Model uncertainty and the Forward Premium Puzzle

• Revisits the forward premium puzzle with an ambiguity averse investor.
• Uses Hansen and Sargent's (2008) approach.
• Shows that model uncertainty premium is more important than a risk premium in explaining the forward premium puzzle.
• Empirically calibrates ambiguity using detection error probability.

This paper studies the Forward Premium Puzzle in a setting where investors doubt the specification of their models, and thus engage in robust portfolio strategies ( Hansen and Sargent, 2008). It shows that an empirically plausible concern for model misspecification can explain the Forward Premium Puzzle. In particular, the paper shows that Hansen and Jagannathan (1991) volatility bounds can be attained with both reasonable degrees of risk aversion and reasonable detection error probabilities. Hence, observed excess returns in the foreign exchange market appear to be primarily driven by a model uncertainty premium.