Analyses On The Predictive Ability Of Corporate Governance To External Auditing Based On The Neural Network Method

Wang Shunian,Yu Bo

2010-01-01

Abstract:Using the classical multi-layer perception (MLP) model of neural network method, the authors conduct empirical analyses on the predictive ability of corporate governance to external auditing based on the neural network method by taking the A-shares of the listed companies of 2000-2007 years as research samples. The research shows that predictive ability of corporate governance indicators to external auditing can be estimated. Especially, the proportion of circulating shares and first shareholders' ownership are more important, while the proportion of state-owned shares is not important. Moreover, it is effective to use corporate governance indicators for predicting the auditing types and opinion changing. However, it is not effective to use the model to predict audit process.