Page 22 - Azerbaijan State University of Economics

Nina Poyda-Nosyk, Serhii Lehenchuk, Victoriia Makarovych, Iryna Polishchuk, Tetiana Zavalii: Analytical
Procedures in Audit As A Tool For Predicting The Risks Of Financial Statement Fraud In Marketing Companies

Recent advancements in digitalization and AI-driven analytics have further expanded
methodological tools. Machine learning (Achakzai & Peng, 2023), artificial neural
networks (Omar et al., 2017), and data mining techniques (Ravisankar et al., 2010)
now enable more sophisticated detection of non-linear relationships and hidden fraud
signals. Despite these innovations, the Beneish and Roxas models remain widely
adopted in auditing practice due to their interpretability and empirical robustness.
Nowadays, the Beneish and Roxas models are among the most widely employed by
researchers to enhance analytical procedures in auditing, particularly for assessing the
reliability of financial reporting indicators across various industries and countries.
Thus, Repousis (2016) studied the activities of 25468 Greek companies for 2011-2012
and found that 33 percent of the sample had a signal that companies are likely to be
manipulators. Erdoğan and Erdoğan (2020) applied the Beneish model to companies
listed on Borsa İstanbul-50 (BIST-50) for 2015-2017, and found a positive
relationship between the probability of manipulating financial information and the
regressors of the model, AQI and SGAI.
Lehenchuk et al. (2021), using the Beneish and Roxas models, studied the activities
of 30 leading Ukrainian corporations for the period 2017-2018. Their findings
confirmed the reliability of financial statements for 10 corporations, while potential
manipulations were identified in 11 cases. Sankar and Bhanawat (2024) analyzed the
financial statements of Indian corporations for the period 2011-2016 and found that
the Days’ Sales in Receivables Index (DSRI), Total Accruals to Total Assets (TATA)
and Sales Growth Index (SGI) indicators were effective in identifying companies
involved in financial statement manipulations.
Ozkan and Alfarhan (2025) analyzed 9,766 non-financial firms across G7 countries
over the period 2006-2022, identifying earnings manipulators using the Beneish M-
Score Model. The authors also conducted cross-country analyses, which revealed
common manipulative practices prevalent among firms in G7 countries. In addition to
large-scale empirical studies, researchers also apply the Beneish M-Score Model in
case study analyses focused on individual companies. Thus, Ramírez-Orellana et al.
(2017) found, using the Beneish model, detected tendencies towards fraud and
earnings management in the Spanish company Pescanova, specifically through the
manipulation of the DSRI and TATA indicators. Similarly, Hariri, Pradana and
Widjajanti (2017) identified potential financial manipulations in the Indonesian
company XYZ, PT during the years 2010, 2012, and 2013.

22