Credit Analysis Interview Questions: One Company, Priced Two Ways

A credit interview, on a trading desk or in a research seat, comes down to one underlying test: can a candidate take a company's own numbers, turn them into a leverage ratio and a coverage ratio, and say out loud whether they would own the bond that comes out the other end. The covenant vocabulary, the capital structure diagram and the recovery estimate are all scaffolding around that one judgment.

This page works that judgment on one invented company, with the arithmetic shown twice so nothing rests on a formula taken on faith. It also pulls the actual grid that turns those numbers into a letter rating straight from S&P Global Ratings' and Fitch Ratings' own published methodology, not a description of either from memory.

Then it works the harder version of the same question: two bonds from that one company, priced differently, and two independent routes to deciding which one is worth owning.

The question families, at a glance

The trading seat and the research seat are not the same interview

Both seats are tested on the same underlying arithmetic, and both expect a candidate to have an opinion, but the shape of the interview differs in a way worth naming before anything else. A leveraged finance or private credit interview leans on a specific, framework-driven question: walk through net debt to EBITDA, explain what happens to coverage if rates rise two hundred basis points, or say whether you would lend to a specific, cyclical borrower at six times leverage.

Credit trading

• Speed and a defensible market view under time pressure
• The output is a filled order or a passed-on trade
• Deep product knowledge is a precondition, not the whole job

Credit research

• A written or presented recommendation, argued in depth
• The output is a call: own it, avoid it, or size it smaller
• The same leverage and coverage arithmetic, defended out loud

A research-track interview often ends with a genuine case study: read a company's financials, key ratios and capital structure over a few hours, then present a recommendation with a best case and a worst case, weighted toward the downside since a bondholder's upside is capped at par. A trading-track interview is more likely to stay conversational, testing whether the same judgment can be produced quickly and defended under a follow-up question.

Investment grade and high yield debt are separated by one line: anything rated below the Baa or BBB threshold is high yield. Distressed is a further, informal step down, and the desk convention many dealers use is debt trading around one thousand basis points over the equivalent Treasury, a market convention that moves with spreads rather than a fixed regulatory line.

Walk me through a credit, on one invented company

Every worked number below belongs to Harbor Point Industries, a company built for this page only. No real business, ticker or bond is used anywhere in it.

5.0x

leverage

net debt to EBITDA

2.86x

coverage

EBITDA to cash interest

$185m

free cash flow

after interest, capex and cash taxes

"Walk me through a credit" is really asking for this table read out loud, in order: what the company earns before financing costs, what it owes, what it pays to service that debt, and what is left over once it has.

Leverage and coverage, computed two ways

Total debt is the three tranches added together, $1,200 million plus $700 million plus $700 million, which is $2,600 million. Net debt subtracts the $100 million of cash, leaving $2,500 million. Leverage and coverage each check out two independent ways from there.

Both routes land on the identical number each time, which is the whole point of showing two: a candidate who reaches the same figure by a second, independent path is not just guessing right once.

A stricter version of coverage, EBIT divided by interest rather than EBITDA divided by interest, gives $380 million divided by $175 million, or 2.17 times, meaningfully lower once real depreciation is subtracted rather than added back. Free cash flow, EBITDA minus interest minus capex minus cash taxes, is $500 million minus $175 million minus $100 million minus $40 million, or $185 million.

The debt itself is not spread evenly across time either. $1,200 million comes due in 2029, another $700 million in 2030, and the last $700 million in 2031: a concentrated wall rather than a smooth ladder, and worth naming as a separate concern from the leverage number itself.

How the rating actually gets built from the numbers

S&P Global Ratings publishes the actual grid that turns a business assessment and a leverage assessment into a rating, and the mechanics are more specific than "strong companies get good ratings." A company is scored on a business risk profile, from excellent down to vulnerable, and separately on a financial risk profile built from leverage and cash flow ratios, from minimal up to highly leveraged. The two scores are combined on a published table to produce an initial anchor rating.

S&P's own worked example inside that methodology states it plainly: a company judged excellent on business risk can still only reach a BB-plus anchor if its debt sits at eight times EBITDA or more with no offsetting factors. Leverage caps the outcome regardless of how good the underlying business is. None of this runs on character, capacity or collateral; it runs on this exact grid.

The same methodology publishes the actual leverage and coverage bands behind each financial risk category, for an industry of ordinary volatility.

Harbor Point Industries sits right at the edge of two categories at once. Its 5.0 times leverage and 2.86 times coverage both land inside "Aggressive." But its funds from operations, a rough measure of cash generation approximated here as EBITDA minus cash interest minus cash taxes, is $285 million against $2,500 million of debt, or 11.4% of debt: just under the "Aggressive" floor of 12% and into "Highly leveraged" territory instead.

S&P's own rule for exactly this situation is to pick whichever ratio better predicts the company's future leverage, not to average the two or default to the weaker one automatically.

Moody's and Fitch each publish their own scorecards, built from a similar but not identical set of ratios and weights, so the same company can land in a slightly different place on each agency's own grid. That is a reason to read the actual methodology a question references rather than assume every agency runs the identical formula.

