Explaining the adoption of machine learning-based financial fraud detection in Arab Gulf family firms

  • 20 Views
  • 2 Downloads

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License

Type of the article: Research Article

Abstract
Financial fraud in Arab Gulf family firms threatens key stakeholders, yet many still rely on informal, trust-based controls. The purpose of this study is to explain why some family firms in this region adopt machine learning-based financial fraud detection systems, while others remain locked into legacy mechanisms of control. Results from a cross-sectional survey of 416 owners, executives, finance, and compliance officers from family businesses in Saudi Arabia, the United Arab Emirates, and Qatar were analyzed using variance-based structural equation modeling and multi-group analysis. The results show that perceived effectiveness (β = 0.08, p < 0.01) and regulatory support (β = 0.07, p < 0.05) are the strongest positive drivers of the likelihood of adopting machine learning-based fraud detection. Awareness of machine learning, ethical concern, and existing fraud detection practices also have significant positive effects (β ≈ 0.03-0.05, p < 0.05), while barriers to adoption exert a significant negative influence (β = −0.05, p < 0.05). The structural model explains 52.7% of the variance in the likelihood of adoption and 46.8% of the variance in perceived effectiveness. Indirect effects indicate that awareness and regulation promote adoption through perceived effectiveness, whereas barriers reduce adoption through heightened ethical concern. The findings suggest that stronger governance, clearer and incentive-aligned regulation, and explainable, well-governed machine learning implementations are essential to shift family businesses in the Arab Gulf region from trust-based to data-driven fraud detection.

view full abstract hide full abstract
    • Table 1. Demographic characteristics of the respondents
    • Table 2. Descriptive analysis
    • Table 3. Reliability analysis
    • Table 4. Convergent and discriminant validity assessment
    • Table 5. Multicollinearity check (Variance inflation factor – VIF)
    • Table 6. Path coefficient significance testing
    • Table 7. Effect size (f²) analysis
    • Table 8. Model fit indices
    • Table 9. Invariance testing – Consistency of relationships across groups
    • Table 10. Multi-group analysis (MGA) – Group comparisons
    • Table 11. Mediating effects testing
    • Table 12. Moderation analysis
    • Conceptualization
      Amer Morshed, Hanadi A. Salhab
    • Data curation
      Amer Morshed
    • Formal Analysis
      Amer Morshed, Hanadi A. Salhab
    • Investigation
      Amer Morshed, Hanadi A. Salhab
    • Methodology
      Amer Morshed, Hanadi A. Salhab
    • Project administration
      Amer Morshed
    • Resources
      Amer Morshed
    • Software
      Amer Morshed, Hanadi A. Salhab
    • Supervision
      Amer Morshed
    • Validation
      Amer Morshed, Hanadi A. Salhab
    • Visualization
      Amer Morshed
    • Writing – original draft
      Amer Morshed
    • Funding acquisition
      Hanadi A. Salhab
    • Writing – review & editing
      Hanadi A. Salhab
Related Articles

  • Machine learning for robo-advisors: testing for neurons specialization

    Roman SemkoORCID Researcher ID doi: http://dx.doi.org/10.21511/imfi.16(4).2019.18
    Investment Management and Financial Innovations Volume 16, 2019         Issue #4         pp. 205-214 Views: 3708 Downloads: 841 TO CITE АНОТАЦІЯ

    The rise of robo-advisor wealth management services, which constitute a key element of fintech revolution, unveils the question whether they can dominate human-based advice, namely how to address the client’s behavioral biases in an automated way. One approach to it would be the application of machine learning tools during client profiling. However, trained neural network is often considered as a black box, which may raise concerns from the customers and regulators in terms of model validity, transparency, and related risks. In order to address these issues and shed more light on how neurons work, especially to figure out how they perform computation at intermediate layers, this paper visualizes and estimates the neurons’ sensitivity to different input parameters. Before it, the comprehensive review of the most popular optimization algorithms is presented and based on them respective data set is generated to train convolutional neural network. It was found that selected hidden units to some extent are not only specializing in the reaction to such features as, for example, risk, return or risk-aversion level but also they are learning more complex concepts like Sharpe ratio. These findings should help to understand robo-advisor mechanics deeper, which finally will provide more room to improve and significantly innovate the automated wealth management process and make it more transparent.

  • An application of Altman Z-score model to analyze the bankruptcy risk: cases of multidisciplinary enterprises in Vietnam

    Diep Thanh Tung , Vo Thi Hoang Phung doi: http://dx.doi.org/10.21511/imfi.16(4).2019.16
    Investment Management and Financial Innovations Volume 16, 2019         Issue #4         pp. 181-191 Views: 3455 Downloads: 3153 TO CITE АНОТАЦІЯ

    This study applied Altman Z-score model to assess the bankruptcy risk of a set of multidisciplinary enterprises of various types, mainly small and medium enterprises, with data taken from official financial reports of 180 enterprises in Soc Trang province. The binary logistic regression was employed to assess the impact of non-financial and financial factors on the bankruptcy risk of enterprises. The research findings showed that both the non-financial factors such as business area, types and size of the business, the educational level of managers and executors and other characteristics, and the financial factors (indicators) such as earnings before tax, net profit/equity ratio, earnings before interest and tax/total assets ratio, equity/total debt ratio, affect the bankruptcy risk of enterprises. Predicting the bankruptcy risk and measuring its determinants play an important role not only as an effective managing tool of the business, but also as evidence for policymakers to support the sustainable development of business.

  • Using the Beneish M-score model: Evidence from non-financial companies listed on the Warsaw Stock Exchange

    Artur HołdaORCID Researcher ID doi: http://dx.doi.org/10.21511/imfi.17(4).2020.33
    Investment Management and Financial Innovations Volume 17, 2020         Issue #4         pp. 389-401 Views: 3040 Downloads: 2288 TO CITE АНОТАЦІЯ

    The risk of distortion of financial statements has been growing. Following the 2008 crisis, recipients of financial information are increasingly focusing on the likelihood of financial statements being distorted through fraudulent presentation of financial information. Therefore, scientific research pays more attention to models capable of detecting financial statement manipulation.
    The paper aims to present the principles of functioning and the possibility of using the Beneish M-score model in Polish realities. It analyzes the history of more than 30 companies listed on the Warsaw Stock Exchange to select those whose history indicates that they can be classified as manipulators, and to select the same number of companies from the control group that are considered as non-manipulators.
    The research method involves the analysis of empirical data on companies listed on the Warsaw Stock Exchange.
    The analysis showed the 8-factor Beneish model identified manipulators with 100% accuracy and succeeded in identifying non-manipulators. The effectiveness of the 5-factor model was much lower.
    To serve the purpose of the study, the effectiveness of the Beneish model was tested on a small sample of Polish listed companies as an introduction to a planned larger scale research. The results obtained are consistent with the results of numerous studies by authors from various countries and confirm the effectiveness of the Beneish model in detecting financial statement manipulation.

    Acknowledgment
    The publication is sponsored by funds from the Cracow University of Economics for the maintenance and development of research potential.

Download Preview
Close
Investment Management and Financial Innovations
Investment Management and Financial Innovations