, supra note 48, at 364-65. Dworkin would allow such preferences to justify regulations on public displays of pornography, but it seems he does so only because it is impossible to separate those preferences "that express moral condemnation or would not exist but for it" from "the remaining strands" of personal "taste[s]." Id. at 356.
104. Introduction to UTILITARIANISM AND BEYOND, supra note 25, at 1, 3-4 (emphasis omitted).
105. RAWLS, supra note 53, at 24 (defining a ideological theory as a theory that holds that "those institutions and acts are right which of the available alternatives produce the most good").
106. Other theories, like the theory of justice advanced by Rawls, are "deontological" theories, which do not design rights to maximize some good specified in advance. Id. at 30.
107. DWORKIN, supra note 40, at 276.
108. Id. at 277.
109. KAPLOW & SHAVELL, supra note 4, at 5.
110. KAPLOW & SHAVELL, supra note 15, at 411-17 (arguing in favor of including all preferences, including objectionable preferences, in the calculation of social welfare).
111. Kaplow & Shavell, supra note 3, at 66 n.5.
112. Id. at 65 n.4.
113. KAPLOW & SHAVELL, supra note 4, at 1.
114. Sen, supra note 1, at 155.
115. Id.
116. Note that we define the Pareto principle based on what individuals actually prefer, including not only their personal preferences but also their external preferences. If we were to define the Pareto principle based on personal preferences only, then the conflict with liberal consequentialism would disappear. I argue, however, that a liberal theory of social welfare should respect a Pareto principle based not only on personal preferences but also external preferences. Infra Part III.
117. Sen, supra note 6, at 479-80. Sen uses the term "liberalism" for this concept in his original article. Sen, supra note 1, at 153-54. In subsequent writings, Sen replaces this "more ambiguous" term with "libertarianism." Sen, supra note 2, at 218.
118. Sen, supra note 1, at 153.
119. Sen, supra note 2, at 218.
120. Sen, supra note 10, at 550.
121. See, e.g., Sen, supra note 9, at 1079 (discussing the requirement of "acyclicity").
122. Sen, supra note 1, at 157.
123. Id.
124. Sen, supra note 2, at 219.
125. Id. at 226.
126. ROBERT NOZICK, ANARCHY, STATE, AND UTOPIA 166 (1974) (emphasis omitted).
127. E.g., Peter Bemholz, Is a Paretian Liberal Really Impossible?, 20 PUB. CHOICE 99, 106 (1974) ("The only . . . conclusion which can . . . be drawn from Sen's proof is that there are situations in which other decision rules than liberalism should be applied.").
128. Kaplow & Shavell, supra note 3, at 64.
129. Id. at 76.
130. Id. at 72 (emphasis added).
131. KAPLOW & SHAVELL, supra note 4, at 4; see Kaplow & Shavell, supra note 3, at 72 (arguing that the Pareto principle "has powerful implications for what criteria for making policy choices one can consistently employ" even if it "may not directly determine policy choices in most real situations").
132. KAPLOW & SHAVELL, supra note 15, at 411-17 (arguing in favor of including all preferences, including objectionable preferences, in the calculation of social welfare).
133. Allan Gibbard, A Pareto-Consistent Libertarian Claim, 7 J. ECON. THEORY 388, 401-06 (1974).
134. Id. at 397.
135. Id. at 400-01.
136. Sen, supra note 1, at 157.
137. RUSSELL HARDIN, MORALITY WITHIN THE LIMITS OF REASON 109 (1988).
138. Barry, supra note 57, at 19.
139. Gibbard, supra note 120, at 397. We may, however, believe in a version of liberalism that can justify such paternalistic interventions when parties would make choices that would not satisfy their "true" preferences. For a discussion of such constraints on individual choice, see infra note 153.
