Empirical Research Of Warning Model Based On Information Disclosure Violation Of Listed Companies

Information disclosure is the foundation for security market to achieve the function of resource allocation. In this paper, the author used information disclosure violation from 1996 for study, and analyzed empirically the market response to the information disclosure violation, the violation companies' industry, size and financial statement. Then it developed a warning model through logistic regression based on financial and governance statement to predict the information disclosure violation possibility.Empirical results showed statistically significant negative CAR around the information disclosure violation announcement. For all samples, CAR in [-2,2] was -0.36%. Divided by year, from 1996 to 2001, CAR in [-2,1] was -0.51%; from 2002 to 2004, CAR in [-1,2] was -0.51%; from 2005-2006, CAR in [-2,4] was -0.52%; from 2007-2011, CAR in [-2,0] was -0.38%. Divided by regulator, CAR of companies punished by CSRC in [-1,2] was -0.34%; CAR of companies punished by SSE was positive; CAR of companies punished by SZSE in [-2,2] was -1.16%.Statistical analysis of the violations showed that, from 2005, the number of information disclosure violation of the year became stable, and the proportion of illegal companies was reducing. In all 371 illegal companies, real estate industry, Miscellaneous industry, agriculture, forestry, animal husbandry and fishery industry had the highest proportion of information disclosure violation. It infers that multi-service company is more likely to violate the information disclosure rules. Illegal companies have generally poor profitability, weak solvency, low operating efficiency, high asset-liability ratio and slow size growth rate.Based on foreign and domestic scholars' study, the author chose financial indicators and corporate governance indicators as early-warning model alternative variables, and then picked out by logistic regression return on total assets, quick ratio, asset liability ratio, proportion of the largest shareholder, the number of the board of directors and audit opinion as early warning indicators of the model. Regression analysis showed that the model can distinguish 70% of violation companies and 76.5% of normal companies. Return on total assets and the largest proportion of the shareholding have negative correlation with the possibility of information disclosure violation, while asset-liability ratio, the number of the board of directors and non-standard audit opinion have positive correlation with the possibility of information disclosure violation.