Multiproduct Optimal Hedging by Time-Varying Correlations in a State Dependent model of Regime-Switching

We determine time-varying hedge ratios in a multiproduct setting using a multivariate state dependent model of regime switching dynamic correlations. The model enables one to depict the time-varying correlations for multiple series of cash and future prices in two or more different regimes (i.e. the conditional correlation is not constant in this multivariate model). This provides an improved characterization of the multiproduct dynamic hedging process as it captures the evolution of the cash/futures correlation matrix when the model switches from a regime of low correlation to one of higher correlation or vice-versa. The model switches regimes according to a Markov chain process and does not have a dimensionality problem for larger numbers of series, as does the more conventional BEKK model. In addition, we introduce fundamental, economically related factors in the regime switching process to assess their effect. These are (weakly) exogenous variables with respect to the markets being considered. Results show that these explicit weakly exogenous variables have an impact on the dynamic process. We determine the optimal hedge ratios for the soybean complex by specifically introducing the stocks-to-use ratio of soybeans as a variable in determining the probability of switching correlation regimes. The stocks-to-use ratio contains specific, up-to-date information on the supply and demand conditions relevant to the soybean markets, and hence has a direct role in determining the price of the commodities. By introducing this variable, our model achieves improvement in the characterization of the process over the case of constant transition probabilities between regimes. Additionally, shocks to these related variables permit us to identify the effect on the hedging ratios and comparison to simpler hedging estimation procedures. The model applied is from Tejeda et al. (2009).