This issue of The Journal of Credit Risk presents two research papers and two technical reports. The first research paper in the issue is "Estimation of risk measures for large credit portfolios" by Johannes Hauptmann, Pablo Olivares and Rudi Zagst. The authors propose a methodology to assess risk measures for portfolio losses in the context of credit risk. The paper outlines and studies a saddlepoint method for the estimation of the value-at-risk and the expected shortfall (which is also known as the conditional value at-risk) of the losses on a credit portfolio. The main novelty of the paper is that it takes into account stochastic and correlated recoveries and migration probabilities. Depending on the departure and arrival rating classes, the corresponding variations in default probabilities and in value (recovery) are settled using a factor model. Different approaches that have been proposed in the literature are then compared and calibrated to real data.
Our second research paper, "A credit value adjustment scheme for bank loan portfolios" by Dror Parnes, proposes a scheme for determining the credit value adjustment for a bank loan portfolio. The author points out how the interdependence of various loans of a bank loan portfolio can be accounted for using the Bayesian belief network. A procedure to determine an individual loan's total default probability with given interdependence is provided. Illustrative examples are also given to demonstrate how this type of approach can be applied in practical applications.
A technical report describes a particular practical technique and enumerates situations in which it works well and others in which it does not. Such reports provide extremely useful information to practitioners in terms of saved time and minimizing duplication of effort. The contents of technical reports complement rigorous conceptual and model developments presented in the research papers and provide a lot of value to practitioners.
This issue's first technical report, "Asset correlation in residential mortgage-backed security reference portfolios" by Marco Geidosch, analyzes a core issue of portfolio credit risk modeling; namely, multivariate default dependence. The author studies the collapse of the US residential mortgage-backed securities (RMBSs) market. This market provides a unique and very interesting sample because the widespread underestimation of default comovement was a key feature of the US subprime crisis. Surprisingly, the estimated asset correlation of the data sample (which includes the most toxic RMBS deals from the subprime crisis and which can therefore be interpreted as being under severe stress conditions) is significantly lower than the corresponding Basel II value of 15% for residential mortgages.
The second technical report, "Asset correlation of retail loans in the context of the new Basel Capital Accord" by Pawel Siarka, focuses on an important issue: estimating asset correlations for retail credit risk exposures. The paper reviews results from past studies, presents several estimation methods, and estimates the asset correlation of a car loan pool originated in Poland. The author frames the topic in the context of retail asset correlations, as recommended by the Basel Capital Accord, and shows that for the analyzed car loan portfolio the real level of correlation is much lower than that recommended by Basel II.
JPMorgan Chase, New York
The approach to the measurement of credit risk recommended by the new Basel Capital Accord (Basel II) gives a wide choice of basic risk estimators. However, the rules for estimating asset correlations are defined in an ambiguous manner.
In this study the authors develop an analytical scheme that integrates a large spectrum of typical bank loans and credits, accommodates common bank loan portfolio chronological interdependencies and allows the necessary credit value adjustments (CVAs)…
In this paper, saddle point techniques are used in the computation of risk measures for large mark-to-market credit portfolios with stochastic recovery and correlation between obligors depending on the state of the economy.
This paper contributes to the literature about estimating asset correlation in two ways. First, we compare the performance of different estimation approaches in a simulation study.