3. Democratic Values
Through Equality to Utility
The rule "to maximize society's utility, equalize incomes" gains validity once incomes have been equal for long enough.
No man has more than one stomach, but this is a thin basis for holding that the more equally all goods are shared, the better.
It is part of our intellectual heritage that whatever else it may do which we hold for or against it, equalization of incomes will maximize their utility. The intuitive support which helps this proposition over the more obvious obstacles is that an extra dollar must mean more to the poor man than to the rich. On reflection, all that intuition really strongly supports is that a given absolute sum increases the poor man's utility relatively more (say, ten-fold) than that of the rich (say, by a tenth). Nothing in these "cardinal" comparisons of the poor man's initial utility with its increase, and the rich man's initial utility with its increase, enables us to compare the two utilities, or the two increases, between them either "ordinally" (in terms of bigger or smaller) or "cardinally" (by how much bigger).
One view of this problem (with which, as chapter 2 has shown, I can't help but concur) is that we cannot do this because conceptually it just cannot be done, because interpersonal comparisons are intrinsically misdirected enterprises. If they are undertaken, all they can possibly be known to express is the preferences of whoever is making the comparison, and that is the end of the matter. Pursuing it beyond this point can take us into the analysis of these preferences. We will then be dealing with questions of ideology, sympathy, compassion, party politics, raison d'état and so forth. These or other elements can perhaps explain why the comparison fell out the way it did. They will not shed any further light on the utilities purported to have been compared.
However, the contrary view seems also to be tenable. It must be, if only because it is held by some of the most incisive minds who have addressed this problem. Thus, Little feels able to make "rough-and-ready," and Sen "partial," interpersonal comparisons of utility. The positive case, as distinct from the normative one, for giving some of the rich man's money to the poor man is that the same money, differently distributed, has more utility. Unless it is granted, for argument's sake, that such comparisons make sense, there is no factual case to prove, only moral judgements to be set one against the other and, as Bentham ruefully put it, "all practical reasoning is at an end."
Yet the intellectual tradition of discovering in equality an enabling cause of greater utility, is a positive one. Central to it is a conviction that we are dealing with matters of fact and not of sympathy. Some such conviction, albeit unconsciously and implicitly, conditions an important strand of the liberal argument about the distribution of the national income and optimum taxation.*20 It seems to me worthwhile to meet it on that ground, as if utilities could be compared and added up to social utility, and as if it was social science which told us that one distribution of income was superior to another.
Let me recapitulate—"retrieve from the political subconscious" would be a truer description—the reasoning behind this conviction. It goes back at least to Edgeworth and Pigou (the former taking a more general, and also more cautious, view) and provides a robust example of the capacity of a dated theory to inspire practical contemporary thought with undiminished vigour.
At bottom the theory rests on a basic convention of economics which gives rise to fruitful theories in various branches of it, labelled the Law of Variable Proportions. The convention consists in assuming that if different combinations of two goods or factors yield the same utility (in consumption) or output (in production), the increments of utility or output obtained from combining increasing quantities of the one with a constant quantity of the other, are a decreasing function of the variable, i.e. each increase in its quantity will yield a smaller increment of utility or output than the preceding one. In theories of consumers' behaviour, this is also described as the "principle of diminishing marginal utility," "the convexity of indifference curves" or "the falling marginal rate of substitution" of the fixed for the variable good.
Now if a person is given more and more tea while his other goods do not increase, the utility, satisfaction or happiness he derives from successive doses of tea diminishes. The intuitive support for presuming this resides in the fixity of his bundle of other goods. ("Presumption" is employed advisedly. A hypothesis framed in terms of utility or satisfaction must be a presumption, as it cannot be disproved by experiment or observation unless the context is one of uncertain alternatives, see below.) The same presumption stands for any single good when all the other goods stay fixed. However, it cannot be aggregated. What is presumably true of any single good is not even presumably true of the sum of goods, i.e. income. As income increases, all goods potentially or actually increase. What, then, is the relevance of "knowing" that the marginal utility of each good falls if the quantity of the others remains fixed? The diminishing marginal utility of tea conditions the mind to acceptance of the diminishing marginal utility of income, but the temptation to argue from one to the other is a trap.
