Adequate risk management of credit portfolios requires modelling the dependence between the default events of different clients. Popular industry models often assume that the default events are conditionally independent given some common underlying risk factor. This allows for modelling indirect default dependence, e.g. via macroeconomic or industry-specific risk drivers. The relative ease of simulating and calibrating such models contributes to their widespread use. Based on empirical evidence (and common sense) however, it is clear that clients can also be connected directly. Examples include economic supply chain dependencies and legal parent subsidiary structures. Ignoring these dependencies can lead to significant underestimation of tail risk. Beyond academic literature, the challenge of incorporating default contagion effects in risk management has also been addressed by the European Banking Authority (EBA), which published guidelines on the identification of ‘Connected Clients’.

This presentation focuses on the challenge of extending the standard models, whilst keeping the usage attractive for practitioners. For example, we discuss parameter calibration and introduce a novel estimation framework for estimating default contagion parameters from historical information. Additionally, we provide insight into what type of network structures lead to the most material default contagion.

Bud works as a Capital Modeller in the Capital Adequacy & Scenario Analyses (CASA) department at Rabobank, where he focuses mostly on the quantification and allocation of Economic Capital. He is also pursuing a part-time PhD at the Korteweg-de Vries Institute (KdVI) for Mathematics at the University of Amsterdam, under the supervision of Prof. dr. Michel Mandjes, Prof. dr. Peter Spreij, and Prof. dr. ir. Erik Winands. The research project currently focuses on modelling default contagion effects for risk management of credit portfolios. He holds multiple degrees from the University of Amsterdam, all awarded with the distinction Cum Laude: an MSc in Stochastics and Financial Mathematics, an MSc in Econometrics, a BSc in Econometrics and Operations Research, and a BSc in Economics and Business.