Postdoctoral Fellow Geng Chunxiao from the School of Economics Publishes Academic Paper in Accounting Research

Recently, the academic paper Do False Accounting Data Affect the Identification Effectiveness of Financial Fraud Models? by Postdoctoral Fellow Geng Chunxiao from the School of Economics at Shandong University was published in Issue 5, 2025 of Accounting Research. Geng Chunxiao is the corresponding author, and the collaborators are Professor Wu Xi from Central University of Finance and Economics and doctoral student Fu Rong.

Financial fraud identification is an important topic for accounting supervision and the high-quality development of the capital market. Identification technologies are usually based on accounting data reported by enterprises, but accounting data reported by fraudulent enterprises have material misstatements. At present, no literature in the relevant research field has examined the impact of accounting data doped with fraud on the effect of financial fraud identification.

This paper systematically explores the impact of false accounting data on the identification effectiveness of financial fraud models. It understands the fraud behaviors of listed companies by reading administrative penalty announcements and accounting error correction announcements to restore the false accounting data of fraudulent companies to the pre-fraud level as much as possible. Meanwhile, according to the logic of identifying fraud based on accounting characteristic variables, it divides the variable design concepts in financial fraud models into two categories: those relying on pre-manipulation accounting data and those relying on post-manipulation accounting data, and theoretically deduces the enhancement and weakening effects of the identification effectiveness of the two types of variables after the restoration of false accounting data. Based on the differential impact of false accounting data restoration on variables in the model, this study also expects and finds that the impact of false accounting data on model identification effectiveness cannot be generalized. A "one-size-fits-all" restoration of false data for all accounting variables may not improve the identification effectiveness of fraud models. Targeted rather than "one-size-fits-all" restoration of false accounting data can more stably improve the identification effectiveness of financial fraud models.

This paper makes a methodological contribution to the literature in the field of financial fraud identification. Financial fraud models involve three elements: indicators, algorithms, and data. There are three paths to improve the identification effectiveness of financial fraud models: optimizing indicators, optimizing algorithms, and consolidating the data foundation for estimating models. Previous literature mainly focused on the first two optimization paths. This paper focuses on the third path for the first time, investigates the potential impact of the authenticity of the data foundation used by the model on the effectiveness of fraud identification models, and proposes coping strategies. The false accounting data restoration method proposed based on the underlying logic of accounting data is applicable to multiple identification indicators and algorithms, has future early warning functions, and good practical application value.

Geng Chunxiao is an "Outstanding Postdoctoral Fellow" at the School of Economics at Shandong University. His research fields include industrial organization structure, corporate finance, and capital markets. He has published many articles in domestic and foreign journals such as Economic Research Journal, Economic Research Journal (Quarterly), China Industrial Economics, and Journal of Banking & Finance.

Ass