The Parameters Of The Listed Companies In China Financial Distress Prediction Model Studies

When a company fall into financial distress, not only economic loss will be caused forinvestors, creditors and other stakeholders, but also social stability be endangered. Asa result, it is significantly important to find the reasons contributing to companies’financial distress and to building prediction models that fit Chinese listed companies.Logistic regression is a popular model in previous researches. However, its impliedassumption that regression coefficients are fixed may contradict with the reality andrequired to be tested. Also, the effects of industrial attributes of the company are ofteningored or avoided in the prediction model. Thus, it is neccessary to includeinsdustrial information into the model and test its explainary ability to the variance ofthe regression coefficients. Consequently, this thesis chooses Multilevel LogisticRegression model to use. This model is sutable to discrete response variable, and isable to deal with multilevel data and test the effects of macrolevel variables as well.Firstly, this thesis conduct theoretical investigation to the financial distress issue forlisted companies, and on the basis of which provides the liquidation ratio-baseddefinition of financial distress. Second, the thesis selects the research sample asChinese A-shares listed companied in2010, and selects both financial andnon-financial factors as prediction variables. Third, Multilevel Logistic Regressionmodel is built, random and fixed effects of regression coefficients are tested. The datais extracted for CSMAR database.The results of Multilevel Logistic prediction model proves the existence ofinter-group variance of microlevel coefficients. The variance can be disintegrated intotwo parts, one is explained by macrolevel variables, the other one is by the randomintercept of micro level coefficients. In the meanwhile, industial prospect, amacrolevel factor, is shown to be influential in enhancing the risk of financial distressalong with microlevel factors such as liquidation ratio, liquidating debt ratio, return ofcapital, liquidation turnover rate and audit attitude. In addition, the effect of return ofcapital is random, while others’ are fixed. In conclusion, Multilevel Logistic modelfits the data well and is suitable to the prediction issue of listed company’s financialdistress.