Time-Series Estimation of Aggregate Corporate Bond Credit Spreads
This paper examines the daily time-series properties of aggregate corporate bond credit spreads, using nine Merrill Lynch option-adjusted spread indices with ratings from AAA to CCC for the period of January 1997 through August 2002. The paper introduces an econometric model of credit spreads that incorporates autocorrelation, conditional heteroscedasticity, time-varying jumps, Treasury bond and/or equity mar- ket factors, and index rebalancing effcts. The time-series of credit spread indices are found to be mean-reverting in the long-run through the index rebalancing effct. We also find that the lagged Russell 2000 index return and the lagged changes in the slope of the Treasury yield curve are predictive in forecasting the conditional distribution of credit spreads. Meanwhile, the lagged level of the CBOE VIX index is found to be a good indicator of the probability of jumps in the logarithm of credit spreads. The model diagnostic test shows that the jump specification is crucial in capturing the lep- tokurtic behavior in the daily time-series of log-credit spreads. Finally, the paper finds that the ARCH-jump specification outperforms the specification without jumps in the out-of-sample, one-step-ahead forecast of credit spreads.
Herman Bierens Jing-zhi Huang Weipeng Kong
Economics Department Penn State University Smeal College of Business Penn State University Bear Stearns
国际会议
西安
英文
1-44
2006-07-17（万方平台首次上网日期，不代表论文的发表时间）