Resumen
The modeling of the extremal dependence structure can be made through parametric models classified in two families: Logistic and Mixed, which contain the symmetric and asymmetric models. The bivariate models are very useful in practical applications on the extreme value theory, in particular in a financial area. Considering the strong influence of the North American market on other financial markets, we investigate how does the dependence structure among the Latin American markets change after filtering the influence of the North American market. To remove that influence, we carry on a polynomial regression with GARCH (1,1) errors, and fit the bivariate extreme value models to the pairs of monthly maxima and minima of the standardized regression residuals.