Bootstrap-based probabilistic analysis of spillover scenarios in economic and financial networks

We apply techniques from the event probability forecasting literature to the analysis of spillover scenarios in economic and financial networks. A simple spillover scenario is expressed as an inequality constraint with respect to a single spillover measure. More complex spillover scenarios can be defined as combinations of simple scenarios. The scenario probabilities are evaluated using a non-parametric bootstrap. We use our technique to study credit risk transmission among a group of 18 countries over the 2006–2010 period. We show that abrupt changes in the probabilities of “crisis scenarios” accurately map on to key events during the Global Financial Crisis.