RDP 9314: The Demand for Money in Australia: New Tests on an old Topic 5. Conclusion

Theories concerning the relationship between money, income and prices have a well-established pedigree in the history of economic analysis. The jump in inflation in the 1970s brought with it the idea that targeting money growth could tie down inflationary expectations and inflation. In the face of financial innovation and regulatory changes, however, the short-run demand for real money balances proved unstable, and the policy unworkable.

Recent econometric developments in time series analysis offer the prospect of testing for the existence of a short and long run demand for real money balances, through tests for cointegration. This has been the subject of a number of applied econometric studies in Australia. This literature is a mixed bag in terms of the definitions of money, income and interest rates used and the econometric techniques applied and the conclusions are also mixed.

The aim of this paper was to put the Australian literature into perspective by assessing the sensitivity of the long-run relationship between money, income and interest rates to sensible alternative definitions of these variables, and to different testing procedures. Overall, evidence of cointegration between money, income and interest rates was not particularly strong, and the results were sensitive to changes in the definition of activity and interest rate and to the testing procedure used. There was no evidence of cointegration between M1 and income, and only scarce evidence for the money base, M3 and Broad Money. Evidence was strongest in the case of currency, though events after the sample period used for the formal tests cast doubt on this. These conclusions hold whether money and income are defined in real or nominal values.

A finding of cointegration between money and income would not, in itself, necessarily establish a paramount role for monetary aggregates in policy-making. It would mean that the variables in question ‘move together’ over time, with change in one associated with particular change in the other. That may mean that monetary aggregates have some useful indicator properties, but it implies nothing about causation between the variables, or about whether money is suitable as a control variable for income. Those issues ultimately turn on views about the transmission mechanism, exogeneity issues, the stability and speed of adjustment paths, acceptable bands for predictive errors and the ease with which substitutes for the control variable can be created.