A Note On Researching Keynes's Work In Probability

Abstract

Perhaps the most actively discussed aspect of Keynes's thought during the last decade has been his concern with uncertainty and probability theory. As the concerns of current macroeconomic theorists have turned increasingly to the effects of expectations and uncertainty, interest has grown in the fact that Keynes was the author of A Treatise on Probability (1921) and that uncertainty plays a prominent role in Chapter 12 of The General Theory, as well as in three 1937 papers in which he summarized The General Theory's main points. Not surprisingly, however, there has been very little agreement in this recent discussion about exactly what the significance of Keynes's early work in probability was to his later work as an economist or about what the roles of expectations and uncertainty are in The General Theory. Several commentators have argued that there is no real relationship between Keynes's earlier work and The General Theory or that uncertainty plays no significant role in his economic theorizing chief among them Coddington (1982, 1983) and Patinkin (1976. 1983). Others have argued that there is a connection between Probability and Keynes's later work, but the nature and significance of the relationship has been interpreted in many different ways. G. L. S. Shackle (1961, 1974) and others have interpreted the connection to demonstrate that Keynes was a post-Keynesian; E. Roy Weintraub (1975, 1979) has used the connection to argue that Keynes's work serves as an example of the appropriateness of general equilibrium (i.e., neo-classical, ArrowDebreu) modelling under uncertainty to macroeconomics; Allan Meltzer (1981) has used the connection to argue that Keynes was a rational expectationist. In response to this great diversity of opinion I began to research Probability and its possible connection to Keynes's other work about three years ago. The purpose of this note, though, is not to adjudicate between the previous commentators, but to report on two findings of my research which have not been considered in their work. Both findings indicate the need for further examination of Keynes's work in probability theory. I. Careful exegesis of Keynes's writings in the 1930's indicates that his fundamental conception of what probability is changed after the publication of Probability; whereas he had advocated a single theory of probability in his book, he advocated at least two different theories by 1938. The theory originally presented in Probability denied that the probability of an occurrence referred to the proportion of times that it occurred, or its relative frequency, and contended instead that it referred to the degree of belief that was held in the occurrence's actually happening. Thus, the probability of rain on a given day was not the proportion of times that rain had occurred on meteorologically similar days, but rather the degree of belief that one held that it would rain given one's information (O = certainty that it would not rain; 1 = certainty that it would). Keynes further contended that these degrees of belief were objective in the sense that all reasonable men would agree on their value given the same information. Drawing on the Greek word for knowledge, episteme, philosphers of science today call such a theory of probability an objective epistemic theory. Keynes apparently dropped his belief in the objectivity of probabilities, however, when he responded to Frank Ramsey's (1931) criticism of Probability. Ramsey contended that such degrees of belief were subjective, or individually determined by each individual, and Keynes accepted this criticism in his 1931 review of the posthumous