Heterogeneity and Dynamics in Spatial Network Models

Abstract

We propose an empirical spatial modeling framework that allows for both heterogeneity and time-variation in economic network connections and spillovers. We establish the model's stationarity and ergodicity properties and also show that the model-implied filter is invertible. While highly flexible, the model is straightforward to estimate. We apply the model to several datasets related to Eurozone sovereign credit risk over the period December 2009 to July 2020. Accounting for both heterogeneity and time-variation turns out to be empirically highly important: the new model uncovers intuitive patterns that would go unnoticed in either homogeneous and/or static spatial network models.

Keywords: dynamic networks, spatial auto-regressions, heterogeneous spatial contagion, network heterogeneity, sovereign risk dynamics

JEL Classification: C58, G15, G01