Abstract: | A large empirical literature has tested the unbiasedness hypothesis in the foreign‐exchange market with the use of forward exchange rates. This article amends the conventional testing framework to exploit the information in currency options, with a newly constructed data set for three major dollar exchange rates. The main results are that (a) tests based on stationary regressions suggest that options provide biased predictions of the future spot exchange rate, and (b) co‐integration–based tests that are robust to several statistical problems afflicting stationary regressions and allow for endogeneity issues arising from a potential omitted risk premium term are supportive of unbiasedness. © 2006 Wiley Periodicals, Inc. Jrl Fut Mark 26:627–656, 2006 |