Abstract
The article applies a Regime Switching Fractionally Integrated Error Correction Generalized Orthogonal (RSFIEC-GO) Generalized Autoregressive Conditional Heteroscedasticity (GARCH) model for optimal futures hedging. RSFIEC-GO captures both the relationships of fractional cointegration and regime shifts between spot and futures returns. Empirical investigation in agricultural commodity markets reveals that RSFIEC-GO provides superior hedging effectiveness compared to its nested models in terms of variance reductions. Results of Diebold, Mariano and West (DMW) test with adjusted McCracken's critical values also show the statistical superiority of RSFIEC-GO. This illustrates the importance of simultaneously modelling the fractional cointegration and regime shifts for dynamic futures hedging.
Notes
1 Lien and Tse (Citation1999) investigate daily data and truncate at lag 500. Since the data investigated are weekly data in this study, we truncate at lag 100. Weekly data is investigated because daily data is noisy for hedging practice and most of important; rebalancing on a daily basis incurs a heavy rebalancing cost.
2 Due to its similarity, to save space, only those figures with corn data estimated from RSFIEC-GO model are illustrated here.