Modeling Systemic Risk: Time-Varying Tail Dependence When Forecasting Marginal Expected Shortfall
提出基于动态混合Copula的模型来估计边际预期短缺，该模型能捕捉时变非线性相依性，在道琼斯工业平均指数机构数据上优于标准基准模型。
In this article, a copula-based model is proposed to estimate the marginal expected shortfall. The model is based on a dynamic mixture copula. The proposed model captures time-varying nonlinear dependence, which is assumed to be constant in alternative approaches. The time-varying copula parameters are endowed with generalized autoregressive score dynamics. For the institutions of the Dow Jones Industrial Average, several variations of the proposed model are considered and compared with alternative, competing models. It is shown that the proposed model outperforms standard benchmarks and produces reasonable findings regarding the risk contributions of the sectors of the Dow Jones Industrial Average.