A presumption can be established for the falling marginal utility of income by defining income as all goods except one (which stays fixed when income rises), e.g. leisure. It is possible to suppose that the more income we have, the less leisure we would give up to earn additional income. However, if the falling marginal utility of income is a consequence of excluding one good from income, then it cannot be applied to a concept of income which excludes no good. If any good can be exchanged at some price against any other including leisure, which is by and large the case in market economies, income is potentially any and all goods, and none can be supposed fixed to give rise to falling marginal utility for the sum of the rest.
It is well established that the realm of certainties—where we are sure to get a pound of tea if only we ask for it and pay the shopkeeper the price—does not lend itself to observation of the marginal utility of income. Meaningful observation of the rate of change of utility as income changes, however, is conceptually possible in the face of risky choices. The pioneer study of lotteries and insurance, as evidence relevant to the shape of the utility function, strongly suggested that the marginal utility of income may be falling in certain income brackets and rising in others, consistent with a hypothesis that changes of income which leave a man in his class have, in a sense, a lesser value than changes giving access to a quite different kind of life: "[a man] may jump at an actuarially fair gamble that offers him a small chance of lifting him out of the class of unskilled workers and into the 'middle' or 'upper' class, even though it is far more likely than the preceding gamble to make him one of the least prosperous unskilled workers."*21 We must note (and mentally carry forward to the next two sections of this chapter) that this is the precise obverse of the type of valuation of income which is supposed to induce rational people to adopt a "maximin" defence of their interest in Rawls's Theory of Justice.*22
Now anyone who carelessly reasons as if there could be a means, independent of the observation of choices involving risk, for ascertaining the marginal utility of income, is apt to say that some positive or negative utility may attach to the taking of risks, so that what risky choices measure is the marginal utility of income plus/minus the utility of being at risk, of gambling. Whether we would like it to mean more, or less, to say that there is positive utility in exposure to risk means to say that the marginal utility of income is rising. That a person is adverse to risk (declines fair gambles or is willing to bear the cost of hedging), is no more, and no less, than evidence in support of the hypothesis that the marginal utility of his income is falling. No other proof, over and above the evidence drawn from risky choices, can be produced for it. People's answers to hypothetical questions about how much "utility" or "importance" they attach to successive tranches of their actual (or prospective) income, are not admissible evidence.*23 It is baffling to be told that the observable evidence (risk-avoidance, or risk-taking) somehow adds to or subtracts from the inferred condition (the falling or rising "marginal utility of income") of which it is the sole symptom and whose existence it alone affirms.
There is no "law" of the diminishing marginal utility of income. Educational and career choices, financial and other futures markets,*24 insurance and gambling provide abundant evidence that all shapes of utility functions may occur, falling, constant or rising; that one and the same person's marginal utility may change direction over different ranges of income, and that there is no obvious predominance of one type of function, the others being freaks. Not surprisingly, no theory of utility maximization by promoting a particular distribution of income could be built on so general and shapeless foundations.
The Edgeworth-Pigou theory in fact stands on a better basis than this, though this goes often unrecognized in bowdlerized accounts. Satisfaction derived from income in the properly stated, full theory depends on income itself and on the capacity for satisfaction. Its dependence on income alone does not really yield the standard conclusion usually associated with the theory; if all goods vary with income, the marginal utility of income need not be falling and we cannot say anything much about what an egalitarian redistribution of incomes would do to "total utility." Its dependence on the capacity for satisfaction, on the contrary, looks like leading to the desired result. As income rises in the face of a fixed capacity for satisfaction, the makings of a law of diminishing returns are all there, with intuitive support provided by the concept of satiety. If we have, then, two forces acting on the marginal utility of income and the effect of the first can go either way without any obvious bias, while the second makes marginal utility diminish, the tendency for a falling marginal utility of income could be taken as established in a probabilistic sense.
The remaining pieces fall easily into place. Only goods which can be brought into relation with the "measuring rod of money" are taken into account. People have the same tastes and pay the same prices for the same goods, hence spend a given money income the same way. For purposes of "practical reasoning," they have the same "appetites," "intensity of wants," "capacity for enjoyment" or "temperament," as the capacity for satisfaction has been interchangeably called. Inherent in the concept of capacity was the idea that it could get filled up. Successive units of income would yield successively smaller increments of utility or satisfaction as the ceiling of capacity was getting closer. Given the total income of society, total utility must obviously be the greater the more nearly equal is the marginal utility of everybody's income, for the total can always be augmented by transferring income from people having a lower marginal utility to people having a higher one. Once marginal utilities are equal all round, no further utilitarian good can be done by income transfers; total "social" utility has been maximized. Utility, satisfaction are intangibles, attributes of the mind. The visible evidence of the all-round equality of marginal utilities is that there are no rich and no poor any more.
