本文透過因素分析(factor analysis) 研究2000年網路崩盤前後影響公司債價差風險的要素，以協助投資人運用相關資訊，更準確的預測或提高判斷標的資產的違約可能性。本研究依據Campbell and Taksler (2003) 的架構，以1999年至2004年間新發行公司債為研究對象，剔除金融保險機構、公用事業公司、資料遺漏之公司，綜合早期文獻所提出影響公司債價差風險之要素，經歸納整理後分為五大類：個別公司債特性、公司財務數據、總體經濟數據、公司信用評分及其他公開資訊(非財務數據)。此外，近年來公司治理議題日受重視，本文於模型中另加入公司治理之代理變數，以觀察該解釋變數對公司債價差風險之影響。
研究結果顯示，歷史股價波動性對公司債的價差有一定的影響力，當公司普通股價格波動提高時，該公司股價風險提高，其公司債的違約風險相對提高；公司信用評分對公司債價差風險的影響在本研究並不顯著，其原因可能是信評公司對發行公司採行長期評估，因此，其評價對短期公司債並無影響，然而財務指標中的總負債比率在景氣蕭條時，其變化對公司債價差有顯著的影響力。 This paper implement the concept and methodology of Campbell and Taksler (2003) in the framework of corporate bond yields with a set of explanatory variables generated from earlier literatures that characterized the different aspects of credit risk. After comparing with earlier literatures, We organized the questionable significant factors into five groups to test which variable(s) will influence corporate bond yield spread. The five groups are: macroeconomic factors, accounting factors, bond features, credit ratings, and others. We focus on newly issued US corporate bonds during 1999 to 2004. In addition, this paper makes a distinction between recession period (2000 -2002) and recovery period (2003 – 2004) of the economy to see whether or not macro economy conditions would cause differences in the explanatory variables. Overall, we are able to pinpoint factors for investors to target as signals for default of issuers using a less structured econometric analysis in various positions. Results show similar receptiveness as Campbell and Taksler (2003), which indicate that historical equity volatility and total debts over total assets are two noteworthy variables to the changes in corporate yield spread. The important predictors of short term model focuses on bond characteristics (coupon rate and bond duration) and GDP. Bonds with different maturity, collateral, creditworthiness, and optionality can be affected by the investors’ level of confidence towards the future. From this research, we conclude that risk premiums for investment grade bonds are difficult to explain by any only one explanatory variable.
