Textual Analysis Of Bond Prospectus And Bond Financing Cost

In recent years,China’s bond market has continued to develop and expand,and has become one of the important ways of direct financing for enterprises.More and more investors will pay more attention to the credit risk of bonds.In practice,investors generally judge the credit risk of bonds in two ways: first,by using the credit rating ratings and reports published by third-party rating agencies;second,by making their own risk assessment of the bond issuer based on various publicly available information.For the first approach,the reliability and accuracy of the bond rating depends on the independence and professionalism of the third-party rating agency,which has been questioned by many parties in the Chinese bond market,mainly because the issuer-paid approach makes the credit rating agency and the bond issuer previously have a deep interest connection.At the same time the imperfect regulatory system for ratings and the lack of a reputation mechanism further exacerbate the market’s doubts about rating ratings.As a result,investors will rely more on information disclosed by the issuer or obtained from public sources to assess the risk of the bond.One of the key disclosure documents is the bond prospectus.In the SHSE’s business guide,it is clearly stated that the bond prospectus should disclose information to investors in clear and concise text.The standardisation of the format will help investors to obtain the necessary information and provide a basis for their risk assessment and to give a reasonable pricing.Based on the above analysis,this paper conducts the following research: Firstly,this paper takes the corporate bonds publicly issued on the SSE and SZSE from 2008 to 2022 as the sample,uses a Python crawler to obtain the bond prospectus,and separates out the part of "Risk Tips and Notes",and constructed three indicators of text similarity,readability and tone by using the text vector method and the lexicon method.Secondly,this paper examines whether the above three text information indicators have an impact on the cost of bond financing,controlling for the basic characteristics of the bond and the financial indicators of the bond issuer.Thirdly,this paper measures the degree of information asymmetry by the proportion of intangible assets and the credit risk by the bankruptcy index,and examines the mechanisms of similarity,readability and tone on the cost of bond financing respectively.Fourthly,this paper analyses the differences in the impact of the three textual features on the cost of bond financing under different credit ratings and underwriter rankings,further demonstrating the impact mechanism.Fifthly,this paper uses entropy value method to construct comprehensive indicators of textual information characteristics,determine the size of the weights of the three indicators and study their impact on the cost of bond financing.The main findings of this paper include: first,the lower the text similarity,the higher the text readability and the more positive the text tone of the bond prospectus,the lower the cost of bond financing,and this finding remains robust after changing the explanatory variables,the way the explanatory variables are measured,and using instrumental variables to eliminate endogeneity.Second,the paper verifies that the degree of information asymmetry mediates the effect of text similarity and readability on the cost of financing,while the tone of the text mainly reflects bond risk information,which in turn affects the cost of financing.Thirdly,the effect of the textual features of the prospectus on the cost of bond financing is stronger in the group with lower debt ratings,while it is weaker in the group with higher underwriter reputation.Fourthly,the comprehensive index of textual information features has a greater impact on the cost of bond financing than each of the three indicators,and the construction of a bond prospectus textual information disclosure evaluation system will help investors obtain more comprehensive information.The innovations of this paper include the following two aspects: firstly,this paper is the first to construct a bond prospectus text similarity index,based on Hanley and Hoberg(2010),which constructs a unique information content index and a cosine similarity index on the basis of a spatial vector model.Secondly,this paper constructs the comprehensive index of textual information characteristics of bond prospectuses for the first time.Based on the analysis of the three textual information characteristics,entropy method is used to determine the weights of these three textual information characteristics variables and construct a comprehensive evaluation index to provide a theoretical basis for the regulator to establish a bond prospectus textual information disclosure evaluation system in the bond issuance stage.