140. Gibbard, supra note 120, at 398.
141. Id. at 406.
142. Sen, supra note 2, at 224.
143. Id. at 226.
144. Sen, supra note 6, at 482 n.35.
145. E.g., Kaushik Basu, The Right To Give Up Rights, 51 ECONOMICA 413 (1984) (exploring problems of incentives under any system in which individuals waive their rights voluntarily);
146. Edi Karni, Collective Rationality, Unanimity and Liberal Ethics, 45 REV. ECON. STUD. 571 (1978) (showing how Gibbard's solution may be manipulated by parties who cheat);
147. Jerry S. Kelly, Rights Exercising and a Pareto-Consistent Libertarian Claim, 13 J. ECON. THEORY 138 (1976) (exploring difficulties parties would encounter in deciding when to waive their rights).
148. Sen, supra note 2, at 224.
149. Sen, supra note 10, at 552 n.2.
150. Peter J. Hammond, Utilitarianism, Uncertainty and Information, in UTILITARIANISM AND BEYOND, supra note 25, at 85, 86 (noting that "many common misconceptions have arisen because of a confusion of objectives and constraints").
151. Sen, supra note 10, at 551.
152. Id. at 551-52.
153. Id. at 552.
154. Id. Thus, whether the refusal to trade is "a violation of the Pareto principle or a consistent application of it depends on whether one regards an individual's liberal values as a part of his preferences or a constraint upon them." DENNIS C. MUELLER, PUBLIC CHOICE II 404 (1989).
155. Sen uses the term "conditional Pareto principle" to describe a Pareto principle that is based on preferences that are conditional on political preferences as well as all other preferences. Sen, supra note 2, at 236. He distinguishes this Pareto principle from "Pareto preferences in the traditional sense," by which he means preferences based on happiness. Id. at 237. It might be equally accurate, however, to refer to the "conditional Pareto principle" as the "traditional" principle, because the "conditional Pareto principle" turns simply on what outcomes individuals actually prefer, all things considered.
156. Amartya Sen, Minimal Liberty, 59 ECONOMICA 139, 145 (1992).
157. Amartya Sen, Minimal Liberty, 59 ECONOMICA 139, 145 (1992). Id.
158. Furthermore, if this particular conflict is all that Sen wants to illustrate, then he can do so using a much simpler hypothetical. It would be sufficient to consider a society of one person who would enjoy reading a book but refrains from doing so in order to comply with a moral commitment. In this case, the desire-based Pareto principle would regard reading the book as the better outcome, but a liberal respect for her right not to read the book would suggest the reverse ranking.
159. Sen, supra note 141, at 146.
160. Id. at 145.
161. Barry, supra note 57, at 14.
162. Id. at 19 (emphasis added). Sen complains that "Barry supposes that the only barrier to such contracts can arise from people not being allowed to enter into such contracts," whereas "[t]he real issue is . . . whether they will seek such a contract." Amartya Sen, Foundations of Social Choice Theory: An Epilogue, in FOUNDATIONS OF SOCIAL CHOICE THEORY, supra note 22, at 213, 225. We can interpret Barry's claim, however, to be that liberalism is silent on the issue of whether people should enter such contracts. If so, then liberalism does not conflict with the Pareto principle on this issue.
163. Sen, supra note 2, at 235.
164. Sen, supra note 1, at 157.
165. See, e.g., Bernholz, supra note 114, at 100 ("[N]o one would dream of applying one decision rule like the rule of liberalism under all circumstances.").
166. D.K. Osborae, On Liberalism and the Pareto Principle, 83 J. POL. ECON. 1283, 1286 (1975). Sen responds that Osborne's claim is "based on an unadulterated piece of logical error." Sen, supra note 2, at 228. Sen explains that his strict preference cycle arises from comparisons between pairs of alternatives: "The Pareto principle can operate over one or more pairs (without conflicting with the liberal principle over those pairs) and the liberal principle can operate over two or more other pairs (without conflicting with the Pareto principle over those pairs), and these choices together can be inconsistent." Id. at 229. Osborne's claim holds true, however, if we interpret it as a challenge to the "liberal principle" that Sen applies to generate the preference cycle. If we apply a more modest liberal principle that declines to rank pairs when the holder of a right wants to waive the right (or perhaps reverses the ranking when the holder wants to waive the right rather than exercise it), then Sen's response to Osborne fails.
