A study of risk premiums in the foreign exchange market

In this study, we estimate the risk premium in the foreign exchange market with various models. We find that consumption risk and market risk can explain the risk premium. When the risk premium is assumed to be a function of the volatility of the spot rate, a model combining the ARCH and the Markov-switching models can produce more desirable estimates of the risk premium than the ARCH and the Markov-switching model do separately. Based on parameter estimates of the models, we generate spot exchange rates using bootstrap and Monte Carlo simulations and apply the moving average trading rule to the actual data and the artificial data. By comparing the conditional mean returns of buy and sell periods from the actual data and from generated data, we test the null models that are employed to generate data. We find that the random walk model is significantly rejected and the SWARCH model can explain substantial parts of returns of the actual data from moving average trading rules. A better empirical specification for fitting market excess returns is needed for the CAPM to explain the actual trading rule profits very well.