Research Of Multi-stage Financial Distress Early Warning On China's Real Estate Listed Companies

As the pillar industry of national economy,the real estate industry involves dozens of industries during the process of production and consumption,which requires extremely large amount of capital and goods.Therefore the development of real estate has a significant impact on the whole national economy.However,according to the financial security assessment report of listed companies in 2014 autumn,the overall financial state of real estate listed companies is serious,and the FSI of the real estate ranked last among all industries while it becomes the industry most at risk.Given the situation,it is necessary to establish an early warning model especially for the real estate listed companies of our country,so that the company can in time discover the crisis and take measures to prevent and control it.Firstly,the literature of concept,method about financial distress and multi-stage financial crisis warning is reviewed.On the basis of literature review,the concept of financial distress is redefined.The financial state of an enterprise is divided into three stages:normal,mild financial crisis and severe financial crisis.Secondly,the causes of financial distress are analyzed from internal and external factors,with the consideration of features of real estate companies and fundamental theories.After that,characteristic variables of real estate listed companies are proposed.Finally,the two-stage early warning models and multi-stage early warning models are established,which is to prove new models are superior to traditional ones.The two types of models are both based on the theory of sliding time window,which consist of the sample data of 2009-2013.Indicators are filtered through the normality test and difference test,and then are extracted principal components by factor analysis.The models are established under the theory of binary logistic regression model and ordinal logistic regression model.Main conclusions are as follows:(1)Most indicators pass the significance test,which implies that there is a significant difference among normal stage,mild crisis stage and severe crisis stage.They reflect the enterprise's solvency,profitability,operation ability,developing competence,the ownership structure and the features of the real estate.(2)As the traditional division lacks accuracy,confusing the information of companies in different situations together,it can't show the difference among different financial stages.So there are more indicators that passed the significance test under the data environment of multi-stage than that of two-stage.Furthermore,the quantity of indicators which pass the test is also different among data blocks from time windows,especially apparent in two-stage models.It shows that financial distress concept drift does exist,and the sliding time window can continuously update the sample set and delete the old historical information.(3)From the aspect of fitting,the R2 of multi-stage models is higher than that of two-stage models,which implies that data are better fit the multi-stage models than two-stage models.The forecast accuracy of multi-stage models is higher than that of two-stage models,and also the probability of type I error of multi-stage models is generally lower.Therefore,compared with two-stage model,multi-stage model can better forecast the occurrence of financial crisis,with lower misclassification cost.