The Monetarist TransmissionMechanism
Monetarists of all types emphasized the heterogeneity of wealth. But they did so for different reasons. Brunner and Meltzer (1976) focused on the specifics of the portfolio-balance effect by which a monetary impulse is transmitted to aggregate spending and production.
They were also concerned with asset market conditions that could stifle fiscal impulses and contain the multiplier. They believed that the IS-LM model is generically unable to accommodate the heterogeneity of assets, and drafted eclectic macro-models instead. Brunner, in particular, was hostile to the IS-LM apparatus, which dominated the macroeconomic discourse during that period. He considered the implicit assumption that equity and other “real assets are frozen into portfolios by forbidding transaction costs” to be backward and even “schizophrenic” (Brunner 1978: 56). He liked to call the IS-LM model the “islamic framework”, and to talk of the “contamination” of post-World War II theory and policy advice (ibid.: 56, 60-62). He rather emphasized that there exists for each asset, including monetary assets, “a general and imperfect range of substitution in all directions over the whole spectrum of assets” (Brunner 1978: 61; original emphasis). Friedman, by contrast, was not interested in the details of the portfolio-balance effect (Friedman 1974: 134-7). For him, the variety of yields affected by the portfolio-balance effect allowed for a complete specification of the money demand function, accounting for a stable functional relationship (Friedman 1956: 15-16).Common to both approaches is the asset space, introduced as a continuum of close substitutes. That is, for each out of an infinite number of assets there exists a neighbourhood of “similar” assets. Assets outside of this local neighbourhood are rather “dissimilar”. In particular, the two polar assets have nothing in common.
Yet, they are connected via infinitely many overlapping neighbourhoods. Although the polar assets, say equity and currency, are highly “dissimilar”, their respective yields are interrelated. They are gross substitutes. The topology (the “neighbourhood-system”) of the asset space is generated by risk characteristics: assets are “similar”, if they expose the asset-holder to similar types of risk. In the example, with equity and money as polar assets, money is characterized by real risk, yet it is nominally riskless, whereas equity is immune to inflation risk, yet it is exposed to all other types of risk associated with a capitalistic society. Nevertheless, the monetarist risk topology suggests that money market conditions codetermine the equity yield in equilibrium.Accordingly, the transmission of a monetary impulse is wave-like, crossing an entire ocean of assets until it hits the facing coast. To simplify things, monetarist analysis reduces this ocean to a few asset classes: equities, bonds, and money. The goal of the monetarist transmission mechanism is to show that a monetary impulse affects the yields of a large variety of assets such that, even if some specific transmission channels are broken (in a Keynesian manner), the impulse is transmitted to the “real” sector. Money, so the argument, is never “trapped”. Further, monetarism is founded on the (false) belief that “proper” micro-foundations for the money demand function could resurrect the predominance of the money supply as determinant of nominal income. In fact, Friedman’s restatement of the quantity theory as a theory of money demand has at its core a risk-averse representative “wealth-owning unit”, introduced as a problem of portfolio optimization (Friedman 1956: 4, 8-9, 14, 1974: 13-14).
In modern parlance, Friedman’s representative portfolio holder maximizes a utility function characterized by constant relative risk aversion subject to a wealth constraint such that the warranted composition of wealth is invariant to its level.
Due to the homothetic property of preferences, the optimal wealth composition is independent of the wealth level (including the discounted perpetual payoff of human capital; Friedman 1956: 10, 1974: 13-14). The representative agent is just concerned with the composition of his portfolio, which includes real as well as nominal assets (real claims to real/human capital payoff versus dollar-denominated government liabilities). Nominal assets, in turn, are composed of monetary and non-monetary assets. The former contains central bank liabilities and other information-insensitive and, thus, liquid assets, while the latter contains those assets exposed to nominal risks. In studying the demand for money as portfolio choice, monetarism accepts Keynes’s liquidity preference theory and, thus, money as a store of value.Given the monetarist insistence of the importance of real assets for the transmission mechanism, providing a direct link between money supply and investment spending, the Fisher equation plays a significant role (Friedman 1956: 6-7, 9-10, 1974: 13, 36; Brunner and Meltzer 1976: 72, 179). It provides the one-period no-arbitrage condition between real and nominal assets. For sufficiently small values, the Fisher relation is approximated by:
where i denotes the nominal rate of interest, r denotes the ex-ante real rate of interest (in monetarist analysis equal to the rate of return on equity), and πe gives the expected inflation rate. Irving Fisher (1896 [1997]) introduced the no-arbitrage condition to “move up one derivative beyond” the long-run neutrality prediction of Hume’s static quantity theory (Dimand 2013: 287; see also Dimand 2012). The real rate of interest is unaffected by expected changes in the value of money, which reformulates “the divorce between monetary and real factors” in the context of a perfect-foresight equilibrium.
