Modeling the Credit Default Swap Spread Curve

The term structure of credit default swap spreads and interest rates shows similar stylized facts, especially in terms of persistency and cross correlations. We implement the Nelson-Siegel yield curve model to fit the term structure of credit default swap spreads. The factors in our model evolve according to a vector autoregressive process. We rewrite the model in state-space form and estimate the parameters in a Bayesian framework, using the multimove Gibbs sampling and extend it by adding Markov switching to the intercept and error covariance matrix of the transition equation. We employ the model to forecast credit default swap spreads, finding that the model outperforms the forecasts of the random walk approach and autoregressive models directly applied to the spreads, in terms of sum of squared prediction errors and the sign forecast performance.