曾郁仁臺灣大學：財務金融學研究所施翔宇Shih, Hsiang-YuHsiang-YuShih2007-11-282018-07-092007-11-282018-07-092006http://ntur.lib.ntu.edu.tw//handle/246246/60611巨災債券投資人享有的利息收入為三個月LIBOR加上一個固定的利差(spreads)，巨災債券從1990年代後期開始發展，相關的研究眾多，探討利差的文章卻付之闕如，亦無完整的交易的資訊。本文透過各種管道蒐集了1997年至2005年的交易，運用兩個迴歸模型：取自然對數後的LFC模型和本文提出的Aggregate Model，並參考相關文獻，希望找出影響利差高低的因子。實證結果顯示，損失的頻率及幅度、發行金額、所連結的天災風險(peril)的數目和非投資等級之信用評等，對利差有顯著的解釋能力。然而，損失攤陪條件(trigger types)雖被眾多文獻認為有影響能力，在本研究中並不被實證結果所支持。而本篇文章提出的模型，其調整後R2明顯比取自然對數後的LFC模型來得高出許多，證明Aggregate Model的確為目前最能解釋巨災債券利差的模型。The study attempts to explain the spreads over three-month LIBOR rates on catastrophe bonds using regression models. Transactions from 1997 to 2005 are all analyzed by two empirical pricing models, namely the log LFC Model and Aggregate Model. Analytical results indicate that spreads are explained by frequency and severity of loss, size of issue, numbers of perils and non-investment grade rating. However, the role of trigger types is not supported sufficiently. The proposed Aggregate Model generates more accurate estimates for actual spreads than log LFC Model.Contents 1. Introduction ……………………………………………………… 1 2. Data, Models, and Methodology ……………………………… 4 2.1 Selection of Factors ………………………………………… 4 2.2 Models ……………………………………………………...... 7 2.3 Data …………………………………………………………... 9 2.4 Summary Statistics ……………………………………………12 3. Empirical Results ………………………………………………16 4. Conclusion ……………………………………………………… 23 References …………………………………………………………. 25 Data Sources …………………………………………………………26181159 bytesapplication/pdfen-US巨災債券利差迴歸模型Catastrophe BondsSpreadsRegression Modelsthesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/60611/1/ntu-95-R93723034-1.pdf