Crisis-Contingent Dynamics of Connectedness: An SVAR-Spatial-Network “Tripod” Model with Thresholds

In recent years a growing number of methodologies have been proposed to empirically measure the connectedness among financial entities. However, few of them capture the dynamics of the financial connectedness. This paper aims to fill this gap and proposes a novel and systematic model to portray not only the connectedness in a given regime but also its transitions across different regimes. The model is based on an improved version of a “tripod” model, which unifies structural vector auto-regressions (SVARs), spatial models, and network models under one framework. I introduce a transition mechanism into my model, thus making it possible to observe how the interconnections among financial entities vary with certain threshold variables. My model may be applied to various issues regarding the financial and non-financial interconnections among certain entities. As an illustration, I show how my model can shed some new light on the modeling of the recent Eurozone contagions. I reveal a clear causal map of the propagation of shocks among the stock markets in five selected Eurozone countries in both contagion and non-contagion periods, which are determined automatically. The unique roles played by each selected market under both the contagion and non-contagion regimes are efficiently summarized.