Other Financial Services and Investment Specialists
This unit group covers Financial Services and Investment Specialists not elsewhere classified. It includes Credit Risk Managers.
Three calls before lunch
You are a credit risk manager at a lender, responsible for how safely the organisation lends to households and businesses. It is Thursday morning. The quarter ends soon, the sales team is chasing a target, and your risk reports have just landed. Several people need decisions from you before lunch.
Choose what you would do. Then see what happens. There are often good reasons for each choice.
We are close to our quarterly target. If you loosen the lending rule on small business loans just for this month, we'll hit it easily. It's a temporary tweak. Can you sign off on it?
Sign off, since the sales team knows the customers well and the change is only temporary, just for this one month
Loans written under a relaxed rule stay on the books long after the month ends. You have taken on risk the policy was designed to avoid, mostly to help a number look better.
Decline, explain which risks the rule protects against, and offer to review whether it is too tight
Dwayne is annoyed but gets a real offer. You agree to look at the data on whether the rule is too strict. Saying no with a way forward keeps the relationship intact.
Say no immediately because policy is policy, and end the conversation
You have followed the rule, but Dwayne feels shut out and starts to go around you. A flat no leaves the next problem to be solved without you in the room.
The model flagged a group of home loan customers as high risk this month, but I can't see why from the inputs. Do we act on it, or ignore it until we understand it?
Ignore it for now since nobody can explain it, and revisit when there is time
If the flag is real, you have let a problem grow. If it is a fault, you have let one sit. Either way, the loose end becomes harder to explain the longer it waits.
Act on it fully, contacting every flagged customer, because the model is usually reliable
Several customers get worrying calls for no reason, and some are upset. Trusting the tool completely is just as risky as ignoring it.
Investigate with Aisha before acting, and put a light watch on the group in the meantime
You trace the flag to a change in how one input is measured. Some customers really are at higher risk, many are not. You act on the real ones, and the model gets a fix.
A cafe owner who has been with us for years has missed two repayments. She says business is slow after roadworks outside her shop. Policy says we should escalate. Can you help me find a fair way through?
Waive the repayments quietly to keep her happy, and note it as an admin matter
It feels generous, but it is outside what you are allowed to do, and the record is untrue. If every case were handled this way, nobody would trust how the lender treats customers.
Escalate it by the book and let the process take its course, without speaking to her
The policy is followed, but the owner receives a formal letter before anyone has spoken with her. She feels dropped. Following the process still leaves room for a human conversation.
Follow the process but ask Ines to talk with her first and explore hardship options that policy allows
The owner is relieved to be asked what is happening. A temporary arrangement within policy is agreed, with notes on file. Fairness and rules can work together.
How it went
You end the day with no dramatic wins, but several quiet ones: a target not bought with extra risk, a flag understood and a customer treated fairly. Credit risk work is about protecting both the lender and the people it lends to. Does balancing those two sound interesting?
This is a simplified, fictional situation to help you explore. It is not professional instruction or a test of your ability.
Case study: The tool that said no to a whole region
A fictional case set in the credit risk team of a fictional lender that makes loans to households and small businesses.
The situation
Samuel was a credit risk manager at Wattlebird Finance. One Monday he looked at the weekly report from the lender's automated scoring tool and noticed that applications from small businesses in one regional area were being declined far more often than usual. Nothing in the local economy had changed. The sales team had not complained yet, but Samuel knew some businesses would be waiting on answers, and turning them away would cost both them and the lender.
What they did
Samuel did not assume the tool was right or wrong. He asked the data team to compare the declined applications with approved ones. They found that a postcode update had been loaded incorrectly, so the tool was reading these businesses as being in a higher risk area. Samuel paused the automatic declines for that region, had the affected applications reviewed by hand, and wrote up what had happened so the update process would get an extra check.
How it turned out
Most of the affected applications were reviewed fairly within a few days, and some were approved. A few were still declined for ordinary reasons. The lender added a check on every data update, and Samuel was asked to review the tool's results each month.
What would you do?
- Why is it risky to trust an automated tool without checking its results?
- Who is affected when a lending decision is wrong, and in what ways?
- How would you explain a complicated problem like this to a sales manager?