167. Sen's impossibility claim is striking only if his notion of liberalism is uniquely plausible. See Osborne, supra note 151, at 1285 n.3 ("[I]f a person is going to insist on claiming 'The Impossibility of a Paretian Liberal,' he must expect people to examine his concept of liberalism.").
168. Preferences revealed by choice would be what Harsanyi calls "manifest" preferences. Supra notes 85-86 and accompanying text. If these preferences are not "true" preferences, then we have another ambiguity regarding what we mean by the Pareto principle. If we want to uphold the Pareto principle as applied to "manifest" preferences, then alienable rights would ensure compliance with that Pareto principle. If, however, we choose to uphold the Pareto principle as applied to "true" preferences, then alienable rights would not ensure compliance with that Pareto principle, because parties may choose not to trade even if their "true" preferences would indicate that they should. This Pareto principle could in theory justify paternalistic intervention to satisfy the parties' "tree" preferences. This intervention could require a trade when one or both parties prefer not to trade, or it could prohibit a trade when both parties would prefer to trade. We should not, however, attribute these restrictions on individual choice to the Pareto principle per se. If we object to these paternalistic restrictions on individual liberty, then we object to the reliance on "true" preferences as the morally relevant preferences, not to the Pareto principle itself. If we instead accept the use of "true" preferences rather than "manifest" preferences, then we would not object to such paternalistic interventions in theory. We would understand libertarian rights to be qualified by our theory of justified paternalism and therefore would not consider it an infringement of such rights for the government to implement a Pareto improvement based on "true" preferences. This conflict between libertarian rights and the Pareto principle should disturb us only if we adopt a definition of the Pareto principle inconsistent with our own beliefs regarding paternalistic interventions.
169. Sen, supra note 2, at 236.
170. Kotaro Suzumura, On the Consistency of Libertarian Claims, 45 REV. ECON. STUD. 329, 332 (1978).
171. Sen, supra note 2, at 237.
172. Sen agrees with Gibbard on this point, stating, "I see no reason why rights of this kind should not in general be taken to be open to contracting and exchange through mutual agreement." Sen, supra note 141, at 145-46.
173. RAWLS, supra note 53, at 42; see, e.g., id. at 302-03 (giving liberty priority over his second principle of justice and giving his principles of justice priority over efficiency and welfare).
174. Kaplow & Shavell, supra note 3, at 72-73.
175. Id. at 72.
176. RAWLS, supra note 53, at 4.
177. T.M. Scanlon, Preference and Urgency, 72 J. PHIL. 655, 660 (1975) (emphasis added).
178. Kaplow & Shavell, supra note 3, at 72 n.20.
179. Fried states this point "in Kantian terms"; that is, to permit a Pareto improvement "means we are not trading off one person's welfare for another's," and thus "we are not using one person for another's ends." Fried, supra note 69, at 24. Where "one person is left worse off," however, "for a Kantian the problem of whether one person's rights are unfairly compromised for the good of others is raised." Id. at 25.
180. It is not unusual that the outcome of our normative analysis might turn on two pennies. It would also be true under utilitarianism, for example, that we might rank alternatives differently in marginal cases if we change one alternative by mere pennies.
181. Thomas M. Scanlon, The Moral Basis of Interpersonal Comparisons, in INTERPERSONAL COMPARISONS OF WELL-BEING 17, 28 (Jon Elster & John E. Roemer eds., 1991).
182. Scanlon, supra note 162, at 659.
183. DWORKIN, supra note 40, at 235.
184. Harsanyi, supra note 63, at 56.
185. In a similar vein, Posner criticizes utilitarians for ascribing value to "envy and cruelty" but defends a "well-regulated market economy" in which an individual cannot "promote his self-interest without benefiting others as well as himself." POSNER, supra note 70, at 83 (noting that "lawfully obtained wealth is created by doing things for other people - offering them advantageous trades"). Posner endorses a "rights system" that would require a sadist to buy his victims' consent and thus "to pay them whatever compensation they demand." Id. at 82.
