LONGITUDINAL ANALYSIS OF PUBLIC BENEFITS FRAUD EVOLUTION IN MINNESOTA'S SOMALI COMMUNITY: A TIME-SERIES EXAMINATION OF CRIMINAL ADAPTATION, POLICY INTERVENTIONS, AND DETECTION PATTERNS (2018-2025)

This longitudinal quantitative study examined the evolution of public benefits fraud in Minnesota's Somali community from 2018 to 2025, analyzing 89 defendants charged with $364.7 million in alleged fraud across multiple federal assistance programs. The research employed interrupted time-series analysis (ITSA) and ARIMA modeling to investigate temporal patterns, criminal adaptation mechanisms, policy intervention effectiveness, and detection patterns using publicly available federal court records, Department of Justice prosecutorial data, and state program expenditure reports. The theoretical framework integrated Fraud Triangle theory, Routine Activity Theory, and Criminal Adaptation Theory to examine how fraud schemes emerged, escalated, and evolved in response to regulatory environments and enforcement pressures. The findings revealed a four-phase evolutionary model: Emergence (2018-2019), Escalation (2020-2021), Detection (2022-2023), and Consolidation (2024-2025). Three primary fraud programs were identified: Feeding Our Future ($247.8 million), Early Intensive Developmental and Behavioral Intervention autism services ($58.3 million), and Housing Stabilization Services ($43.1 million). COVID-19 regulatory waivers implemented in March 2020 served as a critical inflection point, with program expenditures increasing 3,847% within 18 months. Time-series analysis demonstrated significant structural breaks at the COVID-19 waiver implementation (p < .001, Cohen's d = 2.14) and FBI investigation initiation (p < .001, Cohen's d = 1.94). The mean detection lag was 22.4 months (SD = 8.7), with prosecution resulting in a 71.8% conviction rate and mean sentence of 58.4 months. Evidence of criminal adaptation was observed through scheme sophistication increases, network decentralization, and asset concealment evolution. Whistleblower reports proved critical for detection, with 78% of cases initiated through tips rather than algorithmic detection. These findings contribute to theoretical understanding of white-collar crime dynamics, provide empirical support for Routine Activity Theory's capable guardian hypothesis, and offer actionable recommendations for fraud prevention policy during emergency regulatory periods. Keywords: public benefits fraud, longitudinal analysis, Somali community, Minnesota, COVID-19, Feeding Our Future, criminal adaptation