A spatiotemporal approach is proposed for modeling risk spillovers using time-varying proximity matrices based on observable financial networks and a new bilateral Multivariate GARCH specification is introduced. The covariance stationarity and identification of the model is studied, developing the quasi-maximum-likelihood estimator and analysing its consistency and asymptotic normality. Further, it is shown how to isolate risk channels and it is discussed how to compute target exposure in order to reduce the system variance. An empirical analysis on Euro-area sovereign credit default swap data indicates that Italy and Ireland are key players in spreading risk, France and Portugal are major risk receivers, and Spain’s non-trivial role as a risk middleman is uncovered.

提出了一种基于可观测金融网络的时变邻近矩阵对风险溢出进行建模的时空方法，并引入了新的双边多元 GARCH 规范。研究了模型的协方差平稳性此外，还展示了如何隔离风险渠道，并讨论了如何计算目标暴露以减少系统方差。对欧元区主权信用违约掉期数据的实证分析表明，意大利和爱尔兰是风险分散的关键参与者，法国和葡萄牙是主要的风险接收者，而西班牙作为风险中间人的作用也凸显出来。