Non-Gaussian Panel Time Series Model Decomposing Default Risk
We model 1980--2003 rating and cohort specific cumulative default
frequencies. The data is decomposed into systematic and firm-specific
risk components. We have to cope with
(i) the shared exposure of each cohort and rating class to the same
systematic risk factor;
(ii) strongly non-Gaussian features of the individual time series;
(iii) possible dynamics of the unobserved common risk factor;
(iv) changing default probabilities per rating cohort over time
(ageing effects), and
(v) missing observations. We propose a non-Gaussian multivariate state
space model that simultaneously deals with all of this issues.
The model is estimated using importance sampling techniques.
A NON-GAUSSIAN PANEL TIME SERIES MODEL FOR ESTIMATING AND
DECOMPOSING DEFAULT RISK
Session Credit Risk
Field Finance
Session Chair Haibin Zhu, Bank for International Settlements
Presenter(s) Robert J.O. Daniels, De Nederlandsche Bank
Co-Author(s) Andre Lucas, Vrije Universiteit Amsterdam and
Tinbergen Institute and Siem Jan Koopman, Vrije Universiteit Amsterdam
Topics Banking, Empirical Finance, Financial Econometrics and State
Space and Factor Models
Keywords credit risk, importance sampling, multivariate
unobserved components models and non-Gaussian state space models
JEL Codes C32, G21