Abstract
When risk premia are estimated using the standard asset pricing approach of two-pass regressions with the omission of factors, inference of these risk premia can be wrong. In this study, we apply a three-pass methodology developed by Giglio and Xiu (2017) to estimate the risk premia attached two common factors, dollar and carry factors, in the currency market. We find that the omission of factors affects the magnitude of risk premia for these two factors constructed from the currency portfolios. Carry factor is able to explain most of the cross-sectional variation of excess returns at the portfolio level but not in the individual currency level.