The Empirical Research On Financial Crisis Early Warning Of The Listed Companies In The Manufacture Industry Based On KMV Model

The Government recently proposed ’second five’ plan that requires changing the mode of China’s economic development and accelerating the pace of industrial upgrading. This policy will inevitably result in different industries flourished alternately, the manufacturing sector as the dominant force in national economic development will inevitably bear the brunt. This will inevitably result in different policy preferences in deferent manufacturing industry segments, so that the different segments of the industry flourished alternately.The enterprises in the policy supported industry will increase the investment Scale and the risks of the investment projects are rising. During to the shrinking of the market, production capacity reduction, the enterprises in the policy limited industry, will inevitably lead to the decline in business performance, this maybe result to a financial crisis. Based on this, this paper attempts to research financial crisis early warning of listed manufacturing companies in China, expects to make some contribution to China’s manufacturing financial crisis early warning. This has important practical significance for managers, investors, regulatory authorities and many other stakeholders.The main content is as follows:First, this paper analyses the conception of "financial crisis" from the domestic and foreign research, and give the criteria of definition of manufacturing listed companies "financial crisis", as well as crisis time.Second, combing through the existing research literature, we reviewed the classic financial crisis forecasting model, the Z-sore model, the F score model, the logistic regression model, the Probit model, KMV model,default distance and the lack of the model exists in our application.Third, we took a qualitative research of China’s manufacturing industry, manufacturing listed companies’ industry distribution and geographical distribution Fourth, the paper took the listed manufacturing companies in China as the research object, based on a sample screening criteria we selected54"ST" listed companies and51non-ST companies as paired samples in2009to2011. According to the theory of the KMV model, we used MATLAB software programming to calculate the default distance indicator of the sample companies.Furthermore, this paper innovatively proposed to adjust the default distance indicator. By K-S normality test, independent samples T-test, ANOVA analysis and building a single variable logistic regression analysis, we comparatively studied on the default distance indicator and adjusted default distance indicator when it applied to predict financial crisis.Fifth, combined with the financial indicators, the default distance indicator and the adjusted distance default indicator we constructed two financial crisis prediction model based on logistic regression analysis method. From the significance of the variables, the goodness of fit, the forecasting accuracy and the payoff matrix four aspects we comparatively analyzed the two models.Through this study, we obtained the following conclusions:First, the paper Defined "ST" as the definition Standard of financial crisis. This paper used the samples’annual report financial indicators in year t-2and the stocks’transaction data in year t-2to construct model, in order to predict whether the financial crises will happen in year t.Second, from the qualitative research of manufacturing status quo we found that China’s manufacturing enterprises are at the lower end of the value chain; from the geographical distribution of listed manufacturing companies, there is combination effect in industry distribution.Third, by K-S normality test, independent samples T-test, it showed that the KMV model’s output indicators-the default distance director and the adjusted default distance indicator used in the listed companies’financial crisis early warning were feasible. According to contrast the goodness of fit and the forecasting accuracy of the two single variable logistic regression models, we thought that the adjusted default distance indicator is better using to the financial crisis early warning.Fourth, by K-S normality test, independent samples Man-Whitney test, Spearman correlation coefficient analysis, we got the "current ratio","Return on total assets","cash flow to current liabilities ratio","Net profit in cash flow","total asset turnover","The growth rate of main business" a total of six financial indicators to reflect the accounting information in the financial early-warning model.Fifth, combined the above-mentioned six financial indicators, respectively, with the default distance indicator, adjusted default distance indicator we built two financial crisis early-warning models based on logistic regression analysis method. According to contrast the significance of variables, the goodness of fit, the forecasting accuracy, and the payoff matrix of the two models, we got the following conclusions:The significance of variables:in the logistic regression model building on the default distance and financial indicators, the Wald value of default distance is0.755,corresponding to P=0.385; and in the adjusted default distance Logistic regression model, the Wald value of adjusted default distance is2.709,corresponding to P=0.1. The contrast showed that the default distance indicator in the regression model performance is not significant, while the adjusted default distance indicator performed much better, predictive ability is stronger.The goodness of fit:in the logistic regression model building on the default distance and financial indicators, the-2log-likelihood value of this model is66.3571, Nagelkerke R Square is0.755; and in the adjusted default distance Logistic regression model,-21og-likelihood value of this model is64.1823, Nagelkerke R Square is0.7188. The contrast can be seen, the goodness of fit of the logistic regression model based on the adjusted default distance is superior to the latter, it’s better to explain the variation.The forecasting accuracy:the two models’overall prediction accuracy rates are the same, are88.6%, but two models have different type Ⅰ error and type Ⅱ error.The payoff matrix:it can be seen form the two model predictions, the logistic regression model based on the default distance tends to commit Type Ⅰ error, the error of "abandoning the truth"; while the adjusted default distance regression model tends to commit Type Ⅱ error. According to our analysis, when the overall predictive abilities are the same, committing Type II error will cause much fewer losses than committing Type Ⅰ error, whether for company managers, investors or regulatory authorities.Therefore, we thought that the adjusted default distance indicator, combined with financial indicators constructed logistic regression models to predict the financial crisis, can reduce the loss of decision-making, forecast stronger, better adaptability.