Toward a Reconstruction of Utility and Welfare Economics

Economics and Ethics

It is now generally accepted among economists, at least pro forma, that economics per se cannot establish ethical judgments. It is not sufficiently recognized that to accept this need not imply acceptance of the Max Weber position that ethics can never be scientifically or rationally established. Whether we accept the Max Weber position, or we adhere to the older view of Plato and Aristotle that a rational ethics is possible, it should be clear that economics by itself cannot establish an ethical position. If an ethical science is possible, it must be built up out of data supplied by truths established by all of the other sciences.

Medicine can establish the fact that a certain drug can cure a certain disease, while leaving to other disciplines the problem whether the disease should be cured. Similarly, economics can establish that Policy A leads to the advancement of life, prosperity, and peace, while Policy B leads to death, poverty, and war. Both medicine and economics can establish these consequences scientifically, and without introducing ethical judgments into the analysis. It might be protested that doctors would not inquire into possible cures for a disease if they did not want a cure, or economists would not investigate causes of prosperity if they did not want the result. There are two answers to this point: (1) that this is undoubtedly true in almost all cases, but not necessarily so — some doctors or economists may care only about the discovery of truth, and (2) this only establishes the psychological motivation of the scientists; it does not establish that the discipline itself arrives at values. On the contrary, it bolsters the thesis that ethics is arrived at apart from the specific sciences of medicine or economics.

Thus, whether we hold the view that ethics is a matter of non-rational emotions or taste, or whether we believe in a rational ethic, we must agree that economic science per se cannot establish ethical statements. As political policy judgment is a branch of ethics, the same conclusion applies to politics. If prosperity vs. poverty, for example, are political alternatives, economic science cannot decide between them; it simply presents the truth about the consequences of each alternative political decision. As citizens, we take these truths into account when we make our politico-ethical decisions.

The Problem of the New Welfare Economics: The Unanimity Rule

The problem of “welfare economics” has always been to find some way to circumvent this restriction on economics, and to make ethical, and particularly political, statements directly. Since economics discusses individuals’ aiming to maximize their utility or happiness or welfare, the problem may be translated into the following terms: When can economics say that “society is better off” as a result of a certain change? Or alternatively, when can we say that “social utility” has been increased or “maximized”?

Neoclassical economists, led by Professor Pigou, found a simple answer. Economics can establish that a man’s marginal utility of money diminishes as his money-income increases. Therefore, they concluded, the marginal utility of a dollar is less to a rich man than to a poor man. Other things being equal, social utility is maximized by a progressive income tax which takes from the rich and gives to the poor. This was the favorite demonstration of the “old welfare economics,” grounded on Benthamite utilitarian ethics, and brought to fruition by Edgeworth and Pigou.

Economists continued blithely along this path until they were brought up short by Professor Robbins. Robbins showed that this demonstration rested on interpersonal comparisons of utility, and since utility is not a cardinal magnitude, such comparisons involve ethical judgments.46 What Robbins actually accomplished was to reintroduce Pareto’s Unanimity Rule into economics and establish it as the iron gate where welfare economics must test its credentials.47 This Rule runs as follows: We can only say that “social welfare” (or better, “social utility”) has increased due to a change, if no individual is worse off because of the change (and at least one is better off). If one individual is worse off, the fact that interpersonal utilities cannot be added or subtracted prevents economics from saying anything about social utility. Any statement about social utility would, in the absence of unanimity, imply an ethical interpersonal comparison between the gainers and the losers from a change. If X number of individuals gain, and Y number lose, from a change, any weighing to sum up in a “social” conclusion would necessarily imply an ethical judgment on the relative importance of the two groups.48

The Pareto-Robbins Unanimity Rule conquered economics and liquidated the old Pigovian welfare economics almost completely. Since then, an enormous literature known as the “new welfare economics” has flourished, devoting itself to a series of attempts to square the circle: to assert certain political judgments as scientific economics, while still retaining the Unanimity Rule.

Professor Robbins’s Escape Route

Robbins’s own formulation of the Unanimity Rule far undervalues the scope of its restrictive power over the assertions of economists. Robbins stated that only one ethical assertion would be necessary for economists to make interpersonal comparisons: namely, that every man has an “equal capacity for satisfaction” in similar circumstances. To be sure, Robbins grants that this ethical assumption cannot be established by economics; but he implies that since all good democrats are bound to make this egalitarian assumption, we can all pretty well act as if interpersonal comparisons of utility can be made and go on to make ethical judgments.

In the first place, it is difficult, upon analysis, to make sense of the phrase “equal capacity for satisfaction.” Robbins, as we have seen, admits that we cannot scientifically compare utilities or satisfactions between individuals. But since there is no unit of satisfaction by which we can make comparisons, there is no meaning to any assumption that different men’s satisfactions will be “equal” to any circumstances. “Equal” in what way, and in what units? We are not at liberty to make any ethical assumption we please, because even an ethical assumption must be framed meaningfully, and its terms must be definable in a meaningful manner. Since there is no meaning to the term “equality” without some sort of definable unit, and since there is no unit of satisfaction or utility, it follows that there can be no ethical assumption of “equal capacity for satisfaction,” and that this cannot provide a shortcut to permit the economists to make conclusions about public policy.

