Modelling Dependence in Latin American Markets Using Copula Functions

Two important issues in the analysis of association among financial markets are the degree of dependence and the underlying shape commanding the cross-market dependencies, so that any model used to describe this association must cope with both the issues. In the study presented in this article, we approach the modelling of dependence in two stages. The first stage is based on modelling the dependence between the returns of two assets by means of a single Archimedean copula, whereas the second one takes advantage of the mixture between copulas to gain the necessary flexibility to capture different tail dependence patterns. Both stages have been followed in modelling the dependence among the daily returns of seven Latin American country indices, a regional index and a worldwide index. Our findings have important implications for portfolio risk management.JEL Classification: G11, G15