Should preferences count? (resource allocation)

Sen, A. K. 1977. "Rational Fools: A Critique of the Behavioral Foundations of Economic Theory." Philosophy and Public Affairs 6 (4):327-44.

Skinner, B. F. 1953. Science and Human Behavior. New York: Macmillan.

Tversky, A., and D. Kahneman. 1981. "The Framing of Decisions and the Psychology of Choice." Science 211:453-58.

Wright, G. H. von. 1963. The Logic of Preference. Edinburgh: Edinburgh University Press.

Watson, Gary. 1975. "Free Agency." Journal of Philosophy 72 (8):205-20. I. INTRODUCTION

Welfare economics presents itself as a normative discipline: its goal is "to evaluate the social desirability of alternative allocations of resources" (Henderson and Quandt 1958, 222) and "to formulate propositions by which we may rank, on the scale of better or worse, alternative economic situations open to society" (Mishan 1967, 156). Normative welfare economics bases itself on "one fundamental ethical postulate"--that the preferences of individuals are to count in the allocation of resources (Samuelson 1953, 223; Quirk and Saponsik 1968, 104). "In this framework, preferences are treated as data of the most fundamental kind. Value, in the economic sense, is ultimately derived from individual preferences. . . ." (Randall 1981, 156).

How can we justify the postulate that preference-satisfaction should be a principal goal of resource and environmental policy? This essay proposes the modest thesis that this question is harder to answer than one might think.

The Thesis of this Paper

The normative postulate of welfare economics proposes that preferences--simply insofar as or because they are preferences--should count in allocation to the extent people are willing to pay to satisfy them. The postulate distinguishes among preferences not according to their worthiness or reasonableness, but simply on the basis of their intensity as measured by willingness-to-pay.

Why is it a good thing, all else being equal, that preferences are satisfied--preferences taken as they come, on a willingness-to-pay basis, constrained by income and bounded by indifference between alternatives? The answers that welfare economists offer appeal either to the concept of choice or to the concept of well-being. In order to show that these concepts fail to ground economics as a normative science, I shall drive a wedge between preference and choice and between preference-satisfaction and well-being.

II. PREFERENCE AND BEHAVIOR

"I'd rather be sailing," reads the bumper sticker on the car ahead, but it is obvious that the driver is not sailing. He is sitting in a traffic jam. He is listening to the radio. He is daydreaming, changing lanes, driving south, going to Disney World, drinking coffee, and tailgating. He is . . . well, a thousand descriptions--a million stories--are all consistent with the behavior we observe.

Pick one of these descriptions--the man is driving to Disney World--and you may posit any number of explanations. He wants to please the wife, honor a promise he made to his children, get away from the office . . . . Describe his behavior another way--he is sitting in a traffic jam--and other explanations suggest themselves: a wrong turn, bad luck, or, perhaps, a preference for traffic jams. There are a million ways to describe his behavior and thus a million ways to explain it--but no description conforms to the preference he expresses. He is not sailing.

"If I had my druthers . . ." introduces a statement about one's preferences; it is in the subjunctive mood. That is the way it is with preferences--if by "preferences" we refer to the ordinary wants and desires we take pleasure in satisfying. We rarely get to act on our "druthers" because other reasons, motives, and circumstances tend to take precedence over them. And it is a good thing that they do. As cartoonist Jules Feiffer said recently: "I've discovered that the best decisions are the ones that are forced on you, and if you had been given your druthers, your life would have been far more miserable" (Ball 1992, 62).

The choices we make are to be understood in relation to many factors besides our inclinations, for example, personal ties, moral commitments, religious taboos, wrong turns, second thoughts, sheer fatigue, missed opportunities, guilt trips, and contingent events forced upon us, such as our children's bloody noses. How often does a person--a responsible adult with small children, let us say--get to indulge his or her preferences? A little study of sleep deprivation among yuppie parents should answer that question.

This is not to deny that economists can predict where the consumer dollar--for example, the recreational dollar--will be spent. The Disney Corporation, for example, predicts sales at its theme parks on the basis of information about demographics, the economy, market trends, and so on. The preferences of the people who spend the money (parents) have little to do with it. Which room or dad at Disney World, surrounded by screaming children standing in line for the Safari Jungle Ride and importuned for cotton candy, would not at the same price prefer to enjoy a bone-dry martini at a Club Med? Socrates in his speech in the Symposium says, in effect, that once a person gets married and has children, he can pack up all his druthers and kiss them good-bye.

