II. Demonstrated Preference

A Statement of the Concept

Human action is the use of means to arrive at preferred ends. Such action contrasts to the observed behavior of stones and planets, for it implies purpose on the part of the actor. Action implies choice among alternatives. Man has means, or resources, which he uses to arrive at various ends; these resources may be time, money, labor energy, land, capital goods, and so on. He uses these resources to attain his most preferred ends. From his action, we can deduce that he has acted so as to satisfy his most highly valued desires or preferences.

The concept of demonstrated preference is simply this: that actual choice reveals, or demonstrates, a man’s preferences; that is, that his preferences are deducible from what he has chosen in action. Thus, if a man chooses to spend an hour at a concert rather than a movie, we deduce that the former was preferred, or ranked higher on his value scale. Similarly, if a man spends five dollars on a shirt we deduce that he preferred purchasing the shirt to any other uses he could have found for the money. This concept of preference, rooted in real choices, forms the keystone of the logical structure of economic analysis, and particularly of utility and welfare analysis.

While a similar concept played a role in the writings of the early utility economists, it had never received a name, and it therefore remained largely undeveloped and unrecognized as a distinct concept. It was generally discarded in the 1930s, before it had even achieved recognition. This view of preference as derived from choice was present in varying degree in the writings of the early Austrian economists, as well as in the works of Jevons, Fisher, and Fetter. Fetter was the only one who clearly employed the concept in his analysis. The clearest and most thorough formulation of the concept has been the works of Professor Mises.1

Positivism and the Charge of Tautology

Before developing some of the applications of the demonstrated preference principle to utility and welfare theory, we must consider the methodological objections that have been leveled against it. Professor Alan Sweezy, for example, seizes on a sentence of Irving Fisher’s which very succinctly expressed the concept of demonstrated preference: “Each individual acts as he desires.” Sweezy is typical of the majority of present-day economists in not being able to understand how such a statement can be made with absolute validity. To Sweezy, insofar as it is not an empirically testable proposition in psychology, such a sentence must simply reduce to the meaningless tautology: “each individual acts as he acts.”

This criticism is rooted in a fundamental epistemological error that pervades modern thought: the inability of modern methodologists to understand how economic science can yield substantive truths by means of logical deduction (that is, the method of “praxeology”). For they have adopted the epistemology of positivism (now dubbed “logical empiricism” or “scientific empiricism” by its practitioners), which uncritically applies the procedures appropriate in physics to the sciences of human action.2

In physics, simple facts can be isolated in the laboratory. These isolated facts are known directly, but the laws to explain these facts are not. The laws may only be hypothesized. Their validity can only be determined by logically deducing consequents from them which can be verified by appeal to the laboratory facts. Even if the laws explain the facts, however, and their inferences are consistent with them, the laws of physics can never be absolutely established. For some other law may prove more elegant or capable of explaining a wider range of facts. In physics, therefore, postulated explanations have to be hypothesized in such a way that they or their consequents can be empirically tested. Even then, the laws are only tentatively rather than absolutely valid.

In human action, however, the situation is reversed. There is here no laboratory where “facts” can be isolated and broken down into their simple elements. Instead, there are only historical “facts” which are complex phenomena, resultants of many causal factors. These phenomena must be explained, but they cannot be isolated or used to verify or falsify any law. On the other hand, economics, or praxeology, has full and complete knowledge of its original and basic axioms. These are the axioms implicit in the very existence of human action, and they are absolutely valid so long as human beings exist. But if the axioms of praxeology are absolutely valid for human existence, then so are the consequents which can logically be deduced from them. Hence, economics, in contrast to physics, can derive absolutely valid substantive truths about the real world by deductive logic. The axioms of physics are only hypothesized and hence subject to revision; the axioms of economics are already known and hence absolutely true.3 The irritation and bewilderment of positivists over the “dogmatic” pronouncements of praxeology stem, therefore, from their universal application of methods proper only to the physical sciences.4

The suggestion has been made that praxeology is not really scientific, because its logical procedures are verbal (”literary”) rather than mathematical and symbolic.5 But mathematical logic is uniquely appropriate to physics, where the various logical steps along the way are not in themselves meaningful; for the axioms and therefore the deductions of physics are in themselves meaningless, and only take on meaning “operationally,” insofar as they can explain and predict given facts. In praxeology, on the contrary, the axioms themselves are known as true and are therefore meaningful. As a result, each step-by-step deduction is meaningful and true. Meanings are far better expressed verbally than in meaningless formal symbols. Moreover, simply to translate economic analysis from words into symbols, and then to retranslate them so as to explain the conclusions, makes little sense, and violates the great scientific principle of Occam’s Razor that there should be no unnecessary multiplication of entities.

