What Is Wrong with Market-Based Forecasting of Exchange Rates?

Abstract

Market-based forecasting of exchange rates is flawed because it is based on two hypotheses that are not supported by empirical evidence: the simple random walk hypothesis and the unbiased efficiency hypothesis. By using historical data on six currency combinations it is shown that these two hypotheses are rejected because of the presence of a significant time-varying drift factor and what is typically perceived as a risk premium. It is also shown that the model representing the unbiased efficiency hypothesis is misspecified because the relationship between the spot and forward exchange rates is contemporaneous rather than lagged. The results cast doubt on the usefulness of the spot and lagged forward rates as benchmarks for measuring the forecasting power of time series and structural models. It is also demonstrated that market-based forecasting may lead to faulty financial decisions.