Harry Johnson as a Macroeconomist

ed from economic growth. The literature on money and growth meets the second of these objections, but only by completely ignoring the first: that is one reason why it has now fallen out of fashion. Harry's particular contribution to it (1966), in contrast to Tobin's initial work (1965b), "stressed the necessity of allowing money a function in a monetary economy, and therefore of attributing to the presence of money an increase in economic welfare" (1970, p. 109). In short, this paper incorporated Friedman's analysis of money "as if" a consumer durable; it showed that because the rate of inflation influenced holdings of real balances it would also influence "real income" and hence the savings rate, the details depending on the way in which savings behavior was modeled. The conclusion established by this literature-and Harry's contribution sets out all the essential results while managing also to be the most readable paper on the subject-is that the full-employment equilibrium properties of a monetary economy are extremely unlikely to be independent of monetary factors (specifically the rate of change of the nominal money supply). This conclusion holds even if all Keynesian problems about uncertainty are assumed away by making inflation, and everything else for that matter, "fully anticipated."' Since one of the properties of "fullemployment" equilibrium that fully anticipated inflation might affect is presumably the "natural" unemployment rate, the literature in question may be relevant to important policy issues. But such relevance is still open to doubt.'4 In my view based on Harry's judgment of the upshot of the closely related money-and-economic-welfare literature, it is likely to remain so. As is well known, Pesek and Saving (1967) showed that the "insideoutside money" distinction of Gurley and Shaw-to which Harry devoted considerable space in his early 1960s surveys-stemmed from endowing accounting conventions with spurious economic content. Once we leave a commodity money world, the key distinction-and this only became clear after a somewhat convoluted discussion to which Harry made an important contribution (see 1969b, 1969c)-is 3 See Johnson (1966). Some of the analysis there presented is further developed in the latter half'of "Inside Money, Outside Money" (1969b). 14 I am indebted to Douglas Purvis for some discussions on this point. 6o6 JOURNAL OF POLITICAL ECONOMY between money that bears interest at competitive rates and that which does not because its issue is monopolized. But all the results of the money in growth-models literature and the "schizophrenic policy proposals" that emerged from Friedman's "Optimum Quantity of Money" (1969) essay hinge upon variations in the rate of nominal monetary expansion influencing the opportunity cost of holding money. Thus, as Harry was quicker than anyone else to stress (see 1968b), they depend on the assumption of a particular institutional anomaly-namely, that the monetary system is not competitive. Only if "the practical point that the payment of interest on currency holdings is infeasible . . . is taken as being of predominating importance" (1970, pp. 106-7) does the rate of nominal monetary expansion have any long-run significance, and Harry was sufficiently unimpressed with its importance to be willing to abstract from it entirely in his most extensive discussion of these very issues (1968b). Otherwise, as far as he was concerned, the welfare analysis of money was significant in two respects: first, as a component of the case for more competition in the banking system, this was an important policy issue in Britain in the late 1960s; and, second, inasmuch as it dealt with the transition from commodity to fiat and credit money systems, it yielded insights into the problems of reforming the international monetary system.'5 The notion of a fully anticipated inflation rate is fundamental to the literature on money in a growth-model or money-and-welfare literature. The recognition that if agents allow for inflation expectations in their portfolio behavior they are also going to allow for them in their behavior toward money wages and prices underlies the literature on the expectations-augmented Phillips curve. The "accelerationist hypothesis" follows from the proposition that, in the long run, inflation is indeed fully anticipated-that is, it is expected to occur and behavior is completely adapted to the expectation in question. Like everyone else in the 1960s, Harry was slow indeed to incorporate expectations into his treatment of the Phillips curve, and once he had, though he immediately grasped the logic of the accelerationist hypothesis as a piece of economic theory, he never embraced it as a hypothesis relevant to the practical conduct of monetary policy in an inflationary environment. In his article "A Survey of Theories of Inflation" (1963e) Harry provides as good an example as one could find of a man who knows something without