Lay out the parts of a control framework aimed at managing the risks of financial fraud and of money laundering.
The idea
A control framework for financial crime has to deal with two related risks. Fraud is deception for gain, usually explained by the fraud triangle of pressure, opportunity and rationalisation. Controls mainly remove opportunity through segregation of duties, authorisation limits, mandatory leave, reconciliations and a whistleblowing channel. Money laundering moves criminal proceeds through placement, layering and integration. The anti-money-laundering programme rests on an enterprise risk assessment, customer identification and due diligence (enhanced for high-risk customers such as politically exposed persons), transaction monitoring and suspicious activity reporting. It also needs sanctions screening, record keeping, training, a named responsible officer and independent testing. Accountability follows three lines of defence: business owns the risk, compliance oversees it, internal audit tests it. Exam point: collusion defeats segregation of duties, and monitoring only works if alerts are investigated and reported on time.
Formulas
Precision of a transaction-monitoring rule
precision = suspicious alerts ÷ total alerts
- suspicious alerts
- alerts that prove genuinely suspicious
- total alerts
- all alerts the rule generates
Raising a threshold usually lifts precision but loses some genuine cases, so also measure the share of suspicious cases missed.
A practice question
Which sequence correctly lists the three classic stages of money laundering, in the order in which they occur?
- Layering, placement, integrationWhy not: Reverses the first two stages: illicit cash must first enter the financial system before it can be layered.
- Placement, layering, integration - correct
- Placement, integration, layeringWhy not: Puts integration before layering; proceeds are only reintroduced as apparently legitimate wealth after they have been obscured.
- Detection, layering, reportingWhy not: Mixes in compliance responses (detection, reporting) that are not stages of the laundering process itself.
Illicit cash is first placed into the financial system, then layered through transfers and transactions to obscure its origin, and finally integrated into the legitimate economy as apparently clean wealth.
A harder one
A bank onboards a customer who is a senior official of a foreign government. Under a risk-based anti-money-laundering framework, which treatment is most appropriate?
- Rate as high risk, check source of wealth, get senior approval and monitor more closely - correct
- Apply standard due diligence because the customer has no adverse media and no criminal recordWhy not: Ignores that status as a politically exposed person is itself a risk factor requiring enhanced due diligence.
- Decline the relationship automatically because politically exposed persons may never be customersWhy not: Treats a risk factor as a prohibition; the framework requires enhanced controls, not automatic refusal.
- Verify identity at onboarding only and rely on annual sampling thereafterWhy not: Drops ongoing monitoring; due diligence is continuous and its intensity should rise with customer risk.
Politically exposed persons carry higher corruption risk, so the risk-based approach calls for enhanced due diligence, source-of-wealth checks, senior approval and heightened ongoing monitoring rather than standard treatment or an automatic ban.
The hardest tier
A bank's transaction-monitoring rule currently generates 10,000 alerts a month, of which 150 are genuinely suspicious. Raising the threshold would cut alerts to 6,000 a month and retain 138 of the 150 genuinely suspicious cases. Which conclusion is correct?
- Alert precision rises from 1.5% to 2.3% and no suspicious cases are lost, because every one of the removed alerts is a false positiveWhy not: Assumes every removed alert is a false positive, ignoring the 12 genuinely suspicious cases that fall below the new threshold.
- Precision rises from 1.5% to 2.3%, but 8% of suspicious cases are lost, so below-the-line testing is needed - correct
- Alert precision falls from 1.5% to 1.38% because fewer of the genuinely suspicious cases are retained after the changeWhy not: Divides the retained suspicious cases by the old alert count (138 / 10,000) instead of the new count (138 / 6,000).
- Alert precision rises to 2.3%, and the 40% fall in alert volume equals the share of suspicious cases that are lostWhy not: Confuses the 40% reduction in alert volume with the miss rate, which is 12 / 150 = 8%.
Precision before = 150 / 10,000 = 1.5%; after = 138 / 6,000 = 2.3%. Missed suspicious cases = 12 / 150 = 8%. Efficiency gains come at a detection cost, which must be tested below the line and accepted explicitly by management.
In the app: 6 practice questions on this outcome, spaced review that brings it back before you forget it.
Start free →