Menger the Empiricist

14 September 2007 at 9:54 am Peter G. Klein 5 comments

| Peter Klein |

Austrian economists eschew empirical analysis in favor of deductive, a priori reasoning. They don’t believe in prediction. Neoclassical economists, by contrast, endorse the “scientific method” of rigorous empirical testing. You know that, right?

Then you might be surprised to learn that Carl Menger (1840-1921), founder of the Austrian school, called his approach the “empirical method,” as distinguished from Léon Walras’s “rational method.” Menger was a prominent economic journalist before turning to scientific work and his primary interest, as a scholar, was to explain the actual pricing processes he observed in the marketplace, processes that did not at all resemble those described in contemporary textbooks. Menger’s purpose, writes Guido Hülsmann in Mises: Last Knight of Liberalism, was

to demonstrate that the properties and laws of economic phenomena result from these empirically ascertainable “elements of the human economy” such as individual human needs, individual human knowledge, ownership and acquisition of individual quantities of goods, time, and individual error. Menger’s great achievement in [Principles of Economics, 1871] consisted in identifying these elements for analysis and explaining how they cause more-complex market phenomena such as prices. He called this the “empirical method,” emphasizing that it was the same method that worked so well in the natural sciences.

This empirical method does not involve the use of abstract (and deliberately unrealistic) postulates to formulate testable hypotheses, as is appropriate for the study of natural phenomena. Rather, for the social sciences, empirical analysis involves the systematic construction of a causal theory based on observation of basic empirical phenomena such as human wants, stocks of goods, technical knowledge, and so on.

In a 1884 letter to Walras Menger criticized the use of mathematics in economic analysis. The merit of a theory, Menger observed,

always depends on the extent to which it succeeds in determining the true factors (those that correspond to real life) constituting the economic phenomena and the laws according to which the complex phenomena of political economy result from the simple elements. . . .

A researcher who arrives by the way of analysis at such elements that do not correspond to reality or who, without any true analysis, takes his departure from arbitrary axioms — which is only too often the case with the so-called rational method — falls necessarily into error, even if he makes superior use of mathematics.

The quotes (and their translations) are provided by Hülsmann, whose discussion of Menger (pp. 101-40 of Last Knight) provides a valuable overview of, and commentary on, the Austrian approach. Hülsmann is influenced by William Jaffé’s important work on the differences between Menger, Walras, and Jevons but goes beyond Jaffé in important ways. For example, Hülsmann emphasizes the much-overlooked influence of Hermann Heinrich Gossen on Walras’s and Jevons’s systems. While Menger’s approach emphasized causal explanation and real phenomena such as preference, Gossen focused on an abstract, psychological concept of utility (“want-satisfaction”) that is measurable and comparable across individuals. As Hülsmann explains (pp. 131-33):

In Menger’s theory, the term “value” does not refer to a psychological feeling, but rather to the relative importance for an individual of the marginal unit of good X — that is, to the importance of X in comparison to the marginal units of other goods Y and Z. The market price of a good results from the interplay of sellers and buyers, for whom the goods bought and sold have different relative importance. In contrast, in the theories of the other three authors, the price of a good results from the interplay of sellers and buyers whose feelings or well-being are differently affected by control of the good. While Menger explained the pricing process as resulting from the importance of a good relative to the importance of other goods, Gossen, Jevons, and Walras explained the pricing process as the impact of a marginal quantity of a good on the psychology of the actor — an impact they called want-satisfaction (Gossen), utility (Jevons), and satisfied needs (Walras). Jevons’s marginal utility thus played structurally the same role that marginal value played in Menger’s theory — it delivered an explanation of market prices — but where marginal utility explains the price of a good by the good’s direct impact on human feelings, Menger’s marginal value explains the price of a good by how the good ranks in importance compared to other goods, according to the needs of the individuals involved in the pricing process. . . .

