RDP 8708: Risk Effects Versus Monetary Effects in the Determination of Short-Term Interest Rates 1. Introduction

Economic theory offers two distinct approaches to the explanation of movements in interest rates over time. At the microeconomic level, there is a well developed body of theories of asset pricing based on investor optimisation under conditions of risk; these theories can in principle be applied to the pricing of any asset, and thus can be used to price interest bearing assets in particular. At the macroeconomic level, on the other hand, interest rates have traditionally been viewed as being determined in an economy-wide equilibrium in which the money demand function plays a key part in determining short-run behaviour. The two approaches have some important differences in their empirical implications. Perhaps the most important of these arises when slow adjustment of output prices is assumed, thus introducing short-run non-neutrality of money into the macroeconomic approach. This would imply that money supply movements will have temporary effects on both the level of interest rates and the term structure; depending upon the degree of rigidity assumed in output prices, these effects may be highly persistent. Such effects are not present in asset pricing models based purely upon investor optimisation; in these models, real rates of return are determined entirely by the real variables which characterise the risk-return tradeoff.

Recent empirical work on the behaviour of interest rates has strongly emphasised the approach at the microeconomic level, focussing on efficiency considerations related to the term structure of interest rates (for example, Shiller (1979), Mankiw and Summers (1984), Fama (1984)), and attempting to reconcile returns on interest bearing (and other) assets with intertemporal optimising behaviour on the part of consumers and investors (Hansen and Singleton (1983), Hall (1985), Mankiw et.al (1985)). These studies have been largely unsuccessful in explaining the time series behaviour of interest rates during the 1970s and 1980s. Real interest rates during this period have varied over a much wider range than it seems can be explained by variations in expectations or in systematic risk factors.

One explanation for the lack of success of these models is that they may be insufficiently sophisticated. Hence, it is often suggested that there is a “time-varying risk premium” which explains violations of the simple expectations model of the term structure, and which contributes to the variability of real interest rates over time. However, if this risk premium is to be more than a catch-all residual, it is important that it be empirically modelled, and this has not yet been successfully achieved. A second line of explanation is the one suggested in the introductory remarks to this paper: the presence of short-run nominal rigidities in the system may mean that monetary shocks can have persistent effects on equilibrium real rates of return. As yet, there has been no attempt to test these two theories in an integrated framework or to assess their relative contributions to explaining the time series behaviour of interest rates. It is this task which is attempted in the present paper.

Section 2 of the paper sets out a model of asset pricing based on intertemporal optimisation using the “consumption CAPM” model of Lucas (1978, 1982) and Breeden (1979). An equation for nominal interest rates of various maturities is derived, together with exact expressions for the theoretical risk premiums as functions of a risk aversion coefficient and of the variances of the distributions of future prices and consumption. Section 3 examines the alternative approach, taking a simple macroeconomic equation for the determination of nominal interest rates under price rigidity and rational expectations; this is used to show the effects of unanticipated movements in the money supply and in the steady state inflation rate on nominal interest rates. A general model which encompasses the models from Section 2 and 3 is then proposed. Section 4 tests the general model using data on interest rates for four countries: the United States, United Kingdom, West Germany and Switzerland. Restrictions under which the model reduces to one or other of the two special cases are tested.

The main empirical finding is that a reduction of the general model to a sticky-price monetary model with no risk premium, cannot be rejected; thus the risk premium makes no significant contribution to the explanation of interest rate behaviour over the sample period. Section 5 concludes the paper by discussing implications of this finding for the study of other financial markets.