Three essays on credit spreads

This dissertation consists of three essays, which investigate the time series properties of credit spreads and their correlations with other markets.; The first essay empirically examines the common factor in junk bond spreads. The yield of industry bond indices is decomposed into four unobserved factors: permanent term factor, temporary term factor, common risk factor and industry specific risk factor. We find that permanent and temporary term factor are major components of junk bond yields and fit the risk free rate very well. Also our analysis confirms the existence of a common risk factor in junk bond yields and finds that it is well explained by Fama-French bond risk factors and volatility in equity markets. In addition, we provide evidence of co-movement between common risk factor and business cycle.; The second essay suggests a regime switch in the volatility of the factors in bond yields. The regime shifts are governed by a discrete-time Markov process with state-dependent transition probabilities. We estimate this model with a Markov Chain Monte Carlo algorithm and find the common risk factor has the same high and low volatility regimes as term factors. The maturity of bonds plays an important role in identifying the regime shift and characterizing the correlation between common risk factor and term factors. Our results support a high volatility regime during Fed experiment period.; The third essay empirically analyzes the time-varying correlations between changes in credit spreads and changes in interest rates, between changes in credit spread and equity returns and between equity returns and changes in interest rates. A multivariate GARCH model is proposed to facilitate the dynamic correlation process. We find the correlation changes not only quantitatively but also qualitatively, especially those correlations associated with equity returns. The correlations display a hump shape across the quality of bond, in which BB and B grade bond have highest correlation; AAA and BAA grade bonds have the next level correlations; C grade bond has the lowest correlation with all other asset categories. Moreover, we find that business cycle has an impact on equity returns and junk bond spreads associated correlations.; The paper suggests that some historic events such as "911" changes the structure of the correlations. Furthermore, macroeconomic announcements have more impacts on the bond market than the equity market. Individually, FOMC affect more on correlation with the change in the slope of interest rates. GDP and PPI have more impact on correlations associated with change in the credit spreads with lower ratings.