Real yields are equally governed by a no-arbitrage condition, invoking the fundamental theorem of finance - equilibrium returns linearly increase in risk:
with e ≥ 0, where rf gives the risk-free, inflation-adjusted rate of return on bonds with certain nominal payoffs, e denotes the equity premium, that is, the excess returns necessary to compensate for undiversifiable risk.
Friedman assumed that diversification eliminates all risk so that the interrelation of aggregate variables fits a deterministic model: “The society, though stationary, is not static. Aggregates are constant, but individuals are subject to uncertainty and change” (Friedman 1969: 2). In this case, equity is a perfect substitute of an asset that is not exposed to market risk, that is, e = 0. This is also Keynes’s specification in the General Theory. Brunner and Meltzer, however, insisted on imperfect substitutability: the risk topology is such that diversification does not eliminate all risk. Some “systematic risk” remains, differentiating equity from bonds and money. In equilibrium, equity trades at a discount, That is, e > 0.What matters for the monetarist transmission mechanism is that risky and riskless returns co-vary. This they do irrespective of the equity premium level, which is a “deep parameter ”. In contrast to what Brunner and Meltzer believed, their eclectic models are no substantial improvement upon Friedman’s early exposition of the transmission mechanism. Nor is there any substantial difference between the monetarist transmission mechanism and the Keynesian portfolio-balance effect as developed by Tobin (1958 [1987], 1969 [1987]) and Blinder and Solow (1973). To conclude with Benjamin Friedman (1978: 109-10; see also Dornbusch 1976: 123-4):
What is one to make of all this? Perhaps Brunner and Meltzer are not monetarists. Or perhaps Tobin is not a Keynesian. Perhaps. A more likely conclusion, however, is that, once monetarists and Keynesians specify clearly a “transmission mechanism” by which monetary policy has its effect in their respective theoretical models, these alternative mechanisms are by and large identical. On this key issue, which is the essence of the theoretical dimension of the monetarist debate, it is hard to find significant disagreement.
The Monetarist Restatement of the Quantity Theory
The portfolio optimization exemplified above yields Friedman’s restatement of the quantity theory as a theory of money demand. Given the no-arbitrage conditions (4) and (5), the real money demand of the representative agent is:
with
where Ω is the aggregate wealth constraint (a real value) which - in equilibrium - equals the present value of real “permanent income”, denoted by Y∞ (Friedman 1956: 4-5, 10-11, 1974: 11-12).
Permanent income is defined as the mean or mathematical expectation of a stationary real income process (more prominently formulated in Friedman 1957: 21). Finally, | is the portfolio-weighted average over the net returns of the different asset classes.Because preferences are assumed to be homothetic, the demand for any asset class, including money, is linearly increasing in wealth or permanent income. Linking shortrun money demand to a steady-state variable renders it more stable: by specifying CRRA utility, the portfolio weight of liquid assets is unaffected by random income shocks, or by “transitory” deviations of “actual income” from steady state. The first three arguments of the money demand altogether represent the Keynesian liquidity preference theory.
The real demand for money decreases in the nominal interest rate, i = rf + e + πe, the relevant opportunity cost of liquidity. It follows that the money demand function decreases in its first three arguments (enabling the portfolio-balance effect).
Finally, actual or current income Y accounts for the transaction motive (demand is increasing in actual income). In contrast to Tobin (1958 [1987]), Friedman did not provide micro-foundations for transaction demand. He rejected the hypothetical distinction between “active” and “idle” balances. To him, Keynes’s liquidity preference theory in the General Theory is a refinement of the Tract, which is a refinement of Marshall’s Principles of Economics, the latter’s exposition being “precisely the liquidity preference theory of the General Theory: M = M1 + M2 = L1(Y) + L2(r), except that Marshall expressed the part M2 as a fraction of wealth, whereas Keynes expressed it as a function of the interest rate” (Friedman 1974: 170; emphasis added).
Friedman systematically de-emphasized the general-equilibrium implications of Keynes’s liquidity preference theory in a short-run model of output and employment, which is also his setting.
