Research On Financial Distress Early-warning For SMES Based On Artificial Intelligence Methods

On December 11,2016,China has witnessed its fastest-ever economic growth during the 15-year period since its entry into the WTO.In the wave of the economic development of 15 years,China has become the world's second-largest economy community.Enterprise is the lifeblood of the national economic development,China's 13'th Five Year Plan has taken the first step,in response to economy maintained rapid growth in the planning objectives,the role of small and medium-sized enterprises can not be ignored.In recent years,the state introduced a series of polices to encourage the development of small and medium-sized enterprises.Under the environment of policy supporting,the number of small and medium-sized enterprises in China is sustainable growing.At the same time,the small and medium-sized enterprises in China are facing with a series of problems such as fierce market competition,insufficient financial risk managing ability and imperfect corporate governance structure.The occurrence of international financial crisis brings huge impact on Chinese overall economy,and lead to a large number of small and medium-sized enterprise bankruptcy.Obviously,the lack of risk prevention awareness as well as effective financial crisis early warning mechanism are the important factors for common phenomenon of the small and medium-sized enterprise bankruptcy.In fact,the formation of small and medium-sized enterprise financial crisis is not only affected by the financial factors,managers' professional skills and disorder of the corporate governance factors cannot be ignored too.Market operation risk is still significant in post financial crisis era,how small and medium-sized enterprises survive in the complex environment to resist risk,nip in the bud,become the common proposition of the academia and practice.Based on this,this paper revolves the small and medium-sized enterprise financial crisis warning research as the theme,takes 2005-2015,small and medium listed enterprises as the research samples,the integrated use of smalland medium-sized enterprise financial and non-financial data,build small and medium-sized enterprise financial early warning system of artificial intelligence.Help small and medium-sized enterprises find the exception existing in the operation and management activities early,and response to market risk in time.This paper is proposed to solve the four key problems as follows: first,the dilemma sample selection;Second,the choice of the paired samples;Third,the screening of early warning indicators;Fourth,the construction of early warning model.According to the above four key problems,bared on artificial intelligence methods,we build the small and medium-sized enterprise financial early warning system.And the 10—fold cross validation method is adopted to improve the robustness test,the main research results and innovation are summarized as follows:1.Considering the quantity of ST small and medium-sized board listed companies is less,so when we select the dilemma samples,we not only consider the ST and ST* company,but also take the enterprises which facing the annual loss only once for two consecutive years,but the average quarterly net profit margin less than zero in two years into consideration.2.When select the matching samples,this paper considers two piece standards,namely "the same industry,similar assets" and "the same industry,similar number of employees".Empirical study results show that the "assets" rather than "employees" can effectively measure scale,as the proxy variable of enterprise scale.3.Considering small and medium-sized enterprises own characteristics,in the early warning index system,this paper not only considering the financial indicators,including the corporate governance index at the same time.Research results show that corporate governance variables can be integrated into the index system,which play a positive role in promoting the model prediction.4.We use the rough set approach to select features,then constructing C4.5,Bayesnet,RF and ANN artificial intelligence models.And use the 10—fold cross validation as a robustness test criterion to evaluate the early warning results.To test the validity of artificial intelligence methods,this paper also established the Logistic model based on rough set method,the empirical results show that rough set method can effectively select the features,and artificial intelligence early warning methods achieve higher accuracy.Overall,this article considers the financial index and corporate governance index,based on the artificial intelligence method to build the effective financial crisis early warning model of small and medium-sized enterprises.The results is of great importance for the controlling of small and medium-sized enterprises operational risk,and promoting the sustainable development of small and medium-sized enterprises.