RDP 9111: Monthly Movements in the Australian Dollar and Real Short-Term Interest Differentials: An Application of the Kalman Filter 2. Approaches to the Determination of Short-Term Movements in the Real Exchange Rate

If we think of foreign currency as an asset price, then an efficient foreign exchange market would ensure that the observed variation in the Australian dollar reflected the behaviour of fundamental economic variables. We would expect that the stance of Australia's monetary policy relative to that of our trading partners, as embodied in real short-term interest rate differentials, would be reflected in this price. This paper focuses on the role of the real short-term interest differential in determining the monthly variation in Australia's real exchange rate.

Uncovered interest parity theory (UIP) provides a convenient building block from which an investigation of the relationship between real interest and exchange rates may be developed. As mentioned, UIP is based on three theoretical assumptions. Firstly, the market in foreign exchange is efficient (such that market expectations are rational). Secondly, any transactions costs incurred by market participants are negligible, and finally, the actions of these participants are not affected by any risk considerations. While little empirical evidence can be found to support UIP, this may be the result of the inappropriateness of one or more of the theory's underlying assumptions:

If we accept that the first premise, that of an efficient market in foreign exchange, is appropriate, then the failure of UIP may be due to the fact that traders are not risk-neutral. Risk premia which vary over time may exist in and distort the market. Otherwise, transactions costs, although usually maintained to be too small, may effect exchange rate transactions. Baldwin (1990) proposes that such costs occasion “bands of inaction” in the price of foreign exchange.^[11]

Alternatively, market participants may hold an ongoing belief that an infrequently occurring event relevant to the determination of the exchange rate is imminent. The low probability of this event being captured in sample is the so-called “peso” problem. Market participants may incorrectly be labelled “irrational” if the sample period doesn'st contain the infrequent event that is driving their behaviour.

Within this efficient market framework, the literature is divided as to the mechanism by which the observed real exchange rate is predominantly determined.^[12] The empirical content of this paper concentrates on the monetary approach which isolates international investment decisions channelling capital into those countries offering the highest real return on their assets (that is, the highest real rates of interest) as the dominant explanator of exchange rates in the shorter-term. A second “rational” school advocates the real approach, believing observed variation in the Australian dollar to reflect the market modifying its expectations of the equilibrium exchange rate to account for real factors (Coughlin and Koedijk (1990)). As the relative price of foreign goods in terms of domestic goods, the real exchange rate should be expected to reflect shocks to the supply (for example, changes in productivity, technology or the labour supply) and demand (for example, shifts in preferences) for each country's product (Ghosh (1990)). As a small open economy, Australia's terms-of-trade for example, have a substantial effect on our real exchange rate (refer Blundell-Wignall and Gregory (1989)). Meese and Rogoff (1988), investigating the case for the US, find little empirical evidence to support the existence of any stable relationship between real interest rates and real exchange rates. Instead, they propose that real disturbances may be a major source of exchange rate volatility. Such real variables have not typically been considered responsible for the observed short-run movements in real exchange rates.^[13] They are more commonly believed to explain the longer-term value of the dollar.

If, alternatively, we reject the first premise that the foreign exchange market is efficient, then two main theoretical explanations are offered for the generation of inefficiencies. One school of thought believes that the unexplained short-term changes in exchange rates can be accounted for by the existence of rational speculative bubbles in the foreign exchange market.^[14] These speculative bubbles occur when market expectations of the movement in the exchange rate are self-fulfilling. As a result, market expectations alone have a strong influence on the value of the dollar, causing inefficiency in the foreign exchange market, at least for a time. There is little explanation for the origin of such bubbles and their existence is underpinned by the belief that the market will ultimately return to fundamentals.

Another school proposes that the foreign exchange market is inefficient because expectations are not fully rational. That is, heterogeneous groups of market participants may drive the exchange rate away from its fundamental value either through irrational sentiments, investor misperception or “fads” (Miller and Weller (1991)). If the foreign exchange market is inefficient because market expectations are irrational, fundamental economic variables will not figure in the determination of short-run movements in the dollar. Therefore, interest differentials would not be expected to have any explanatory power. Theory in this area identifies two specific groups of “irrational” traders responsible for destabilizing exchange rate movements.

The first group, the feedback traders, base their demand for foreign exchange on historical returns rather than on any expectation of future fundamentals (Cutler et al. (1990)). Therefore, their forecasts over short time horizons (1 week to 3 months) employ extrapolative rules.^[15]

The second group, the noise traders, irrationally perceive random fluctuations in the exchange rate as a viable source of information on which to make a profitable trade (Black (1986)). The very existence of these unpredictable traders introduces a risk in the price of foreign exchange that discourages rational arbitrageurs from the market. De Long et al. (1990) describe the process by which generally “bullish” noise traders can earn higher expected returns than sophisticated traders: despite the absence of any fundamental risk, the observed exchange rate diverges from its fundamental value as the risk generated by the presence of the noise traders depresses its price.^[16]

Questions concerning the cost-benefit analysis of actively acquiring information about the equilibrium value of the exchange rate may also have implications for explaining inefficiency in this market. If participants perceive the foreign exchange market to be efficient (that is, they believe that the current real exchange rate is a price signal that reflects all the private and public information available in the market) then they may not believe that research into the equilibrium real exchange rate is worthwhile. These issues are beyond the scope of this paper and we recognise the technical difficulties associated with their empirical estimation.

This paper now undertakes the specific technical application of a rational expectations, efficient market model of the real Australian exchange rate as estimated by Campbell and Clarida for the United States. Based on the ambiguous results returned by this application, the alternative explanations for the short-run activity of the dollar, as canvassed in this section, must be seen as live possibilities.

Footnotes

That is, transactions costs accompanied with uncertainty, may result in interest differentials unmatched with any expected exchange rate change (allowing for the possibility of a risk premium). Within a small band then, interest differentials do not induce capital flows towards the country offering the highest return because the transactions costs make such movement of capital suboptimal. [11]

See Coughlin and Koedijk (1990) and Dornbusch (1989) for more comprehensive reviews of the different theoretical approaches to real exchange rate determination. [12]

Explanations have more commonly depended on the Dornbusch (1976) overshooting model. However, Coughlin and Koedijk (1990) and Stockman (1987) mention the potential importance of random real shocks for moving the real exchange rate. [13]

Refer, for example, to Flood (1987) for evidence of rational bubbles in the foreign exchange market. “Rational” bubbles occur when the market does indeed know the true model describing the exchange rate. That is, the market makes a “rational” response to irrational factors and whilst the path of the exchange rate may deviate from that consistent with fundamentals, rational arbitrage conditions are satisfied along this path (Miller and Weller (1991)). [14]

Over long time horizons (1 or more years), feedback traders assume regressive expectations in order to make their forecasts. See Cutler et al. (1990) for a discussion of the role of feedback traders in increasing the variability of the exchange rate. [15]

Miller and Weller (1991) present a comprehensive survey of the literature in this area. They note that three assumptions are crucial to the results of the De Long et al. model:
(i) the sophisticated investors, taking positions on the basis of fundamental indicators, have time horizons that are not “too long”;
(ii) the erroneous beliefs of the noise traders are positively correlated; and
(iii) the expectations of noise traders are generally “bullish”. [16]