Research On Financial Early Warning Model Of Listed Companies On GEM With Neural Network

Since October 2009,the growth enterprises market(GEM)has been run more than eight years.During the 8-year period,the controversy of the high P/E and stock price has not ever been discontinuous.Some listed companies in the growth enterprise market have greater volatility result in the doubt about their profitability.Presently,the study of financial early warning model of listed companies on GEM is far from comprehensive and mature.This paper selects annual report of the GEM listed companies from 2010-2016.Early-warning indexes from the two sides of finance and non-finance have been chosen to construct financial prewarning model with neural network.Firstly,this paper introduces GEM listed company financial early warning research background,significance,research status at home and abroad,innovation place insufficiency.Then it introduces the financial early warning theory,especially the qualitative and quantitative forecast method.This paper conducts a statistic analysis of GEM status and features.An analysis on the main factors influencing the financial risks of listed companies on GEM is made in this article.Based on the researches of other scholars,this paper defines the first year of loss as the sign of financial risk.72 risk samples and 1438 risk-free samples are filtered out.48 risk samples and 959 resk-free samples are chosen as the training samples in the neural network.24 risk samples and 479 resk-free samples are chosen as the verifying samples in the neural network.This paper selects 18 preliminary warning indexes from profitability,solvency,capacity development,operating capabilities,cash flow and equity structure of companies.Six factors as the final warning indexes are determined through significance analysis and factor analysis by SPSS.The BP neural network and RBF neural network are established from the training samples by MATLAB.The accuracy of the models are verified with the verifying data.The result of the test indicates that the precision of BP neural network model is 88.87%and RBF neural network model is 80.12%.The empirical result shows that the neural network has high accuracy in financial early-warning of listed companies on GEM.BP neural network is better than RBF neural network.