Private Credit Covenants: The Ratios, the Headroom and the Breach

Ask what a covenant is worth and most candidates reach for a definition. Moody's has a better answer: nine points of recovery. Lenders holding a first-lien loan with a maintenance covenant recovered an average of 66 cents on the dollar between 2023 and mid-2025. Lenders in the identical seat, on an otherwise identical loan with no maintenance test, recovered 57. A longer nineteen-year study from S&P puts the same gap at roughly eleven points.

That is not a rounding difference. It is the entire commercial argument for the most technical subject in this asset class, stated as a number rather than a definition, and it is where a covenant question in an interview is actually pointing.

9-11 pts

recovery gap

Maintenance covenant vs cov-lite, first-lien loans

21%

of 2025 deals were cov-lite

Up from 4% in 2023, direct lending

1.2x

extra debt turns already built in

Average debt capacity vs closing leverage, 2025

Maintenance against incurrence, and what the difference is worth

A maintenance covenant is tested on a schedule, usually quarterly, whether the borrower does anything or not. An incurrence covenant only has to be satisfied at the moment the borrower takes a specific action — raising debt, paying a dividend, buying something. One is a trip wire. The other is a gate that only matters to someone walking through it.

A maintenance covenant

• Tested every quarter, whether the borrower acts or not
• A breach gives the lender an immediate right to engage
• Restricts the revolver on trip, before cash runs out
• The thing private credit sells against the cov-lite bank market

An incurrence covenant

• Bites only when the borrower takes a specific action
• No trigger at all while the borrower does nothing
• Silent while performance quietly deteriorates underneath it
• The syndicated-market default, now spreading into direct lending

Why the trip wire pays for itself

A borrower on a covenant-lite loan can miss its plan by a wide margin and still be in perfect compliance, because nothing in the document is watching. A borrower with a maintenance covenant trips it, and tripping it hands the lender a right most people underrate: the right to restrict the revolver and get a seat at the table before the business has burned through its liquidity. That earlier seat is what shows up in the recovery data.

The gap holds across two independent studies covering different periods, which is worth more than either number alone. Neither claims the covenant is the whole story — leverage and debt cushion still do most of the work in a recovery — but a nine-to-eleven-point gap that appears twice, in two different windows, is a pattern worth having in your answer rather than a coincidence worth explaining away.

The ratios that actually get tested

Everyone arrives able to define leverage. What separates candidates is knowing what each ratio is structurally unable to see.

The leverage ratio is the one that carries the most weight in the document, because its definition drives far more than the covenant itself. It sets the incurrence baskets, the pricing mechanics, the excess cash flow sweep, and often the terms of any future amendment. Get the definition wrong and every other number in the agreement inherits the mistake.

The covenant nobody profiles

Interview prep tends to stop at leverage and coverage, but a growing share of deals now carry something else entirely. Where a business is driven by working capital cycles, heavy capital spending, or asset quality rather than steady earnings, lenders increasingly pair the leverage test with a minimum liquidity threshold or a borrowing-base construct.

In software and other high-growth, low-current-profitability sectors, that has gone a step further: an annual recurring revenue covenant, adjusted for churn and retention, tests whether the business itself is durable rather than whether this quarter's EBITDA looks acceptable. A candidate who can name this and explain why leverage would be the wrong test for that borrower is answering a harder question than the one that was asked.

The reason lenders reach for a second or third ratio at all is that no single test is blind in the same place as the others. A business can hold flat leverage while its cash conversion quietly worsens, pass fixed-charge coverage while its revenue base is eroding underneath a still-healthy margin, or clear a liquidity floor while burning cash every single quarter. Two ratios that are blind to different things catch more than either one twice.

Headroom is two different numbers, not one

Ask how much room a borrower has and most candidates answer with a single figure. There are two, and they measure completely different things.

The first is EBITDA-decline headroom: how far earnings can fall before the leverage covenant breaks. The market convention sits around a 25–35% cushion to the sponsor's projected EBITDA, and the credit case study walks through exactly how that cushion has drifted wider over time. The second is debt-capacity headroom, and it is rarely discussed at all: how much more debt the same document already permits a sponsor to add without breaching anything.

5.1x

average closing leverage

Up from 4.8x in 2024

6.3x

average debt capacity

What the document already allows

1.2x

turns of headroom given away at signing

Before a sponsor even negotiates an amendment

That gap between closing leverage and debt capacity, about 1.2 turns on average, is capacity a sponsor can draw down through an acquisition facility or an incremental basket without ever going back to the lender for a waiver. It is headroom the covenant does nothing to protect, because the covenant was never testing it.