186. MILL, supra note 82, at 6.
187. Kaplow & Shavell, supra note 3, at 75.
188. Sen & Williams, supra note 91, at 17 (noting that "completeness is often seen as a merit" and that with full interpersonal comparison of cardinal utility, "utilitarianism yields a complete ordering").
189. See ROBIN BOADWAY & NEIL BRUCE, WELFARE ECONOMICS 34 (1984) (defining a "complete" ordering);
190. JOHN VON NEUMANN & OSKAR MORGENSTERN, THEORY OF GAMES AND ECONOMIC BEHAVIOR 26 (3d ed. 1953) (same). The Pareto principle by itself, on the other hand, does not provide a complete ranking of alternatives. The Pareto principle can rank alternatives only if one alternative is Pareto superior to the other. If neither alternative in a pair Pareto-dominates the other, then the Pareto principle does not tell us whether one alternative should get a higher rank or whether the two alternatives are of equal rank.
191. I assume that the purpose of our social welfare function is to guide social choices, and we can make such choices only among feasible alternatives. See Kevin W.S. Roberts, Interpersonal Comparability and Social Choice Theory, 47 REV. ECON. STUD. 421, 422 (1980) ("We are interested in obtaining a social ordering . . . defined over the set of feasible social states X.") (emphasis omitted). The "unrestricted domain" condition then requires that our social choice function generate a ranking of those feasible alternatives for any possible set of individual preferences.
192. Kevin W.S. Roberts, Interpersonal Comparability and Social Choice Theory, 47 REV. ECON. STUD. 421, (1980) Id.
193. A social choice can have implications for the feasibility of alternatives in the future. To analyze a sequence of social choices over time, it is useful to conceive of alternatives as a game theorist conceives of strategies chosen by a player in a game. Game theorists define a "strategy" as "a complete plan of action," which "specifies a feasible action for the player in every contingency in which the player might be called on to act." ROBERT GIBBONS, GAME THEORY FOR APPLIED ECONOMISTS 117 (1992) (emphasis omitted);
194. see also ERIC RASMUSEN, GAMES AND INFORMATION: AN INTRODUCTION TO GAME THEORY 24 (1989) (defining a strategy for any player as "a rule that tells him which action to choose at each instant of the game, given his information set"); VON NEUMANN & MORGENSTERN, supra note 174, at 79 (defining a strategy as " a plan which specifies what choices [a player] will make in every possible situation, for every possible actual information which he may possess at that moment").
195. We can similarly define an alternative as "a complete contingent plan in advance." DREW FUDENBERG & JEAN TIROLE, GAME THEORY 85 (1991). At any given decision node, we may also be uncertain regarding whether a specific result is feasible. We can analyze the social choice at that node by considering the attempt to achieve that result as a feasible action with uncertain consequences. We can evaluate actions with uncertain consequences according to the expected values for utility and for fairness, taking the average value over the possible results, with each result given weight in proportion to the probability of that particular outcome.
196. See KENNETH J. ARROW, ESSAYS IN THE THEORY OF RISK-BEARING 53 (1971) ("One action will be preferred to another if and only if the expected value of the utility of its consequences is greater.").
197. See generally LEONARD J. SAVAGE, THE FOUNDATIONS OF STATISTICS 263 (1954) (defining "expected value").
198. Each individual enjoys the same alienable right to the fair outcome and can waive this right only with the consent of all other individuals. Because this alienable right is enjoyed by all on an equal basis, it differs from the alienable right proposed by Gibbard, which "gives a person a special voice on an issue . . . [and thus] cannot be accorded indiscriminately." Gibbard, supra note 120, at 404. A right to fairness that requires unanimous consent for waiver is sufficiently alienable to ensure compliance with the Pareto principle. We can relax this unanimity requirement to allow more frequent waivers of the right to the fair outcome. Infra Section IV.C.