The crucial concept of the positivists, and the one that forms the basis for their attack on demonstrated preference, is that of “operational meaning.” Indeed, their favorite critical epithet is that such and such a formulation or law is “operationally meaningless.”6 The test of “operationally meaningful” is derived strictly from the procedures of physics as outlined above. An explanatory law must be framed so that it can be tested and found empirically false. Any law which claims to be absolutely true and not empirically capable of being falsified is therefore “dogmatic” and operationally meaningless — hence, the positivist’s view that if a statement or law is not capable of being falsified empirically, it must simply be a tautologous definition. And consequently, Sweezy’s attempted reduction of Fisher’s sentence to a meaningless identity.7

Sweezy objects that Fisher’s “each man acts as he desires” is circular reasoning, because action implies desire, and yet desires are not arrived at independently, but are only discoverable through the action itself. Yet this is not circular. For desires exist by virtue of the concept of human action and of the existence of action. It is precisely the characteristic of human action that it is motivated by desires and ends, in contrast to the unmotivated bodies studied by physics. Hence, we can say validly that action is motivated by desires and yet confine ourselves to deducing the specific desires from the real actions.

Professor Samuelson and “Revealed Preference”

“Revealed preference” — preference revealed through choice — would have been an apt term for our concept. It has, however, been preempted by Samuelson for a seemingly similar but actually quite different concept of his own. The critical difference is this: Samuelson assumes the existence of an underlying preference scale that forms the basis of a man’s actions and that remains constant in the course of his actions over time. Samuelson then uses complex mathematical procedures in an attempt to “map” the individual’s preference scale on the basis of his numerous actions.

The prime error here is the assumption that the preference scale remains constant over time. There is no reason whatever for making any such assumption. All we can say is that an action, at a specific point of time, reveals part of a man’s preference scale at that time. There is no warrant for assuming that it remains constant from one point of time to another.8

The “revealed preference” theorists do not recognize that they are assuming constancy; they believe that their assumption is simply that of consistent behavior, which they identify with “rationality.” They will admit that people are not always “rational,” but uphold their theory as being a good first approximation or even as having normative value. However, as Mises has pointed out, constancy and consistency are two entirely different things. Consistency means that a person maintains a transitive order of rank on his preference scale (if A is preferred to B and B is preferred to C, then A is preferred to C). But the revealed preference procedure does not rest on this assumption so much as on an assumption of constancy — that an individual maintains the same value scale over time. While the former might be called irrational, there is certainly nothing irrational about someone’s value scales changing through time. Hence, no valid theory can be built on a constancy assumption.9

One of the most absurd procedures based on a constancy assumption has been the attempt to arrive at a consumer’s preference scale not through observed real action, but through quizzing him by questionnaires. In vacuo, a few consumers are questioned at length on which abstract bundle of commodities they would prefer to another abstract bundle, and so on. Not only does this suffer from the constancy error, no assurance can be attached to the mere questioning of people when they are not confronted with the choices in actual practice. Not only will a person’s valuation differ when talking about them from when he is actually choosing, but there is also no guarantee that he is telling the truth.10

The bankruptcy of the revealed-preference approach has never been better portrayed than by a prominent follower, Professor Kennedy. Says Kennedy: “In what respectable science would the assumption of consistency (that is, constancy) be accepted for one moment?”11 But he asserts it must be retained anyway, else utility theory could not serve any useful purpose. The abandonment of truth for the sake of a spurious usefulness is a hallmark of the positivist-pragmatist tradition. Except for certain auxiliary constructions, it should be clear that the false cannot be useful in constructing a true theory. This is particularly the case in economics, which is explicitly built on true axioms.12

Psychologizing and Behaviorism: Twin Pitfalls

The revealed-preference doctrine is one example of what we may call the fallacy of “psychologizing,” the treatment of preference scales as if they existed as separate entities apart from real action. Psychologizing is a common error in utility analysis. It is based on the assumption that utility analysis is a kind of “psychology,” and that, therefore, economics must enter into psychological analysis in laying the foundations of its theoretical structure.

