RDP 8607: The Expectations Theory of the Term Structure and Short-Term Interest Rates in Australia 1. Introduction

The relationship between yields on financial assets of different maturities is a subject that has interested economists and policy makers for decades. The most commonly discussed explanation of this relationship is the expectations theory of the term structure. The “pure” expectations hypothesis (PEH) states that, in equilibrium, the expected returns from different investment strategies with the same horizon should be equal. For example, the expected return from investing in an n-period bond should equal the expected return from investing in a one-period bond over n successive periods. If this theory holds then long-term rates can be (approximately) expressed as a weighted average of current and expected short-term rates. More importantly, it suggests that if policy makers wish to alter long-term rates through their influence on short-term rates they must succeed in altering the market's expectations of future interest rates.

The expectations theory has recently been subject to extensive empirical scrutiny in the United States. On the basis of this empirical work, the expectations theory of the term structure has been rejected in various studies. Despite this consistent rejection, Shiller, Campbell and Schoenholtz (1983) note that the theory continually reappears in policy debates. They liken this “superficially appealing” theory to the indefatigable Tom of Tom and Jerry cartoons; “The villain, Tom the cat, may be buried under a ton of boulders, blasted through a brick wall (leaving a cat-shaped hole), or flattened by a steamroller. Yet seconds later he is up again plotting his evil deeds”.^[1]

In addition to these extensive empirical rejections of the expectations theory, the logical consistency of many of the economic propositions derived from it have been questioned. In an important paper Cox, Ingersoll and Ross (1981) [hereafter (CIR)] re-examine several propositions about the relation between long and short rates typically associated with the expectations theory.^[2] CIR show that, when interest rates are random, these different propositions are inconsistent with each other and all but one are incompatible with any continuous time rational expectations equilibrium.^[3] The single proposition which obtains in continuous-time rational expectations equilibrium is the proposition that the instantaneous expected rates of return on all bonds are equal to the prevailing spot interest rate. CIR call this the Local Expectations Hypothesis. CIR also show that the various propositions are inconsistent with each other in discrete time but are compatible with arbitrage pricing equilibrium. These arguments suggest that traditional tests of the expectations hypothesis may be incorrectly specified.

Campbell (1986), however, has defended the empirical applications of the expectations theory on two grounds. First, he argues that CIR consider a more restrictive form of the theory than is considered in the empirical literature. In particular, CIR's discussion is directed to the “pure” expectations theory which states that risk premia are zero whereas most empirical applications consider the less restrictive expectations hypothesis (EH) which allows for constant risk premia. Campbell shows that the propositions derived from this less restrictive theory are not necessarily incompatible with each other or with arbitrage pricing equilibrium. Furthermore, Campbell shows that any inconsistencies are of second order and may often be ignored in empirical studies.

Previous Australian studies on this topic were conducted at a time when yields on government securities were largely set by the authorities.^[4] Since that time, a move to a more market-oriented system of interest rate determination has occurred. In particular, tender systems for the sale of Treasury notes and Government bonds were introduced in 1979 and 1982 respectively.

The purpose of this paper is to test the expectations theory using data on Australian short-term financial assets for the period since 1979. The sample period encompasses two significantly different policy regimes; namely, managed and floating exchange rate regimes. Because of this, inferences can be drawn as to whether the relationships between interest rates are altered by policy regimes. A recent paper by Mankiw and Miron (1985) on U.S. data found that the empirical tests of the expectations hypothesis are sensitive to the policy regime of the time.

The findings of this paper are at variance with the U.S. results.^[5] It is found that the data are consistent with the pure expectations hypothesis for the whole period. Furthermore, in most cases, the pure expectations hypothesis cannot be rejected in either the period of managed or floating exchange rates. In no case can the less restrictive expectations hypothesis (i.e., constant risk premium) be rejected. However, there is some evidence of parameter instability in the reported equations. Although cross country and cross time comparisons of empirical results are difficult, one explanation of these results may lie in the difference in monetary regimes and the pattern of financial flows in Australia and the United States. Another explanation may lie in the different risks perceived by agents in the U.S. financial markets. These issues will be discussed in Section 3.

The paper is structured as follows. Section 2 outlines the expectations hypothesis and surveys the results of recent U.S. studies. Section 3 reports the results of the various tests of the expectations theory employed in this paper while Section 4 contains some concluding comments.

Footnotes

Shiller, Campbell and Shoenholtz (1983, p.175). [1]

Campbell (1986), notes that these various propositions can be expressed as different definitions of the term premium. [2]

Because of the non-linearities in the term structure equation the alternative definitions of the term premium cannot simultaneously hold given Jensen's Inequality. [3]

See Bloch (1974) and Jüttner, Madden and Tuckwell (1975). Bloch, for instance, found that a version of the expectations hypothesis held in the short term. However, the relationship between short-term yields and much longer-term yields was not consistent with the expectations hypothesis. However, both Bloch and Jüttner et.al. caution against drawing strong inferences from the results because of the dominance of the Reserve Bank in the market at that time. [4]

In the U.S. the expectations theory is rejected at each end of the maturity spectrum. The U.S. results which are most directly comparable to those reported here are those dealing with the relationship between various short-term interest rates. There are numerous rejections of the expectations theory at the short end of the market in the U.S. See, for example, Friedman (1979), Shiller, Campbell and Schoenholtz (1983), Mankiw and Summers (1984), Jones and Roley (1983) and Mankiw and Miron (1985). [5]