| With the requirements of reform and opening up and high-quality economic development,the importance of the bond market has become increasingly prominent.As the main financing method of companies,the issuer and issuance volume of corporate bonds have increased rapidly,but corporate bonds have high credit risks.Since the first default in 2014,the number of default events has increased year by year,with a normalized trend,so it is of great significance to the accurate pricing of bonds,which is not only conducive to the construction of bond portfolios,but also reflects the company’s risk level.Considering the high degree of clustering of bond default events,this paper establishes a default intensity model under the framework of affine jump diffusion,assuming that default intensity is an affine function of a series of state variables,so as to reflect the relationship between default intensity and macro variables and firm-level variables.At the same time,a random process with Hawkes jumps is used to characterize the behavior of state variables to reflect the self-exciting characteristics of default intensity.This paper presents a semi-analytical formula for the pricing of defaultable bonds and analyzes the impact of parameters on bond prices through numerical simulation.Based on the evidence from the U.S.market,this paper further analyzes the performance of the proposed model by comparing it with the benchmark models including the model without jump and the model with Poisson jumps,and finds that the Hawkes model has strong in-sample pricing power,and has superior out-of-sample predictive power.Compared with the benchmark model,it is found that the Hawkes model is more suitable for high-risk bonds.Finally,based on the macro market,this paper extracts jump information such as jump intensity,magnitude and direction from macro variables,and analyzes the ability of jump information to explain the credit spread of corporate bonds with different credit ratings and the ability to predict the overall market default rate in the future.It is found that the jump variable contains effective information for credit risk pricing,and the longer the time horizon,the more information it contains. |
