Detecting earnings management in bank merger targets

The purpose of this study is to test whether merged banks have previously engaged in earnings management using the Modified Jones model and an industry-specific model. Comparisons were made between the two models to determine which model is the better model to detect earnings management. The research differs from prior research in two ways: (1) the Modified Jones model had not been used in a one-industry environment such as financial institutions and (2) an industry specific model was compared to the Modified Jones model.; A sample of merged or acquired banks from the Standard Industrial Classification code 602, state and national commercial banks, is used to test the two models. I hypothesize that banks manipulate earnings in the pre-merger year and/or the year prior to the pre-merger. In addition, I expect to find that the industry-specific model is a stronger model than the Modified Jones model, in detecting earnings management.; A firm-specific multivariate time-series estimation is used to test the hypotheses. A dummy variable (PART) in the model separates the event period from the estimation period. A Z-statistic is used to test the significance of earnings management.; Results based on both models indicate that banks managed earnings in the year prior to the pre merger year while banks did not manage earnings in the pre-merger year. A test of year -3 also concludes that banks managed earnings based on both models. Results of split sample indicate that state banks drive the results. Therefore, state banks manage earnings in the year prior to the pre merger year and year -3 while national banks do not manage earnings in either year.; One of the main conclusions is that the Modified Jones model can be used to detect earnings management in a one-industry environment such as financial institutions, specifically, merged banks. In addition, the results are mixed on which model is the better model in detecting earnings management. The results also confirm along with other studies such as Key (1997) and Cahan et al. (1997) that industry-specific models can detect earnings management in one-industry samples. Moreover, evidence indicates that the Modified Jones model is not misspecified because both models reach the same conclusions in each year tested.