Reporting Just to Be Safe: Reducing Defensive Reporting in Anti-Money Laundering
This thesis examines defensive reporting at the Raiffeisen Group: why compliance officers file suspicious transaction reports to avoid sanctions rather than from substantiated suspicion, and how targeted training can reduce this without weakening compliance standards.
Topic
Banks must report suspicious transactions to the authorities, but a report is only useful if it is well founded. This thesis examines defensive reporting at the Raiffeisen Group: The tendency of compliance officers to file reports mainly to avoid personal or institutional sanctions rather than on the basis of substantiated suspicion. Focusing on the human decision rather than the technical monitoring system, it asks which factors drive this behavior and how a targeted training program could reduce it while keeping compliance standards fully intact.
Relevance
Suspicious-activity reports to Switzerland's Money Laundering Reporting Office have risen nearly fourfold since 2020, reaching 21,087 in 2025. When many reports are precautionary rather than substantiated, they lower the signal-to-noise ratio of financial intelligence and tie up scarce compliance and law-enforcement resources. For banks, understanding why officers over-report, and addressing it through training rather than additional rules, is a practical lever to improve report quality, use resources efficiently, and strengthen the overall effectiveness of the anti-money-laundering system.
Results
Across eleven expert interviews and nine internal documents, the study identifies eight interlocking drivers of defensive reporting. At the core, case-assessment uncertainty is resolved, under fear of personal and institutional consequences, through a perceived asymmetry: Over-reporting feels individually safer than under-reporting. Missing operational decision aids, defensive escalation routines, system-generated over-alerting, and the absence of regulatory feedback sustain the pattern, while the existing curriculum reinforces it. The pattern holds consistently across analysts, officers, and managers, including the counter-cases.
Implications for practitioners
- Train judgment, not the rule: Operationalize the "reasonable suspicion" threshold through case-based, gray-area scenario exercises rather than restating legal definitions.
- Give officers a concrete decision aid (e.g., a one-page decision algorithm) for borderline cases.
- Make a documented, well-reasoned non-report a visibly legitimate outcome, not only the report.
- Use structured peer calibration across functions to substitute for the missing regulatory feedback loop.
- Treat training as a complement to, not a replacement for, regulator-side reforms that address the underlying incentive asymmetry.
Methods
The thesis uses a qualitative single-case design at the Raiffeisen Group. Eleven semi-structured expert interviews with AML professionals across three role groups were combined with an analysis of nine internal compliance and training documents. The material was examined through structured qualitative content analysis (Mayring), combining theory-derived categories with categories emerging from the data. Three counter-cases were deliberately retained and analyzed to test the pattern. From the empirical drivers, the thesis derives a concrete five-module training concept, each module linked to a specific driver and a pedagogical rationale, together with an implementation plan and an evaluation concept.