The Robbins position, moreover, embodies a highly oversimplified view of ethics and its relation to politico-economic affairs. The problem of interpersonal comparisons of utility is only one of the very many ethical problems which must at least be discussed before any policy conclusions can rationally be framed. Suppose, for example, that two social changes take place, each of which causes 99 percent of the people to gain in utility and one percent to lose. Surely no assumption about the interpersonal comparison of utility can suffice to establish an ethical judgment, divorced from the content of the change itself. If, for example, one change was the enslavement of the one percent by the 99 percent, and the other was the removal of a governmental subsidy to the one percent, there is apt to be a great deal of difference in our ethical pronouncements on the two cases, even if the assumed “social utility” in the two cases is approximately the same.

The Compensation Principle

“There is no meaning to the term ‘equality’ without some sort of definable unit.”

A particularly notable attempt to make policy conclusions within the framework of the Unanimity Rule was the Kaldor-Hicks “compensation principle,” which stated that “social utility” may scientifically be said to increase, if the winners may be able to compensate the losers and still remain winners.49 There are many fatal errors in this approach. In the first place, since the compensation principle is supposed to help economists form policy judgments, it is evident that we must be able to compare, at least in principle, actual social states. We are therefore always concerned with actual, and not potential, winners and losers from any change. Whether or not the winners may compensate the losers is therefore irrelevant; the important question is whether the compensation does, in fact take place. Only if the compensation is actually carried out so that not a single person remains a loser, can we still assert a gain in social utility. But can this compensation ever be carried out? In order to do so, everybody’s utility scale would have to be investigated by the compensators. But from the very nature of utility scales this is an impossibility. Who knows what has happened to anyone’s utility scale? The compensation principle is necessarily divorced from demonstrated preference, and once this occurs, it is impossible to find out what has happened to anyone’s utility. The reason for the divorce is that the act of compensation is, necessarily, a unilateral gift to a person rather than an act of that person, and therefore it is impossible to estimate how much his utility has increased as compared to its decrease in some other situation. Only if a person is actually confronted with a choice between two alternatives can we say that he prefers one to the other.

Certainly, the compensators could not rely on questionnaires in a situation where everyone need only say that he has lost utility in order to receive compensation. And suppose someone proclaims that his sensibilities are so hurt by a certain change that no monetary reward could ever compensate him? The existence of one such person would annul any compensation attempt. But these problems necessarily occur when we leave the realm of demonstrated preference.

The Social Welfare Function

Under the impact of criticisms far less thoroughgoing than the above, the compensation principle has been abandoned by most economists. There have been recent attempts to substitute another device — the “Social Welfare Function.” But after a flurry of activity, this concept, originated by Professors Bergson and Samuelson, quickly struck rocky waters, and virtually sank under the impact of various criticisms. It came to be regarded as an empty and therefore meaningless concept. Even its founders have given up the struggle and concede that economists must import ethical judgments from outside economics in order to make policy conclusions.50

Professor Rothenberg has made a desperate attempt to salvage the social welfare function by radically changing its nature, that is, by identifying it with an existing “social decision-making process.” To uphold this shift, Rothenberg must make the false assumption that “society” exists apart from individuals and makes “its” own valuation. Furthermore, as Bergson has pointed out, this procedure abolishes welfare economics, since the function of the economist would be to observe empirically the social decision-making process at work and to pronounce its decisions as gains in “social utility.”

The Economist as Adviser

Failing the establishment of policy conclusions through the compensation principle or the social welfare function, there is another very popular route to enable the economist to participate in policy formation while still remaining an ethically neutral scientist. This view holds that someone else may set the ends, while the economist is justified in telling that person (and in being hired by that person) the correct means for attaining these desired ends. Since the economist takes someone else’s hierarchy of ends as given and only points out the means to attain them, he is alleged to remain ethically neutral and strictly scientific. This viewpoint, however, is a misleading and fallacious one. Let us take an example suggested by a passage in Professor Philbrook’s seminal article; a monetary economist advising the Federal Reserve System.51 Can this economist simply take the ends set by the heads of this System and advise on the most efficient means to attain them? Not unless the economist affirms these ends as being positively good, that is, not unless he makes an ethical judgment. For suppose that the economist is convinced that the entire Federal Reserve System is pernicious. In that case, his best course may well be to advise that policy which would make the System highly inefficient in the pursuit of its ends. The economist employed by the System cannot, therefore, give any advice whatever without abandoning ethical neutrality. If he advises the System on the best way to achieve its ends, it must be logically inferred that he supports these ends. His advice involves no less an ethical judgment on his part if he chooses to “tacitly accept the decisions of the community as expressed through the political machinery.”52

The End of Welfare Economics?