199. Sen, supra note 9, at 1079.
200. Id.
201. The rankings under F* depend on fairness considerations. One can of course contrive special examples in which the rankings under F* will not depend on fairness information. For example, consider a set S containing only two alternatives, one Pareto superior to the other. Any Paretian theory, including not only welfarism but also F*, will rank the Pareto superior alternative higher than the Pareto inferior alternative, regardless of fairness considerations. Such examples do not show that F* is welfarist, however, because F* would be welfarist only if its rankings were independent of fairness information for any set S. As long as F* incorporates fairness considerations anywhere, under any circumstances, it is a fairness theory and not a welfarist theory. Kaplow and Shavell define "conceptions of fairness" to be principles based "at least in part" on factors unrelated to individual utility, Kaplow & Shavell, supra note 3, at 65, so that "fairness" theories include "mixed views," which give weight not only to notions of fairness but also to individual utility, id. at 67.
202. The F* described above is only one example of a fair Paretian social welfare function. We can describe other fair ranking procedures that we can apply to finite sets of alternatives while complying with the Pareto principle. Consider, for example, the following procedure for ranking alternatives in set S. Rank all Pareto optimal alternatives according to our fairness criterion F. Give the highest rank to those Pareto optimal alternatives that rank highest according to this fairness criterion and declare them to be socially optimal. To rank the remaining alternatives, delete these socially optimal alternatives from set S, and repeat the procedure. That is, apply the procedure to the remaining alternatives, assuming that the higher ranking alternatives are unavailable. Give the next highest rank to the socially optimal alternatives for the remaining set. Continue until we have ranked all alternatives in set S.
203. Finite sets of alternatives have been the traditional focus of social choice theorists. Graciela Chichilnisky, Social Aggregation Rules and Continuity, 97 Q.J. ECON. 337, 337 (1982) (noting that in the "vast literature" on social choice, "most existing work has focused on problems where the individuals face a finite number of choices").
204. For example, a technical problem arises if set D does not contain its own boundary, such that we cannot identify an optimal alternative within that set. Savage describes this problem: "Suppose, for example, that you were free to choose any income, provided it be definitely less than $100,000 per year. Precisely which income would you choose, abstracting from the indivisibility of pennies?" SAVAGE, supra note 176, at 18; see also id. ("If infinite sets of available acts are set up and interpreted without some mathematical tact, unrealistic conclusions are likely to follow.").
205. The welfarist principle does not have to be utilitarian. For example, it can be the "leximin" rule. Infra note 198. Allowing a welfarist principle to play some limited role in determining rankings does not render this procedure a welfarist social welfare function. Supra note 180.
206. These new rankings will not rank any Pareto superior alternative below a Pareto inferior alternative, because Pareto superiority is transitive. Thus, when we move an alternative up the rankings to outrank all Pareto inferior alternatives, the alternative in question will not outrank any Pareto superior alternative, because any Pareto superior alternative would have to move at least as high in the rankings. Any Pareto superior alternative would be Pareto superior to any alternative that is Pareto inferior to the alternative in question. Similarly, any Pareto inferior alternative would receive a rank no higher than the alternative in question, even if the Pareto inferior alternative moves up the ranks, because the alternative in question would have to move at least as high in the rankings. The alternative in question would be Pareto superior to any alternative that is Pareto inferior to the Pareto inferior alternative.
207. Ranking an infinite number of alternatives under this procedure could require an infinite number of steps. This feature, however, does not distinguish this procedure from welfarist ranking procedures. A utilitarian, for example, would have to calculate average or total utility for an infinite number of alternatives in order to rank them. Thus, the proposed procedure can provide a complete ranking of an infinite number of alternatives in the same sense that a utilitarian theory can. Using the proposed procedure to rank a finite number of alternatives within set S, however, could also require an infinite number of steps if set S contains an infinite number of alternatives. This feature distinguishes this procedure from welfarist ranking procedures and raises practical problems in actually implementing the procedure. See infra Section IV.B (discussing the implications of this analysis for making social choices in practice).
208. The F** procedure gives welfarism a somewhat larger role than the F* procedure does, insofar as F** uses a welfarist principle to break ties under criterion F. Nevertheless, F** never ranks a less fair alternative above a more fair alternative unless the Pareto principle requires this ranking. In this sense, F** respects each individual's alienable right to a fair alternative, because when there are multiple fairness optima, no one enjoys an entitlement to any particular optimum on fairness grounds.
