Regulatory capital, credit risk and tail dependence are at the center of this issue of The Journal of Risk. The rapidly evolving Fundamental Review of the Trading Book (FRTB) is the revised framework within which internal models are to be validated. It is evaluated and further enhanced in the issue’s first paper. This is followed by a study on Japanese corporate defaults that relies on data mining. Third comes a paper that advances a new method for backtesting the expected shortfall risk measure. We conclude with a paper on safe haven assets in the presence of extreme tail comovements.
In our first paper, “The implicit constraints of Fundamental Review of the Trading Book profit-and-loss attribution testing and a possible alternative framework”, Alessandro Pogliani, Federico Paganini and Marilena Rata highlight the fact that the central tool of the FRTB, namely profit-and-loss attribution, relies on implicit constraints that are unlikely to be fulfilled and is very sensitive to sample size and outliers. The authors illustrate an alternative approach based on the well-developed ZC test in a framework with minimal assumptions and greater robustness.
In the issue’s second paper, “Loss given default estimation: a two-stage model with classification tree-based boosting and support vector logistic regression”, Yuta Tanoue and Satoshi Yamashita propose the use of classification tree-based boosting and support vector regression to build a predictive model for loss given default. Their study contributes a new methodology and an empirical credit risk analysis that enriches the limited body of research involving Japanese corporations.
Robert Löser, Dominik Wied and Daniel Ziggel develop a test that extends – to multiple dimensions – the backtesting of the unconditional coverage of the expected shortfall risk measure in the third paper in this issue: “New backtests for unconditional coverage of expected shortfall”. This test is based on the so-called cumulative violation process, having the property that its distribution is known for finite out-of-sample sizes.
Safe havens are risky assets known to exhibit negative or zero extreme tail comovements with the market. In “Could holding multiple safe havens improve diversification in a portfolio? The extended skew-t vine copula approach”, Meng-Shiuh Chang, Jing Yuan and Jing Xu use the vine copula approach based on a bivariate extended skew-t distribution to assess whether US dollars (US$) and gold together can be safe havens against stocks. This is the first study to consider the simultaneous use of both assets as safe havens, leading to a trivariate tail setup. With data covering the United Kingdom, Germany, Switzerland and Australia, the authors show that adding both US$ and gold to a portfolio of stock indexes does not provide better safe haven properties than either US$ or gold separately.
Warrington College of Business, University of Florida
The implicit constraints of Fundamental Review of the Trading Book profit-and-loss-attribution testing and a possible alternative framework
This paper presents the constraints embedded in the the profit-and-loss-attribution test and explores a possible alternative framework.
Loss given default estimation: a two-stage model with classification tree-based boosting and support vector logistic regression
In this paper, the authors using a data set composed of five Japanese regional banks, propose an loss given default estimation model using a two-stage model, classification tree-based boosting and support vector regression (SVR).
In this paper, the authors present a new backtest for the unconditional coverage property of expected shortfall.
Could holding multiple safe havens improve diversification in a portfolio? The extended skew-t vine copula approach
In this paper, the authors propose a vine copula model based on a bivariate extended skew-t distribution and derive its corresponding multivariate tail dependence function.