Praxeology, the basis of economic theory, differs from psychology, however. Psychology analyzes the how and the why of people forming values. It treats the concrete content of ends and values. Economics, on the other hand, rests simply on the assumption of the existence of ends, and then deduces its valid theory from such a general assumption.13 It therefore has nothing to do with the content of ends or with the internal operations of the mind of the acting man.14

If psychologizing is to be avoided, so is the opposite error of behaviorism. The behaviorist wishes to expunge “subjectivism,” that is, motivated action, completely from economics, since he believes that any trace of subjectivism is unscientific. His ideal is the method of physics in treating observed movements of unmotivated, inorganic matter. In adopting this method, he throws away the subjective knowledge of action upon which economic science is founded; indeed, he is making any scientific investigation of human beings impossible. The behaviorist approach in economics began with Cassel, and its most prominent modern practitioner is Professor Little. Little rejects the demonstrated preference theory because it assumes the existence of preference. He glories in the fact that, in his analysis, the maximizing individual “at last disappears” which means, of course, that economics disappears as well.15

The errors of psychologizing and of behaviorism have in common a desire by their practitioners to endow their concepts and procedures with “operational meaning,” either in the areas of observed behavior or in mental operations. Vilfredo Pareto, perhaps the founder of an explicitly positivist approach in economics, championed both errors. Discarding a demonstrated preference approach as “tautologous,” Pareto, on the one hand, sought to eliminate subjective preferences from economics and, on the other, to investigate and measure preference scales apart from real action. Pareto was, in more ways than one, the spiritual ancestor of most current utility theorists.16, 17

A Note on Professor Armstrong’s Criticism

Professor Armstrong has delivered a criticism of the revealed-preference approach which he would undoubtedly apply to demonstrated preference as well. He asserts that when more than one commodity is being ranked, individual preference scales cannot be unitary, and we cannot postulate the ranking of the commodities on one scale.18 On the contrary, it is precisely the characteristic of a deduced preference scale that it is unitary. Only if a man ranks two alternatives as more and less valuable on one scale can he choose between them. Any of his means will be allocated to his more preferred use. Real choice therefore always demonstrates relevant preferences ranked on a unitary scale.

