Debt Capital Markets interview preparation

Bond mechanics, duration, credit spreads, ratings, primary issuance, syndicated loans, structured credit, covenants and liability management, plus the Indian debt market. Every question is either traced to a named firm from a public candidate report, or tagged at desk level when we could not trace it.

Jump to the question bank

Question bank

100 questions, mapped to the firms that asked them

Questions: 100
Traced to a firm: 45
Firms: 26
Updated: September 2026

Showing 1–1 of 1 · filtered from 100Clear filters

075How would you approach building a delinquency model?Structured creditHardcase studyNeuberger BermanRisk · Chicago · 2024
Say this
I would build it as a roll-rate transition model on vintage cohorts, then overlay macro sensitivity. Group loans by origination vintage and current bucket — current, 30, 60, 90 plus, charge-off — estimate the monthly transition probabilities from history, and project forward. Then stress the transition matrix against unemployment rather than adding a flat haircut.
Then walk it
1. Vintage cohorts are the critical design choice. Delinquency depends heavily on seasoning — losses peak 18 to 30 months after origination for consumer loans — so a portfolio-level rate mixes young and mature cohorts and tells you nothing. Vintage curves separate credit quality from portfolio growth.
2. Roll rates: for each bucket, the probability of moving to the next bucket, staying, or curing back. Estimate from history by segment, because a prime and subprime cohort have completely different cure rates. The 30-to-60 roll rate is usually the most informative early indicator.
3. Then charge-off and recovery: a loan that reaches 90-plus rolls to charge-off with some probability and lag, and you apply a recovery assumption net of collection costs and time. Loss equals default rate times loss given default, and both need to be modelled, not assumed.
4. Macro overlay: regress historical roll rates on unemployment, real income and, for secured pools, collateral values. Then run scenarios. This is what regulators require for CECL and IFRS 9 expected credit loss, so the framework is standard rather than exotic.
5. Validation is where the marks are: back-test on a holdout period, check stability of the transition matrix over time, and watch for the growth illusion — a rapidly growing book has an artificially low delinquency rate because the denominator is full of young loans. That single effect has masked deterioration in countless portfolios.
6. The honest limitations: roll rates are unstable through structural breaks, and the 2020 payment deferral programmes broke every consumer model because delinquency artificially vanished. So I would report the model output alongside the vintage curves themselves, because the raw curves are harder to fool than the projection.
Where candidates lose it
Proposing a single portfolio-level delinquency rate or a regression on borrower characteristics alone. The two things that make this a real answer are vintage cohorts, which control for seasoning, and roll-rate transitions. And naming the growth illusion — a fast-growing book looks clean — is what shows you have seen this go wrong.
Expect next
- Why does a growing book understate delinquency?
- How would you handle the 2020 deferral distortion?
- Which roll rate is the best early warning?
Reported by candidates at Neuberger Berman (Risk, Chicago, 2024). Source: Wall Street Oasis.

Firm tags come from public, anonymous candidate reports on Wall Street Oasis: strong signal, not sworn testimony. Firms are named as the places a question was reported, not as partners of Fin Maverick. Answers are written for this page to show how to think out loud; they are not scripts to recite.

100 questions, mapped to the firms that asked them

Say this

Then walk it

Where candidates lose it

Expect next