Return to search

A framework for the separation of PD and LGD of CDS using stock options

Summary I propose a framework to separate the probability of default (PD) and loss given default (LGD) of credit default swap (CDS). While the no arbitrage CDS spread is determined given PD and LGD, the separation of PD and LGD is impossible solely with CDS spread. Thus I develop a joint estimation procedure of PD with stock optiCJ'riS;a:s the probability that the underlying stock price becomes zero. As the option pricing model under credit risk is one of the jump diffusion option pricing model, the calibration of the option pricing model to estimate the PD implied in stock options require the identification of jump diffusion intensity and the diffusion coefficient of Brownian motions. For the stable and precise identification in the calibration, I separate the calibration procedure into two steps, non credit risk factors and credit risk factors. Since the behavior of the term structure model is similar to that of the jump diffusion intensity in the option pricing model, the choice of the term structure model influences the calibration performance. I demonstrate that the adoption of Hull and White (1990) model improve the fitness of the option pricing model and the stability and precision of the PD estimator. While I propose an alternative approach of the PD estimation, I also develop a procedure of LGD derivation of CDS given PD when CDS contract is subject to counterparty risk since CDS pricing model with counterparty risk implies counterparty risk can bias the LGD estimation. Finally I analyze the CDS spread of major UK and US financial institutions in 2008-2009. The results indicate that the market wide systemic risk factors drive the implied LGD although the CDS spread of the large banks reflected strong impact of the government intervention.

Identiferoai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:573072
Date January 2012
CreatorsTakeyama, Azusa
PublisherUniversity of Essex
Source SetsEthos UK
Detected LanguageEnglish
TypeElectronic Thesis or Dissertation

Page generated in 0.0018 seconds