A downside case: EBITDA falls twenty percent

A common version of this question hands a candidate the same company and asks what happens if EBITDA falls twenty percent. Take Harbor Point Industries down from $500 million to $400 million, holding interest, debt and cash fixed since all three tranches carry fixed coupons rather than a floating rate.

That asymmetry is the actual point of the question. A twenty percent hit to earnings barely dents coverage, because interest is a fixed dollar amount either way, but it pushes leverage a full category weaker, because EBITDA sits in the denominator of that ratio specifically. A strong answer names both moves and explains why they are not the same size.

S&P's own criteria build a version of this stress test directly into the rating. A company judged "volatile" on its cash flow and leverage measures is expected to see its ratios move one or two categories worse in a downturn, which the criteria describe as typically equivalent to about a thirty percent decline in EBITDA, and the final assessment gets knocked down a full category to reflect that risk before anything has actually gone wrong.

A twenty percent EBITDA decline, in other words, is a realistic stress case rather than a worst-case invention, and a rating already prices in the possibility of seeing one.

The same stress case is also where a covenant question and a leverage question meet. If Harbor Point Industries had a maintenance covenant capping leverage at, say, 5.5 times, the base case at 5.0 times has room to spare, but the twenty percent downside case at 6.25 times breaches it outright. Naming that breach, not just the new leverage number, is what separates a complete downside answer from a partial one.

Where a bond sits in the capital structure

Fitch Ratings' own published criteria set out how much an instrument's rating moves off the company's own issuer default rating, purely based on where that instrument sits in the capital structure. For a company already rated below investment grade, secured debt can be notched up as much as two full levels above the issuer's own rating, while subordinated debt is notched down instead.

Repaid from the top down. Each band is sized by its share of the structure.

Term Loan B, secured, due 202946% · paid first, secured

Senior Secured Notes, 6.50%, due 2030← Bond A, priced below27% · secured, junior only to the term loan

Senior Unsecured Notes, 8.00%, due 203127% · unsecured, behind both secured tranches

Repaid from the top down. The two secured tranches stand ahead of the unsecured notes for any collateral, which is the whole reason the two bonds below do not trade at the same spread.

Two bonds issued by the same company, with the same underlying risk of the company defaulting at all, can still carry different ratings and trade at different prices, purely because one has a claim on specific collateral ahead of the other. That distinction is the entire subject of the relative value question two sections below.

The covenant question, in plain English

A leveraged loan's maintenance covenants are tested on a fixed schedule, typically every quarter, whether or not the borrower has done anything unusual. A high yield bond's covenants work differently: a leading law firm's own high yield covenant handbook states plainly that high yield covenants are incurrence tests, checked only when the issuer or a subsidiary actually does something, such as paying a dividend, taking on new debt or granting a lien on an asset.

That difference exists for a structural reason, not just a looser negotiating stance:

• A bank credit agreement is comparatively easy and cheap to amend with lender consent, so its maintenance covenants can afford to be strict and checked every quarter.
• A high yield indenture is expensive and slow to amend, usually requiring a formal process to reach bondholders who are widely dispersed and rarely paying close attention, so its incurrence covenants are built once, loosely, to last the seven to ten years the bond is typically outstanding.

The covenant package itself only reaches a defined "credit group," the issuer and its restricted subsidiaries. A company can designate a subsidiary "unrestricted," which removes most of that subsidiary's cash and assets from the covenants entirely, and a strong answer to a covenant question knows to ask which subsidiaries are actually inside the credit group before assuming a covenant protects anything.

The single most negotiated covenant in a high yield package is the restricted payments covenant, which limits how much cash can leave the credit group as dividends, stock buybacks or early repayment of subordinated debt. It typically carries three kinds of room to work within:

• A "builder basket" that grows the company's capacity to make restricted payments over time, usually tied to retained earnings.
• Separately negotiated baskets the company can use even without having built capacity under the builder basket.
• A list of investments the covenant always permits regardless of either basket.

A candidate who can explain why bond covenants are looser than loan covenants, rather than just naming the difference, is answering the question the interviewer actually wants answered.

The recovery question

Recovery is not a vague guess about how much comes back after a default. Fitch's own published recovery framework assigns a named band to each estimate, from RR1, "outstanding," at ninety-one to one hundred percent recovered, down to RR6, "poor," at zero to ten percent, with each band tied to a specific number of notches added to or subtracted from an instrument's rating.

A short recovery answer names the mechanism rather than a number pulled from nowhere: recovery in a default depends on the value of the company relative to its total debt, and on where a specific instrument sits in the queue for that value. The full mechanics of estimating recovery in an actual distressed capital structure, including how to find the specific class where the value runs out, sit on Distressed Debt Trading.

The relative value question: two bonds, one company

Harbor Point Industries has two bonds outstanding at different points in its capital structure. Bond A is the Senior Secured Notes: a 6.50% coupon, four years to maturity, priced at 98.00. Bond B is the Senior Unsecured Notes: an 8.00% coupon, five years to maturity, priced at 95.00. Which one is worth owning.

