Mean-variance hedging in the presence of estimation risk

Abstract

The mean-variance hedging (MVH) with a significant risk-aversion coefficient is approximately equal to the minimum-variance (MV) hedge. However, how large the risk-aversion coefficient should be in practice? We determine the boundaries of risk-aversion coefficients that significantly distinguish the MV hedge and the MVH based on the different magnitudes of statistical errors in the presence of estimation risk. Based on the hedged variance, hedged return, and hedge ratio, we show that the MV hedge is statistically justified for MVH investor with an extensive range of risk-aversion coefficients. Additionally, the upper bound of the significant risk-aversion coefficient is positively related to the squared information ratio of futures.