- Advanced
- About 16 min
- 5 practice questions
The risk-based approach in depth
Work through an invented risk-scoring model, apply override rules, and judge how risk ratings drive due diligence, monitoring and updation.
If something looks wrong, please tell us.
At the advanced level, questions about risk categorisation move beyond 'which customer is high risk?' to how a rating is actually built and what follows from it. Banks combine several risk factors into a rating, and they add override rules for situations that must always be treated as high risk. This lesson uses an invented scoring model to practise that reasoning. The scores and bands here are made up for learning; real banks design their own models under their regulator's rules.
Why a score, and why overrides
A score turns many judgements into one consistent rating, so that two officers looking at the same customer reach the same result. But a pure score can hide a single serious problem behind several harmless factors. That is why models add overrides: rules that put a customer straight into the high category whatever the total, for example where the customer is a politically exposed person or where the beneficial owner cannot be identified. Overrides work only in one direction, upward; a good score never cancels a serious red flag.
The invented model used in this lesson
| Factor | Score 1 | Score 2 | Score 3 |
|---|---|---|---|
| Customer type | Salaried individual | Small local business | Entity with layered ownership |
| Geography | Local | Other states | Links to places with weak controls |
| Product/channel | Basic deposit account | Standard current account | Products moving funds fast or across borders |
| Activity | Predictable, documented | Some variation, explained | Cash-heavy or unexplained |
Suppose the total of the four scores places the customer in a band: 4 to 6 low, 7 to 9 medium, 10 to 12 high. Suppose also one override: any politically exposed person is rated high regardless of the total. All of this is invented for practice.
Add the factor scores to get the band, then apply overrides: an override can only raise a rating, never lower it.
Worked example 1: two customers
Customer A is a salaried individual (1), local (1), with a standard current account (2) and predictable activity (1). Customer B is an entity with layered ownership (3), links to places with weak controls (3), using a product that moves funds across borders (3), with some explained variation in activity (2). Rate both. Working: A's total is 1 + 1 + 2 + 1 = 5, in the 4 to 6 band, so low. B's total is 3 + 3 + 3 + 2 = 11, in the 10 to 12 band, so high. Neither is a politically exposed person, so no override applies. Answer: A low, B high.
Worked example 2: the override
Customer C is a former senior public official (a politically exposed person) with a salaried profile (1), local (1), basic deposit account (1) and predictable activity (2). Rate C. Working: the total is 1 + 1 + 1 + 2 = 5, which by score alone is low. But the override for politically exposed persons applies and lifts the rating to high. Answer: high, because of the override; the low score does not cancel it.
What the rating drives
Once a rating is set, three things follow from it. The depth of due diligence: simplified checks may be allowed for low ratings, enhanced checks are applied for high ratings. The intensity of monitoring: high-rated accounts are watched more closely, with tighter alert rules. The frequency of periodic updation: higher ratings are refreshed more often. A rating is also not permanent. A trigger event, such as a sudden change in activity, an adverse news report or a change of ownership, should prompt a fresh rating at once, without waiting for the next scheduled updation.
Advanced questions often test the direction of these links. If a statement says that a customer moved from high to low should now be updated more often, it reverses the rule. If it says that a strong score lets a bank ignore an override, it misunderstands overrides. And if it says that the rating is fixed at account opening, it ignores trigger events.
Key takeaways: combine factor scores into a band, then apply overrides that only raise the rating; the rating sets the depth of due diligence, the intensity of monitoring and the frequency of updation; and trigger events demand an immediate re-rating.
Check your understanding
-
1. Using the lesson's invented model, a customer scores 2, 3, 1 and 2 on the four factors and is not a politically exposed person. What is the rating?
- Low
- High
- Medium
- Cannot be rated without an override
Show answer
Answer: C. Total = 2 + 3 + 1 + 2 = 8, which is in the 7 to 9 band, so medium.
-
2. Using the invented model, a customer scores 3, 3, 3 and 2. What is the rating?
- Low
- Medium
- Low, because one factor scored only 2
- High
Show answer
Answer: D. Total = 3 + 3 + 3 + 2 = 11, in the 10 to 12 band, so high.
-
3. Using the invented model, a politically exposed person scores 1, 1, 2 and 1. What is the rating?
- Low, because the total is 5
- Medium, as the average of low and high
- High, because the override applies
- Low, because overrides apply only to companies
Show answer
Answer: C. The total is 5, which alone would be low, but the override rates every politically exposed person high.
-
4. Which statement about override rules is correct?
- They lower the rating of customers with good scores
- They replace the need for any scoring
- They can raise a rating but never lower it
- They apply only at account opening
Show answer
Answer: C. Overrides exist so that one serious factor cannot be hidden by a good total; they push ratings upward only.
-
5. A low-rated customer's ownership suddenly changes and an adverse news report appears. What should the bank do?
- Wait for the next scheduled periodic updation
- Re-rate the customer at once as a trigger event
- Keep the rating because ratings are fixed at opening
- Lower the rating because the customer is in the news
Show answer
Answer: B. Trigger events such as an ownership change or adverse news call for an immediate re-rating.
Sign in to keep track of the lessons you finish.