"Lustig and Verdelhan (2007) argue that the excess returns to borrowing US dollars and lending in foreign currency "compensate US investors for taking on more US consumption growth risk," yet these excess returns are all approximately uncorrelated with the consumption risk factors they study. Hence, their model cannot explain the cross-sectional variation of the returns. Their positive assessment results from allowing for a large constant in the model, and from ignoring sampling uncertainty in estimated betas used as explanatory variables in cross-sectional regressions that determine estimated consumption risk premia."--abstract.