Measures to assess the discriminatory power of loss given default models

Date

2022

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

The purpose of this master’s thesis is to explore measures that can be used to assess the discriminatory power of loss given default models, which are used for the quantification of unexpected losses by financial institutions. In the first chapter, a general overview of the Basel Accords, the requirements related to regulatory capital calculations, the estimation of unexpected losses and the validation of internal risk estimates are provided. The second chapter highlights various measures that can be used to assess the discriminatory power of loss given default models, including a measure defined in the European Central Bank’s instructions for reporting the validation results of internal models, which is referred to as the generalized AUC. The mathematical properties of the measures are analyzed and comparisons between the measures are provided. Two complementary measures are introduced in the thesis, which can be used to support validation conclusions. In the third chapter, a simulation is presented that illustrates a situation where most of the observations in the validation sample carry relatively low losses and a comparison across the measures defined in the second chapter is provided.

Description

Keywords

Baseli Akordid, sisereitingutel põhinevad mudelid, maksejõuetusest tingitud kahju prognoosivad mudelid, järjestusvõime, astakkorrelatsioon, üldistatud AUC, generalized AUC, rank correlation, discriminatory power, loss given default models, internal ratings-based models, Basel Accords

Citation