The Bankruptcy of Economics
Rational choice theory dominates orthodox microeconomics, and as we have already mentioned, it is having an increasing influence on the social sciences in general. The broad rational choice approach explains social phenomena by reference to specific individual psychological properties, in particular: (1) it assumes that agents are self-regarding in their desires and actions and (2) in acting to satisfy these self-regarding desires agents will act in accordance with the formal theory of rationality represented by modern decision theory — typically Bayesian decision theory (Pettit, 1993, 265). In this chapter we will attack the behavioral foundations of modern economics, primarily by an attack upon decision theory. Our aim in this chapter and the next is to show that the theory of “rational economic man” is untenable. We hope to show that orthodox economics is damaged at the most fundamental epistemological level.
KeywordsCoherence Smoke Nash Kelly Defend
Unable to display preview. Download preview PDF.
- 3.Homa Katouzian (1980, 60–7) has argued that the ordinal theory of utility is ultimately dependent upon the measurability of utility and hence upon the cardinal theory of utility.Google Scholar
- 4.Arrow’s proof has recently been challenged by Howard De Long in his A Refutation of Arrow’s Theorem (1991). He attempts to undermine Arrow’s Impossibility Theorem by a self-referential argument, by letting an object of social choice be the repeal of Arrow’s axioms such as the Independence of Irrelevant Alternatives (De Long, 1991, 41–6). Now by this strategy one could undermine any theorem in social choice theory. What this argument establishes is that a self-referential paradox can be generated if one violates the condition of Unrestricted Scope and considers logically incoherent sets of individual preference rankings. De Long’s self-referential argument will be rejected by Arrowian theorists, precisely because he violates the condition of Unrestricted Scope in setting up his self-referential argument.Google Scholar
- 5.It could be objected here that the circularity at the heart of the frequency definition of probability is not problematic because we have a clear pre-analytic understanding of probability. Bart Kosko in Fuzzy Thinking (Kosko, 1994, 44–64) argues that we do not. It is difficult to accept the axioms of probability as unexplained primitives in his opinion because we do not “catch probability in the act” (Kosko, 1994, 52). Kosko’s position is that fuzzy logic is a more basic theory than probability: the universe is deterministic but fuzzy; probability is a special part of fuzziness (Kosko, 1990).Google Scholar
- 7.One notable and still unresolved problem for Bayesian confirmation theory is the problem of “old evidence” (Glymour, 1980, 85–93). In science evidential support for a theory often exists long before the theory is proposed. For example, the anomalous advance of the perihelion of Mercury was known before Einstein’s General Relativity. Now according to Bayesian confirmation theory, Pr(h, e) > Pr(h). But if e is known to be true before h then Pr(e) equals 1 at that later time, so that at a later time Pr(h, e) must equal Pr(h), by the definition of Pr(h, e) and the axioms of the probability calculus. Hence known evidence cannot support newly invented theories. This problem has generated a debate eg (Eells, 1985; Earman, 1989; Howson, 1991). Zynda (1995) after surveying the main arguments in the field has concluded that the problem is still unresolved. Again, we take this to be an objection to Bayesianism.Google Scholar
- 8.Adopting a slightly wider definition of coherence (this being preferences that conform to the axioms of expected utility theory) assume that each person has coherent preferences. According to Bayesianism, we cannot reasonably assume that agents will make the same initial assessment of the probability of every event. However it is reasonable according to received economics to assume that rational agents are Paretian; that is, if everyone is indifferent between two objects of choice, then the social preference should also be indifferent and if one person prefers A to B and no one else prefers B to A, then the social preference should be that A is preferred to B. But there is an inconsistency which exists between all of these conditions, captured in the Probability Agreement Theorem: if each person has coherent preferences then social preferences cannot be both coherent and Paretian unless everybody agrees about the probability of every event (Broome, 1989b, 8). According to Broome, this conflict arises because the coherence condition enables us to attribute probabilities to social preferences but the Paretian condition allows individual preferences to reflect different probability assessments, so social preferences cannot reflect consistent probability assessments. Broome notes that the Probability Agreement Theorem can be proved without even the coherence condition, requiring only that people have some preferences (Broome, 1989b, 9). The paradox cannot be resolved by abandoning the coherence assumption. It is, in our opinion, yet another nail in the coffin of orthodox economics and decision theory.Google Scholar
- 11.In response to those who claim that it does not make sense to compare certain things with respect to preference or indifference, the economist may dismiss such concerns, insisting that since every commodity has a price or a cost (even if this is only an opportunity cost) all commodities can be compared. As Marc Linder (1977, 137) has shown, this argument is circular because the very point of formulating a consumer’s preference ranking was to account for phenomena such as prices and income, not the other way around.Google Scholar