Bits About Money: Fraud Investigation Is Believing Your Lying Eyes

Patrick McKenzie of Bits About Money delves into the intricate world of fraud investigation, particularly highlighting the stark differences in approach between the financial industry and government social programs. He argues that political sensitivities and an unwillingness to confront uncomfortable truths often enable large-scale fraud to thrive.

• The Minnesota Case Study: The article centers on the decade-long, industrial-scale fraud against Minnesota's social programs, especially the daycare reimbursement scheme, where state auditors found over 50% of reimbursements to be fraudulent. Despite clear evidence like empty daycare centers, official action was slow and politically muddled, often dismissed with specious accusations of racism.
• Financial Industry vs. Government: McKenzie contrasts the financial sector's pragmatic, P&L-driven approach to fraud (where the "optimal amount of fraud is not zero," but it must be managed aggressively) with the government's often reactive, politically constrained, and less effective methods.
• Common Fraud Signals: He outlines several patterns common to large-scale frauds:
  • Iterated Games: Fraudsters are repeat offenders; tracking them is crucial.
  • Fraud Supply Chains: Frauds are often structured businesses with specialized infrastructure and shared service providers, which can be profiled.
  • Ethnic Clustering: Fraud rings frequently form within existing community trust networks (affinity fraud), which defenders must acknowledge without succumbing to prejudice.
  • High Growth Rates: Fraudulent enterprises often exhibit implausible growth rates compared to legitimate businesses, acting as a red flag.
  • Weak Links: Fraudsters gravitate towards financial institutions or programs with lax controls.
  • Suborned Identities: Frauds often rely on recruiting individuals to use their identities, sometimes openly advertised on social media.
• Asymmetric Burdens of Proof: Government investigations often bear an overwhelming burden of proof, requiring extensive, costly investigations to refute minimal fraudulent evidence, unlike the financial industry which can more easily cut off suspected bad actors.
• Machine Learning's Role: ML can provide a powerful, adaptive tool for early fraud detection, allowing for more nuanced interventions and potentially reducing application burdens for legitimate beneficiaries.
• Fraud Lifecycle: Frauds follow a predictable lifecycle from creation to detection and resurrection, emphasizing the need to track individuals and their business associations across different schemes.

Ultimately, McKenzie asserts that responsible civil actors must aggressively detect fraud to prevent demagogues from exploiting the issue, stressing that the public will always "believe its lying eyes" if institutional players refuse to.