Resumen

MENDOZA VELAZQUEZ, Alfonso  y  GALVANOVSKIS, Evalds. La cópula GED bivariada. Una aplicación en entornos de crisis. El trimestre econ [online]. 2014, vol.81, n.323, pp.721-746. ISSN 2448-718X.

The General Error Distribution (GED) has been extensively used in time series econometrics applications, due to its great flexibility in the estimation of financial stylized facts. However, there has been no attempt to employ this statistical distribution in the construction of copulas. Copulas are probability functions that link one multivariate distribution to univariate distribution functions called marginals. These marginals are continuous and follow a uniform behavior within [0,1]. In this paper we introduce the bivariate GED copula to investigate financial contagion in Latinamerica during the 2008 crisis. We examine contagion in foreign exchange, equity, bonds and sovereign markets in Latinamerica. Standard decision criteria provide strong evidence in favor of the GED copula, against other widely used elliptical and arquimidean alternatives.

Palabras llave : distribución GED; cópula; distribución bivariada; contagio; mercados financieros; GARCH.