Maintenance covenants and incurrence covenants are the two fundamental ways a lender can constrain a borrower, and the difference between them is the single most useful thing to understand about how debt documentation works. A maintenance covenant is tested at fixed intervals whether or not the borrower does anything; an incurrence covenant is tested only when the borrower takes a specific action. That distinction — a recurring test versus a conditional one — explains almost everything about why bank loans, high yield bonds and investment grade bonds look so different from one another.
The one-sentence difference
A maintenance covenant requires the borrower to maintain a financial ratio inside a defined limit at every test date, typically the end of each quarter. If leverage drifts above the cap because earnings fell, that is a breach, even if management did nothing at all. An incurrence covenant requires the borrower to satisfy a ratio before it is permitted to do something — raise new debt, pay a dividend, make an acquisition, sell an asset. If the ratio is not met, the action is simply blocked. Nothing has been breached; the borrower just cannot proceed.
The practical consequence is enormous. Under a maintenance covenant, a bad quarter can hand the lenders the keys: a breach is an event of default, which gives them the right to accelerate the loan, and in practice the right to sit down and renegotiate terms from a position of strength. Under an incurrence covenant, the same bad quarter produces nothing. The company keeps operating, the bonds keep paying, and the only thing that has changed is that certain doors are temporarily shut.
| Feature | Maintenance Covenant | Incurrence Covenant |
|---|---|---|
| When tested | Every quarter, automatically | Only when the borrower acts |
| Trigger for a breach | Passage of time plus a bad ratio | Cannot be breached passively |
| Consequence of failure | Event of default | The proposed action is blocked |
| Typical instrument | Senior secured bank loans, revolving credit facilities | High yield bonds, covenant-lite term loans |
| Typical holder | Relationship banks, direct lenders | Dispersed bondholders, CLOs, credit funds |
| Remedy available to the borrower | Equity cure, waiver, amendment | Consent solicitation, refinancing |
Why the holder base drives the design
The reason bank loans carry maintenance tests and bonds carry incurrence tests has less to do with credit quality than with who is on the other side of the paper. A syndicated bank loan is held by a manageable number of institutions with a credit committee, a relationship history and the capacity to negotiate. When a maintenance covenant trips, those lenders can convene, assess the situation and grant a waiver in exchange for a fee, a higher margin or tighter terms. The covenant is not really a tripwire for acceleration — it is a scheduled opportunity to reprice risk.
A high yield bond, by contrast, may be held by several hundred accounts that have never met the issuer and have no mechanism for coordinated negotiation. Amending an indenture requires a consent solicitation, which is slow, expensive and public. A maintenance covenant in that setting would be actively destructive: a routine cyclical dip would put a functioning business into technical default with no practical way to fix it. So the bond market took the opposite approach — no periodic testing, but hard conditions on the actions that would genuinely harm creditors. If you want to understand why high yield and investment grade instruments carry such different covenant packages, the answer starts with this coordination problem rather than with the ratings.
Investment grade: almost no covenants at all
Investment grade issuers occupy a third position that surprises people the first time they see it. A typical IG bond has essentially no financial covenants — no leverage test, no coverage test, no restricted payments basket. What remains is a thin package: a negative pledge preventing the issuer from granting security to other creditors without equally securing the bonds, a cross-default provision, and often a change-of-control put that lets holders sell the bonds back at 101 if the company is acquired and downgraded.
The logic is that the rating does the work the documentation would otherwise do. For an investment grade company, a downgrade to sub-investment grade is genuinely expensive: spreads widen across the entire curve, commercial paper access can disappear, some index-tracking funds are forced sellers, and certain counterparties will not face the credit at all. That threat disciplines behaviour more reliably than any leverage test, and it operates continuously rather than quarterly. Adding maintenance covenants would also be impractical for an issuer that refinances constantly across dozens of instruments and jurisdictions — a single quarterly test capable of tripping the whole structure would introduce a fragility neither side wants.
High yield covenants exist precisely because that rating cushion is absent. The issuer has already been judged to carry meaningful default risk, so the protection has to be written into the contract rather than borrowed from the rating agencies.
The four families of high yield incurrence covenants
Incurrence packages are more varied than maintenance packages, but almost everything in a standard high yield indenture falls into one of four buckets.
