Addressing the issue of conservatism in probability of default estimates: a validation tool

Carlos Branco

Abstract

ABSTRACT

The revision of the Basel Capital Accord has sparked renewed interest in credit risk, in particular because it has enabled a link to be established between capital requirements and risk measures resulting from models developed by banks. Further research into the implementation of the recommendations of the Basel Committee has been undertaken, specifically around the estimation and validation methodologies used in credit risk models. One theme of this strand of research is the margin of conservatism included in probability of default (PD) estimates. The purpose of this paper is to define a methodology to support the assessment of the margin of conservatism included in PD estimates. This methodology can seen as a benchmark from which comparisons can be drawn rather than an attempt to quantify the margin of conservatism. As such, the evaluation of the margin of conservatism included in PD estimates cannot be isolated from the characteristics of the sample (in terms of nature, availability, timeliness and representativeness), nor from the assumptions assumed, namely, calibration, business cycle, borrower dependencies and results of stress-testing exercises. In the same vein, it is important to position the criteria under which the margin of conservatism is supported into the context of internal policies, regarding, for example, risk appetite.