A Study On Prediction Of Financial Distress Of Listed Companies

With the rapid development of China’s economy,the competition among enterprises has become increasingly fierce.As of November 22,2022,with the listing of "Dingtai Hi-Tech" and "Matrix" in A-share,the number of listed companies in China’s A-share market officially exceeded 5000.With the increasing number of listed companies in China,there are constantly companies in financial difficulties and even facing the danger of bankruptcy,which brings great losses to the economic development of China and the interests of investors.There is no doubt that accurate prediction of the risk of financial distress of listed companies can not only maintain the stability of the financial market and reduce the systemic risk,but also prevent the loss to investors caused by the company’s financial crisis,so accurate prediction of financial distress is of great importance.In the previous studies,most of the predictions of corporate financial distress were made from the perspective of financial information,and most of them were based on logistic regression and other models using financial data of the same frequency.These studies ignore other information that may have an impact on corporate financial distress and have major shortcomings.In order to solve the above problems,this paper firstly analyzes theoretically the influencing factors of financial distress of listed companies,including financial indicators,macroeconomics and investor sentiment.Secondly,this paper constructs investor sentiment indicators based on Baidu index,and selects suitable macroeconomic and financial indicators,and innovatively proposes a mixed-frequency prediction model of listed companies’ financial distress based on multidimensional data.Subsequently,this paper conducts an empirical study on the prediction model of financial distress of listed companies based on the proposed model.Specifically,the paper screens all companies in China that are ST for a total of 11 years from 2012-2022 based on the definition of financial distress commonly used in the literature through this paper’s multi-factor screening approach.Based on the annual macroeconomic indicators obtained from the National Bureau of Statistics and Guotaian database,we also combined monthly investor sentiment indicators with annual financial indicators to fit the prediction model of listed companies’ financial distress.In order to examine the predictive power of the constructed model more comprehensively,this paper considers an industry-wide sample and a single manufacturing sample,respectively,and fits a one-step-forward and two-step-forward financial distress mixed-frequency prediction model,and compares the predictive power of the mixed-frequency prediction model proposed in this paper with the traditional logistic regression model using metrics such as accuracy and recall.The empirical results show that: first,in the industry-wide forecasting analysis,the forecasting ability of the mixed-frequency model based on macroeconomic variables,investor sentiment indicators and financial indicators is the best regardless of the one-step or two-step forecasting model,while the forecasting ability of the logistic regression forecasting model based on financial indicators alone is relatively the worst.Second: In the comparison between the whole industry and the manufacturing industry,although the forecasting ability of the model for the manufacturing industry also improved after adding macroeconomic variables and investor sentiment indicators respectively,the forecasting effect of the model for the whole industry was better than that of the model for the single manufacturing industry in all aspects,regardless of any of the model evaluation indicators.