The Effectiveness of Flagging Systems in Detecting Fraud among Registered Micro-Finance Institutions in Lusaka, Zambia

Abstract

This study investigated the effectiveness of flagging systems in detecting fraud among registered microfinance institutions in Lusaka. Microfinance institutions play a critical role in promoting financial inclusion; however, persistent fraud and high non-performing loans suggest weaknesses in existing detection mechanisms. The study aimed to determine the standard flagging systems used, assess their effectiveness, and examine the challenges affecting their performance. A pragmatic research philosophy was adopted, utilising a mixed research approach. A cross-sectional survey design was employed, with primary data collected from 108 employees across two microfinance institutions using structured questionnaires and semi structured interviews. Quantitative data were analysed using SPSS through descriptive and inferential statistics, while qualitative data were analysed thematically to provide contextual insights. The findings revealed that institutions relied on a hybrid system combining integrated core banking modules, manual processes, and limited advanced technologies. Flagging systems contributed to fraud detection, although effectiveness was generally moderate, with inconsistencies in detection speed, accuracy, and fraud reduction outcomes. The regression analysis further showed that the model was statistically significant (F = 5.232, p = 0.000) and explained 23.70 percent of the variation in fraud detection effectiveness. Reliance on automated digital flagging systems had a significant positive effect on effectiveness (B = 0.126, p = 0.004), while technical system limitations had a significant negative effect (B = -0.131, p = 0.004). Formal training also had a significant positive influence (B = 0.085, p = 0.048), while other variables such as system interaction frequency, perceived training insufficiency, and policy constraints showed weaker or insignificant effects. Key challenges identified included system limitations, poor data quality, insufficient staff training, false alerts, and institutional constraints. The results indicated that while flagging systems improved fraud detection, their effectiveness was strongly shaped by technology quality, automation, and staff capacity. The study concluded that although flagging systems enhance fraud detection, their effectiveness is constrained by technological and operational factors. The study recommends system integration, adoption of advanced technologies, regular system updates, improved data management, and continuous staff training to strengthen fraud detection and institutional resilience.