RDP 8603: Risk Premia, Market Efficiency and the Exchange Rate: Some Evidence Since the Float 1. Introduction

There is an extensive literature testing the “efficiency” of foreign exchange markets – that is, whether available information is optimally used in the determination of exchange rates. This widespread interest in foreign exchange market efficiency can be attributed to two factors. First, information on forward and spot exchange rates provide a rich source of data which can be used to (indirectly) test whether agents form their expectations rationally. Second, deviations from the efficiency hypothesis have important policy implications. In particular, if the foreign exchange market is inefficient the authorities can, in principle, intervene successfully to prevent or burst price “bubbles” or to assist the market to move quickly to a new equilibrium. Such intervention can help to smooth fluctuations in the exchange rate.

Tests of market efficiency used in much of this literature are actually tests of the joint hypotheses that markets are efficient and (at least some) agents are risk neutral. The assumption of risk neutrality is needed to give empirical content to the familiar notion of market efficiency discussed in Fama (1976). The market efficiency hypothesis states that prices fully reflect available information. By imposing the assumption of risk neutrality, testable implications of the efficiency hypothesis can be derived.^[1] These testable versions of the model will be discussed in the following section.

Many recent studies of exchange markets overseas including Hansen and Hodrick (1980), Hodrick and Srivastava (1984), Hakkio (1981), Hsieh (1982), Baillie, Lippens and McMahon (1983) and Korajczyk (1985) all find evidence of inefficiency. That is, other available information improves upon the forecast of the future spot exchange rate that is provided by the current forward rate.

For Australia, on the other hand, several studies including Levis (1982) and Turnovsky and Ball (1983) have found weak support for the speculative efficiency hypothesis. Levis found that, for the period 1974–1981, the 90-day US$/$A forward rate was an unbiased predictor of future US$/$A spot rates and that US$/$A forward premiums contained no unexploited information about future US$/$A spot rates. Turnovsky and Ball (1983), conducting similar tests over the same sample period, found that the speculative efficiency hypothesis could not be rejected at the one per cent level of confidence but could be rejected at the five per cent level.

These two studies were conducted at a time when Australia had a managed exchange rate system. The introduction of a floating exchange rate system in December 1983 and the associated increase in exchange rate volatility^[2] may have altered behaviour in the foreign exchange market. The purpose of this paper is to re-examine the speculative efficiency hypothesis for the period of floating exchange rates. This paper extends previous Australian studies in three areas. First, data on forward and spot exchange rates are sampled weekly, thus giving more precision to the parameter estimates. Secondly the paper pays more attention to tests of semi-strong form versions of the hypothesis than earlier papers. Finally, the speculative efficiency hypothesis is examined for forward rates of different maturities.

The earlier Australian studies failed to reject the speculative efficiency hypothesis on the ground that the forward rate was an unbiased predictor of future spot exchange rates. On this criterion alone, the present paper also fails to reject the speculative efficiency hypothesis. However, the more extensive tests of the semi-strong form of the speculative efficiency hypothesis (i.e., the orthogonality of other information) suggest that the forecast errors in the 30-day market are serially correlated and that there was a behavioural change in the market after the depreciation of February 1985. In particular, the null hypothesis of speculative efficiency is rejected for each market after the February 1985 depreciation. Readily available information (on earlier expectation errors and forward premiums) improves the forecast of the future spot exchange rate that is provided by the current forward rate. Since the tests involve a joint hypothesis, this rejection could be due to market inefficiency and/or time varying risk premia. Because of the limited sample size since the float it was not possible to conduct more sophisticated tests of the existence of time varying risk premia.^[3]

The paper is structured as follows. Section 2 outlines the joint hypothesis and examines its testable implications.Section 3 discusses some econometric issues while section 4 discusses the data and results. Section 5 offers some concluding thoughts.

Footnotes

Bilson (1981) has popularised the name “speculative efficiency” for this joint hypothesis. He defines the market to be speculatively efficient if the supply of speculative funds is infinitely elastic at the forward price that equals the expected spot price. To avoid confusion I shall use this terminology. [1]

See Trevor and Donald (forthcoming). [2]

For instance, Cosset (1984) uses a measure of risk premium derived in Graver, Litzenberger and Stehle (1976) which is a function of world prices and nominal world wealth, while Mark (1985) using a model of intertemporal asset pricing derives a risk premium which is a function of real consumption, amongst other things. There are very few observations on these variables since the float. Thus, further testing in this area will be limited until the data become available. [3]