Following the influential work by Bilson (1981) and Fama (1984), an enormous amount of literature on testing whether the forward rate is an unbiased predictor of the future spot rate accumulated over the last years. Excellent surveys are Hodrick (1987), Engel (1996), and Sarno and Taylor (2002).
Although results vary depending on how exchange rates are modeled, the common finding in the overwhelming majority of past research is that the forward rate is not an unbiased predictor of the future spot rate. This forward bias implies the apparent predictability of excess returns over uncovered interest rate parity (UIP).
Tests of the UIP are frequently based on the ‘Fama regression’ which relates the change in the spot rate to a constant and the lagged forward premium. The null hypothesis is that the constant should equal zero and the slope coefficient unity, i.e. that the movement in the exchange rate compensates for the (interest rate) differential. However, the common finding is that the slope coefficient is less than unity and often negative, indicating not only that UIP does not hold but also that the higher interest rate currency tends to appreciate rather than depreciate.
Motivated by the limited success of statistical or economical explanations, Lyons (2001) proposes an approach based on limited trader participation. The limits to speculation hypothesis (LSH) postulates that deviations from UIP (might) occur and persist because nobody is willing to trade on these deviations since other investment opportunities yield higher Sharpe ratios. Recent papers provide evidence consistent with the LSH. Villanueva (2005) argues that exchange rate undershooting is an empirical implication of the LSH. Inspired by the LSH, Sarno, Valente, and Leon (Forthcoming) and Baillie and Kiliç (2006) model exchange rates in a smooth transition regression (STR) framework and both report strong evidence for such non-linearity in the relationship between spot and forward rates.
However, the LSH suggestion that deviations from UIP are too small to attract speculative capital seems to be in contrast to what can be observed in the foreign exchange market. In fact, major financial institutions try to generate excess returns from exploiting the anomaly and market surveys provide evidence that the use of such trading rules is a key driver for the surge in foreign exchange trading. While most papers investigating the forward bias are more concerned with modeling the puzzle rather than exploring how deviations from UIP can be exploited, papers dealing with trading strategies in the foreign exchange market are not directly related to UIP and mainly focus on neural network, genetic programming, or simple technical trading rule approaches; see e.g. Neely, Weller, and Dittmar (1997), Okunev and White (2003), and Olson (2004).
In the present paper we investigate whether trading rules explicitly aimed at exploiting deviations from UIP have the potential to attract speculative capital. In particular, we develop two classes of strategies. For the first we take the perspective of a levered proprietary trader who can choose to allocate funds to a bias exploiting strategy or some other investment opportunity. Recall, that the LSH postulates that speculative capital can only be attracted by strategies with Sharpe ratios higher than a certain threshold.
However, we stress that for the comparison of investment opportunities downside risk should be taken into account as well, especially, in the light of the capital requirements financial institutions have to fulfill for market risks of positions in their trading books. Thus, we analyze downside risk (and hence capital charge) constrained strategies based on taking positions in deposits in different currencies. We apply a heuristic portfolio optimization algorithm in which the trader chooses a certain level of volatility and determines expected returns either through the differentials or on the basis of forecasts for the future spot rate generated from a vector error correction model (VECM).
Download
PDF Ebook Trading the Forward Bias: Are there Limits to Speculation?
