The Predictive Audit Opinion Of Chinese Listed Corporate Based On Financial Indicators

Financial reports released by the listed companies are important basis for enterprises stakeholders to make decisions. In order to enhance its credibility, a "third person"–the CPA, who independent of listed companies and stakeholders is necessary. The financial reports of listed companies audited by the CPAs are more reliable, make the financial reports'users have more confidence, and reduce the decision-making errors. So the type of audit report of the list company is highly regarded by all kinds of outsiders, and may have important effects on their decision behaviors. Therefore it has strong theoretical and empirical implications to predict the type of audit report of the list company by some appropriate ways. The main purpose of this paper is to investigate the finance indicators which have affect on the audit opinion, and to establish audit opinion prediction models.This paper samples from 309 firms (Clean samples) which have received the standard qualified audit opinions and 162 firms (unclean samples) which receive the non-standard audit opinions. The firms are all manufacturing A-share listed companies.In the predictive models, the dependent variable y is a standard binary variables, when the company received a non-standard audit opinion, y take 1, else y take 0.We choose 16 financial indicators which can basically reflect the company's financial condition and operating performance as explanatory variables. They are current ratio, ;quick ratio, capital and liability ratio, cash flow to current liabilities ratio, net profit growth variables, total assets growth variables, the main business income growth variables, Total net assets ratio , net profit interests of the shareholders variable rate, inventory turnover rate, the main cash operating income ratio variables ,and ln(total assets ).First, we use descriptive statistics to compare financial indicators of companies that received standard audit opinion with that of companies which received non-standard audit opinion. We find that the former indicators'averages are obverse better than the latter's. The financial indicators of the standard qualified companies are different from that of standard unqualified companies significantly. It is feasible to predict audit opinion type with financial indicators.Second, we establish the Logistic model; these papers defines the financial variables reflecting the financial conditions and business performance of the listed company as the independent invariables, and choose the variables reflecting whether a listed company has received the standard unqualified opinion about its financial report as the dependent invariable. Logistic regression analysis required the independent variables. So before regression analysis, the correlations between variables are tested.We found there was a significant correlation between the variables and quick-moving ratio and liquidity ratio. We removed the liquidity ratio test again, finally adopted a total of 11 variables to linear test. In this paper, 11 variables entered the regression analysis, parameter estimation using the Forward conditions: conditional to report the test results to be the ultimate quick-moving ratio, the debt-to-asset ratio. Cash flow to current liabilities ratio, the main business income growth rate of total net assets. Ln (inventory) turnover rate and the total assets of the seven variables prediction model .We choose 0.3 as the best judge of points. We got clean sample predictive accuracy is 73.2%. Clean-sample predictive accuracy of is 81.3%, the prediction accuracy rate of overall is 76.1%.In the process of empirical analysis ,as the combination of variables model are random ,a single certain prediction model to make a judgment aren't accurate. In this paper we use a triple cross validation techniques Logistic models of clean samples company forecast accuracy. Unclean samples company forecasts accurate and the overall accuracy were 71.54%, 73.08% and 72.06%, we can see a better model prediction capabilities.Thirdly, we choose Fisher discriminate analysis to compared with the logistic model, and study the seven variables selected by Logistic regression as the sample data The clean sample for the accuracy of the model is 85.7% For non-clean samples accurately predict the rate of 68.8%, overall accuracy is 78.1%. Triple Cross forecasts accuracies are: 79.84%, 68.48% and 75.98% for the corresponding rate.As the integrated forecasting accuracy, Fisher model is better than the logistic model. But the unclean samples forecast accuracy of Fisher model is lower than that of the logistic model. .The proportion of the misjudgment of standard audit opinion is type I error, and the proportion of the misjudgment of Non-standard audit opinion is the Type II error. In general opinion, the investors and the CPA are risk aversion.If a company financial report should to provide non-standard audit opinion, but for some reasons, the CPA issued a standard audit opinion. The investors are doubtful of it but the Type II error occurred when using the model. It may cause investment losses. While Type I may be cause investors not to invest but it won't lead to the losses. In contrast, the type II error rate lower than the type I is below expectations of the results we want to obtain. Therefore, from the cost of misjudgment, the logistic model's predictive ability is superior to that of the Fisher model.The models developed in the paper have good predictive capability. It can serve as a decision aid for the CPA when issue an audit opinion, and as a defense in lawsuits. As for the investors, they can use the conclusion to judge the accuracy of the audit opinion when they are doubtful. The Regulators can use the model to supervise the audit quality.Finally, we discuss the research deficiencies and the future research directions.