| As of December 2021,Chinese bond market custody balance has reached 133.5trillion yuan,ranking second in the world.The outline of the "14th Five-Year Plan" points out that it is necessary to steadily expand the scale of the bond market and improve the market-oriented bond issuance mechanism.As an important part of Chinese capital market,the bond market is increasingly important and is injecting a strong impetus into economic development.However,at the time of the rapid development of the bond market,many problems have also been exposed.The default of the "11 Superday Bonds" in 2014 represented that the "rigid redemption" of Chinese bond market was broken,and bond defaults were gradually normalized,reducing the trust of investors.Improving bond pricing efficiency and determining reasonable credit spreads have become issues that need to be solved urgently.The research on analyzing the influencing factors of bond credit spreads from structured data has been quite sufficient.Limited by the technical level,the value of unstructured data has been ignored.As text information directly related to the issuance of bonds,the unstructured data in the bond prospectus clarifies the issuer’s credit status,investment risks,and use of raised funds,etc.,and is an important reference for investors in the primary market to determine risk compensation.The China Securities Regulatory Commission stipulates that issuers should use plain,factual descriptions and disclose risk factors related to the bond and the issuer.Therefore,based on the disclosure provisions,this paper selects two text feature indexes,text professionalism and risk information disclosure content,from the text features of the bond offering prospectus,studies its impact on the bond credit spread,and puts forward corresponding suggestions.This paper takes the corporate bonds issued by A-share listed companies from2010 to 2020 as the research sample,uses the bond offering prospectus downloaded from Wind database as the data source,constructs financial professional dictionaries and risk-related dictionaries with reference to authoritative literatures,and uses Python technology to quantify the textual characteristic data as explanatory variables.And from the two dimensions of corporate financial characteristics and bond attributes selected control variables for empirical regression analysis,study the text features of bond prospectus on bond credit spread.In the robustness test,the robustness of empirical results is tested by changing the definition of explanatory variables and verifying the impact of modifying the disclosure policy of bond prospectus.In the test of heterogeneity,this paper studies the nature of property rights and the reputation of underwriters.In addition,in the further analysis,this paper also analyzes the influence of the above text features of the annual report,another important channel for bond investors to obtain information,on the bond credit spread.This paper draws the following conclusions: the degree of text professionalism and risk information disclosure of the bond prospectus are significantly positively correlated with the bond credit spread;this conclusion still holds when the bond issuer is a non-state-owned enterprise or the underwriter has a high reputation;the above textual features of corporate annual reports are significantly and positively correlated with bond credit spread.This paper puts forward the following suggestions: investors should improve their professionalism,tap the value of unstructured data,and establish correct risk awareness;relevant departments should strengthen the supervision of disclosed text information,set up text professionalism measurement indicators,and specify information disclosure,and refine the types of disclosed risks;information disclosing parties should strengthen internal control management,reduce agency costs,and truthfully disclose risk-related information in simple and easy-to-understand language. |
