本文以Jarrow, Lando & Turnbull (1997)為基本架構評價風險性債券,將信用評等納入債券評價,破產過程以有限狀態的馬可夫鍊來表示。模型中修正JLT假設破產時回收率為常數及破產過程僅與信用評等變動有關。模型一修正回收率可能受不同破產等級及隨機變數影響,推移矩陣也受隨機變數影響。模型二假設回收率與違約機率受景氣循環影響,利用一階線性迴歸表示景氣循環對回收率與違約機率的影響。實證部分利用COMPUSTATE資料庫中美國長期信用評等低於BBB、Z數值小於1.81,負債權益比大於1、流動比率小於0.8的公司為研究對象,研究景氣循環對償債能力的影響,償債能力指標以流動比率、率速動比率與負債比率表示,得到景氣上升有助於償債能力增加及負債比例下降。同時針對1990年至2003年投機級債券的違約機率分析,景氣循環對投機級債券的違約機率有顯著負相關,顯示景氣上升會降低投機級債券破產機率。回收率與景氣指標也有顯著正相關。故本文評價美國投機級債券時加入景氣循環考量,以景氣指標估計違約率與回收率,在評價上確實得到比JLT模型更接近市場價格表現的結果。 台灣債券由於推移矩陣與回收率資料獲得不易,故利用TEJ資料庫中TCRI信用評等自行推算1999-2003年間之推移矩陣,包括一年、兩年、三年、四年、五年的推移矩陣。回收率利用已知之債券價格及評價公式反推估計。不過推移矩陣的估計期間不夠長,無法完整考量景氣循環影響,所以最後僅以JLT模型試算台灣公司債的價格。
Jarrow, Lando & Turnbull (1997) is the fundamental framework employed in this article to consider the bond pricing problem incorporating credit rating, and represents default process as finite states Markov chain. We adjusted the assumption in JLT that recovery rate is constant when a company goes bankruptcy and default process is only related to credit rating. Model 1 assumes that recovery rate can be affected by different default grade and other variable, and transition matrix can be affected by other variables. Model 2 assumes that recovery rate and default probability can be influenced by macro-economy situation, and we use simple linear regression to demonstrate the relationship between macro-economy situation and default probability. In empirical studies, we use the U.S. corporations in COMPUSTATE data base, which’s long term credit rating is below BBB, Z-score measure less than 1.81, debt-to-equity larger than 1, current ratio less 0.8 as targets to study how do macro-economy affect solvency. Solvency indicator can be represented as current ratio, quick ratio and debt-to-equity, and assume that when macro-economy situation becomes better can make solvency improved and debt-to-equity lowered. We did an analysis of default probability of investment-grade bond during 1990 to 2003 also, and we found there is obvious negative correlation between macro-economy situation and default probability. The result shows the booming helps lowering the default probability investment-grade bond. Moreover, recovery rate and macro-economy indicator has obvious positive correlation too. Therefore, the result of adding macro-economy indicator as a factor in estimation of default probability and recovery rate can really make the pricing result be closer to the real result observed in market than JLT model. It is hard to find data of transition matrix and recovery rate in Taiwan market, so we calculate transition matrix ourselves by the TCRI credit rating in TEJ database from 1999 to 2003, including one, two, three, four and five years transition matrix, and trace back the recovery rate by the observed bond price and pricing formula. Nevertheless, the estimation term structure of transition matrix is not long enough, we can’t consider the macro-economy situation completely, so only used JLT model trying to calculate corporate bond price in Taiwan in the end.