RDP 8605: On Some Recent Developments in Monetary Economics 3. The Stability of Money Demand

The centrality of the stability question to monetary economics warrants that the issue be considered at some length. There is, however, no readily available theoretical framework in which to analyse the issue. Standard demand theory does not generate money demand as naturally as the demand for bread; artificial devices, such as the adding of money balances to consumers' utility functions, are unsatisfactory. Modern theories which attempt to explain from microfoundations the motives to hold money are theories of general disequilibrium; here, stability questions remain largely unanswered.

Although the question of stability has been studied empirically by a wide range of researchers using an even wider range of techniques, the most ambitious (and possibly the most carefully executed) have been the studies by Friedman and Schwartz.

Critics of their work have been numerous, but one particular stream of criticism deserves closer inspection. This is the argument – advanced for example in Hall (1982) – that the more or less stable relationship between money and money income is a by-product of the system of financial regulations in force over the period examined.

How could this come about? If there are direct controls on interest rates (either fully determining rates or setting permitted ranges for them), the scope for interest rates to adjust will be limited. Observed interest rates will not be market clearing values, so there is likely to be excess demand or supply in money markets. Quantity rationing in contractionary periods, and the unwinding of rationing in expansionary periods, will establish a positive relationship between the money stock and money income. At the same time, the relative stickiness of the interest rate during this process provides no (or at best a poor) correlation between measured rates of interest and other key economic variables.

The apparent relationships between the money stock and nominal income may have masked (or proxied) a true relationship between the effective (but unmeasured) rate of interest and nominal income. Observed rates of interest did not reflect the effective rate: movements in the effective rate (which includes the cost imposed by rationing) may have been correlated (inversely) with movements in the money stock itself.[9]

Regulations which apply directly to quantities will work in a similar way, by changing the money stock rather than its price. In fact, where the regulatory system contains reserve requirements on financial institutions, these may be the most important of all. Changes in reserve ratios will have a direct impact on institutional balance sheets, and hence on the money stock; they may, therefore, speed the process of adjustment by acting directly on supply as well as on demand (through the effects on interest rates).

Of course, this raises the possibility that changes in regulations may be used as policy instruments in such a system. But it is the continuity of the overall framework that will contribute to the stability of the money stock – money income relationship.

These relationships can perhaps be better understood with reference to a quite simple view of the transmission mechanism. Changes in the supply of cash to the financial system will lead to changes in cash rates and exchange rates, and to banks' demands for and supplies of funds. These will lead to changes in supply of money. At the same time, the interest rate and exchange rate changes will impact on the economy directly.

The “transmission mechanism” from monetary policy to the economy is, on this view, one in which the behaviour of money and the behaviour of income are simultaneously determined. The simple money stock – money income relationship is therefore not open to a causal interpretation. Rather it is merely a statistical correlation of two variables determined by the behaviour of financial markets. Any change in the structure of the financial markets will act potentially to alter the relative outcomes and hence the correlation as well.

Stability, in the sense of a function which fitted over some historical period, was in some ways a historical accident. Definitions were found (often with some difficulty and disagreement, as in the case of the money stock itself) which were consistent with the simple money-income relationship.

But this backward-looking stability underplayed the sometimes large errors in the predictive power of the money demand equation for short periods. These errors, themselves a reflection of the lack of short-run dynamics in the money demand function view of the transmission mechanism, were subsumed into the variability of the “long and variable lags” of the monetarists. Deregulation and innovation have increased the frequency of these periods of errors – and therefore the need for some greater insight into the dynamic process of adjustment.[10]

Although currently somewhat speculative, a strand of the literature has begun to look also at the longer run by asking what would happen in the complete absence of regulations. The literature in this area is as yet relatively undeveloped; some of the main models are critically surveyed in McCallum (1985). These models typically assume that the transactions technology is refined to allow the transaction costs to be effectively ignored, so that all money becomes inside money, bearing full competitive rates of interest. There is therefore no longer any role for a specific transactions asset. In such models, there ceases to be a money stock; “monetary policy” has a role only in influencing the prices of assets (that is, the spectrum of interest rates) in the economy. A sufficient condition for the reappearance of a money stock in these models is a minimal level of regulation and/or an irreducible demand for a particular asset part of whose rate of return would therefore be non-pecuniary – this asset would thus have a below market rate of interest and in that sense be a monetary asset.[11]

The above discussion suggests that the simple aggregate relationship between the money stock and money income, which is central to the received monetary macroeconomics of the 1960s and 1970s, may be merely a correlation produced by a specific set of regulations on the financial system.

This does not mean that “money” (or monetary policy) does not matter for macroeconomics. It does mean that the indicators of money's effects will be different in different regulatory systems.


This was a theme in Davis & Lewis (1977) with respect to the role of monetary disequilibrium in the RBII model of the Australian economy. [9]

The apparent breakdown of the money demand function is discussed at length in these terms in Laidler (1986). [10]

Fama (1980), Harper (1984). [11]