Are Currency Returns Really Predictable?: Unforeseeable Structural Change and the Forward Rate Anomaly

Abstract

This paper provides evidence suggesting that currency returns are not predictable. We find that the Bilson-Fama regression is not only unstable, but the instability is triggered by novel historical events. The novelty of the events implies that the structural change underpinning returns cannot be characterized ex ante by a probability rule. We show that allowing for unforeseeable structural change leads to considerably more evidence of a time-varying risk premium. It also reveals temporary correlations involving the market's forecast error. These correlations, like the bias of forward rate predictions, change at times and in ways that are not predictable.

Keywords: Asset Returns, Predictability, Structural Change, Risk, Knightian Uncertainty

JEL Classification: F31