Efficiency, risk premia, error correction models and conditional heteroscedasticity in foreign exchange markets

In foreign exchange markets, efficiency tests have typically rejected the hypothesis of unbiasedness of the forward rate as a predictor of the future spot rate. Since the 'simple efficiency' hypothesis is a joint hypothesis of no risk premium and expectations being rational, several authors have tried to explain the rejection of the hypothesis in terms of a non-zero time-varying risk premium. In the present study, we investigate the existence of a time-varying risk premium in the foreign exchange market, based on the conditional variance of market forecast errors. The forecast errors are assumed to follow the ARCH process introduced by Engle (23). The present study offers innovations in two directions. First, we extend the ARCH-M framework used by Domowitz and Hakkio (20) to investigate the presence of risk premium in the three-month market using an overlapping data set. Second, and more importantly, based on the time series properties of the exchange rates we use the Error Correction framework developed by Engle and Granger (28) to test efficiency and investigate the presence of time-varying risk premium where the risk premium is once again a function of the conditional variance of the errors which follow an ARCH process. Results are presented for the West German Mark, the Japanese Yen, the French Franc, the Canadian Dollar, the Italian Lira and the UK Pound for the 1975-1987 period. We find evidence of the existence of risk premium for the German Mark, the Lira and the Pound.