Re-Interpreting the Failure of Foreign Exchange Market Efficiency Tests: Small Transaction Costs, Big Hysteresis Bands

Small transaction costs and uncertainty imply that optimal cross-currency interest rate speculation is marked by a first-order hysteresis band. Consequently uncovered interest parity does not hold and market-efficiency tests based on it are mis-specified. Indeed, measured prediction errors are a combination of true prediction errors and a wedge that consists of the 'option value' of being in foreign currency and either plus or minus the transaction cost. Due to the nature of this wedge, we should expect measured prediction errors to be serially correlated, correlated with the current forward rate and perhaps have a non-zero mean, if the interest differential itself is serially correlated. The existence of the wedge helps account both for the failure of market-efficiency tests and for the difficulties in finding an empirically successful model of the risk premium.