He, like other monetarists, regarded Keynes’s “momentous deviation” from the neoclassical Cambridge tradition “in reversing the roles assigned to price and quantity” in Marshall’s market period, thereby “[merging] the market period and the short-run period.” “At least for changes in aggregate demand, quantity was the variable that adjusted rapidly, while price was the variable that adjusted slowly” (Friedman 1974: 18). By introducing short-run income as an endogenous variable, monetarism accepts the possibility of demand-constrained equilibrium income (the principle of effective demand). “Friedman, like Keynes, offers a theory of the level of income.” (Meltzer 1983: 7). Then, however, there are two unknowns - the real risk-free rate of interest (rf) and actual income (Y) - which require two market-clearing conditions for a short-run equilibrium to exist. Yet, monetarism insists on one equation, which is the gist of Friedman’s restatement:I regard the description of our position [Friedman and Schwartz (1963)] as “money is all that matters for changes in nominal income and for short-run changes in real income” as an exaggeration, but one that gives the right flavor of our conclusions. I regard the statement that “money is all that matters,” period, as a basic misrepresentation of our conclusions. (Friedman 1974: 27; original emphasis)
To demonstrate the compatibility between the quantity theory as a theory of income and Keynes’s liquidity preference theory, Friedman formulates a “simply common” IS-LM model (1974: 29-30). For a closed economy with a public sector (that borrows to consume), the monetarist money market equilibrium condition is:
and a plausible monetarist IS condition is given by:
where G denotes government consumption, I is the investment function, and S the saving function. The investment function is standard, decreasing in required rate of equity return. In this short-run model, the exogenous variables are M, P, Y∞, G, πe and e.
The saving function is specified according to Friedman’s permanent income hypothesis (PIH), which was an essential leverage point of the monetarist counterrevolution (Friedman 1957). The hypothesis dissolves the tight link between consumption and actual income, defined to include unpredictable and transitory deviations from the mean. Assuming risk aversion, optimization dampens the impact of transitory income fluctuations on consumption (precautionary savings). Current saving is increasing in actual income, because current consumption is not. It is increasing in interest rates. Friedman’s restatement suggests that the use of the money market condition (8) to determine real income is a first step. Then, in a separate step, the real interest rate is determined via (9), given the value of income derived in the first step (or vice versa). This procedure is only feasible, if and only if the marginal propensity to consume out of transitory income and the interest-elasticity of money demand are both zero. The first condition is substantiated by the PIH, which suspends the Keynesian multiplier.
The zero interest-elasticity of money demand, however, involves the major analytical contradiction of monetarism. It is a necessary assumption for any variant of the quantity theory, as in fact acknowledged by Friedman (1966 [1969]: 146). A positive interestelasticity, however small, suggests that some increase in the money supply is absorbed by higher money demand, which would violate the equi-proportionality between money and nominal income. This, however, means that Keynes’s liquidity theory does not apply (Johnson 1965: 396), an implication Friedman denied (Friedman 1966 [1969]: 142). He did not see this contradiction. Since money demand is a stable functional relationship (derived from deep parameters), the stability of its arguments would translate into a stable functional value and, therewith, guarantee the independence of the demand for real balances from money supply (A2). Most importantly, “if interest rates are stable, knowledge of interest rates is not necessary to predict changes in nominal income or in prices, so exclusion of interest rates is not even a necessary condition for a divorce” of money and commodity markets (Friedman 1966 [1969]: 146; emphasis added). His reliance on a stable interest rate, however, involves a presumption about the commodity market clearing condition. According to the IS condition, a stable short-run interest rate suggests a stable level of actual income. Friedman’s argument for stable income, in turn, is a stable money supply, which however involves (A2). Friedman argued in a circle.
This analytical contradiction had far-reaching consequences for the empirical controversies that ensued after Friedman’s restatement (excellently summarized by Laidler 1993). His Keynesian opponents brought forward overwhelming evidence of positive interest-rate elasticity, yet Friedman remained unimpressed and thought he had been misunderstood. Unaware of the circularity of his theoretical reasoning, he denied that a positive interest elasticity of money demand could falsify his theory. The resurrection of the quantity theory tradition, so he believed, would rather depend on a positive but “low” semi-elasticity of money demand with respect to short-run interest rates, that is, an interest rate elasticity smaller one (Friedman 1966 [1969]: 142-4). The same empirical evidence that led his opponents to reject the monetarist quantity theory convinced Friedman that he was right.