1 See Alan R. Sweezy, “The Interpretation of Subjective Value Theory in the Writings of the Austrian Economists,” Review of Economic Studies (June 1934): 176–85, for an historical survey. Sweezy devotes a good part of the article to a criticism of Mises as the leading exponent of the demonstrated preference approach. For Mises’s views, see Human Action (New Haven, Conn.: Yale University Press, 1949), pp. 94–96, 102–3; Theory of Money and Credit (1912, 3 rd ed; New Haven: Yale University Press, 1951), pp. 46ff. Also see Frank A. Fetter, Economic Principles (New York: The Century Co., 1915), pp. 14–21.
2 See the methodological treatises of Kaufman, Hutchison, Souter, Stonier, Myrdal, Morgenstern, and so on.
3 On the methodology of praxeology and physics, see Mises, Human Action, and F.A. Hayek, The Counter Revolution of Science (Glencoe, Ill.: The Free Press, 1952), pt 1.
4 It is even dubious that positivists accurately interpret the proper methodology of physics itself. On the widespread positivist misuse of the Heisenberg Uncertainty Principle in physics as well as in other disciplines, cf. Albert H. Hobbs, Social Problems and Scientism (Harrisburg, Penn.: The Stackpole Co., 1953), pp. 220–32.
5 For a typical suggestion, cf. George J. Schuller, “Rejoinder,” American Economic Review (March 1951): 188. For realization that mathematical logic is essentially subsidiary to basic verbal logic, cf. the remarks of Andre Lalande and Rene Poirier, on “Logique” and “Logistique,” in Vocabulaire technique et critique de la philosophie, Andre Lalande, ed., 6th ed. (Paris: Presses Universitaires de France, 1951), pp. 574, 579.
6 Paul Samuelson has added the weight of his authority to Sweezy’s criticism of Mises and demonstrated preference, and has couched his endorsement in terms of “operational meaning.” Samuelson explicitly rejects the idea of a true utility theory in favor of one that is merely hypothetical. See Paul A. Samuelson, “The Empirical Implications of Utility Analysis,” Econometrica (1938):344ff; and Samuelson, Foundations of Economic Analysis (Cambridge, Mass.:Harvard University Press, 1947), pp. 91–92. The concept of operational meaning was originated by the physicist Percy W. Bridgman explicitly to explain the methodology of physics. Cf. Bridgman, The Logic of Modern Physics (New York: Macmillan, 1927). Many founders of modern positivism, such as Mach and Boltzmann, were also physicists.
7 The heroes of positivism, Rudolf Carnap and Ludwig Wittgenstein, disparaged deductive inference as merely drawing out “tautologies” from the axioms. Yet all reasoning is deductive, and this process is peculiarly vital to arriving at truth. For a critique of Carnap and Wittgenstein, and a demonstration that inference is not merely identity to “tautology,” cf. Lalande, “Tautoglie,” in Vocabulaire, pp. 1103–4.
8 Samuelson’s analysis suffers from other errors as well, such as the use of invalid “index number” procedures. On the theoretical fallacies of index numbers, cf. Mises, Theory of Money and Credit, pp. 187–94.
9 See Mises, Human Action, pp. 102–3. Mises demonstrates that Wicksteed and Robbins committed a similar error.
10 It is Samuelson’s credit that he rejects the questionnaire approach. Professors Kennedy and Keckskemeti, for different reasons, defend the questionnaire method. Kennedy simply says, rather illogically, that in vacuo procedures are being used anyway, when the theorist states that more of a good is preferred to less. But this is not in vacuo; it is a conclusion based on the praxeological knowledge that since a good is any object of action, more must be preferred to less while it remains a good. Kennedy is wrong, therefore, when he asserts that this is a circular argument, for the fact that action exists is not “circular.” Keckskemeti actually asserts that the questionnaire method is preferable to observing behavior in discovering preferences. The basis of his arguments is a spurious dichotomy between utility and ethical valuations. Ethical valuations may be considered either as identical with, or a subset of, utility judgments, but they can not be separated. Cf. Charles Kennedy, “The Common Sense of Indifference Curves,” Oxford Economic Papers (January 1950): 123–31; Kenneth J. Arrow, “Review of Paul Keckskemeti’s Meaning, Communication, and Value,” Econometrica (January 1955): 103.
11 Kennedy, “The Common Sense of Indifference Curves.” Kennedy’s article furnishes the best brief explanation of the revealed-preference approach.
12 This error again stems from physics, where such assumptions as absence of friction are useful as first approximations — to known facts from unknown explanatory laws! For a refreshing skepticism on the value of false axioms, cf. Martin Bronfenbrenner, “Contemporary Economics Resurveyed,” Journal of Political Economy (April 1953).
13 The axiom of the existence of ends may be considered a proposition in philosophical psychology. In that sense, praxeology is grounded in psychology, but its development then completely diverges from psychology proper. On the question of purpose, praxeology takes its stand squarely with the Leibnizian tradition of philosophical psychology as opposed to the Lockean tradition upheld by positivists, behavorists, and associationists. For an illuminating discussion of this issue, cf. Gordon W. Allport, Becoming (New Haven, Conn.: Yale University Press, 1955), pp. 6–17.
14 Thus, the law of diminishing marginal utility does not at all rest on some postulated psychological law of satiety of wants, but on the praxeological truth that the first units of a good will be allocated to the most valuable uses, the next units to the next-most valuable uses, and so on.
15 I.M.D. Little, “A Reformulation of the Theory of Consumers’ Behavior,” Oxford Economic Papers (January 1949): 90–99.
16 Vilfredo Pareto, “On the Economic Phenomenon,” International Economic Papers 3 (1953): 188–94. For an excellent rebuttal, cf. Benedetto Croce, “On the Economic Principle, Parts I and II,” ibid.: 175–76. 201. The famous Croce-Pareto debate is an illuminating example of early debate between praxeological and positivist views in economics.
17 Vivian C. Walsh is an interesting current example of the combinations of both types of error. On the one hand, he is an extreme behaviorist, who refuses to recognize thatany preferences are relevant to, or can be demonstrated by, action. On the other hand, he also takes the extreme psychologizing view that psychological states per se can be directly observed. For this, he falls back on “common sense.” But this position fails because Walsh’s psychological “observations” are ideal types and not analytic categories. Thus, Walsh says that: “saying that someone is a smoker is different from saying that he is smoking now,” upholding the former type of statement for economics. But such statements are historical ideal types, relevant to history and psychology, but not to economic analysis. Cf. Vivian C. Walsh, “On Descriptions of Consumers’ Behavior,” Economica (August 1954): 244–52. On ideal types and relation to praxeology, cf. Mises, Human Action, pp. 59–64.
18 Wallace E. Armstrong, “A Note on the Theory of Consumer’s Behavior,” Oxford Economic Papers (January 1950): 199ff. On this point, cf. Little’s rebuttal, in I.M.D. Little, “The Theory of Consumer’s Behavior — A Comment,” ibid., 132–35.