Abstract:

Present literatures on corporate financial early warning are mainly based on financial index to build model to forest financial situation, so it is difficult to further explain the financial difficulties and hardly be applied to forecast the early financial distress. Based on the analysis of financial index, the paper introduces corporate governance, EVA, etc. non-financial indicators to build the financial early warning index system of listed companies after selecting the candidate index with statistical analysis method; by building PSO - SVM prediction model to find the optimal combination of parameters automatically with PSO algorithm, the paper tests validation of PSO-SVM prediction model based on non- financial index. Research results show the prediction accuracy of this model has been significantly improved compared with other methods, which can provide theoretical basis for the financial early warning of listed companies in China.

Key words:

non-financial measures, financial early warning, listed company, Support Vector Machine, Particle Swarm Optimization