RDP 8708: Risk Effects Versus Monetary Effects in the Determination of Short-Term Interest Rates 5. Conclusions

Most theories of asset pricing can be thought of as determining the rates of return on risky assets relative to the risk free rate (assuming that a risk free asset exists), and so an understanding of what causes the risk free rate to vary must be central to an understanding of the time series behaviour of asset prices in general. For this reason, the study of short-term interest rates seems a good starting point in identifying the most important determining factors for this behaviour. As the estimates in Section 4 indicate, short-term deposits are to a good approximation risk free, since short-term uncertainty about the future price level is empirically negligible.

The paper has argued that there are two major theoretical approaches to the study of short-term interest rates. These were characterised as models based on the microeconomics of risk, and models based at the macroeconomic level on the money demand function. In recent empirical work, explanations based on the consumption risk model can probably be said to have featured the most prominently.

In examining the time series data on short-term interest rates, both nominal and real, one of the features most immediately apparent is the high degree of serial correlation in these series. From a purely econometric point of view this suggests that interest rates are linked to some variable that is slow adjusting; but the consumption growth index, which is the basic determinant of the real interest rate in the consumption risk model, does not have this property. On the other hand, serial correlation of interest rates is exactly what is predicted by the sticky-price monetary model. Consistent with these stylised facts, the empirical results reported in Section 4 point strongly towards the rejection of the consumption risk model as an explanation for interest rate behaviour, in favour of the sticky-price monetary model. Of course, the latter model is not without defects: some of the coefficients were not of the expected sign or significance, and the specification used in this paper ignored a number of important macroeconomic variables. Nonetheless, the use of models in which money plays a non-trivial role in short-run behaviour, seems to offer a more fertile ground for improving our understanding of asset price behaviour, than does persistence with refinements of the consumption risk approach.