Research On Finance Warning Of China’s Foreign Trade Enterprises Based On Neural Network Model

Today the entire world economy is in a big wave of globalization, the graduateexpanding of the market and the continuous deepening of the international division oflabor as well as scientific advances in technology have significantly contributed to theworldwide economic integration, more and more enterprises develop foreign tradebusiness. China’s import and export trade then presents the spectacle of a prosperityand development. However, in the rapid development of China’s foreign trade at thesame time as the main body of the foreign trade business-the foreign tradeenterprises are in more competitive environment of uncertainty compared withgeneral trade enterprises.Some companies will inevitably find themselves in financialdifficulties or financial crisis and ultimately affect survival. Foreign trade enterprisesare facing with both internal and external complex environment. Therefore it is verysignificant to researching financial crisis warning of the foreign trade enterprises.Based on review of the research of scholars, the paper has a comprehensiveanalysis of the financial risks of foreign trade enterprises combined with theiroperating characteristics, elaborate mainly from the macro and micro level,whichconstitutes the theoretical basis of the foreign trade enterprises in the financial earlywarning. Then it is a core part of this article that is to build a financial early-warningsystem for foreign trade enterprises. Firstly, it is the Financial Early Warning IndexSystem, micro-financial indicators selected to reflect the business situation andmacroeconomic indicators selected to reflect the environment of the enterprise in;then the paper uses SPSS18.0software to test for significant initial selected indexexcluding the swap is not significant indicators, to retain a significant indicator,during the main component analysis, in order to build the index system. Then thepaper uses SPSS18.0software and Matlab7.10software to establish a logisticregression model based on different index system and the BP neural network model,respectively horizontal and vertical comparative analysis of the validity of the indexsystem and early warning models; results show that the final empirical analysis basedon macro micro-index system of the BP neural network model of early-warningcapability, the accuracy rate of prediction samples of88.1%. Lastly, the paperdescribes the overall conclusion of this article, as well as the limitations and a simpleoutlook of the topics for possible future research directions.