Delisting Risk Research Based On Multidimensional Risk Index

China’s Shanghai and Shenzhen exchanges were founded in 1990.Limited by the imperfect delisting system of establishment and the shortage of listed places,the situation of listed companies in China is serious.Under the supervision of The State Council and China Securities Regulatory Commission,Stock Exchanges have continuously revised the delisting system and gradually established a multidimensional delisting index system.Since 2019,the number of delisting companies has increased rapidly.The launch of the comprehensive registration system reform in February 2023 also marks the further improvement of the survival of the fittest mechanism in the capital market,and the number of delisted companies will be at a high level in the future.However,passive delisting often causes a company’s stock price to suffer a severe impact.Based on the protection of investors’ legitimate earnings,this paper expects to provide investors with a reference for investment selection by predicting the risks of passive delisting of listed companies.By summarizing domestic and foreign literature on delisting risk prediction,this paper reviews the reform and optimization process of the delisting system of various levels of capital markets in Shanghai and Shenzhen Exchanges and the policy regulations of the 2020 revised version of the "New Delisting Regulations",and establishes a multidimensional risk index system for delisting risk prediction.The logistic regression model,random forest model and extreme random tree model were used to conduct empirical analysis,rank the importance of variables and test the accuracy of the model,and the following conclusions were obtained:First,the delisting ratio of listed companies in Shanghai and Shenzhen stock exchanges is on the rise,but still at a low level.From 2019 to 2021,the ratio of passive delisting is 0.37%,0.72% and 0.68% respectively.Only in 2022,the ratio of passive delisting is 1.42%,which marks the first year that the delisting ratio of Chinese capital market exceeds 1%.However,there is still a big gap between the delisting ratio of Chinese capital market and that of developed countries,which is7% to 8%.However,with the comprehensive launch of the registration system reform,the normal delisting mechanism will be further improved,and the exit channels will be smooth and diversified.Driven by the optimization of the system,the delisting ratio will rise steadily.Secondly,taking 78 listed companies delisted from 2019 to 2022 as research samples,this paper found that 21 companies were delisted due to face value,53 companies were delisted because they simultaneously or separately hit the upper limit of financial indicators such as net profit,operating income,audit report opinions and net assets,1 company was delisted due to violation of exchange regulations,and 3 companies were delisted due to major violations of laws.The reasons for delisting in previous cases correspond to the delisting provisions of trading,financial,compliance and judicial delisting in the New Delisting Regulations.Meanwhile,many scholars have proved that both financial and nonfinancial information can help predict company delisting.Therefore,this paper will screen and transform indicators based on the New Delisting Regulations,and establish a multidimensional index system for predicting delisting risks covering transaction risks,financial risks,compliance risks and judicial risks.Thirdly,this study found that the delisted companies from 2019 to 2021 from reaching the delisted index to the termination of listing focused on 3-4 years,while in 2022 after the promulgation of the New Delisted regulations,the delisted companies from reaching the delisted index to the termination of listing shortened to 2 years.Therefore,this paper will carry out delisting risk prediction research on companies delisted from 2019 to 2021 and 2022 respectively.Through the empirical study and robustness test of the three models,it is found that for the companies delisted in 2022,there is little difference in the classification correctness of the three models.Compared with the random forest model,the prediction effect of financial,judicial risk and other related indicators 2 years in advance is slightly better.For companies delisted from 2019 to 2021,the classification correctness of logistic regression model is significantly higher than that of the other two models.Therefore,applying logistic regression model and using relevant indicators of financial and judicial risks to forecast three years in advance has the best effect.Further,this paper believes that when the sample is in imbalance,after taking the oversampling algorithm for the training set,using the logistic regression model to get the dichotomy result more robust,the prediction accuracy can keep at a good level.This article speculates that it may have to do with the fact that SMOTE oversampling balanced the imbalance of the training set sample to compensate for the deficiency of logistic regression,and that the low proportion of delisting firms during 2019-2021 made it more difficult for the Random Forest model to capture the heterogeneity of characteristics.Finally,this paper suggests to enrich the dimensions of financial indicators,refine major illegal indicators and define clear indicator measurement caliber,in order to provide some references for the optimization and improvement of the delisting system.