• Login
• Admin

Kosoglu, Elif (2010) CDS Implied Default Probabilities: A Study of a Reduced Form Model for Credit Risk. [Dissertation (University of Nottingham only)] (Unpublished)

PDF - Registered users only - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader Download (1MB)

Abstract

ABSTRACT

This study estimates risk-neutral probability of default from quoted Credit Default Swap (CDS) spreads by employing Hull and White (2000 and 2003) reduced form model for pricing CDS. Once fixing the recovery rate at a predetermined level, we introduce default probability density, q(t) in two forms; a continuous piecewise linear function and a step function. Considering CDS maturity times we extract default probability densities recursively from the model based on the CDS spreads quoted for each maturity. To apply the method to real life data, we construct term structure of default probability densities of 10 UK firms operating in various sectors and with different credit ratings over the three measurement dates: May 1, 2007, December 31, 2008 and December 31, 2009. These dates are believed to reflect changes in market conditions during the period 2007-2009 which covers subprime mortgage crisis. For this analysis, we use one year forward rates estimated from treasury rates as reference risk-free rate. The results suggest that the default probability densities record lowest values in 2007 for most of the firms and the typical positive relation between the default probability and maturity reverses in 2008 when default probability densities usually decline with longer maturities. We also test the model’s performance in terms of discriminating defaulters from non-defaulters by means of Receiver Operating Characteristics (ROC). We use U.S. firms for this analysis and compare performances of different default probability measures calculated based on the CDS implied default probabilities. We find that our model has a very strong discriminatory power in terms of firm defaults, despite a weakness arisen from period dependent analysis due to insufficient CDS data.

Item Type:	Dissertation (University of Nottingham only)
Depositing User:	EP, Services
Date Deposited:	18 Jan 2011 16:19
Last Modified:	15 Feb 2018 15:14
URI:	https://eprints.nottingham.ac.uk/id/eprint/23993

Actions (Archive Staff Only)

Edit View