| Summary: | 碩士 === 銘傳大學 === 財務金融學系 === 86 === Existing hedging models do not perform quite well in reducing spot risk . For example , Myers(1991) applies OLS and GARCH model on wheat market . Both models reduce only about 45% of risk of unhedged position. Another example is the bivariate GARCH estimation for the optimal commodity futures hedge proposed by Baillie and Myers (1991), the dynamic hedging model is slightly better than the static hedge. Part of the unsatisfactory performance results from the fact that both models are expost. Ex-post models are data fitting based and do not reflect the information of future basis change. The weekness of ex-post models motivates us to propose a dynamic hedging strategy which uses the forecasting data of spot and futures to project the optimal hedge ratio in the later period. The primary characteristic of this research is that price forecasts of spot and futures are employed for deriving hedge ratio. This forward-looking mechanism allows the optimal hedge ratio to be ex-ante. Resultantly, hedgers can rebalance the hedged portfolio ahead of relative price changes, with the knowledge of future basis obtained from the forward-looking mechanism, the hedge ratio is determined such that the expected return of hedged portfolio atnext period is suppressed into zero. In other words, the value of hedged portfolio is restricted to follow a zero-return/zero-risk martingale time series . Consequently, the dynamic hedging strategy makes the portfolio payoff a fair game martingale process, i.e., a expected zero risk/zero return strategy, via ex-ante adjustment of hedge ratio. Empirical research firstly employ different forecasting models (State Space , Kalman Filter) to test the robustness of our expected zero risk dynamic hedge. Secondly, the commodity, foreign exchange and index futures are employed to perform the comparison among expected zero risk dynamic hedge, OLS, GARCH and error correction hedging models in the ways of hedging effectiveness and cost of hedge.
|
|---|
