A New Form of Farlie-Gumbel-Morgenstern Copula: A Comparative Simulation Study and Application

Abstract

In this study a new mixture copula has been obtained by creating a linear combination of product copula with copula which is obtained with marginal distributions of order statistics which was proposed by Dolati and Úbeda-Flores, Kybernetika. 45 (2009), 992–1002. In addition, a new form of Farlie-Gumbel-Morgenstern copula is obtained by applying Farlie-Gumbel-Morgenstern copula to new mixture copula. The new form of the Farlie-Gumbel-Morgenstern copula has been compared to Farlie-Gumbel-Morgenstern, Gaussian and Frank copulas. To find the best-fit copula, Akaike information criterion and Bayesian information criterion are used.

Keywords

• Mixture copula
• Farlie-Gumbel-Morgenstern copula
• Negative quadrant
• Dependency order statistics