RDP 9009: An Empirical Model of Australian Interest Rates, Exchange Rates and Monetary Policy 1. Introduction

Ever since the collapse of the Bretton Woods regime of quasi-fixed exchange rates in the early 1970's, the relationship between interest rates and exchange rates has been subject to a great deal of theoretical and empirical scrutiny. While the literature is now voluminous, one general theme has emerged: single equation structural models of exchange rate determination, which include interest rates as exogenous regressors, are abysmal failures when confronted with the data.[1]

Irrational (i.e. not profit-maximizing) behaviour by participants in the foreign exchange market, non-linearities in the structural relationships, and the existence of amorphous risk premia have all been advanced as potential reasons to account for these failures. As significant as these explanations might be, we believe that a more fundamental influence is at work, viz. the policy reaction of the monetary authorities.

For example, while a simple portfolio balance model predicts that a decrease in domestic interest rates will depreciate the exchange rate, such a depreciation might also elicit a tightening of monetary policy, which will increase interest rates. In other words, the relationship between interest rates and exchange rates is simultaneous, and single equation estimation of the parameters in an exchange rate equation is likely to result in biased and inconsistent estimates. A further complication arises from the fact that the direction of the relationship running from interest rates to exchange rates depends on the source of the change in the interest rates. For example, if nominal interest rates increase because inflation is expected to increase, the exchange rate will probably depreciate, contrary to the predictions of the portfolio balance model.

There are several approaches that researchers can take in response to these problems. One is to attempt to build structural models in which exchange rates and interest rates (including policy instruments) are determined endogenously.[2] Another is to eschew structural models, and to estimate reduced forms. We choose the latter method of investigation in this paper.

Specifically, we estimate a five-equation vector autoregression using the following variables: the $A/$US exchange rate, the unofficial cash rate, the Australian 10 year bond rate, the United States federal funds rate, and the United States 10 year bond rate. The U.S. variables are included because the international mobility of capital means that domestic interest rates are linked, in theory, to foreign interest rates, as is the exchange rate. (The extent and nature of those linkages are, of course, empirical questions.)

Using this model, we can then examine the effects of unanticipated shocks to each of these variables on the exchange rate and interest rates. The advantage of using a reduced form model is that we can do so without having to specify any structural relationships.

The rest of the paper is organized as follows: Section 2 presents some stylized facts; Section 3 discusses the VAR methodology and its application to this problem. Section 4 contains the results and Section 5 contains some concluding remarks.

Footnotes

See Macdonald (1988) and Meese (1990) for a review of the theory and evidence on floating exchange rates. [1]

One subset of this approach is the voluminous literature on money supply announcements, which attempts to model the reaction of the financial markets in response to the latest information on the money supply, conditional on an expected policy response to that information. For a recent contribution, see Strongin and Tarhin (1990). [2]