Return predictability and intertemporal asset allocation: Evidence from a bias-adjusted VAR model

We extend the VAR based intertemporal asset allocation approach from Campbell et al. (2003) to the case where the VAR parameter estimates are adjusted for small-sample bias. We apply the analytical bias formula from Pope (1990) using both Campbell et al.'s dataset, and an extended dataset with quarterly data from 1952 to 2006. The results show that correcting the VAR parameters for small-sample bias has both quantitatively and qualitatively important e¤ects on the strategic intertemporal part of optimal portfolio choice, especially for bonds: for intermediate values of risk-aversion, the intertemporal hedging demand for bonds - and thereby the total demand for bonds - is strongly reduced by the bias-adjustment. We also investigate the robustness of the results by changing the lag-length and one of the state variables of the VAR.