A new tool known as abnormalest has been developed to help researchers more easily identify unusual financial patterns, a task that plays an important role in both accounting and social science research. By automating the estimation of unusual accruals, the tool, discussed in the International Journal of Data Analysis Techniques and Strategies, could overcome the inefficiencies of traditional methods, which are often time-consuming and error-prone.
Francesca Rossignoli and Nicola Tommasi of the University of Verona, Italy, explain that abnormal accruals refer to the discrepancies between what is reported in a financial statement and what the actual finances are. Such discrepancies can be indicative of manipulation, where managers alter financial reports for personal benefit. The conventional approach to estimating abnormal accruals is a complex process involving manual calculations, the selection of control samples, and the application of specific conditions to detect the problems. abnormalest has been developed to automate the key steps.
The tool can select appropriate control samples, carry out entropy balance pre-processing, and then use predictive models such as regression-based techniques to estimate abnormal accruals. This kind of automation is much faster than manual approaches and makes far fewer mistakes.
The abnormalest system provides a more detailed output than traditional methods. It includes valuable information such as the abnormal accrual measure, degrees of freedom, and explanations for any estimation failures. This is information that conventional approaches cannot easily provide.
Although originally designed with accounting research in mind, abnormalest might also be used in the social sciences. Its flexibility allows it to be used in a variety of contexts, from identifying fraud to assessing unusual business performance or examining behaviour that deviates from the norm.
The researchers have successfully tested the system using real-world financial datasets. They found that it performs better than existing models used in academic research.
Rossignoli, F. and Tommasi, N. (2025) ‘Abnormal accrual estimation: an automation data analysis technique’, Int. J. Data Analysis Techniques and Strategies, Vol. 17, No. 5, pp.1–18.