Comparison of Financial Distress Prediction Models Accuracy and its Effect on Earnings Management Tools

The main purpose of this paper is to investigate financial distress prediction models accuracy and earnings management approaches. Thus, primarily model was selected by comparing financial distress prediction models and its relation was analised through earnings management tools. In order to predict financial distress the comparison of machine learning and statistical models were considered for 312 listed companies at the Tehran Stock Exchange (TSE) during 2006 to 2015 and the result determined by comparing mean test shows that machine learning models can predict financial distress more accuracy than statistical models. Then, the relation between the best model resulted from previous section and earnings management tools was investigated by multiple linear regressions and the result shows that relation between financial distress prediction and operating cash flows earnings management was negative and significant and this relation with earnings management for manufacturing costs and accrual items was positive and significant.