Early Warning Research On Financial Crisis Of Manufacturing Listed Companies Based On Bayesian Model Average

With the implementation of the comprehensive deepening reform policy,the market is more and more open and the competition is more and more fierce.The data show that the proportion of listed manufacturing enterprises treated by ST is much higher than that of financial industry.Therefore,it is important to accurately predict the possible financial risks of manufacturing enterprises.However,the financial situation of enterprises is often determined by many factors,How to choose the appropriate index to establish the model is a difficult problem.For the first time,the Bayesian Model Average(BMA)method is introduced in the enterprise financial crisis early warning field.All possible financial indicators can be incorporated into the indicator system construction model,and the posterior probability of the entire model space can be calculated.Obtaining posterior inclusion probability,posterior mean,posterior standard deviation,T statistic significant rate,etc of each indicator,comprehensively display the importance degree of each indicator,and provide effective evaluation and early warning for possible financial crisis and solve the problem of model uncertainty.The BMA method is used to predict the risk probability of a financial crisis,and it is compared with the conventional stepwise regression method.Through empirical research,the following conclusions are drawn:among the six ways to realize BMA method,the model built by the integral likelihood approximation calculation method combined with theMC~3 model screening algorithm has a better comprehensive effect,the prediction accuracy rate is up to 82%,and the prediction accuracy of BMA method is better than the stepwise regression method in the prediction of enterprise financial crisis risk.Comparing the index posterior inclusion probability obtained by the BMA method with the index coefficient and significance included in the model constructed by the stepwise regression method,it is found that the BMA method can more accurately identify the key financial indicators and is not easy to lose information.after the BMA method,the current ratio,fixed assets turnover,receivable turnover ratio,rate of return on total assets,net profit rate of total assets,net operating interest rate play an extremely important role in predicting the financial crisis risk of enterprises,which posterior probability is greater than 0.5,It represents four aspects of corporate solvency,operational capability,development capability and indicators per share.Moreover,these indicators have a great contribution to the model and belong to the indicators that should be focused on,which plays an extremely important role in preventing corporate financial risks.