Friedman’s Views On The Methodology of Economics

Milton Friedman Methodology of Positive Economics

Friedman's methodological position was to be elaborated at a relatively early stage in his work. It was to provide a vital underpinning for much of his major theoretical and empirical research to follow.

According to Friedman (1953b), the appropriate criterion for establishing the fruitfulness' of a theory is not the 'realism' of its assumptions, but the degree of empirical corroboration attained by its predictions. His argument bases itself on the proposition that the assumptions of any theory are inevitably - if it is to be scientifically fruitful - 'unreal' in some degree. This is so because the purpose of science is not to replicate the 'real world', in all its manifold complexity, but to provide simple models which abstract from the inessentials, and allow us to predict its behaviour.

This stress on the role of abstraction and predictive power leads to an important corollary implication. Friedman rejects both large-scale theoretical (e.g. general equilibrium) and econometric , models, in favour of small-scale empirical models derived from relatively simple but predictively powerful theories. This ap­proach has been much evident in his own empirical work (e.g. Friedman, 1957, 1959; Friedman and Meiselman, 1963).

Friedman's approach to the methodology of positive econ­omics has given rise, as so commonly with his work, to controversy (e.g. Samuelson, 1963, Wong, 1973). Simply put, much of this argument boils down to the proposition that Friedman's stress upon scientific prediction leads him to ignore the importance of scientific explanation, in the sense defined by the classic statement of the concept of causal explanation provided by Hempel and Oppenheim (1948).

I would, however, argue that Friedman's methodological prescriptions have had a generally beneficial effect upon the orientation of subsequent economic research. First, the Hempel-Oppenheim conditions for the adequacy of a causal scientific explanation of the deductive-nomological variety are so stringent that it is difficult to visualize how, if ever, theories that meet these conditions can be generated in economics. Friedman's emphasis upon predictive power would thus seem the more realistic criterion for economists to utilize for the foreseeable future. Secondly, in the current state of econometric 'technology' and data availability, an emphasis upon empirically-fruitful small-scale models would likewise seem the more realistic research strategy. You have to be able to walk before you can run. Friedman has at least provided us with a reminder of that pertinent fact, and a reasoned rationale for doing so, in the context of economic research.

The Nature and Significance Of Friedman’s Macroeconomic Analysis, Milton Friedman Positive Economics

Friedman's work on the role of money in the economy-his 'major professional interest for many years' (Friedman, 1975, p. 56)-and macro-economics generally, has touched on so many topics, and is so central to contemporary debates in macro-economic analysis and policy, that it is exceedingly difficult to encapsulate even this one aspect of his work in the space available. I therefore confine myself to a discussion of Friedman's contributions to the positive side of macro-economics, ignoring his (closely-related) work on the implications for macro-econ­omic policy.

A central 'strategem' which Friedman has employed in found­ing an alternative framework for macro-economic analysis to that of the prevailing textbook Keynesianism (epitomised by the income-expenditure and IS-LM models) has been that of, in fact, building upon the Keynesian revolution, in order to erect a more theoretically-sophisticated alternative. A second 'strategem', closely related to his methodological perspective, has been that of developing simplified specifications of his own alternative hypotheses that make them eminently suitable for empirical testing against large bodies of data relating to a wide variety of contexts.

These two talents were to be displayed most impressively in Friedman's (1957) presentation of his permanent income hypo­thesis of consumption. Note that in this work Friedman did not reject the concept of an aggregate consumption function - which is the central strut of Keynesian analysis because it underpins the multiplier process -but rather sought to provide a more theoret­ically-sophisticated analysis of it. On the basis of Fisherian capital theory, Friedman argued that neither the dependent variable (current consumer expenditure) nor the independent variable (current income) were specified properly in the Keynesian consumption function. Friedman's alternative hypo­thesis was that only the permanent component of current consumption is functionally related to the permanent component of income, where the qualifying adjective 'permanent' is taken to refer to the magnitudes of these variables viewed in a long-term, intertemporal choice context. Expressed at its simplest, the underlying idea is that choices about consumption 'now' are influenced not only by current (e.g. today's, this year's) income, but also by longer-term income expectations.

This simple idea proved to be empirically powerful, because it enabled Friedman (1957) to explain much wider classes of consumption phenomena than the Keynesian hypothesis - such as the apparent paradox in the form of the estimated secular and cyclical consumption functions, and observed differences in the savings behaviour of such groups as business entrepreneurs, farmers, and wage earners.

Friedman's contributions to this particular branch of macroeconomics were largely to lapse after this seminal publication, but the permanent income hypothesis has exerted a profound influence on the later direction of research in this area.

Friedman's own attentions have been drawn the more heavily to the monetary aspects of macro-economics. A major step in the evolution of his work in this area came with his restatement of the quantity theory (Friedman, 1956). Over the two previous decades, the quantity theory of money had been jettisoned, as a result of the effective demolition job done on it by Keynes and his disciples. Their central argument was that the relationship between the demand for money and nominal income is not one of proportionality (as assumed by the textbook 'straw-man' version of the quantity theory) - that the 'Cambridge A:' in the formulation Md = k PY, is not a constant but a variable, influenced by
the level of interest rates.

Friedman re-established the acceptability of the quantity theory by the simple device of taking these Keynesian criticisms on board in his re-statement of it. Friedman, in essence, re-defined the quantity theory as a theory of the demand for money - as a theory of the determinants of k:, in other words. Furthermore, byJ once again turning to Fisherian capital theory, Friedman derived a (supposedly quantity-theoretic) analysis of the demand for money which is far more sophisticated than the original Keynesian hypothesis, setting the demand for money decision in a much broader portfolio choice context. There is thus clear justice in Patinkin's (1969, p. 47) contention that:

what Friedman . . . actually presented [in his 'restatement' of the quantity theory] is a statement of the modern portfolio approach to the demand for money which . . . can only be seen as a continuation of the Keynesian theory of liquidity preference.

In his restatement paper, Friedman derives a (real) demand-fop money function, the arguments of which include not only the yield on bonds (as in the original Keynesian liquidity preference function), but also the yield on equities, physical capital and human capital, and the expected rate of inflation - a theoretical­ly-relevant variable (as inflation acts as a tax on money holdings) which Keynes had ignored. Friedman's theoretical treatment of the demand for money is further differentiated from its Keynesian predecessor by assuming that permanent income, and not current income, is the relevant income constraint variable in the money-demand function. Finally, the Friedman specification does not, as in the Keynesian treatment, partition the demand for money function into a demand for 'active' balances and a (money-illusioned) demand for 'idle' balances.

In his later empirical work on the demand for money, however, this elaborate theoretical model is considerably pared down. Friedman (1959) found that the demand for money over a long span of time in the US is primarily explained by permanent income, and that alone. Subsequent research has confirmed generally that permanent income (or some similar proxy for wealth) performs better than current income in estimated de-mand-for-money functions. The same body of research, however, has generally corroborated the postulate that the level of interest rates does have an independent and significant effect on the demand for money, contrary to Friedman's (1959) own find­ings. In the light of the gathering evidence, Friedman has subsequently reformulated his stated position on this issue, now accepting that no fundamental significance attaches to the matter of the actual interest-elasticity of the demand for money, provided that it is not minus infinity (i.e. a Keynesian 'liquidity trap' situation - the existence of which he denies). It is, in any case, quite in accordance with Friedman's (1956) theoretical specification that the (real) return on bonds affects the demand for money.