Downturn LGD, Best Estimate of Expected Loss, and Potential LGD under Basel II : Korean Experience
- Authors
- 김명직
- Issue Date
- Nov-2006
- Publisher
- 한양대학교 경제연구소
- Keywords
- Advanced-IRB Approach; Best Estimate of Expected Loss; Downturn LGD; Potential LGD; Single-Factor Model
- Citation
- Journal of Economic Research (JER), v.11, no.2, pp.203 - 223
- Indexed
- KCI
- Journal Title
- Journal of Economic Research (JER)
- Volume
- 11
- Number
- 2
- Start Page
- 203
- End Page
- 223
- URI
- https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/180745
- ISSN
- 1226-4261
- Abstract
- The Advanced-IRB banks should be able to demonstrate to the regulatory supervisors that the long-run LGDs and downturn LGDs are validated with their own data on historical recovery rates. For defaulted exposures, the CRD (Capital Requirements Directive) requires the use of the best estimate of expected loss (BEEL) and of the potential LGD (PLGD) that reflects possible additional unexpected losses during the recovery period. This paper attempts to provide a concrete illustration of the method that allows one to compute various LGD estimates, particularly the BEEL and PLGD, the latter of which have been often overlooked in the previous studies. The proposed LGD model is essentially the same as the single-factor model developed by such authors as Frye (2000) and Dullmann and Trapp (2004), among others, and is applicable to banks which experience the data scarcity in preparing the adoption of the Advanced-IRB approach. By considering the extreme quantile value of the latent factor implied from the model, this model can also be used to generate stress LGD for a stress testing purpose dictated in Pillar II of the New Accord.
- Files in This Item
-
Go to Link
- Appears in
Collections - 서울 경제금융대학 > 서울 경제금융학부 > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/180745)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.