CREDIT DEFAULT SWAP SPREAD FORECASTING USING THE LINEAR BAYESIAN RANDOM COEFFICIENTS MODEL WITH BALANCED PANELS

Access to IIT electronic theses and dissertations is restricted to IIT community members with a valid iit.edu email address. Please log in using your iit.edu email address and password. Requests for external users are handled through inter-library loan and can be emailed to iitmyillms@iit.edu.

Description

This study (thesis) predicts out of sample one to five year quarterly credit default swap spread curves for subsets of a population comprised of 308 companies via the linear Bayesian Random Coefficient Model (RMC) with balanced panel construction, capturing over 80% of reference entities with... Show moreThis study (thesis) predicts out of sample one to five year quarterly credit default swap spread curves for subsets of a population comprised of 308 companies via the linear Bayesian Random Coefficient Model (RMC) with balanced panel construction, capturing over 80% of reference entities with liquid CDS term structures. The use of scoring, structural and reduced form model variations generates credit spread tenure points and curves at the company level. The Altman Z-score and classic Merton structural framework explain too little of the credit default swap spreads out of sample. However, The Merton structural framework works well in predicting out of sample credit default swap spreads when modified by deriving the implied leverage ratio via market spreads. The widely used, Bloomberg implemented, JPMorgan 2001(CDSW) model works well for the period the study covers. The Bayesian Random Coefficients model explains 87% of observed credit default swap spreads one quarter out of sample, substantially exceeding any published research on the credit spread forecasting subject.
PH.D in Management Science, May 2014 Show less

Details
Department

SSB / Stuart School of Business

Title

CREDIT DEFAULT SWAP SPREAD FORECASTING USING THE LINEAR BAYESIAN RANDOM COEFFICIENTS MODEL WITH BALANCED PANELS

Institution

Illinois Institute of Technology

Creator(s)

Arifi, Imir

Advisor(s)

Cooper, Ricky

Date

2014
2014-05

Description

This study (thesis) predicts out of sample one to five year quarterly credit default swap spread curves for subsets of a population comprised of 308 companies via the linear Bayesian Random... Show moreThis study (thesis) predicts out of sample one to five year quarterly credit default swap spread curves for subsets of a population comprised of 308 companies via the linear Bayesian Random Coefficient Model (RMC) with balanced panel construction, capturing over 80% of reference entities with liquid CDS term structures. The use of scoring, structural and reduced form model variations generates credit spread tenure points and curves at the company level. The Altman Z-score and classic Merton structural framework explain too little of the credit default swap spreads out of sample. However, The Merton structural framework works well in predicting out of sample credit default swap spreads when modified by deriving the implied leverage ratio via market spreads. The widely used, Bloomberg implemented, JPMorgan 2001(CDSW) model works well for the period the study covers. The Bayesian Random Coefficients model explains 87% of observed credit default swap spreads one quarter out of sample, substantially exceeding any published research on the credit spread forecasting subject. Show less

Type

Dissertation

File type

application/pdf

Access

Restricted Access

Rights

In Copyright
http://rightsstatements.org/page/InC/1.0/

Identifier

http://hdl.handle.net/10560/3327

<p></p><p></p>