And as Jules Feiffer adds, it's a good thing--for otherwise the "lightness of being," which is to say, the continuous opportunity to indulge preferences, would make a person's life even more miserable. Disney pitches its advertising not to parents but to children. It knows whose preferences count--and how to manipulate those preferences.

My neighbor wants to be on time, but she is always late when she drives the car pool. This is not her preference; it's a bad habit. We refer to habits, character traits, education, socioeconomic background, ethnic identity, age, and other such factors to predict how people will behave. Market researchers and advertisers depend on these variables--along with figures about past performance--to predict what people will buy. I doubt that "preference schedules" or "maps," as these terms occur in the theory of welfare economics, are useful in market research. Zip codes may have more predictive value.

Ad Hoc Explanations in Psychology

Empirical psychologists, who try to explain behavior in terms of its causal conditions, do not give preferences a prominent position among them. Some psychologists--B. F. Skinner is an example--argue that preferences, being private mental states, are not observable and therefore cannot enter testable scientific explanations. Skinner (1953, 72) wrote that anyone "who readily engages in a given activity is not showing an interest, he is showing the effect of reinforcement." He argued mental states such as desires and preferences are "invented on the spot to provide spurious explanations" for behavior (Skinner 1953, 30). Skinner disparaged preference-based explanations of behavior not only because preferences, being mental states, are unobservable, but also because explanations based on them are ad hoc. You have first to describe a person's behavior--and the description you choose already implies the "preferences" that then "explain" it.

Critics lined up to lodge the same charge against Skinner's invocation of reinforcing events as explanatory causes of what he called operant behavior (Scriven 1956; Chomsky 1959). These critics pointed out that Skinner and his colleagues successfully explained and predicted the behavior of laboratory rats because there was so little of it. Two psychologists watching the behavior of a laboratory rat would agree in their description of the reinforcing event (a food pellet), the stimulus (a flashing light), and the behavior (treadle pushing) they observe. In situations this simple--this well controlled--scientific inquiry can succeed whether or not it invokes "mental" states. What is crucial is that the behavior to be explained (treadle pushing) is identifiable separately from the conditions that are supposed to explain it (food pellets)--that the cause does not vary with (and is not determined by) the way we happen to describe the effect.

In contrast to the limited number of ways we may describe behavior in laboratory rats, there are at least a million ways to characterize the plight of parents in line for the Jungle Safari. On the basis of naive observation appropriate to laboratory rats, one might say that these parents either prefer to stand in line rather than walk around or that they want to go on the Safari Ride--otherwise why would they line up for it? We should then say that Dad prefers the Safari Ride to a sailboat (in spite of his bumper sticker) and cotton candy to a dry martini. But nothing could be more implausible than that.

If we describe what is going on differently, we can infer a very different set of preferences. Our description of the behavior in question might refer not to Jungle Safaris but to promises; we may recall that Dad, feeling guilty after a week-long business trip, hastily said he would take the children to Disney World. Accordingly, the behavior we observe is that of "keeping promises" rather than "standing in line." And we may infer that Dad prefers keeping his promises to going sailing.

In examples involving human beings rather than laboratory rats, raw or naive observation will not secure agreement about the nature of the behavior to be explained. Rather, we have to interpret what we see--we have to tell a story--and the way we describe the event determines the story we tell about its causal conditions. Dad chooses to go on the Safari Ride rather than on a sailboat: does that mean he prefers the Safari to sailing? Shall we say instead that he prefers to keep his promises than to indulge his druthers? Maybe Dad prefers cotton candy to dry martinis. Maybe he just can't say "no" to his kids. We can tell what laboratory rats are doing, but who knows what reasons lurk in the heart of man? Only the Shadow knows. Everyone else proposes an interpretation.

There are a million ways to describe what a person is doing: standing in line, minding the children, buying cotton candy, keeping a promise, avoiding the office, gritting his teeth, daydreaming about sailing, doing Disney World; and whichever description we choose determines the "reinforcing conditions" or, if you like, the "preference ordering" we invoke to explain it. Describe the individual's behavior one way--as "appeasing one's spouse"--and you then must specify one set of preferences, reinforcers, …