The Research On Financial Crisis Warning Based On Earnings Management

Earnings management behavior would make the financial information distorted. Many companies will conduct relatively large number of earnings management behavior driven by the financial crisis motive, so from this point of view, the indicators of earnings management can predict the financial crisis. The indicators of earning management can make certain contribution to the traditional financial model and improve the effectiveness of the original model's prediction by adding variables of measuring earnings management. We choose discretionary accrual (|DA|) as ameasure of earnings management indicators which is added to the traditional financial model constructed in the financial indicators to observe its improving effectiveness.This study used the frequency distribution method putted forth by Burgstahler &Dichev(1997) to analysis the earnings management phenomenon of the non-listed companies and observe the obvious "zero-surplus" earnings management phenomenon. For the listed companies, this study used the Healy(1985) and DeAngelo(1986)'model to calculate the discretionary accruals in different years. The statistical analysis shows that the financial crisis companies conducted relatively large number of earnings management behavior before the crisis. In this study we still use the stylebook comprised by ST companies and normal companies to build the five financial crisis warning models. Model 1's regression result shows the probability of financial crisis occurred has a positive correlation with the indicator of measuring the earnings management behavior. Model 2 and model 3's regression result show that adding the earnings management indicators can improve the explanation and prediction effectiveness of the model 2. To compare the indicators of earnings management with other non-financial indicators, we establish model 4 and model 5. The regression result shows the indicator of earnings management has a new improving effect but not very significant, indicating these improving effect contributed by the non-financial indicators due to the correlation between each other has much overlap effect.