Too much headroom is not the same as safety

A wider cushion sounds like a safer loan, and often is not. Set a covenant too tight and it forces frequent amendments over ordinary operating noise, which is its own cost. Set it too loose and the borrower can exhaust its actual liquidity well before the covenant ever trips — the trip wire has been placed so far back that it stops doing the one job it exists for.

A covenant set with too much headroom is decorative: it will show green on every compliance certificate right up until the point where there is genuinely nothing left to protect.

Private credit's own cov-lite creep

Direct lending built its identity on retaining what the broadly syndicated loan market gave away. That identity is now under pressure from inside.

2023

4%

2025

21%

Two measured years, shown side by side rather than joined — there is no data between them to draw a line through. Neither a benchmark nor a forecast, and the erosion is concentrated in the largest borrowers.

Covenant-lite direct lending deals went from 4% of the market in 2023 to 21% in 2025, and the shift is not evenly spread. Nine in ten of those cov-lite deals belong to borrowers with more than $50 million of EBITDA — precisely the segment competing hardest against the broadly syndicated market, where banks can offer terms private credit has to match to keep the deal.

Average closing leverage moved the same direction over the same two years, from 4.8x to 5.1x.

None of this means maintenance covenants are disappearing. Four in five 2025 deals still carried one, and the smallest borrowers — the ones the firm guides mostly cover — are barely touched by the trend at all. But a candidate who answers a covenant question as though cov-lite is purely a syndicated-market phenomenon is answering with information that is two years out of date.

EBITDA, and the definition fight underneath every ratio

Every ratio on the page above is only as strong as the EBITDA underneath it, and that number is negotiated, not reported.

What gets added back

The standard categories are restructuring and integration costs, transaction expenses, one-off or extraordinary losses, projected synergies, run-rate cost savings not yet realised, and a catch-all for company- or industry-specific items. Some of that is defensible — a genuine one-off cost distorts a single quarter's earnings and a lender is right to look past it. Some of it is a sponsor's growth story dressed up as a fact.

Europe shows where the fight is currently moving. A majority of European deals in 2025, 57%, now include revenue adjustments to the EBITDA definition, not just cost add-backs. That is a meaningfully more aggressive ask than the traditional cost-side synergy, because it lets a sponsor take credit for revenue that has not yet shown up anywhere in the accounts.

Caps and sunsets, the shape of the compromise

Lenders rarely refuse add-backs outright. Instead they cap them, often as a shared percentage of EBITDA across all the adjustment categories together, and they time-limit them with a sunset so a projected saving cannot inflate leverage forever if it never actually shows up.

Neither the cap nor the sunset is a fixed market number — they are negotiated deal by deal, which is exactly why the skill being tested is reading the specific number in front of you rather than reciting a convention.

State leverage twice wherever the add-backs are genuinely contested: once on the number as presented, once on your own view of what should survive. The gap between the two is usually the most useful sentence you say about any credit.

What happens when a covenant breaks

Acceleration — demanding full and immediate repayment — is the outcome candidates picture and the one that almost never happens first.

1. A waiver, for a breach that looks like a one-off. The lender formally forgives it and the rest of the document is untouched.
2. An amendment, for a breach that reflects something structural. The covenant level resets going forward, usually alongside a fee, a margin increase, tighter or more frequent reporting, or some combination of the three.
3. An equity cure, where the sponsor injects cash to bring the borrower back into compliance rather than accept either of the above.

Roughly two in three corporate defaults in 2025 were distressed restructurings of exactly this kind rather than a borrower simply failing to pay — a negotiated fix, not a collapse.

⚠️ That share is measured across corporate credit broadly, not private credit alone, and anyone quoting it as a private credit figure is over-reading it. What it supports is narrower and still useful: negotiated outcomes now dominate, and a facility held by one lender or a small club is the structure where such an outcome is easiest to reach.

Counting those restructurings as ordinary defaults overstates how bad the underlying losses actually are — worth knowing before quoting a headline default rate in a room.

The equity cure, and its limits

A sponsor putting new cash in sounds generous. The documents are written on the assumption that it should not become a habit.

