Dependence Analysis for the Exchange Rate Data using Extreme Value Copulas

บทคัดย่อ

Abstract

This article considers the bivariate generalized extreme value (BGEV) distribution and the bivariate generalized Pareto distribution (BGPD) to model the tail probability and tail dependence of the financial return series based on monthly and daily maxima of BHT/USD, EUR/USD foreign exchange data, respectively. The selection and estimation of the copula is based on the maximum likelihood estimation(MLE) approach which is proposed for nine parametric models of dependence function for both distributions. The copula parameters are estimated by Inference For Margins(IFM)approach and then select best fitting model by Akaike Information Criterion (AIC) value.

Keyword: bivariate generalized extreme value distribution, bivariate generalized pareto distribution, parameter estimation, extreme value copulas, dependence function, tail probability, tail dependence.