Financial monitoring effectiveness in Kazakhstan’s bank investment operations: A mixed-methods evaluation

  • 13 Views
  • 1 Downloads

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

Type of the article: Research Article

Abstract
Investment activities of banks are a key driver in financial sector development, yet their effectiveness largely depends on the quality of financial monitoring, which can detect, diagnose, and correct anti-money laundering and countering the financing of terrorism (AML/CFT) weaknesses. This article aims to assess the effectiveness of financial monitoring in the investment operations of Kazakhstani banks and to identify transaction-level risk indicators that can support data-driven AML/CFT supervision. The study employs a mixed methods design that combines analysis of national AML/CFT legislation and supervisory guidance with semi-structured expert interviews and case studies of three major institutions (Halyk Bank, Kaspi Bank, and ForteBank). In the quantitative component, a synthetic dataset of 1,000 investment-related transactions, calibrated to 2019–2024 statistics from the National Bank of Kazakhstan and the Committee for Financial Monitoring, is analyzed using logistic, multilevel, and Bayesian logistic regression with cross-validation and bootstrapped confidence intervals. The models achieve high predictive performance (AUC up to 0.93, test accuracy up to 93.2%, sensitivity 82.0%, specificity 94.1%) and show that higher transaction amounts, cross-border origin, SWIFT channel use, and investment-linked operations increase the odds of being flagged as suspicious by factors of roughly 1.5-3.5. Qualitative evidence reveals uneven digitalization, fragmented data integration, and capacity gaps, especially in mid-sized banks, which limit the practical implementation of these risk-sensitive tools. The results justify targeted regulatory support for advanced analytics and provide a replicable framework for strengthening investment-related financial monitoring in Kazakhstan and comparable emerging markets.

view full abstract hide full abstract
    • Table 1. Key variables used in the modeling process
    • Table 2. Software packages
    • Table 3. Model variables and evaluation metrics
    • Table 4. Case study analysis
    • Table 5. Thematic insights on challenges and developments in financial crime monitoring
    • Table 6. Variables and distribution targets
    • Table 7. Logistic and multilevel logistic regression estimates for suspicious transaction prediction
    • Table 8. Model fit statistics
    • Table 9. Logistic and multilevel logistic regression results: suspicious transaction classification in Kazakhstani banks
    • Table 10. Model performance & diagnostics
    • Table 11. Bayesian model
    • Conceptualization
      Sholpan Abzhalelova, Anar Kurmanalina
    • Methodology
      Sholpan Abzhalelova, Olga Tyan
    • Visualization
      Sholpan Abzhalelova
    • Writing – original draft
      Sholpan Abzhalelova, Elvira Ruziyeva, Gaukhar Kalkabayeva
    • Formal Analysis
      Elvira Ruziyeva
    • Project administration
      Elvira Ruziyeva
    • Validation
      Elvira Ruziyeva
    • Data curation
      Gaukhar Kalkabayeva
    • Investigation
      Gaukhar Kalkabayeva
    • Supervision
      Gaukhar Kalkabayeva, Anar Kurmanalina
    • Funding acquisition
      Olga Tyan
    • Software
      Olga Tyan
    • Writing – review & editing
      Olga Tyan, Anar Kurmanalina
    • Resources
      Anar Kurmanalina