Abstract
We revisit the controversial risk premium hypothesis (RPH) using survey data on exchange rate expectations, and the Phillips-Hansen fully modified ordinary least squares procedure. This allows us to conduct an unrestricted cointegration analysis and also differentiate between strong (bivariate) and weak (multivariate) from of RPH. The no risk premium hypothesis is rejected across the board for both data sets and over all forecast horizons. [F31]
*We wish to acknowledge the referees' comments, But we are responsible for all remaining errors. We are grateful to Kenneth A. Froot for the survey data and Peter C. B. Phillips for the Phillips-Hansen software.
*We wish to acknowledge the referees' comments, But we are responsible for all remaining errors. We are grateful to Kenneth A. Froot for the survey data and Peter C. B. Phillips for the Phillips-Hansen software.
Notes
*We wish to acknowledge the referees' comments, But we are responsible for all remaining errors. We are grateful to Kenneth A. Froot for the survey data and Peter C. B. Phillips for the Phillips-Hansen software.