erc/metu
INTERNATIONAL CONFERENCE IN
ECONOMICS IV
September 13-16, 2000, Ankara
Bivariate GARCH Estimation of Time-Varying Optimal Hedge Ratios on Japan's Precious Metals Futures Markets
Nakatani Tomoaki (Obihiro University
of Agriculture & Veterinary Medicine, Japan)
Sasaki Jun (Japan Society for the Promotion of Science)
Abstract
In this study, we estimate time-varying optimal hedge ratios on Japan's precious metals futures markets, and analyze the effectiveness of the dynamic hedging strategy in comparison with the naive (one to one) hedge ratios and the constant hedge ratios in terms of variance reduction of returns from spot and futures portfolio. The models we use are three types of bivariate generalized autoregressive conditional heteroscedastic (BGARCH) model, which is recognized by many authors to be a suitable procedure for estimating hedge ratios because it accounts for time-varying second moments, or conditional heteroscedasticity peculiar to financial time series. In addition, we evaluate alternative BGARCH models with error correction terms in the conditional mean equations (ECM-BGARCH). The preliminary statistical tests show that daily data of spot and futures prices of gold, silver, palladium and platinum exhibit the co-integration relations and the conditional heteroscedasticity. Consequently, ECM-BGARCH models are appeared to be appropriate. The examination of various hedging strategies reveals several interesting points. Among three ECM-BGARCH models, the diagonal vector half parameterization scores the greatest reduction in the portfolio variance, followed by the positive definite and the constant conditional correlation representations. However, the time-varying optimal hedge ratios, which substantially outperform the naive hedge ratios, give a slight improvement on average in the portfolio risk reduction compared to the constant hedge ratios. Also, including error correction terms in the conditional mean equations provides mixed results: the hedging performance becomes worse in gold and silver markets, and better in palladium and platinum markets compared to the cases where BGARCH models are used. Overall, it can be concluded that the conventional constant optimal hedge ratios seem to perform relatively well in hedging practice if we consider the cost of computation: the constant hedge ratios are easy to compute, and the time-varying hedge ratios are not.
Economic Research Center
Middle East Technical University
06531 Ankara Turkey
Phone: + 90 312 210 3044, 210 2003
Fax: +90 312 210 1244
e-mail: metuerc@metu.edu.tr