Whatever else one might think of the merits of the psychological approach, it had at least one great attraction, namely, that it allowed the possibility of a mathematical price theory based on marginal utility. With the human psyche as the common denominator of all economic values, it became conceivable to represent the want-satisfaction or utility derived from the consumption of a good as a mathematical function of the quantities consumed; it became conceivable to scale satisfaction and utility into units with which one could perform economic calculation completely disconnected from market prices. . . .

These considerations probably played a role in prompting Gossen, Jevons, and Walras to choose the psychological approach. They did not begin with observation and then adopt algebraic and geometric techniques as the most adequate tools for representing what they observed. Rather, they began with an agenda — the need to apply mathematics in economics to make it more “scientific” — and were looking for a plausible hypothesis to justify their preferred approach.

Entry filed under: - Klein -, Austrian Economics, Methods/Methodology/Theory of Science.

More Presentation Tips Social Science Factoid of the Day

5 Comments Add your own

1. srp | 14 September 2007 at 4:23 pm

I’m afraid this passage from Hulsmann encapsulates the sterility of much Austrian method-mongering.

1) What, pray tell, is the non-psychological meaning of “the relative importance for an individual” of an increment of X versus an increment of Y? Maybe this is intended to be a Samuelsonian argument that only revealed preferences are meaningful–it only means that the individual shows a stable propensity to buy so much of X and so much of Y in a given situation–but that kind of operationalism is usually thought to be an anathema to Austrians. In reality, this concept of “relative importance” is just as “psychological” as marginal utility. (Of course, we all know about ordinal versus cardinal utilities, but mathematical theory can cover both, so that can’t be a point of comparison.)

2) Why is Menger’s approach inherently non-mathematical? For example, I often employ the concept of “willingness to pay” for a given product as a buyer characteristic that can be compared across rival products. That characteristic might be psychological, but it could also be based on the derived demand of a business or the consumption technology of an end user. In either case, it can be thought of as a mathematical function mapping product characteristics and other transaction features to dollars per unit. Similarly, willingness to trade off X for Y at various endowment levels is eminently mathematical.

3) What deep empirical research, other than basic introspection, did Menger do to come up with his assumptions? My wild guess–practically none. Both approaches are “armchair economics”–perfectly justified in this case, I might add–and to suggest that “relative importance of X versus Y” is more empirical than “utility of X versus Y” is farcical.

4) Note the anti-formalization sneering at the end of the passage. Gossen, Jevons, and Walras were allegedly trying to puff themselves up and pretend to be “scientific” by applying mathematical techniques, unlike the humbly empirical Austrians. Please. This is just as odious (if possibly less justified) than accusing anti-mathemetizers of not knowing/enjoying math.
2. Peter Klein | 14 September 2007 at 4:47 pm

Steve, it’s hard to know how to address your comment, because you and Menger are not only speaking different languages, you seem to be living on different planets. The best I can suggest is that you try reading Menger’s Principles, or for that matter any of the basic works in Mengerian price theory written since 1871 (Bohm-Bawker, Fetter, Wicksteed, Davenport, Clark, Mises, Rothbard). I’ll make one remark about utility theory nonetheless.

Menger infers preference orderings from observed choices. It is like revealed preference in that sense (Rothbard uses the term “demonstrated preference” to distinguish his approach from Samuelson’s), though it assumes nothing about stability (i.e., one cannot construct a utility function by assuming constancy of preferences over time). To see the difference between this and a “psychological” approach, compare Menger’s explanation of diminishing marginal utility (a term he doesn’t use) — each additional unit of a homogenous good is assigned to a lower-valued use — with the standard one (‘the second scoop of ice cream doesn’t taste as good as the first”). Indeed, look at the breezy and superficial treatment in Mas-Colell et al., p. 44: “Convexity can . . . be viewed as the formal expression of a basic inclination of economic agents for diversification. . . . A taste for diversification is a realistic trait of economic life.” Menger assumes no such “taste,” only that economic agents employ a (subjective) ranking of ends, to which they allocate scarce means. Mas-Colell et al. continue: “Economic theory would be in serious difficulty if this postulated propsensity for diversification did not have significant descriptive content. But there is no doubt that one can easily think of choice situations where it is violated. For example, you may like milk and orange juice but get less pleasure from a mixture of the two.” Huh? This constitutes a robust, scientific approach to diminishing marginal utility? Puh-lease.
3. Rafe Champion | 14 September 2007 at 5:09 pm