1. Debt incurrence — the ratio debt test
The core provision permits new debt only if a specified ratio is satisfied on a pro forma basis. The most common formulation is a Fixed Charge Coverage Ratio, typically requiring Consolidated EBITDA to cover fixed charges at least 2.00x after giving effect to the new borrowing. Many indentures add a parallel secured leverage test — new secured debt only if Consolidated Secured Net Leverage stays below, say, 3.00x — which exists specifically to stop the unsecured noteholders from being structurally subordinated by a wave of new secured borrowing that would rank ahead of them in a recovery.
Alongside the ratio test sits a set of permitted debt baskets: fixed euro amounts, a credit facilities basket, a general basket, capital lease capacity. These do not require the ratio to be met at all. In practice a sophisticated issuer often has more capacity through the baskets than through the ratio test, and the practical answer to "how much can this company borrow" is always the sum of what the baskets allow plus whatever the binding ratio permits. Working through a covenant analysis case that tests both the maintenance and incurrence sides of a real capital structure is the fastest way to see how those layers interact, because the arithmetic makes the layering obvious in a way that reading an indenture does not.
2. Restricted payments
The restricted payments covenant governs cash leaving the restricted group: dividends to shareholders, share buybacks, and voluntary payments on subordinated or junior debt. Its signature feature is the builder basket, which accumulates as a percentage — conventionally 50% — of cumulative Consolidated Net Income since the issue date, less any payments already made. It grows only when the business actually earns money, which ties the sponsor's ability to extract cash to performance rather than granting it upfront.
A general or starter basket sits alongside it to provide day-one flexibility before any income has built up. Restricted payments are also usually conditional on the ratio debt test being satisfied, so a company that cannot borrow generally cannot pay a dividend either. This is exactly the constraint that determines whether a sponsor can execute a dividend recapitalisation out of existing capacity or needs a full refinancing to do it.
3. Liens and asset sales
The lien covenant limits the granting of security, protecting the priority position of existing creditors. The asset sale covenant requires disposals to be at fair market value, mostly for cash, and requires the proceeds to be applied in a defined order: reinvested in the business within a set window, used to repay senior debt, or offered to bondholders at par through an asset sale offer. Without it, an issuer could sell the crown jewels and distribute the proceeds, leaving bondholders with a claim on a hollowed-out group.
4. Change of control and anti-layering
Change of control provisions give holders a put at 101 if the company changes hands, which matters because the credit they underwrote may no longer exist. Anti-layering provisions prevent the issuer from creating debt that sits between the notes and the senior facilities in the payment waterfall. Understanding where each instrument sits in that hierarchy is the foundation for the whole analysis, which is why it is worth being fluent in the ranking of senior, mezzanine and PIK instruments in a leveraged capital structure before you try to read a covenant package.
Cure mechanisms: the borrower's escape hatch
Because maintenance covenants can be breached passively, loan documentation almost always gives the borrower a way to repair a breach without going through a formal waiver. This is the equity cure, and it is one of the most heavily negotiated provisions in sponsor-backed deals.
The mechanic is straightforward: the sponsor injects fresh equity, and the documentation deems the covenant to have been satisfied. What varies — and what is worth real money — is how the injection is treated. Under an EBITDA cure, the cash is added to Consolidated EBITDA for covenant purposes, which increases the denominator of the leverage ratio. Under a debt paydown cure, the cash must be applied to reduce debt, which decreases the numerator. Because leverage is a ratio, the EBITDA route is dramatically more capital-efficient, and the efficiency multiple is precisely the covenant level itself. At a 3.50x covenant, curing through EBITDA costs roughly one third and a half of what curing through debt paydown costs for the same effect.
Lenders therefore push back on three fronts. They cap the frequency — commonly no more than two cures in any four consecutive quarters and five over the life of the facility — on the theory that a cure should bridge a one-off shortfall rather than substitute for operating cash flow. They prohibit over-curing, meaning an injection larger than the minimum needed to pass, because excess cure proceeds build artificial headroom into future periods and mask a deteriorating business. And they resist letting cured amounts roll forward into subsequent test periods for the same reason.
Repeated cures send a mixed signal. On one reading, a sponsor willing to put fresh money into a struggling asset still believes the equity is worth defending. On another, a company that needs cash injections to pass its tests is one whose original credit case has broken. Credit committees generally weight the second reading more heavily after the second cure.
