Warning For Corporate Financial Distress Based On BP Neural Network Optimized By Particle Swarm

Financial distress is a world problem. Because it greatly a?ects the financial decisionmaking of investors, credits, and bank o?cers. Auditors also need judgment and predictionon financial distress for the financial information. It is a gradual process for a companygetting into financial distress, so it must have some signs and thus could be forecasted.Forecasting corporate financial distress accurately makes much real sense. For example,the exact forecast could increase interests of investors and debtors, could protect managersfrom financial crisis, and could help government to monitor listed company's quality andsecurities market's risk.There are many statistical procedures to handle this financial distress predictionproblem. The most widely used classification technique is statistical methods includingMDA, Logit, and Probit methods. The violation of the underlying Normality assumptionof independent variables causes the biased results. The decision tree, Logit, and Probitmethods have been used as alternative statistical methods. However, they also requiresamples satisfied with di?erent kinds of statistical assumptions which limit the usefulnessof their application.Firstly, According to the concept of financial distress, this paper selects 60 com-panies in financial distress and 60 contemporaneous companies in regular status as thestudy samples. Based on the domestic and foreign current research models, the test fornormalized distribution, the nonparametric test for mean value di?erence and Spearmannonparametric analysis of correlation of the regular financial indexes and cash ?ow finan-cial datas of the t-3 year are separately putted up, this paper chooses 14 financial indexeswith stronger distinguished ability out of the 28 as input variables of the model. Secondly,in view of the limitation of the traditional statistical method we establish the model forpredicting financial distress using BP neural network. Finally, aiming at the shortcom- ings of BP neural network, financial predicting model of BP neural network optimized byParticle Swarm is applied. It showed that the 14 index variables had a stronger timelycharacteristic to predict the Special Treatment of Listed Companies three years earlierthan it happened, the two predicting models we mentioned above have obtained 80% ac-curate rate and 85% accurate rate to the examination samples. The demonstration resultsshow that both models can predict very well, the prediction ability of model of BP neuralnetwork optimized by Particle Swarm is superior than BP neural network model. So themodel of BP neural network optimized by Particle Swarm this paper mentioned is fit forresolving distinguishable and prediction problem of the corporation financial distress, andit has preferable application prospect and application value in the aspect of analyzing andstudying company's financial distress prediction.