文过饰非还是秉笔直书:中国债券评级报告文本信息的价值相关性  

Concealing or Confessing:Value Relevance of the Textual Information in Chinese Bond Rating Reports

作　　者：潘怡麟 张舒怡 朱凯[2] Pan Yilin;Zhang Shuyi;Zhu Kai(Business School,Nanjing University;School of Accountancy,Shanghai University of Finance&Economics)

机构地区：[1]南京大学商学院 [2]上海财经大学会计学院

出　　处：《南开管理评论》2023年第4期101-110,I0020,I0021,共12页Nankai Business Review

基　　金：国家自然科学基金重点项目(71632006);国家自然科学基金项目(72072107)资助。

摘　　要：在我国信用评级行业由政府管制主导、声誉机制不健全的背景下,本文研究了债券评级报告文本信息的价值。以公司债券首次评级报告为样本,本文发现:第一,在控制评级结果的情况下,评级报告文本语调越消极,债券发行价格越高,即具有价值相关性;第二,发行人调增利润的应计盈余管理越多,评级报告文本语调与债券发行价格的关系越弱;第三,当年被证监会采取行政监管措施的评级机构发布的评级报告文本语调与债券发行价格的关系更强;第四,评级报告文本语调越消极,债券违约的概率越大。这些实证结果说明我国评级报告中的文本信息虽然能弥补评级结果划分粗糙的不足,传递增量信息,但是仍受到评级机构与发行人合谋的影响。本文深化了对于我国评级机构行为动机的理解。The credit rating industry in China has long been subject to intense regulation since its inception,and to this day,a robust reputation mechanism that effectively constrains issuer-paid credit rating agencies is yet to be established.Issues such as coarse rating classifications and collusion between rating agencies and issuers are common in corporate credit ratings.Existing research mostly analyzes the motivations of rating agencies by studying credit rating results,while paying insufficient attention to the textual information contained in rating reports that are released concurrently with the rating results.In light of such practice and research background,this study examines the value relevance of texts in corporate bond rating reports,specifically its relationship with bond issue prices,as well as the primary determinants of this value relevance.Texts in rating reports typically contain extensive information about the issuer’s industry,financial position,investment projects,and debt repayment safeguards.Therefore,in cases where the classification of rating results is crude and bonds share similar rating results despite variations in risk,we expect that texts in rating reports can help investors in ranking the credit risks of bonds,thus demon-strating a degree of value relevance.However,the high processing costs associated with textual information pose challenges to reg-ulators in effectively monitoring texts in rating reports.Consequently,we expect that collusive rating agencies may manipulate these texts to‘sugarcoat’the issuer’s risks rather than presenting an objective depiction.The value relevance of texts in rating reports will thus decline when issuers engage in opportunistic behaviors such as earnings management.We utilize rating reports from six issuer-paid rating agencies,covering corporate bonds issued by non-financial firms between 2007 and 2017.A sentiment word dictionary specific to the sample is constructed to measure the linguistic tone of texts in rating reports.The main regression res

关 键 词：信用评级 文本分析 债券定价 管制 

分 类 号：F832.51[经济管理—金融学]