Modeling Dependence of Agricultural Commodity Futures through Markov Switching Copula with Mixture Distribution Regimes

Woraphon Yamaka, Rungrapee Phadkantha, Songsak Sriboonchitta

Abstract

Proposed is a Markov Switching copula with mixture distribution regimes for modeling the dependence of agricultural commodity futures.  This model involves different dependence structures that can characterize the dependence behaviors in different regimes as the copula function in each regime can be different from that in another regime. By permitting different copula structure, this model is able to capture more complex dynamic patterns of daily movement of agricultural commodity futures (sugar, coffee, corn, wheat and soybean). The criteria as Akaike Information Criterion(AIC), Bayesian Information Criterion (BIC) and Log-Likelihood (LL) are based in-sample statistical performance have suggested that our model is superior to the single regime copula and two-regime Markov Switching copula in 9 out of 10 cases. This result reveals that the high and the low dependence of agricultural commodity futures exhibit a different dependence structure.