Gradual learning about shocks and the forward premium puzzle

When interest rates are higher in one’s home country than they are abroad, standard arbitrage arguments suggest this signals that the home currency will depreciate in the future. However, empirical evidence has regularly been found to be strongly at odds with this intuition. This is the “forward premium puzzle”. This paper proposes a learning-based explanation for this puzzle. Specifically, we assume economic agents cannot ascertain whether the monetary policy or technology shocks affecting the economy are persistent or transitory but only gradually infer this persistence using Kalman filtering. We embed this information problem in a two-country open-economy DSGE model with nominal rigidities and simulate the model with and without this informational friction. We find that our incomplete information with learning framework, combined with data generating processes dominated by the persistent monetary policy shifts or by slightly distorted beliefs as in Gourinchas and Tornell (2004), lead our artificial data to replicate features of the forward premium puzzle.