4

Credit Portfolio Correlations and Uncertainty

Steffi Höse and Stefan Huschens

Technische Universität Dresden

4.1 INTRODUCTION

In light of the current financial crisis, financial institutions are under pressure to implement a meaningful sensitivity analysis and a sound stress-testing practice for the parameters of the credit portfolio models used. Although these models differ in their assumptions and complexity, default probabilities and asset (return) correlations are typical input parameters. The literature focuses mainly on the point estimation of these parameters, see Gordy (2000, p. 146), Frey and McNeil (2003), de Servigny and Renault (2004, pp. 185f.), McNeil and Wendin (2007), and on the interval estimation of default probabilities, see Höse and Huschens (2003), Christensen et al. (2004), Trück and Rachev (2005), Pluto and Tasche (2005), Hanson and Schuermann (2006), Pluto and Tasche (2006). In contrast, this chapter concentrates on the statistical interval estimation of asset correlations. There are at least six methods for the estimation of asset correlations depending on the type of available data: (a) the estimation of default correlations from default data and the derivation of implicit asset correlations by using a factor model; (b) the estimation of asset correlations from time series of transformed default rates, e.g. probits of default rates, see Höse and Huschens (2011a); (c) the estimation of asset correlations from time series of asset returns, sometimes approximated ...

Get Credit Securitisations and Derivatives: Challenges for the Global Markets now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.