209. For simplicity of exposition, assume that alternative o is not a feasible social choice. Adding a third alternative like o would make the example more complicated without affecting any important results.
210. There would be no conflict between criterion F and the Pareto principle within this set B, because all alternatives within this set would feature outcome p. Thus, we would rank alternatives within set B simply according to our fairness criterion F, which would favor an equal division of wealth.
211. This feature is inherent in welfarism, because any welfarist theory must ignore all fairness considerations and select a social optimum based only on utility information. Given this constraint, no welfarist theory can ensure that the optimum it selects will have any particular relationship with the fairness optimum. Given utility information, the welfarist is committed to the same social choice, regardless of fairness information. Thus, the welfarist cannot adjust this social choice in light of new information regarding the fairness optimum. Only a fairness theory can make its social choice a function of such information.
212. I have described F* and F** as giving the Pareto principle priority over fairness principles, but we can also describe F* or F** as giving alienable rights to fairness priority over welfarist considerations. Both characterizations describe the same theory.
213. Ranking alternatives in set D under F* in our example requires us to move all alternatives featuring outcome p up the rankings so that they outrank any Pareto inferior alternatives featuring outcome l. The alternatives featuring outcome l would not change rankings with respect to one another, but all alternatives featuring outcome p would move up the rankings with respect to at least some alternatives featuring outcome l. With each iteration of our refinement procedure, set C for the current iteration moves further down the rankings under criterion F, further away from the fairness optimum for set S as a whole. As we move away from the fairness optimum for set S, each iteration allows a greater range of possible levels of utility for each individual in set C. This greater range implies that criterion F in this example will provide a lower constraint on the utility to which each individual is entitled as we move away from the fairness optimum for set S. The rankings under F** for the alternatives in set D are similar but complicated further by the use of a welfarist principle as a tiebreaker. The rankings within set D, however, would become relevant for the social choices we actually make only if the alternatives in higher ranking sets are not available. If those alternatives are not available, however, then set S is not as we have assumed, and we would have to apply our refinement procedure to a different set S. A different set S, however, would generally imply different rankings for alternatives within set D. See infra Section IV.B (discussing how the rankings under F* are generally a function of the entire feasible set S). Our rankings within our original set D would become relevant only in the very special case that the actual set S coincides exactly with this set D. In general, however, we would expect criterion F and the constraints on set 5 to produce a unique fairness optimum such that fairness principles would give each individual a substantial entitlement. Thus, under F* or F**, fairness considerations generally exert the most force precisely where it counts the most as a practical matter: in actually making a social choice from any given set S.
214. KAPLOW & SHAVELL, supra note 4, at 2. There is some ambiguity regarding what Kaplow and Shavell mean by "individualistic." In general, they appear to use the term to refer to welfarist social welfare functions. They also use the term, however, to refer to all social welfare functions that comply with the Pareto indifference rule, which holds that if every individual is indifferent between two alternatives, then we should be socially indifferent between those two alternatives. See Sen, supra note 9, at 1075. Kaplow and Shavell assume that compliance with the Pareto indifference rule is equivalent to welfarism, which requires " strong neutrality," such that "the utility information regarding any two social states is all that is needed for ranking that pair." Id. at 1121. These two concepts, however, are in general not equivalent. Id. at 1154-55. Strong neutrality implies the Pareto indifference rule, but the converse implication does not hold unless we make "quite demanding" assumptions. Id. at 1155. That is, social welfare functions may comply with the Pareto indifference rule without being welfarist. For example, we could have qualified our fairness principle F further, so as to comply with not only the weak Pareto principle but also the Pareto indifference rule. To implement this additional qualification, we could modify steps (2) and (4) in the F* procedure to expand sets A and C accordingly. For example, if each individual is indifferent between a "fair" alternative in set C and another alternative, then we would include both alternatives in set C. The rationale would be that if all are willing to exchange the "fair" alternative (to which they are entitled) for another alternative, but no one would strictly prefer to make this trade, then we should assign the same level of social welfare to these two alternatives. Similarly, we could modify F** to use the Pareto indifference rule to move alternatives up the rankings. I have, however, designed both F* and F** to be "non-individualistic" in both senses used by Kaplow and Shavell, so that they respond specifically to their formal claim without any ambiguity.
