Research On The Construction Of Financial Crisis Early Warning Model Based On Logistic Regression Analysis

With the rapid development of market economy in our country,information technology(IT)industry is developing rapidly.The IT industrial structure will have a new change traditional IT industry continuously integrating with new technologies,new ideas,and new business models.With the concept of Internet plus,big data,cloud computing,Internet of things being put forward,IT industry will continue rapid development.However,it has an important problem that how to make companies operate well especially IT companies in the rapidly changing market environment.As a result of IT companies' own characteristics(high risk,high investment,and high return),they have high probability of financial crisis compared to other industry companies.How to avoid companies falling into financial crisis?How to make companies effectively prevent falling into financial crisis?These problems become into urgent problems.Therefore,it has the great practical significance to make a financial crisis warning model construction(FCWMC)for IT industry companies and the development of economic environment.Based on the previous research results by reading related essays,the research of FCWMC shows diversified tread including statistical prediction methods and non-statistical prediction methods.The statistical prediction methods include single-index model,multivariate linear model(Z-score model and F-score model),multiple Logistic regression model(Logistic model),and probability ratio model(Probit model).Non-statistical prediction methods include neural network model,genetic algorithm,SVM model and etc.From the view of current research status,most of researchers make FCWMC using the data of whole industries or manufacturing Industry,however,few scholars pay attention to IT companies' own characteristics in our country.So based in IT industry the essay adds to non-financial indicators(proportion technicians,R&D intensity,the number of patents applications)to measure the innovation capacity of companies in order to improve index system,make FCWMC and improve predict accuracy.The essay selects 70 listed IT companies from Shanghai and Shenzhen Stock Exchange as samples including 35 ST companies and 35 normal financial companies.Sample data are these listed companies' annual financial statements data for the previous third year(T-3 year).The essay selects 20 financial indicators(solvency,operating capacity,profit ability and development capacity)and 10 non-financial indicators(corporate government ability,audit opinion and innovation capacity).Data are analyzed by SPSS20.0 and Excel 2007.There are 18 indicators screened out by test of normality,two independent samples T-test and Mann-Whitney U test.Because of too many indicators,it will use the way of factor analysis to reduce indicators' dimensions and eliminate multicollinearity between indicators.Test statistics of KMO is 0.666 and the probability P is 0.000.It indicates that the correlation between variables and the data are suitable for factor analysis.It has extracted 6 factors by factor analysis and made them as independent variable building binary Logistic regression model.The model prediction correct accuracy is 92.9%and error rate accuracy is 7.1%by putting the sample back to financial crisis warning model.In order to verify the effectiveness of FCWMC,the essay randomly selects 30 listed companies in Shanghai and Shenzhen Stock Exchange as the predicted sample for the year of 2014 and 2015.The research results:correct accuracy is 76.7%and error rate accuracy is 23.3%.It shows that the model construction effect is well and has high predict accuracy.By empirical analysis of the information technology industry financial crisis early warning and combined with the results of empirical research,the essay puts forward some suggestions to prevent companies falling into financial crisis.Firstly,managers should improve enterprises'core competitiveness-innovation capacity.Secondly,establishing a financial crisis early warning defense system in order to achieve the effects of advance prevention,in-process control and remedy afterwards and reduce the probability of financial crisis in an affordable range.Thirdly,strengthen corporate government ability,the board of directors and supervisors play their roles in the company,take incentive and restraint mechanisms to the enterprise managers,improve the internal control system and corporate governance efficiency,the company will be developed better and better.The innovation of essay is that making FCWMC for IT industry.Combined with IT industry character the model was made by factor analysis and binary Logistic regression added to non-financial indicators(corporate government ability,audit opinion and innovation capacity).Then author tested the accuracy of FCWMC.The research shows that model prediction correct accuracy is 92.9%and error rate accuracy is 7.1%.The addition of non-financial indicators improved the prediction accuracy of the model,and the prediction result was ideal,which provided some reference for financial crisis early warning.