Predicting Qualified Audit Opinions Using Financial Ratios: Evidence from the Istanbul Stock Exchange
Alpaslan Yaşar, Emre Yakut, M. Murat Gutnu

Abstract
The purpose of this study is to predict qualified audit opinions by using discriminant, logit and C5.0 decision tree based on twelve financial ratios. The sample consists of 110 firm-years data that includes 55 qualified opinions firm-year observations and 55 unqualified opinions firm-year observations listed in the industry index of Istanbul Stock Exchange (ISE) for the period 2010-2013. The results show that the variables X10 (retained earnings to total assets) is the significantly most effective variables to identify audit opinions by all of the models used in the study. Other significant variables in the analysis are found to be X6 (equity to total liabilities), X5 (total liabilities to total assets), X12 (net income to equity), X9 (net income to total assets), X3 (Working capital to total assets), X7 (net sales to total assets). The classification results of the models indicate that C5.0 algorithm of decision tree has the greatest classification accuracy rate for explaining unqulified and qualified opinions of the firms, compared to discriminanat and logit models.

Full Text: PDF