Price efficiency and return predictability in international currency markets

One of the most well documented anomalies in international finance is that high interest rates predict excess returns in currencies, and often predict currency appreciation. A recent paper by Grossman (1995) explains this anomaly as the result of an upward sloping supply of capital and imperfect information by investors, together with changing demand for capital. Under this framework a country which wishes to attract foreign capital must raise its real interest rate because of the currency risk premium demanded by foreign investors.; The first contribution of this paper is to set out a mathematical model of Grossman's approach using the Wang (1994) heterogeneous investor framework, and to contrast it with the rational learning under regime shifts model of Lewis (1989). Although both are imperfect information models, they may be differentiated by some conflicting empirical conclusions. The second contribution is to incorporate potential measures of expected depreciation into the return predictability equation to investigate whether the anomaly is due to "peso problems". The final contribution is to estimate the return predictability relation using a dynamic linear model. This time-varying parameter model is consistent with the Bayesian updating present in the imperfect information frameworks.; The empirical results support that return predictability is best explained by an upward sloping supply of funds in the Grossman/Wang framework. This framework explains high interest rates predicting excess returns and currency appreciation, and interest rate increases leading to excess returns. This last result is in contradiction to the rational learning about regime shifts story, where increasing interest rates go along with currency depreciation and negative excess returns. As for the peso problem story, it is shown that excess returns are not increasing with an increased probability of devaluation. Using the time-varying parameter model, almost all countries in the sample show either non-existent or diminishing return predictability over time. This is consistent with increasing financial market integration over the period.