Research On The Financial Crisis Early-warning Model Of Textile And Garment Enterprises Based On Evidence Theory

In the fierce domestic and foreign market competition,China's textile and garment industry are facing many serious problems such as the gradual weakening of labor cost advantages,increase in production costs,and continuous growth in inventory.Enterprises may be caught into financial crisis at any time.Establishing an effective early-warning model can enable managers to discover financial crisis in advance and formulate appropriate treatment measures as soon as possible,which is conducive to the long-term healthy development of listed companies.However,the construction method of the model is still in continuous development.This article takes textile and garment industry enterprises as research objects and builds a model based on evidence theory.The results of the early-warning models established using Logistic regression and support vector machines are integrated in order to bring forward new ideas for the study of the early-warning model of financial crisis.Through literature and theoretical research,enterprises which were specially dealt with are defined financial crisis enterprises.Then taking the status of the textile and garment industry into consideration,the internal and external causes of the financial crisis were analyzed to select early-warning indicators effectively.Next,through the comparison and analysis of several models,Logistic and support vector machines were selected for evidence synthesis,which is expected to improve the accuracy of model prediction.In the empirical design part,for the textile and garment industry,10 ST companies and 40 non-ST companies were selected as research samples for modeling,5 ST companies and 12 non-ST companies were selected as test samples to test the model's accuracy.According to the pertinence principle,comparability principle,scientific principle,system principle and availability principle,48 financial indicators and 8 non-financial indicators were initially selected.Seven public factors which are extracted through the screening of financial indicators together with three non-financial indicators that passed the significance test constitute the financial early-warning index system.According to the empirical test,the classification accuracy rates of the two individual models for the research samples were 96% and 98% respectively,and the accuracy rates for the test samples were 94.12% and 88.24% respectively.The accuracy rates of the combined early-warning model based on evidence theory for research samples and test samples were as high as 98% and 100%.Compared with the two individual models,the overall false positive rate is reduced from 4.48% to1.49%,which shows that evidence theory can improve the accuracy.Applying theevidence theory to the financial crisis early-warning field has effectiveness and feasibility.