The forward unbiasedness regression is revisited by varying the prediction horizons from 1 day to 1 year. The panel data suggests some possibility of a positive slope coefficient at a short horizon while the negative coefficient improves forecasting performance at longer horizons

Motivated by a consumption-based asset pricing model, this paper checks the time horizon sensitivity of the forward premium puzzle from 1994 to 2004. The in-sample estimation shows that the longer the time horizon (up to 1 year), the more the future spot exchange return deviates from its...

This paper examines the gains of using high-frequency data in detecting volatility spillovers, risk hedging, and asset allocation. In particular, a realized covariation method based on intra-day US and Japan index funds data is compared with a bi-variate GARCH method based on the same data...