The first route is yield and spread, computed two ways each. A quick mental-math approximation for Bond A, coupon plus the discount to par divided by years, all over the average of face and price, gives 7.07%. Solving the bond's exact pricing equation numerically gives 7.09%, close enough that either is a fair verbal answer. The same two methods on Bond B give 9.23% and 9.30%.

Bond A, secured

309bp

Bond B, unsecured

530bp

An illustrative flat benchmark, used only to size the spread. Bond B pays about 220 more basis points of spread than Bond A.

Read only on this route, Bond B looks like the obviously cheap bond: roughly 220 more basis points of spread for one extra year of maturity. A candidate who stops here has answered the question the way most candidates do, and it is not the whole answer.

Why the wider-yielding bond can still be the wrong buy

Expected loss on a bond is, in its simplest form, the probability of default multiplied by the loss given default, the share of exposure an investor does not get back once a default happens. Both bonds sit at the same company, so the probability of default is the same number for each one.

The only thing that differs between them is loss given default, which is set almost entirely by where each bond sits in the capital structure worked out two sections above.

Assign illustrative recovery assumptions consistent with that capital structure: 70% recovery for the secured Bond A, and 35% for the unsecured Bond B, giving a loss given default of 30% for Bond A and 65% for Bond B. Dividing each bond's spread by its own loss given default measures how much compensation a bondholder is actually paid per unit of risk taken, holding the shared default probability equal across both bonds.

Bond A, secured

1,031bp

Bond B, unsecured

815bp

Bond A pays more spread per unit of loss severity than Bond B does, even though Bond B pays the higher absolute yield.

On this second route, Bond A is the better risk-adjusted value: 1,031 basis points of spread per unit of loss given default, against 815 for Bond B. Bond B's extra 220 basis points of raw spread does not fully make up for going from 70% down to 35% assumed recovery. A complete answer to this question shows both routes and explains why they disagree, rather than defending only the first number reached.

Mistakes candidates make

• Reciting the five Cs of lending in a capital markets interview. Character, capacity, collateral, capital and conditions are a retail and commercial underwriting checklist, and appear in neither S&P's nor Fitch's own published rating methodology, which run on the leverage and coverage grid worked through above instead.
• Confusing leverage with coverage. A company at 5.0 times leverage and 2.86 times coverage are two different statements about two different risks, and mixing them up under pressure is one of the fastest ways to lose credibility in a technical round.
• Comparing two bonds on yield alone, without adjusting for where each one sits in the capital structure, which is exactly the trap the relative value question above is built to catch.
• Treating a covenant question as a vocabulary quiz. Naming cov-lite or a restricted payments basket without explaining what either one actually lets a company do answers the word, not the question.
• Stating a leverage or coverage threshold as a universal rule. The bands above apply to an industry of ordinary cash flow volatility; a low-volatility industry such as a regulated utility gets materially looser bands on the same rating agency's own published grid.

When the interview cycle actually runs

A credit seat is filled through the same broad recruiting calendar as the rest of the trading floor. US bulge-bracket summer applications typically open in the December about eighteen months before the internship itself; for the 2027 class, BNP Paribas's New York posting is dated 16 December 2025, well before most candidates start preparing the technical material on this page.

A dedicated credit research seat is somewhat more likely than a rotational summer program to be filled through a desk-specific process rather than a broad generalist class, so a candidate applying to a named research seat should expect the interview to be scoped to that seat from the first conversation rather than assigned later. The full calendar, bank by bank, sits on Fixed Income Trading Desks.

Preparing for the round

A candidate walking into a credit interview, on either seat, should be able to do the following without notes:

• Build a leverage ratio and a coverage ratio from a short set of financials, and say which category each one falls into on a rating agency's published grid.
• Explain, in one sentence, why two bonds from the same company can carry different ratings.
• State the difference between a maintenance covenant and an incurrence covenant, and why bond covenants lean on the second kind.
• Describe what a recovery rating actually estimates, rather than treating recovery as a single guessed number.
• Compare two bonds on both yield and loss-given-default-adjusted spread, and explain out loud why the two routes can disagree.

The interview readiness check turns this list into a timed self-test, which is a better use of the last hour before an interview than another read-through of definitions.

A candidate preparing for a specific desk should pair this page with the desk-by-desk depth on Credit Trading, High Yield Trading and Distressed Debt Trading, and with the bond math underneath all of it on Bond Math Interview Questions.

The bottom line

Every number on this page comes from one invented company and two published rating-agency documents, and every ratio was computed once, then computed again by an independent method, so nothing here asks to be taken on faith. That is also what a strong interview answer does: state the method, show the arithmetic, and be ready to explain why a second, independent route to the same number lands in the same place.

A credit analyst interview is not a harder version of a bond math interview. It adds one more layer on top: judgment about a specific company's numbers, defended out loud, on either seat that asks for it.