文章基本信息

标题：Advances in multivariate back-testing for credit risk underestimation
本地全文：下载
作者：F. Coppens ; M. Mayer ; L. Millischer 等
期刊名称：Euro Area Balance of Payments and International Investment Position Statistics
印刷版ISSN：1830-3420
电子版ISSN：1830-3439
出版年度：2016
出版社：European Central Bank
摘要：When back-testing the calibration quality of rating systems two-sided sta-tistical tests can detect over- and underestimation of credit risk. Someusers though, such as risk-averse investors and regulators, are primarilyinterested in the underestimation of risk only, and thus require one-sidedtests. The established one-sided tests are multiple tests, which assesseach rating class of the rating system separately and then combine theresults to an overall assessment. However, these multiple tests may failto detect underperformance of the whole rating system. Aiming to im-prove the overall assessment of rating systems, this paper presents a set ofone-sided tests, which assess the performance of all rating classes jointly.These joint tests build on the method of Sterne [1954] for ranking possibleoutcomes by probability, which allows to extend back-testing to a settingof multiple rating classes. The new joint tests are compared to the mostestablished one-sided multiple test and are further shown to outperformthis benchmark in terms of power and size of the acceptance region.
关键词：credit ratings; probability of default; back-testing; one-sided;tests; minP approach; Sterne test