本論文之動機，在於了解公司債信用利差(corporate credit spread)之組成特質，如何藉由對信用利差之分解，改善在長期間信用風險之分散成效。經由對信用利差中受總體經濟變數影響的市場利差(market spread)，以及對受信用等級或產業因素影響之業務特性利差(idiosyncratic spread)之差異分析，了解利差之分解在公司債投資組合風險分散上之意義。本論文之研究方向與Fama（1986）相吻合，該研究提出違約風險溢酬(default premium)在長期間之重要性，相對於期限溢酬(term premium)在傳統流動性偏好假說上之推論，違約風險溢酬或信用利差具有對經濟衰退徵兆之高敏感性，特別是高收益公司債之利差較投資級公司債對衰退有更高之預測能力，亦即不同信用評等下信用利差應以不同方式看待。目前學理或實務上均採用同一模型分析不同信用等級公司債之信用利差，對業務特性利差長期中的差別效果不易分辨。因此本論文之研究目的，在於找出將信用利差分解為市場利差與業務特性利差之方法，並提供實證研究上之根據以及應考慮之相關因素。根據公司債之可分散信用風險(diversifiable credit risk)之理論，本論文提出實證上簡易可靠之風險分解方法。並且在資料檢驗中證明，只有信用利差中之市場利差以外之部份，方為衡量個別公司債信用風險之重點。管理公司債組合時，為調整組合權重之主要參考，亦是個別公司債定價差異之主要來源。依傳統沿用完整之信用利差作為風險分散或差別定價之基礎，將會產生風險配置上之扭曲，此乃本論文最主要精神。本論文之貢獻，在學理上驗證了業務特性信用利差之重要性、提供了對其做具體估計之方法、以其預測信用風險之可行方法、提出其在景氣循環中之非對稱特性，以及應用於個別公司債時適用之計量模型。在實務上對公司債組合之風險分散提供風險配置之準則，並以業務特性信用利差作為各信用等級公司債之訂價之基礎。由於對股票投資組合之風險分散技巧，並無法全然應用於固定收益組合上，因此本論文將使公司債組合中之風險分散更為精確，更有效率地反映公司債市場之風險報酬特性。 本論文對信用利差分解之分析，反映了對公司債各風險因素作正確風險補償之重要性，亦彌補了傳統結構式模型對信用利差估計之缺陷。對於信用利差中系統與個別風險之區隔，與股票價格模型在精神上相似，卻須以幾乎完全不同之方法分析。更精確完善之利差分解，將可應用於更廣泛之固定收益證券及其組合。對世界各先進國家以及我國之資本市場而言，可發揮擴大並促進資本市場有效配置之影響。 The motivation of this dissertation is to explore the characteristics of components of corporate credit spreads and how to improve long run credit risk diversification through the decomposition of credit spreads. Specifically, the implications for risk allocation in holding bond portfolio by decomposing credit spread into a market spread and an idiosyncratic component respectively. The former is driven by macroeconomic variables and the latter is related to credit rating or industry factors. This motivation is also in the spirit of Fama (1986), which argued against the traditional preferred habitat hypothesis of term structure stressing a more important role of default premium than that of term premium in predicting economic recession. His findings of a stronger effect from the high-yield issues than the investment grade ones suggest the need to differentiate how we look at credit risks across different credit ratings or firms. Currently credit spreads of different credit ratings are analyzed without distinguishing between general fixed income market risks and credit rating or idiosyncratic credit risks. So the purpose of this thesis is specifically to identify diversifiable risks by way of decomposing yield spreads of fixed income into systematic and idiosyncratic spreads. This dissertation proposes a practically simple decomposition scheme based on relevant theories and verify its validity through series of solid empirical examinations. Methodology in this dissertation provides a tractable way of measuring diversifiable risks which is crucial for portfolio composition and related adjustments. The ability of idiosyncratic spread to predict future defaults is another contribution. The asymmetric adjustment of credit spread suggests the very drawback of conventional structural model of credit spreads. The applicability of decomposition on individual corporate credit spreads is of substantial importance as well. To the extent that diversification technique in equity portfolio does not apply well to fixed income portfolios, this dissertation provides a solution to more accurately diversify bond portfolio risks and reflect efficiently the risk-return characteristics of corporate liabilities. The analysis of credit spread decomposition in this dissertation reflects the importance of proper risk compensation in corporate bond pricing. It provides a remedy to inconsistencies exhibited in estimates from the conventional structural approach. Although the identification of non-systematic risk within credit spread is similar in spirit to that in equity pricing, methods of entirely different nature have to be adopted. Decomposition conducted with more accuracy could be applied extensively in fixed income securities and their portfolios. Efficient allocation of risks in the market can be extended within the capital markets of developed economy as well as that in Taiwan.