Analytical value-at-risk with jumps and credit risk (Q5936313)

An analytical approximation for the Value at Risk (VaR) and other risk measures of over-the counter portfolio are provided. The portfolio may include options and other derivatives with defaultable counterparties or borrowers. The risk setting is that of a classical multi-factor jump-diffusion for default intensities and asset returns under which between-jump returns are correlated Brownian motions, with return jumps at Poisson arrivals that are jointly normally distributed. This allows for fat-tailed and skewed return distributions. It is found that, at least for the cases that are examined, extreme tail losses are dominated by either credit risk (for example in loan portfolios) or by the underlying market price risk (for example with equity option portfolios), but that only rarely do both sources of risk make a large contribution to tail risk. Using the delta-gamma approximation and taking advantage of the Fourier transform technique, the analytical treatment of the VaR calculation is developed. This analytical approximation technique is illustrated in two examples: one focuses on jump risk in market returns and the other on credit risk exposures. Through these examples it is shown that, for a given level of accuracy, the analytical VaR approximation provides a substantial reduction in computational effort over the alternative of Monte-Carlo simulation. In their previous work in this direction [\textit{D. Duffie} and \textit{J. Pan}, J. Derivatives 4, No. 3, 7-49 (1997)], a relatively standard Monte-Carlo procedure was used under which returns were simulated, and delta-gamma approximations to derivative prices were used to estimate marks to market for each scenario. Repeated simulation is used to estimate a confidence interval on losses that is typically known as ``value at risk'', or VaR.