Specification and power of cross-sectional abnormal working capital accruals models in the Spanish context

Following previous research (e.g. Dechow et al., 1995; Peasnell et al., 2000), we use a simulation procedure to evaluate the specification and power of different working capital abnormal accruals models in the Spanish context. Specifically, we assess the Jones Standard model, the Modified Jones model, the Kang and Sivaramakrishnan model, the Margin model, the Jones Cash Flow model, and the Accounting Process model. Consistent with previous research, our results suggest that all the models analysed are well specified in a random sample of firms. Regarding the models' ability to detect artificially induced earnings management, the Jones Cash Flow and the Accounting Process models produce the most powerful tests in all three types of manipulation studied. In line with Peasnell et al. (2000), we find that the Margin model is more powerful than the Jones Standard and the Modified Jones models at detecting non-bad debt expense manipulation, while these two models perform better than the Margin model for revenue and bad debt manipulation. Our results also support the finding by Dechow et al. (1995) that their modified version of the Jones model is more powerful than its standard version at detecting revenue-based manipulation.