215. KAPLOW & SHAVELL, supra note 4, at 3. Kaplow and Shavell interpret their proof as if it applied to any social welfare function that incorporates notions of fairness. This proof, however, assumes that all such functions violate the Pareto indifference rule, whereas only a subset of all social welfare functions incorporating notions of fairness violate the Pareto indifference rule. Supra note 193. Thus, Kaplow and Shavell's formal proof is not as general as they seem to claim.
216. KAPLOW & SHAVELL, supra note 4, at 3.
217. See BOADWAY & BRUCE, supra note 174, at 147 (noting that "lexicographic orderings . . . are precluded by the requirement of continuity").
218. A survey of the literature on social choice theory reveals few examples of theorists who impose any continuity assumptions on social welfare functions. Chichilnisky, supra note 182, at 337 ("The vast literature on Social Choice makes little use of the continuity properties of the social aggregation rule."); see also Sen, supra note 9 (surveying the literature on social choice theory). In this literature, theorists have used continuity assumptions to derive welfarist social welfare functions. E.g., Eric Maskin, A Theorem on Utilitarianism, 45 REV. ECON. STUD. 93, 94 (1978) (showing that a social welfare function must be utilitarian if it features several properties, including a continuity property); Roberts, supra note 175, at 428 (showing that a social welfare function must be welfarist if it satisfies various conditions, including a continuity condition). Imposing continuity assumptions in conjunction with other conditions is often tantamount to assuming that only welfarist theories are reasonable. Chichilnisky defends her continuity assumption as "natural," arguing that "it is desirable for the social rule to be relatively insensitive to small changes in individual preferences" so that "mistakes in identifying preferences" are "less crucial." Chichilnisky, supra note 182, at 337. This pragmatic defense of continuity, however, lacks force as a reason to impose a restriction on our ideal social welfare function, which indicates what welfare we would assign to
219. It would, for example, rule out a utilitarian theory that first applied the criterion of average utility to rank alternatives and then applied the criterion of total utility as a tiebreaker when the principle of average utility yields a tie. Suppose two states y and z yield the same average level of utility, but state y features a larger population, so that y yields greater total utility than z. Our tiebreaking rule would rank y higher than z. Now consider a third state x, which is just like z, except it includes a slightly higher level of consumption of some good and thus yields a slightly higher level of utility for each individual in z. Our hypothetical theory would rank x above not only z but also y, no matter how small the increase in average utility in x over z, in violation of Kaplow and Shavell's continuity assumption. As another example, consider the "lexical difference principle" proposed by Rawls. RAWLS, supra note 53, at 83. Sen applies this principle to individual utilities and refers to the result as the "leximin" rule. Sen, supra note 9, at 1115-21. This rule first ranks states according to the utility of the worst-off individual. If that individual is indifferent among some states, this rule then ranks those states according to the utility of the second-worst-off individual, and so forth. The leximin rule also violates Kaplow and Shavell's continuity assumption. Consider a society with two individuals and two states y and z. Suppose the worst-off person is indifferent between these two states, but the best-off person prefers y over z. The leximin rule will rank y higher than z. Now consider a third state x, which is just like z except that a slightly higher level of consumption of some good produces a slight increase in utility for each individual. The leximin rule will rank x higher than not only z but also y, no matter how small the increase in utility in x over z. Thus, Kaplow and Shavell's continuity assumption implies that "their real quarry is much broader" than their focus on fairness principles would suggest. Fried, supra note 69, at 39.
220. See Fried, supra note 69, at 21-32 (questioning Kaplow and Shavell's continuity assumption).
221. I use the term "fairness information" here to refer to the information necessary to rank alternatives under criterion F.
