Analysis of Convertible Bond Value Based on Integration of Support Vector Machine and Copula Function

Abstract

The pricing of convertible bonds (CB) is still a problem that needs to be addressed because it is a kind of hybrid financial instrument. This article proposed a novel method with support vector machine (SVM) integrated to copula function. Unlike existing single-factor or bi-factor pricing models based on corporate value and the underlying stock price, respectively, this model can cope with many limitations on the pricing of CB, such as nonlinearity, the departure from normality, multivariate joint normality distribution, market incompleteness, and so on. And above all, the new model exhibited great flexibility in that copula function can portray dependence structure between the underlying stock price and interest rate and that SVM can further tackle nonlinear relationship among variables. Moreover, the integration of SVM and copula function rendered the sensitivity analysis more convenient and accurate. Empirical analysis showed that the proposed model enhanced generation ability of out-of-sample, with satisfactory robustness and mark increase in pricing accuracy and hedging effectiveness compared with the traditional models.

Keywords:

• Bivariate dependence structure
• Convertible bond (CB)
• Copula function
• Statistical learning theory
• Support vector machine (SVM)

Mathematics Subject Classification:

• 68Q32
• 68T05
• 62H20

Acknowledgment

This article was sponsored by the National Natural Science Fund (No. 70971079).