"Vote with Feet"or"Vote with Hands":Institutional Shareholding and Corporate Fraud
Corporate fraud has always been a challenging issue in the capital market.After reaching a low point in 2010,the probability of corporate fraud in China A-share listed companies gradually increased.Is this due to the failure of supervisory measures by institutional investors?This study examines the impact of institutional investor ownership on corporate fraud using data from China A-share listed companies from year 2002 to 2018.The research findings indicate that the higher the proportion of institutional investor ownership,the lower the probability of corporate fraud.Even after robustness and endogeneity tests,the conclusion still holds.Furthermore,the study reveals that the restraining effect of institutional investors on corporate fraud has weakened since 2010,which may have contributed to the increase in the probability of corporate fraud.Mechanism analysis show that institutional investors have a stronger supervisory role in regions with higher levels of education and marketization.Analyst attention,firm profitability,and financing constraints are identified as the main channels through which institutional investors exert their influence.Heterogeneity analysis indicates that securities investment funds demonstrate stronger supervisory capabilities,and institutional investor ownership has a more pronounced restraining effect on corporate fraud in companies headquartered in the eastern and central regions,non-state-owned enterprises,and companies involved in mergers and acquisitions during the analyzed period.There are two cultures in statistical modeling for drawing conclusions from data:"data modeling"and"algorithmic modeling".Data modeling assumes the data is generated by a given model,whereas algorithmic modeling views the data mechanism as unknown.Based on the core principles that"algorithmic modeling"and"data modeling"are complementary,this paper conducts cross-validation of six machine learning models on a benchmark regression model.The results show that the random forest model has the best identification ability(86％),slightly outperforming logistic regression(85％).Considering that the other five models,except for the logistic regression model,support nonlinear relationships.The machine learning results validate the effectiveness and reliability of the benchmark regression model.The conclusions of this study can be summarized in three main points.Firstly,institutional investors in the China A-share market have the capacity to effectively supervise corporate fraud.However,their impact has diminished after 2010,coinciding with an increasing trend of corporate fraud,which partially explains the frequent occurrence of corporate fraud cases in the Chinese capital market.Secondly,institutional investors'influence on corporate fraud exhibits significant variations,which can be summarized as regional differences,institutional differences,and corporate differences.From a regional perspective,the impact of institutional investors differs significantly between economically developed enterprises in the central and eastern regions and those in the western regions.Additionally,the level of education and marketization in different regions also affects the supervisory capacity of institutional investors.Analyzing institutional differences,the impact on corporate fraud varies between securities investment funds and other types of investment institutions.From a corporate perspective,different ownership structures and whether a company is involved in mergers and acquisitions can influence the restraining effect of institutional investors on corporate fraud.Lastly,institutional investors can restrain corporate fraud through multiple channels,including increasing analyst coverage,enhancing firm profitability,and alleviating financing constraints.These measures not only reduce the risk of corporate fraud but also contribute to improving the operational and governance capabilities of companies,ultimately enhancing the quality of listed companies.The reduction of fraudulent behavior is a"by-product"of the improvement in the quality of listed companies,and suppressing the occurrence of corporate fraud requires starting from enhancing the quality of listed company development.
NETL
NSTL
万方数据