222. KENNETH J. ARROW, SOCIAL CHOICE AND INDIVIDUAL VALUES 26 (2d ed. 1963). Others have since restated Arrow's formal condition in informal terms similar to those used in our independence condition.
223. E.g., ALFRED F. MACKAY, ARROW'S THEOREM: THE PARADOX OF SOCIAL CHOICE 8 (1980) ("This requires that the social ordering of a given set of alternatives depend only on the individuals' preference orderings of those alternatives."); MUELLER, supra note 139, at 386 ("The social choice between any two alternatives must depend only on the orderings of individuals over these two alternatives, and not on their orderings over other alternatives."); Sen, supra note 10, at 539 ("The social ranking of any pair of states must be the same as long as the individual utility information about the pair remains the same . . . ."). Our condition is more general in that it allows the use of more information in ranking a pair of alternatives.
224. In her critique of Kaplow and Shavell, Fried also notes that combining fairness principles with the Pareto principle can produce conflicts with the requirement of transitivity. Fried, supra note 69, at 20-21, 33-35. Fried proposes to solve this problem by ruling out any preferences that produce conflicts between the Pareto principle and fairness principles. Id. at 35-36. The solution that I propose solves the same problem without ruling out such preferences.
225. ARROW, supra note 201, at 27.
226. Id.
227. Karni, supra note 132, at 573 (emphasis omitted).
228. MUELLER, supra note 139, at 393; see also MACKAY, supra note 201, at 48 (describing Arrow's independence condition as "the most controversial of the lot").
229. Sen, supra note 1, at 156.
230. MACKAY, supra note 201, at 92-93.
231. ARROW, supra note 201, at 110.
232. MUELLER, supra note 139, at 394. Mueller stresses that social choice processes that violate Arrow's condition might be subject to "abuse" or "strategic misrepresentation of preferences." Id. at 395. Similarly, MacKay notes that Arrow's condition requires a social welfare function to use only information that can be "reliably ascertained" and avoids "strategic manipulation" of the social choice system. MACKAY, supra note 201, at 36.
233. We would first identify the alternatives that are optimal under F. Second, if there is more than one optimum, then we would apply a welfarist principle to rank the fairness optima and to identify a smaller subset of optima. If any of these optima are also Pareto optimal, then declare these alternatives to be our optimal social choices, much as F* would choose an alternative in set A. If none are Pareto optimal, then this subset corresponds to set C under F*. Move all alternatives that Pareto-dominate at least one of these alternatives in this subset to a higher ranking set, corresponding to set B under F*, and repeat the entire procedure on this Pareto superior set. Continue these iterations until an optimum is also Pareto optimal.
234. Kelly, supra note 132, at 139.
235. The probability of an exact tie for the optimum is virtually zero.
236. KAPLOW & SHAVELL, supra note 4, at 4.
237. As Gibbard points out, libertarian rights may be inconsistent with one another and we have to qualify these rights in some way to ensure a system of rights that is consistent with itself even before we consider the conflict between these rights and the Pareto principle. Gibbard supra note 120, at 388-97. Given that "[t]he problem of internal consistency of the kind with which Gibbard is concerned anses only with preference configurations requiring rather other-oriented motivations, Sen agrees that "the weakening of these 'rights' in the presence of other-oriented motivations would not seem to involve any great violations of libertarianism." Sen, supra note 2, at 235. Christian Seidl explores the necessary conditions for liberalism, which he takes to be a system that allows any "coalition" to do what it wants such that "the joint outcome can actually be put into action." Christian Seidl, On Liberal Values, 35 ZEITSCHRIFT FUR NATIONALÖKONOMIE 257, 279 (1975).
238. RAWLS, supra note 53, at 48.
239. Id. at 50.
240. Any Paretian social welfare function that incorporates fairness notions must violate independence or continuity conditions or both over an unrestricted domain. See Roberts, supra note 175, at 428 (showing that imposing independence and weak continuity conditions along with the weak Pareto principle over an unrestricted domain implies a welfarist social welfare function).
241. RAWLS, supra note 53, at 50.