Regulatory use of system-wide estimations of PD, LGD and EAD

Introduction

The objective of prudential regulation has for a long time been the solvency of individual entities and hence a vast range of prudential tools were developed to address this priority. Most recently, due to the period of financial stress and the failure of seemingly solvent institutions, the international supervisory community has expanded the relevance of prudential tools in promoting the stability of the financial system as a whole in addition to individual institutions.

In this sense the Basel Committee has concluded that the issue of systemic risk is probably the most important and most difficult one confronted by the international regulatory community and that progress requires, among other things, a combination of better regulation and the inclusion of a macro perspective into prudential tools.¹ With this in mind, the aim of this paper is to extend the use of a prudential tool typically used to cope with the solvency of individual institutions in order to estimate risk parameters that measure systemic risk.

This objective is achieved by estimating system-wide Probability of Default (PD), Loss Given Default (LGD), and Exposure at Default (EAD) parameters for a retail portfolio with information that is representative of the system, both cross-sectionally and for a relevant part of the economic cycle.

This paper intends to generate a prudential tool that (i) encompasses both micro and macro prudential supervision concerns and (ii) sheds light on the adequacy of banks' individual reserves and their sufficiency to cover systemic expected losses. The tool also seeks to disentangle the nature of exposure of the system to risk, in terms of its dependency on systemic factors, as opposed to idiosyncratic ones.

The paper draws strongly from the recommendations to enhance the resilience of the financial system issued by the Basel Committee in December 2009.²

Particularly, on the loan loss provisioning principles highlighted by the document, in which, among others, it is proposed to: (i) use robust and sound methodologies that reflect expected credit losses in the banks' existing loan portfolio over the life of the portfolio and (ii) the incorporation of a broader range of available credit information than the one presently included in the incurred loss model to achieve early identification and recognition of losses.

In this paper, the second section defines what is understood by system-wide PD, LGD, and EAD and examines the relevance of its use as a regulatory tool. The third section explains the models used to estimate system-wide parameters and the information used in them. The fourth section provides empirical results. The final section provides practical applications of the regulatory tool for both micro and macroprudential dimensions.

¹ Caruana (2010).

² Basel (2009).