It is possible that Menger the empiricist got into trouble with the twin problems of demarcation and induction (sometimes called respectively Kant and Hume’s problems). This is the message that Bill Bartley picked up from Karl Milford and he suggested that was a reason for Menger’s modest output in the last decades of his life. Popper picked up these problems later on and his solutions can perhaps be modified to fit the Austrian praxeology developed by Mises (and with the “action frame of reference” developed by Talcott Parsons “The Sructure of Social Action”.)

http://oysterium.blogspot.com/2006/07/success-and-failure-of-talcott-parsons.html

Milford and Birner contributed papers to the collection edited by Bruce Caldwell “Carl Menger and his legacy in economics” and this is a gloss on Birner’s paper, undortunately missing the introduction on Menger and proceeding to a modified version of Popper’s situational analysis.

http://oysterium.blogspot.com/2006/05/jack-birner-on-situational-analysis.html

It may help to think in terms of two different kinds of natural laws (each of which are different from the manmade laws , norms and regularities which are subject to conscious and unconscious change). There are the unchanging natural laws of the natural sciences and similarly unchanging laws of the catallaxy. Praxeology is conerned with the second kind of laws.

The laws of each kind are used by historians and everyone else to explain why things happen the way they do (and not some other way). Traditional epistemology has been conferned with the justification of those laws, however they cannot be justified (in the strong logical sense that was desired) however they can be tested and if they stand up then that is about as good as it gets.

There is more to be said about these things, but not until after breakfast.
4. srp | 14 September 2007 at 5:49 pm

Peter: You’re doing it again. What is a “lower-valued” use (for an end-user, not a business customer) if not a psychological concept? The “lower-valued use” of a second scoop of ice cream” is also eating it–it’s probably lower valued because of psychological satiation factors, although maybe your cup is too small or something.

Value is a subjective thing–that’s supposed to be the whole point of the Austrian approach. It’s not just subjective because of beliefs, or life situations, but also because of tastes. And tastes are largely psychological things (actually, so are beliefs).

It’s fine to say that we want a theory that is robust to particular psychological theories of preference formation. It’s not fine to pretend that such theories have no psychological underpinnings at all. And it seems odd to dogmatically ignore basic principles of psychophysics such as satiation and contrast effects that are directly relevant to the probable nature of preferences.

I’m not going to defend Mas-Colell’s exposition–the orange-juice/milk example suffers from a lack of specification of the time-period of consumption (i.e. you might want both in your refigerator even if you don’t consume them simultaneously). But the “psychological” aspect is equally present either way you look at it.

And of course, my points 2) to 4) above still stand.
5. Peter Klein | 14 September 2007 at 6:08 pm

Steve, you can call it “psychological” if you want, but psychology is introduced only in the trivial sense of being implied by choice. For Menger and his followers, economic analysis starts with action and deals with the implications of _the fact that_ an agent prefers a unit of X to a unit of Y. It is agnosic about the reasons _why_ X is preferred to Y. As you point out, various psychological concepts of satiation and the like are consistent with diminishing marginal utility, and that’s fine. But Menger’s approach — unlike that of Gossen, Walras, and Jevons — does not require any particular psychological underpinnings.

Anyway, there is a huge literature on all of this, including your points 2 and 3 (your point 4 seems like an editorial comment, not a statement inviting a reply). Rafe’s post above provides some references. I suggest that you try reading some of this literature, which is difficult to summarize in a few sentences. Try to have an open mind.

The point of my post, in any case, was not to demonstrate the superiority of Menger’s approach, but to point out that the conventional categories used to contrast Austrian and neoclassical approaches may be too crude to capture the relevant differences between them. There is an important sense in which Menger’s approach — which, after all, is often termed “causal-realist,” in the Aristotelian sense of realism — is strongly empirical.