This evidence is persuasive if we admit the requisite meaning of interpersonal comparisons (which I have decided to do for purposes of argument, to see where it gets us) and if we interpret the capacity for satisfaction (as it used to be interpreted) as physical appetite for standard goods, or as "the lower order of wants" which are the same for rich and poor, for "nobody can eat more than three meals a day," "no man has more than one stomach," etc. When, however, the capacity for satisfaction is not, or no longer, viewed in the early textbook sense of a few basic physical needs, all bets are off.*25 Though it came straight from the horse's mouth, opinion-makers and utility-maximizers never took enough notice of Edgeworth's warning: "The Benthamite argument that equality of means tends to maximum happiness, presupposes a certain equality of natures; but if the capacity for happiness of different classes is different, the argument leads not to equal, but to unequal distribution."*26
With the admission that capacities for deriving satisfaction from income may well be widely different, what is left of the injunction to take money, say, from rich fat white men and give it to poor thin brown men? Equality ceases to be the direct command of rationality, for it can no longer be identified as the road to maximum utility. Admittedly, redistributive policies could be based on differential patterns of the capacity for satisfaction while rejecting elusive utility as the end to be maximized. In the well-known example of the manic-depressive cripple, utility-maximization would call for taking money away from him since he does not get much satisfaction out of it. An alternative maximand might require throwing a million dollars at him, because it would take that much to raise his satisfaction to the level of that of the average sane and healthy person. The latter policy has the equalization of happiness (and not its maximization) as its end. It makes sense if (in order to rise to the rank of an end) equality need not be derived from the good, but is postulated to be the good.
Under the utility-maximizing tradition, two possible positions seem to remain open. One is to posit that the capacity for satisfaction is a random endowment like the ear for music or the photographic memory, and there is no sensible way to reason about where in a population it is most likely to be concentrated. If so, there is also no sensible way to judge which distribution of income is most likely to maximize utility.
The other position is to assume that although the capacity for satisfaction is not spread evenly, it is not distributed randomly either, but forms patterns which can be inferred from people's other, statistically visible characteristics, e.g. it is concentrated in the under-eighteens, in the old, in those having and in those not having an academic education, etc. Discerning the pattern restores the utilitarian rationale of recommendations to distribute society's income one way rather than another. Happily, scope is thus found again for social engineers to devise redistributive policies which increase total utility and political support for the proponent of the policy, though the coincidence of the two is probably less assured than it would be in the straightforward and classic case of rich-to-poor redistribution.
Is it not, however, reasonable to act on the assumption that the young, with their appetite for leisure, clothes and travel, music and parties have more capacity for satisfaction than the old with their weaker lusts and saturated wants? A policy of making tax rates progressive not only with income, as at present, but also with age, might be a good one both for social utility and for getting the youth vote. By the same token, since the old, with their mature culture and greater experience, are cet. par. likely to have a greater capacity for satisfaction, tax rates declining with age could both increase utility and earn the senior citizen vote. There may also be a case, on plausible grounds, for increasing the income of teachers and decreasing that of plumbers as well as for proceeding the other way round.
Moreover, it stands to reason that the intensity of wants is liable to increase with exposure to temptation, so that total utility could probably be enhanced by subsidizing, for example, readers of Sears Roebuck catalogues. On the other hand, since their enhanced capacity for satisfaction is to some extent its own reward, it would also be a good idea to tax the subsidy and distribute the proceeds among non-readers of advertisements. On balance, benefits in terms of welfare and political consent could perhaps be drawn from adopting all of these policies at the same time or in turn, although careful sample surveys would be required to make the underlying social engineering really precise.
This of course is just being unkind to the kind of earnest and well-meaning officiousness which the majority of politically aware people used to indulge in until quite recently and which some, for a variety of reasons, still practise. It deserves being made fun of. However, more serious reasoning remains to be done.
The rule "to each in proportion to his wants" as a sufficient condition of utility maximization, does not simply translate into the equalization of incomes. People's wants run to many things money can buy over and above bread and dripping, beer and pizza. It is preposterous to interpret their capacity for satisfaction in the physical sense of one man, one stomach. They are much too different for the levelling of their incomes to represent a plausible approximation to solving any maximum problem. Is there any other simple redistributive policy which would look more plausible?
