Uncovered interest rate parity redux: Non-uniform effects

Based on the /3-estimate that captures the interest rate differential effect in uncovered interest rate parity (UIP) regressions, we show that an empirical model that includes proxy variables for unobservable factors and allows for variables to have time-varying weights and parameters can reduce the UIP deviation. However, the specification that alleviated UIP failure does not reduce the variability of the /3-estimate. The explanatory variables exhibit time-varying coefficient estimates and shifting importance across exchange rates. These findings corroborate the scapegoat theory and suggest that the difficulty of rectifying the empirical UIP failure can be attributable to the shifting roles of explanatory variables and time-varying parameter effects.