The cap and the frequency

Almost every deal caps the total number of cures a sponsor can use over the life of the facility, typically at four uses, with five reserved for the largest transactions. On top of that lifetime cap sits a frequency rule, and market practice splits three ways:

• No cures in consecutive quarters — the most common position in the European mid-market.
• No more than two cures in any twelve-month period — more common in the US.
• No more than two consecutive cures — reserved for the largest deals.

A sponsor cannot simply keep paying its way out of a weak business quarter after quarter.

Debt reduction against the EBITDA cure

A cure amount can be treated two ways once it lands. The traditional approach deems it a reduction in debt when the leverage ratio is recalculated. The more aggressive approach, the EBITDA cure, instead deems the cash to have increased EBITDA — which moves the ratio by far more than the same dollar amount would through debt reduction, because EBITDA sits in the denominator of a number that is itself a large multiple.

European lenders resisted the EBITDA cure for years and have increasingly conceded it on larger deals, but typically only with its own sub-cap — often limited to once or twice over the life of the facility even where it is permitted at all. In the US it is closer to universal.

That transatlantic gap is worth having ready, because it is exactly the kind of specific, dated fact that separates a candidate who has read a document from one who has read a glossary.

Repricing: the economics of a breach handled well

A covenant breach is not only a risk event for a lender. Handled well, it is also where private credit earns some of its return.

Default interest typically steps up by around two percentage points above the original spread once a covenant is breached, compensating the lender for the additional risk it is now carrying and the additional work the workout requires.

Where the situation is bad enough to need new money, a priming facility — capital that sits ahead of the existing debt in the queue — has been structured with a liquidity preference of two times or more, meaning the lender providing that new money recovers twice its investment before anyone junior to it sees a cent.

None of this happens automatically. It is negotiated at the point of the breach, alongside the amendment or the waiver, which is why the amendment process is often where a direct lender's return on a difficult credit is actually made rather than lost.

Who decides, and why so few breaches ever become public

A restructuring adviser recommends a course of action. A direct lender decides one. That asymmetry is the structural reason private credit workouts are so much harder to observe from outside than a public default.

Most private credit facilities sit with one lender or a small club rather than being distributed across dozens of unrelated buyers, the way a broadly syndicated loan trades through a CLO market. Fewer parties means faster alignment, which is exactly why the waiver-or-amendment path above resolves most stress before it ever becomes a formal, disclosed event.

A syndicated default shows up in index data the day it happens. A bilateral amendment can happen quietly, get reported nowhere outside the two parties' own books, and leave the credit performing again within a quarter.

Reading a single default rate

The 2025 default rate for private credit is estimated at somewhere between 1.6% and 4.7%, and almost the entire width of that range is the distressed-exchange question above: include the negotiated fixes and the number looks much worse than excluding them, even though the underlying loss to lenders is nowhere near that different.

A published default rate here is a range, not a fact. Knowing why the range exists is a stronger answer than quoting either end of it.

How to answer a covenant question in the room

The pattern that scores well has four parts, in order.

• Name the type. Maintenance or incurrence, and say what that implies about who had negotiating leverage when the deal was struck.
• Place the headroom. Roughly where the EBITDA cushion sits against the market range, and — if it is relevant — whether the debt-capacity headroom tells a different story from the covenant headroom.
• Interrogate the EBITDA underneath it. A wide cushion on an aggressively adjusted number is not really a cushion, and saying so out loud is worth more than any amount of ratio arithmetic.
• Finish on the consequence. What happens the day it breaks — who has to act, roughly what it costs, and what would change your view.

A definition with no consequence attached is half an answer. The candidates who stand out are the ones who treat a covenant not as a fact to recite but as a lever someone, at some point, is going to have to pull.

What that sounds like in the room

This is a maintenance leverage covenant, set at what looks like a market cushion, maybe thirty per cent, but the EBITDA underneath it carries uncapped run-rate savings that haven't been actioned yet — so the real cushion is tighter than the document suggests. If EBITDA misses by half that amount, the sponsor is looking at a waiver conversation within two quarters, and I'd want a fee and a tighter reporting package attached to any reset rather than just a friendlier number.

Four sentences. The type, the headroom, the EBITDA problem underneath it, and what happens next with a price attached. That is the whole exercise.

The bottom line

A covenant is not the paperwork around a loan. It is the mechanism that decides how early a lender finds out something is wrong, and the recovery data says that timing is worth real money — nine to eleven points of it, on the numbers available today.

Know the type, place both kinds of headroom, read the EBITDA definition rather than accept it, and be ready to say what happens the day the covenant actually breaks. That is most of what the question is checking for.