Waiting in the utilitarian wings for this stage of the play are such notions as "learning by doing," "l'appétit vient en mangeant," "tastes depend on consumption" or, perhaps, "the utility of income is an increasing function of past income." They strain the conventional limits of economics, just as the notion that preferences for political arrangements are heavily conditioned by the very arrangements that actually prevail (cf. pp. 18-19), strains those of political theory. The usual and time-tested approach of these disciplines is to take tastes, preferences as given. Treating them as part of the problem may, nevertheless, be worth an occasional attempt.
Rather than assuming, too implausibly, that capacities for satisfaction are given and are much the same all round, let us therefore assume that they are conditioned by people's actual satisfactions, their culture, experience and habitual standards of living which taught them to cut their coat to their cloth, to adjust their wants and to feel relatively comfortable with the things that go with that standard. The greater have been people's incomes for some learning period, the greater will have become their capacity to derive satisfaction from them, and vice versa, though it might be wise to suppose that in the reverse direction, the learning period needed to reduce the capacity for satisfaction is much longer.
If interpersonal comparisons were "on," the impartial spectator might find that there was little to choose between the happiness gained by giving a dollar to the representative underprivileged man and the happiness lost by taking a dollar from the representative well-to-do one (before counting the happiness the one loses by being coerced and the other gains by feeling the state's helping hand under his elbow, and the impartial spectator, to do his job properly, must count these gains and losses, too). Barring the new poor and the new rich, at the end of the day there is probably no utilitarian case left for tampering with the incomes people actually have. If any policy conclusion can be supported by abstract reasoning of this sort, it may well be that the existing distribution of income, if it has prevailed for some time, is more likely than any other to maximize total utility (and if such issue to the argument disgusted people sufficiently to make them stop thinking, however unconsciously, about how to maximize social utility, the quality of political debate would no doubt improve).
Stated otherwise, if income distribution were a means to a society's greater or lesser aggregate satisfaction, the least harmful policy rule to adopt would be that every society "ought" to get the income distribution to which its members are geared by past experience. An egalitarian society, the sort Tocqueville resignedly expected to issue from democracy, where people's natures are similar, their tastes and thoughts conform to agreed norms and their economic status is uniform, "ought" in all probability to be given an egalitarian income distribution—except that it has already got it.
Levelling in a society which was inegalitarian to begin with would quite probably violate the utility-maximization criterion which it was supposed to serve. This is not, in itself, a very good argument against levelling unless one were to take social utility maximization seriously, and despite its great influence on the public subconscious, there is no really strong case for doing so. Whether for or against, arguments about the merits of levelling seem to me to need some other basis. Democratic values cannot be derived, as it were, from the rational man's guide to utility; equality is not rendered valuable by virtue of its purported contribution to the greatest happiness of the greatest number. Whether the democratic values are contained in the rational man's guide to social justice, is the question to be addressed next.
Notes for this chapter
Other liberal arguments about redistribution are not positive but normative; they deal with values, not facts; their recommendations are supported by appeals to social justice rather than social utility.
M. Friedman and L. J. Savage, "The Utility Analysis of Choices Involving Risk," in American Economic Association, Readings in Price Theory, 1953, p. 88. First published in Journal of Political Economy, 56, 1948.
J. Rawls, Theory of Justice, 1972, p. 156. The second and third "features" invoked by Rawls to explain why his people do what they do mean, respectively, that a rise in his "index of primary goods" (which is stated to be co-variant with his income tout court) would not make the Rawls man significantly better off, and a fall would make him intolerably worse off.
"Not even the chooser himself knows his preference until he is confronted with an actual choice, and his understanding of his own preferences is to be doubted unless he is in a real choice situation." (Charles E. Lindblom, Politics and Markets, 1977, p. 103.) If this stand looks a little too severe with regard to the simplest, tea-rather-than-coffee preference relation, it is no more than properly cautious when applied to whole modes of life.
I say "other" futures markets to stress that financial markets are ipso facto markets in futures, e.g. in future interest and dividends.
Thus Robert Wolff in Understanding Rawls, 1977, p. 173: "A full belly of beer and pizza requires very little money, but a cultivated, tasteful, elegant lifestyle, rationally managed in order to 'schedule activities so that various desires can be fulfilled without interference' costs a bundle."
F. Y. Edgeworth, The Pure Theory of Taxation, 1897, reprinted in Edgeworth, Papers Relating to Political Economy, 1925, p. 114, my italics.
End of Notes
Return to top