How Confusing Keynes's Normative Decision Theory in the a Treatise on Probability and General Theory with Kahneman and Tversky's 'Heuristics and Biases (Errors)', Descriptive Theory Leads to Complete Confusion

Abstract

Economists, working in the Heterodox schools of economics, have severely confused Keynes’s interval valued probability–weight of the evidence approach to decision making from the A Treatise on Probability, that Keynes integrated into the General Theory by way of his definition of uncertainty on p.148 in chapter 12 of the General Theory, with the “Heuristics and Biases (errors)” approach of Kahneman and Tversky.

Uncertainty was defined in the General Theory as an inverse function of the weight of the evidence concept from the A Treatise on Probability. Liquidity Preference was then defined as a positive function of uncertainty and confidence was defined as an inverse function of uncertainty. The probabilities used in the General Theory are automatically intervals except in the very extreme cases of no uncertainty (Modern Bayesian, Subjectivist economists) or total and complete uncertainty (ignorance) used by the Heterodox schools under the name of 'fundamental uncertainty' or 'irreducible uncertainty' by way of GLS Shackle.

The Kahneman –Tversky “Heuristics and Biases(errors)” approach is purely descriptive. The normative theory recommended and supported by Kahneman and Tversky is the Subjectivist, Bayesian approach of Ramsey, de Finetti and Savage. The Tversky and Kahneman claim, rejected by L Jonathan Cohen, a logical probability theorist like Keynes, is that decision makers are horribly ignorant of the laws of the probability calculus when making decisions and thus make all kinds of errors and mistakes that could be corrected if the decision makers would restudy basic probability and statistical theory. Keynes and Cohen argue, in direct conflict with Tversky and Kahneman, that the laws of the probability calculus can’t be applied in general because there is a substantial amount of relevant evidence that is missing or unavailable to the decision maker at the time that he must make a decision. Keynes argues that the decision maker usually relies on interval valued probability, combined with assessments of weight as they relate to liquidity that are tied in to the confidence the decision maker has in his own estimates of probability.

Attempts to reinterpret Keynes’s approach in terms of Kahneman and Tversky must fail because there is no weight of the evidence variable incorporated into the normative approach of Tversky and Kahneman.

Keywords: weight, interval vaued probability, upper and lower probabilities, nonnumerical probabilites, liquidity preference

JEL Classification: B10, B12, B14, B16, B20, B22