JPMorgan Chase, New York
In this issue we present three research papers and one technical report. The first research paper, "Evaluating the performance of static versus dynamic models of credit default: evidence from long-term Small Business Administrationguaranteed loans", is by Dennis Glennon and Peter Negro. In this paper the authors focus on the assumption that the probability of default is constant over the full performance horizon. This assumption is often implicit in the static consumer scoring models commonly used in practice, but is least valid for longer-term amortizing loans (eg, mortgages and small business loans) because the likelihood of default is likely to change over time due to factors such as loan seasoning, changes in borrowers' leverage and changing collateral values.
The paper constructs default models using both a conventional scoring (ie, static) approach and survival analysis methods that are specifically designed to capture the time-sensitive (ie, dynamic) nature of the default process. The authors showthat survival-time models consistently generate more accurate dollar-loss forecasts over multiple time periods and performance horizons. They therefore argue that loan seasoning is a key factor in the development of accurate loan-loss forecasts for longer-term amortizing loans.
The second research paper, “Transfer risk under Basel Pillar 1”, is by Amit Agarwal, Paul Harrald and Peter Thompson. The authors propose an approach that takes account of transfer risk: the risk that a borrower will not be able to repay its foreign currency obligations due to foreign currency restrictions. The paper investigates the role of transfer risk in the Basel capital adequacy regulations, and analyzes how an obligor’s default probability should be adjusted to account for transfer risk. The Basel II framework explicitly recognizes the incremental default risk on a foreign currency obligation of a particular obligor due to transfer risk.However, it is relatively rare to find the issue of transfer risk captured in either academic or practitioner-oriented credit risk literature.
The third research paper, “Benchmarking the incremental risk charge”, is by Christopher Finger. The paper describes the incremental risk charge (IRC), discusses some of its key parameters and assumptions and provides some IRC estimates for a few different portfolios. The paper sets out to evaluate the impact of various potential modeling assumptions on the ultimate capital charges produced. In particular, the paper discusses the interpretation and operationalization of the constant level of risk assumption within the IRC. The author seeks to deconstruct the meaning of the constant level of risk rebalancing, and he provides some financial intuition as to why rebalancing has the effect that it does on capital. The work related to the IRC (in particular, once migration risks are included) has focused on specific bank portfolios and taken the use of the convolution approach as given. While providing some sense of the magnitude of the charges, these studies do not add to our understanding of how the IRC would operate as a model.
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 duplication of efforts. The contents of technical reports complement rigorous conceptual and model developments presented in the research papers and provide a lot of value to practitioners.
The technical report in this issue, “A brief note on implied historical loss given default”, is by Rogerio F. Porto. The paper describes two methods for evaluating historical loss given default (LGD) for a portfolio: the implied historical method used to evaluate LGD and the workout LGD method. It then determines the conditions under which the two methods produce equivalent results, while briefly addressing the factors that impact the different results when the equivalence conditions are not fulfilled. While these generally apply to LGDs of any portfolio, Basel II only permits the first method for retail portfolios. A brief comparison between ex post implied historical LGD and ex ante LGD evaluations is also provided in the paper.