Prescriptive Definitions of Money
An entirely different approach is to decide on the theoretical nature of the relationship of money with other important variables in the economy and then to define money so as to show that this relationship exists in practice.
In other words, one starts with a model of the economy in which ‘money’ plays a clear role, with the nature of that role depending on the assumptions underlying the model.
Once we have the model, we seek a definition of money that validates it. Let us assume that we believe money to be neutral. That is, money is a set of assets changes in the value of which have no impact on real variables such as output and employment.
We then hope to define money in such a way that empirical tests show this neutrality. Alternatively, we might begin with the view that inflation Opens in new window is caused by a too rapid growth of the money supply. A corollary of this is that the demand for money is stably related to real income.
Thus, a demand for money function is constructed using what seems to be the most likely definition of money for the purpose. However, if the function turns out not to be stable, the definition of money may be changed until a definition of money is found for which demand does appear to be stable.
This extremely pragmatic approach is strongly supported by Milton Friedman and reflects the statement quoted above that ‘money is what money does’. The underlying belief is that there must be something, the rate of growth of which is closely linked to the rate of inflation, and we may call this something ‘money’.
The only problem is the practical one of finding an empirical counterpart to the theoretical idea. Money is thus defined in the way that yields the most accurate predictions. This is an example of a general approach to economics that sees that predictive power of models as all-important.
We are told not to worry unduly if the assumptions made in the construction of models that predict accurately appear to be unrealistic.
It remains that this approach to defining money can lead to frequent ad hoc adjustments to the definition in an attempt to produce the correct answer.
Given the problems associated with definitions of other variables in the functions, the dubious quality of many of the statistics being employed and the complexity of the time lags involved, there is a danger of exercises of this kind becoming more interesting for the range of econometric techniques used than for any light shed on important economic relationships. It follows naturally that economists who do not start by believing in the theory that inflation is caused by excess growth of an exogenous money supply regard all such empirical exercises with skepticism.
In any case, there is a potential practical difficulty. No simple combination of assets might produce the desired results. Alternatively, a particular definition may seem to give favorable results when applied to the data for one period but the apparent relationships may break down in later periods.
One reason for difficulties of this kind might be that the different assets included in a definition of money may have different relationships with income. For example, notes and coins, sight bank deposits and other bank deposits may each have a stable relationship with nominal income but the relationship may be different in each case.
Then, if our definition of money includes all three of these assets and the proportion each forms of the money supply changes over time, the relationship between money as a whole and nominal income will change.
Thus, a development of the empirical approach has been to weight the various types of deposit to try to take account of different velocities of circulation. Alternatively, as in the case of Divisia indexes, different degrees of liquidity are measured by rates of interest (the lower the interest rate payable on an asset, the more liquid it is assumed to be).