Bayesian Networks and Stochastic Factor Models
27 Pages Posted: 13 Nov 2015 Last revised: 5 Sep 2017
Date Written: August 14, 2017
Abstract
This paper seeks to establish both direct connections and similarities between the Bayesian network approach and stochastic factor modelling in quantitative risk management. The discussion covers comparison between Bayesian networks as used for financial stress testing and portfolio management with models for calculation of potential future exposure (PFE), and to a lesser degree, of xVA and market risk measures such as VAR, ES, IRC, CRM, DRC (formerly IDR). The paper also considers whether in some common form Bayesian networks could be expressed as a particular re-formulation of a stochastic factor model. Appendix considers potential impact of machine learning on market and counterparty risk modelling within the constraints of the current regulatory climate.
Keywords: Bayesian networks, Counterparty risk, Market risk, PFE, xVA, VAR, IRC, CRM, Basel 2.5, Basel 3, FRTB, machine learning
JEL Classification: G13, G17, G18
Suggested Citation: Suggested Citation