Covenant-lite and what happened to the middle ground
The clean split between loans with maintenance tests and bonds with incurrence tests has eroded substantially. Covenant-lite loans — term loans that carry incurrence-style protection rather than quarterly maintenance tests — went from a pre-crisis novelty to the large majority of the institutional leveraged loan market. The driver was structural: term loans are increasingly held by CLOs and credit funds rather than by relationship banks, and those holders have neither the appetite nor the machinery to run a waiver negotiation. They behave like bondholders, so they accept bond-style protection.
Where a maintenance test survives in these structures, it is often a springing covenant on the revolving credit facility only, tested exclusively when the revolver is drawn beyond a threshold such as 35% or 40% of commitments. That gives the revolving lenders — usually the relationship banks who face the most immediate liquidity risk — a seat at the table, while leaving the term loan holders outside the negotiation entirely. The result is a structure where different creditor groups have genuinely different rights, which becomes decisive when things go wrong and the recovery analysis and creditor hierarchy in a restructuring come into play.
What actually gets negotiated
Once you know which type of covenant you are looking at, the substance sits in the definitions rather than in the headline ratio. A 3.50x leverage covenant sounds like a hard number, but it is only as tight as the definition of Consolidated EBITDA it is measured against.
Add-back definitions are where the flexibility hides. Non-recurring restructuring costs, transaction expenses, and run-rate cost savings from initiatives not yet implemented all get added back to reported EBITDA in most leveraged loan documentation. The cap on synergy add-backs — often 20% or 25% of Consolidated EBITDA in aggressive deals, occasionally uncapped, sometimes with no deadline for realisation — can create a full turn or more of phantom headroom. Every euro of permitted add-back mechanically lowers every leverage ratio and raises every coverage ratio simultaneously. This is why the first thing a credit analyst does with a covenant calculation is rebuild it from reported figures, and why knowing which items legitimately come out of EBITDA and which do not is a prerequisite for any covenant work.
The second area is the perimeter. Covenants apply to the restricted group, and the definition of unrestricted subsidiaries determines what sits outside it. Assets moved to an unrestricted subsidiary escape the covenant package entirely, which is the mechanism behind the well-known transactions where valuable collateral migrated beyond the reach of existing lenders. The third is the netting convention: whether cash is netted against debt, whether there is a cap on how much cash can be netted, and whether the ratio is calculated against last-twelve-months or annualised earnings.
A final point that catches candidates out is the difference between covenant headroom expressed as debt capacity and headroom expressed as an earnings cushion. A company at 3.20x against a 3.50x covenant has meaningful absolute debt headroom, but the more revealing number is how far EBITDA can fall before the ratio trips — often well under 10%, which is inside the range of a normal cyclical downturn. Being able to move between the two framings, and knowing which one a lender cares about, is the mark of someone who has actually looked at a credit agreement. The underlying skill is the same one tested in a debt capacity analysis, where EBITDA-based leverage, covenant constraints and free cash flow serviceability all have to be reconciled.
How this comes up in interviews
Covenant questions appear in leveraged finance, debt capital markets, restructuring and private equity interviews, usually in one of four forms. The definitional version — "what is the difference between a maintenance and an incurrence covenant" — is a screening question and should take fifteen seconds. The applied version asks you to calculate a ratio, test it and state whether it passes. The reasoning version asks why the structures differ, and rewards the coordination-problem answer rather than a recitation of credit quality. The judgement version presents a stressed situation and asks what the borrower's options are, where the expected answer walks through cure, waiver, amendment and refinancing in order of cost and difficulty.
The mistake that separates candidates is treating an incurrence covenant as though it were tested quarterly. If a company's Fixed Charge Coverage Ratio falls below 2.00x and it has not incurred debt or paid a dividend, nothing has happened. There is no breach, no default and no lender remedy. Candidates who say "the company is in default" have revealed that they have never read an indenture. The same holds in reverse: a company can comfortably pass every incurrence test and still be in default under a maintenance covenant in its bank facility, because the two packages are independent and the loan tests run on their own schedule regardless of how healthy the bond covenants look.
The second common error is answering "how much can the company borrow" with the first test that produces a number. Capacity is always the minimum across every applicable test plus any available baskets, and an interviewer setting the question will usually construct it so that the obvious test is not the binding one.