CFP Risk Assessment & Underwriting 3 — Questions and Answers
Question 1: Under the FCRA, when a fintech lender uses an alternative data model to deny credit, what is the applicant's right?
- No rights, as alternative data is not regulated
- Right to receive an adverse action notice listing the key factors (Correct answer)
- Right to immediate credit approval on appeal
- Right to request the lender's full model code
Correct answer: Right to receive an adverse action notice listing the key factors
FCRA and Regulation B require adverse action notices that disclose the principal reasons credit was denied, even when AI models are used.
Question 2: A fintech insurer prices cyber liability policies using real-time network vulnerability scans. This is BEST described as:
- Actuarial table pricing
- Dynamic risk-based pricing (Correct answer)
- Community rating
- Flat-rate underwriting
Correct answer: Dynamic risk-based pricing
Dynamic risk-based pricing adjusts premiums in real time based on current risk signals rather than static actuarial tables.
Question 3: Which metric BEST measures the financial cost of misclassifying a defaulting borrower as creditworthy in an underwriting model?
- False negative rate (Correct answer)
- False positive rate
- Precision
- Specificity
Correct answer: False negative rate
A false negative (bad borrower approved) represents the cost of missed default risk, which is the primary credit loss concern in underwriting.
Question 4: In insurance underwriting, what does 'moral hazard' specifically refer to in the context of fintech-enabled parametric policies?
- The risk that the insured takes on more risk because they are protected (Correct answer)
- The risk that data feeds trigger incorrect payouts
- The risk of technology failure in claims processing
- The risk of adverse selection in online enrollment
Correct answer: The risk that the insured takes on more risk because they are protected
Moral hazard occurs when insurance coverage reduces the insured's incentive to avoid the insured risk, a persistent concern even in automated parametric products.
Question 5: A BNPL provider assesses creditworthiness at checkout in under 2 seconds. Which approach makes this speed possible while maintaining risk accuracy?
- Manual underwriting by a credit officer
- Pre-scored decisioning models using cached bureau data and behavioral signals (Correct answer)
- Full credit report pulls for every transaction
- Requiring collateral for all purchases
Correct answer: Pre-scored decisioning models using cached bureau data and behavioral signals
Pre-scored models with cached data and real-time behavioral inputs enable millisecond decisioning without sacrificing predictive accuracy.
Question 6: Which concentration risk scenario would be MOST concerning for a fintech marketplace lender's loan portfolio?
- Loans spread evenly across 50 states
- 80% of loans originated to gig economy workers in a single platform (Correct answer)
- Loans diversified across 10 industry sectors
- Mixed prime and near-prime credit quality borrowers
Correct answer: 80% of loans originated to gig economy workers in a single platform
Heavy concentration in a single borrower segment and platform creates correlated default risk if that platform suffers disruption or sector downturn.
Question 7: What is the key advantage of using a scorecard model over a black-box neural network in consumer credit underwriting for a US lender?
- Higher predictive accuracy in all cases
- Greater interpretability supporting adverse action reason codes (Correct answer)
- Faster processing speed
- Lower data requirements
Correct answer: Greater interpretability supporting adverse action reason codes
Scorecard models produce interpretable point contributions per variable, making it straightforward to generate FCRA-compliant adverse action reason codes.
CFP Risk Assessment & Underwriting 3
Under the FCRA, when a fintech lender uses an alternative data model to deny credit, what is the applicant's right?