knowing that he knows it. He criticized "Keynesian models" that analyze inflation in terms of competition between various social groups for greater shares in national income because, unlike the "quantity theory approach," which he identified with the 15 On the international monetary system see Johnson (1968a). On competition in banking see Johnson (1 967b). HARRY G. JOHNSON 607 Cagan-Bailey analysis of the inflation tax, they failed to recognize "that the processes of determining wages and prices are fundamentally real processes." Protection against inflationary redistributions "is available to all groups in a freely competitive system, through negotiation of contracts to take account of expected inflation, and is increasingly resorted to as the fact of inflation is recognized" (1963e, p. 122). He even pointed out that "there are some serious doubts about the applicability of the Phillips curve to the formulation of economic policy. . . . it may reasonably be doubted whether the curve would continue to hold its shape if an attempt were made by economic policy to pin the economy down to a point on it" (pp. 132-33). Nevertheless, "the Phillips curve appears to be far the most reliable of [the] relationships" describing statistical trade-offs between policy goals (p. 133). It formed the basis for his subsequent discussion of policy trade-offs; and Reuber's pioneering (1964) attempt to quantify, in dollars and cents terms, the trade-off between inflation and unemployment was held up as an example of the type of work which must be further developed so that the curve could be used as a basis "for intelligent policy-making" (1963e, p. 141). It is not difficult to understand why Harry fell into this particular inconsistency. His writings over the years make it clear that he thought the rapid inflations experienced by underdeveloped countries presented a very different set of problems from the mild inflations of the pre-1970s in the developed world (see 1967a). The Phillips curve was a potentially interesting tool for analyzing the latter, while the quantity theory approach had proved its worth empirically in dealing with the former. "For the mild type of inflation typical of the United States and other advanced countries . . . the [quantity theory] approach has not proved nearly so useful . . . the expected rate of price change has not appeared as a significant determinant of the quantity of money demanded" (1963e, p. 126). Because the empirical evidence suggested to him that the expected inflation rate was unimportant in such a context, there was no reason for Harry to incorporate it systematically into his thinking about the Phillips curve. Thus even when, in the Introduction to the second edition of Essays in Monetary Economics (1969a), he noted the inconsistency in question, he argued that it was of more theoretical than practical significance. The importance of incorporating inflation expectations into the Phillips curve "'. . . for the theory of macro-economic policy . . . depends on the length of time it takes for changes in expectations generated by experience to begin to affect behaviour significantly; and the empirical evidence is that the lags in adjustment of expectations are sufficiently long for contemporary policy makers safely to disregard them-even though the cumulative effect may be substantial" (p. x). I have already said that I have found no statement in any of Harry's 6o8 JOURNAL OF POLITICAL ECONOMY writings of a belief in the empirical, as opposed to theoretical, relevance of a vertical long-run Phillips curve. However, in his later work he certainly did pay increasing attention to the role of expectations in determining the economy's response to stabilization policy and expressed increasing doubts about the theoretical foundations of the Phillips curve itself. Thus, he argued (1 968b) that "stabilization operations [themselves might disturb] expectations derived from previous experience" by way of "mechanisms [that] cannot be . . . satisfactorily dealt with by compressing them into the distributed lag structure of the economy" (p. 301). By 1972 he had come to the view that "contrary to the standard assumptions of economic theory, the economic public does not simply respond mechanically . . . to signals reaching it through the blind and impersonal operations of . . . competitive markets. Instead [it] engages in two kinds of political transactions with . . . policy-makers. . . . First, one of its major concerns is to guess how determined the government is about implementing its announced economic policies. . . . Second, the relevant economic public, aware of the sensitivity of government to political pressures, has an incentive to generate such pressures in its own favour" (1972, pp. 14-15).16 As to the Phillips curve itself, new work "broadening the macroeconomic foundations of the relationship in terms of the influence of information and adjustment costs and other real factors" (1972, p. 66) was required, a them