Pricing the callable perpetual bonds with least squares Monte Carlo & artificial neural network method

• Main Authors: Tsai, Wei-Hao, 蔡維豪
• Other Authors: Lin, Shih-Kuei
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/s5c6f5

碩士 === 國立政治大學 === 金融學系 === 106 === This study takes callable perpetual bonds as evaluation target, using the Hull & White (1990) model to characterize the dynamic process of short-term interest rates. Firstly, using the Least Squares Monte Carlo simulation approach proposed by Longstaff & Schwartz (2001), it is simple and intuitive, and can effectively evaluate financial instruments with path-dependent characteristics. Then replace the multiple regression model used in the original method with the back-propagation neural network model, calculate the continuing holding value under the nonlinear relationship and carry out subsequent evaluation based on the model estimation results, to provide another evaluation method of callable perpetual bonds. It is expected that through the results of this research, investors and issuers will have a basic understanding of the evaluation of callable perpetual bonds.