After twenty years of florid growth, welfare economics is once more confined to an even tighter Unanimity Rule. Its attempts to say anything about political affairs within the confines of this rule have been in vain.

The death of the New Welfare Economics has begun to be reluctantly recognized by all of its supporters, and each has taken turns in pronouncing its demise.53 If the strictures advanced in this paper are conceded, the burial rites will be accelerated, and the corpse decently interred. Many New Welfare Economists understandably continue to grope for some way of salvaging something out of the wreckage. Thus, Reder suggests that economics make specific, piecemeal policy recommendations anyway. But surely this is only a despairing refusal to take the fundamental problems into account. Rothenberg tries to inaugurate a constancy assumption based on psychologizing about underlying basic personalities.54 Aside from the fact that “basic” changes can take place at any time, economics deals with marginal changes, and a change is no less a change for being marginal. In fact, whether changes are marginal or basic is a problem for psychology, not praxeology. Bergson tries the mystical route of denying demonstrated preference, and claiming it to be possible that people’s values “really differed” from what they chose in action. He does this by adopting the “consistency”-constancy fallacy.

Does the Unanimity Rule then spell the end of all possible welfare economics, as well as the “old” and the “new” versions? Superficially, it would seem so. For if all changes must injure nobody, that is, if no people must feel worse off as a result of a change, what changes could pass muster as socially useful within the Unanimity Rule? As Reder laments:

Consideration of the welfare implications of envy, for example, make it impossible even to say that welfare will be increased by everyone having more of every commodity.55

• 46 Lionel Robbins, “Interpersonal Comparisons of Utility,” Economic Journal (December 1938): 635–41; and Robbins, An Essay on the Nature and Significance of Economic Science, 2nd ed. (London: Macmillan, 1935), pp. 138–41.
• 47 Vilfredo Pareto, Manuel d’va^conomie Politique, 2nd ed. (Paris: Marcel Giard, 1927), p. 617.
• 48 Kemp tries to alter the Unanimity Rule to read that social utility is only increased if everyone is better off, non being worse off or indifferent. But, as we have seen, indifference cannot be demonstrated in action, and therefore this alteration is invalid. Murray C. Kemp, “Welfare Economics: A Stocktaking,” Economic Record (November 1954): 245.
• 49 On the compensation principle, see Nicholas Kaldor, “Welfare Propositions in Economics,” Economic Journal (September 1939): 549; John R. Hicks, “The Foundations of Welfare Economics,” Economic Journal (December 1939): 706. For a criticism, see William J. Baumol, “Community Indifference,” Review of Economic Studies (1946–1947): 44–48; Baumol, Welfare Economics and the Theory of the State, pp. 12 ff; Kemp, “Welfare Economics: A Stocktaking,” pp. 246–50. For a summary of the discussion, see D.H. Robertson, Utility and All That (London: Allen and Unwin, 1952): pp. 29–35. The weakness in Robbin’s accession to the Unanimity Rule is demonstrated by his endorsement of the compensating principle. Robbins, “Robertson on Utility and Scope.”
• 50 See Abram Bergson, “On the Concept of Social Welfare,” Quarterly Journal of Economics (May 1954): 249; Paul A. Samuelson, “Welfare Economics; Comment,” in A Survey of Contemporary Economics, Vol. II, B.F. Haley, ed. (Homewood, Ill.: R.D. Irwin, 1952), 2, p. 37. Also Jerome Rothenberg, “Conditions for a Social Welfare Function,” Journal of Political Economy (October 1953): 397; Sidney Schoeffler, “Note on Modern Welfare Economics,” American Economic Review (December 1952): 881; I.M.D. Little, “Social Choice and Individual Values,” Journal of Political Economy (October 1952): 422–32.
• 51 Clarence Philbrook, “ ‘Realism’ in Policy Espousal,” American Economic Review (December 1953): 846–59. The entire article is of fundamental importance in the study of economics and its relation to public policy.
• 52 E.J. Mishan, “The Principle of Compensation Reconsidered,” Journal of Political Economy (August 1952): 312. See especially the excellent note of I.M.D. Little, “The Scientist and the State,” Review of Economic Studies (1949–50): 75–76.
• 53 Thus, see the rather mournful discussion in the American Economic Association’s second volume of the Survey of Contemporary Economics; Kenneth E. Boulding, “Welfare Economics,” pp. 1–34; Melvin W. Reder, “Comment,” pp. 34–36; and Samuelson, The Empirical Implications of Utility Analysis. Also see the articles by Schoeffler, Bergson, and Kemp cited Above.
• 54 Jerome Rothenberg, “Welfare Comparisons and changes in Tastes,” American Economic Review (December 1953): 888–90.
• 55 Reder, “Comment,” p. 35.