社会关系与企业信息披露质量——基于中国上市公司年报的文本分析

Guanxi and Information Disclosure Quality: An Empirical Study Based on Textual Analysis of Chinese Listed Firms' Annual Reports

本文采用计算机语言技术针对2002-2013年我国上市公司年报文本进行分析得出企业信息披露的质量,在此基础上研究了企业对社会网络关系的依赖与信息披露质量之间的关系。研究结果表明,依赖社会关系获取资源的企业(关系型企业)对外公开的年报信息披露质量显著低于其他企业。进一步分析发现,股权分置改革在一定程度上缓解了这一问题。具体而言,我们采用双重差分模型,对比关系型企业与其他企业信息披露质量差异在股权分置改革前后的变化。结果显示,关系型企业的信息披露质量相比其他上市公司在股权分置改革后有了显著的提升。本文将社会网络理论引入会计学的研究范畴,对理解中国企业信息披露决策具有重要的参考价值。

Guanxi is an important way for firms to obtain resources, especially in the market in which the legal protection is relatively weak. This paper studies the effect of Guanxi on firms＇ information disclosure quality, which is generated from annual reports of Chinese listed companies from 2002 to 2013 using an advanced machine learning technology. Since disclosure is a mechanism to reduce information asymmetry between firms and the market, those rely more on social connections（hereinafter referred to as relation-based firms） have weaker incentive to provide high quality information disclosure. In comparison, firms without connections have to disclose high quality information in order to gain the trust of investees and therefore reduce the cost of capital. Our results confirm this prediction by finding that the disclosure quality of relation-based firms is significantly lower than that of other firms. We next show the split share reform alleviates this effect to a certain extent. After the reform, controlling shareholders＇ stocks become tradable and therefore their wealth is more affected by the public. Even if relation-based firms could still achieve resources through their connections, controlling shareholders could benefit from increasing the disclosure quality to improve firm value. Using a difference-in-difference（DID） model, we compare the quality of information disclosure between relation-based firms and other firms before and after the reform. The results show that information disclosure quality of relation-based firms has improved significantly compared to other companies after the reform. Taken together, our empirical findings support our prediction that relation-based firms have weaker incentive in information disclosure than other firms, and the stock split reform could alleviate the effect. Literature argues that Chinese listed companies are opaque in public disclosure; our study provides a new angle on the demand of disclosure quality. Specifically, we introduce social network theory into accounting research to have an additional understanding of information disclosure decision of listed firms. Our empirical results could be helpful to government policy.