Modeling asymmetric tail dependence in a non-Gaussian framework

This article models dependence among multiple random variables in a heavy-tailed and asymmetric setting using the Bad Environment-Good Environment (BEGE) model introduced by Bekaert, Engstrom, and Ermolov (2015). The approach estimates covariances between the bad- and good-environments through a two-step procedure. In the first step, the BEGE model is estimated for individual random variables using closed-form conditional characteristic functions. The second step applies the simulated method of moments to estimate covariances. The asymptotic properties of the estimators are studied, stationarity conditions for the BEGE model are derived, and finite-sample evidence is presented. Empirical results reveal strong time-varying bad and good environment covariances across North American and European equity markets, suggesting a common underlying risk factor.