Social welfare state intervention, and value judgments.

• Author: Lemieux, Pierre

The state can promote efficiency--that is, the production of more goods and services; in a second stage, it can, if deemed necessary, redistribute the supplementary output made possible by its efficiency-enhancing interventions. Efficiency and distribution are two different issues, from both a conceptual and a public-policy viewpoint. If redistribution obviously requires value judgments, wealth creation does not because having more goods is always desirable) As the conventional wisdom puts it, we have to bake the cake first and cut the slices later.

So thought many economists until the 1950s. By that time, however, the "new welfare economics" had all but destroyed these conclusions and shown that value judgments are also required for creating wealth--indeed, for even defining it. Although the bases of this conclusion is now unquestioned among scholars cognizant of neoclassical welfare economics, its implications seem to have escaped most people, including many economists.

These implications are monumental because if wealth creation by the state is not neutral, but requires values judgments as much as income redistribution, the justification for all economic interventions by the state is questionable. Interestingly, this conclusion is not what many, perhaps most, developers of welfare economics thought they were proving: they thought they were demonstrating that the free market cannot maximize real income, and they were trying to establish the conditions under which the state can intervene to do so. As Melvin Reder has observed, "To a considerable extent, welfare (and related) theorizing in the 1930's and '40's was an attempt to show the variety and importance of the circumstances under which laissez-faire was inappropriate" (qtd. in Rothbard 1956, 249). Whatever their conclusions about market efficiency, the theorists of the new welfare economics quite clearly demonstrated that the state must necessarily fail in its efficiency-maximizing task unless it rules out failure by resorting to its own arbitrary value judgments. In other words, if the state promotes efficiency, it is only efficiency as the state arbitrarily defines it by taking from Peter and giving to Paul. To demonstrate this conclusion is my objective in this essay.

The New Welfare Economics

The new welfare economics, which is now orthodox welfare economics, was born in the late 1930s, with the work of theorists such as Nicholas Kaldor (1908-86), Tibor Skitovsky (1910-2002), and John Hicks (1904-89). Paul Samuelson (1915-), with his extraordinary analytical skills (whatever one thinks of his politics), was soon to join the demolition team of the new welfare theorists. The new welfare economics brought two major changes in the economic study of social welfare (Blaug 1968, 592-99; Mishan [1960] 1968).

First, this analysis abandoned any notion of measurable, cardinal "utility" (or satisfaction), relying instead on Hicks's (1939) reformulation of neoclassical economics in terms of purely ordinal utility and Vilfredo Pareto's definition of social welfare. Ordinal utility refers to a mere ranking of different bundles of goods and services in an individual's preferences. An individual's utility function gives this ranking by using an index in which the numbers have no meaning other than "more preferred" (for higher numbers), "less preferred" (for lower numbers), or "indifferent" (for equal numbers). We can say only that a given bundle of goods is preferred to another given bundle, or the other way around, or that the two bundles are indifferent, and the numbers used to express the relation do not matter, provided that higher numbers attach to more preferred bundles. Neoclassical economists express this condition by saying that an individual's utility function is "unique except for a monotonic transformation" (Henderson and Quandt 1971, 46). Because utility is only a ranking, interpersonal utility comparisons are meaningless. (2) When two individuals exchange an orange for an apple, we can say only that the one who parts with his apple to get an orange ranked the latter more highly than the former, whereas the ranking was the reverse for the other party. There is no way to know whether one gets more or less satisfaction than the other, whether one is made "better off" more than the other: we know only that each benefited from the exchange according to his own preferences.

The impossibility of interpersonal comparisons of utility immediately raises a major problem: How can we talk of "social welfare" if we can't add the utility of the individuals who gain (from, say, new electronic calculators) and deduct the utility of those who lose (the previous manufacturers of slide-rules)? How can we calculate social welfare from the welfare or utility of the individuals who make up society? Because we can't compare or, a fortiori, add utilities, we can define society's welfare (also called "society's real income") only when there is unanimity: social welfare increases when at least one individual is better off and no one is worse off, a condition known as Pareto improvement, in recognition of Italian economist Vilfredo Pareto (1848-1923), who invented this approach. Alternatively, social welfare decreases when at least one individual is worse off and nobody is better off--a Pareto deterioration. When at least one individual gains and at least one other loses, we cannot make any welfare judgment in the Pareto sense. The Paretian concepts allow us, albeit in a restricted way, to make some individualistic sense of the concept of "society's welfare," an otherwise ascientific and troubling concept inasmuch as society is nothing apart from the individuals who compose it. The modern notion of economic efficiency is borrowed from Pareto (Blaug 1968, chap. 13): an action or a policy is efficient if it brings about a Pareto improvement; a situation is efficient if all possible Pareto improvements have been made so that nobody can be made better off without at least one individual being made worse off. An efficient situation is often called Pareto optimal. Thus, as a first approximation, a policy that benefits some individuals and harms others cannot be said to be a Pareto improvement, to be efficient.

However--and this is the second change the new welfare economics made--we might want to extend the concept of Pareto improvement to potential Pareto improvement. We then say, a la new welfare economics, that a public policy increases social welfare, even if some individuals have lost, provided that the ones who gained have gained enough to compensate the losers for their losses. If the users and makers of electronic calculators gain so much that they can buy off the unemployed makers of slide-rules, we have a potential Pareto improvement and thus an increase in social welfare, even if the compensation is not actually paid. This example suggests one reason for using potential Pareto improvement and thus potential compensation instead of the actual thing: in many cases, it is difficult or impossible to determine who has gained and who has lost following a change in social and economic configuration. For example, has a former employee of a slide-rule manufacturer lost if he has found a new job or if he hasn't found one because he knows that compensation is available?

Reflecting on the notion of "real income" provides another path to the heart of our topic. What is an individual's real income? The most general concept of real income is related to his utility: it moves up or down as he can consume more preferred or less preferred bundles of goods and services. In general, only an individual knows whether his utility, or real income, has increased or decreased after his money income and the prices of the goods and services have changed. Utility is subjective. An individual's preferences can be fuzzy for the individual himself, but he certainly knows more about them than any external observer. In certain cases, though, an external observer can discover part of an individual's preferences by observing the bundles of goods chosen at different prices--that is, through revealed preference. If, following income and price changes, a given individual chooses a bundle that was not available to him before, even when he could still choose the bundle that he was consuming previously, an argument by revealed preference allows us to say that the individual's real income has increased. He has revealed that he prefers the actual bundle to the one he consumed previously. If, on the contrary, the individual now consumes a bundle that he could afford before but chose not to consume, while the former bundle is not affordable in the new income and price situation, his revealed preference tells us that his real income has decreased. If you bought a BMW when you could have afforded a Mercedes, your choice has revealed that the BMW (together with the other goods in the bundle you consume) is higher in your preference index than the latter; but if you bought a Mercedes when you could not afford a BMW, your choice reveals nothing about your preferences. An individual's preferences are revealed only with regard to what he chooses and what he could have chosen but declined to. In many cases, comparing two bundles of goods at different sets of prices does not reveal preferences because neither bundle could have been chosen at the prices under which the other was chosen. Think about the consumers who could afford a house but not a computer at the prevailing set of relative prices of thirty years ago and now cannot afford a house but can buy a computer at today's prices: their choices have revealed nothing about whether they prefer a house or a computer.

In the case of a society--that is, of a group of individuals--we cannot use revealed-preference arguments to determine if a change in nominal national income or gross domestic product (GDP) (which is to society what an individual's income is to him)...