Insights · Guide
Know when the covenant cushion is real
The ratio is not the hard part. The definition is. Capital Refinery reads the covenant compliance certificate to its terms as filed, so the headroom the deal partner sees is the one the lender actually signed.
A leverage covenant reported with a comfortable cushion, say 4.78x against a 6.0x covenant, can still sit on an EBITDA definition that has drifted: add-backs that recur past the quarter they were called one-time, cash interest converted to PIK, or a covenant reset through amend-and-extend. None of those moves breach the covenant on paper. Capital Refinery reads the compliance certificate to its clause as filed, with the covenant, threshold, and headroom traced to the document.
The headline ratio is the report. The definition is the risk.
Read a private-credit covenant compliance certificate. Leverage at 4.78x against a 6.0x covenant, a 25% cushion, looks healthy. Now read the EBITDA definition behind it: add-backs for restructuring, for transaction expenses, for extraordinary and non-recurring losses that have recurred for three consecutive quarters. Pro-forma adjustments for synergies that have not closed. PIK interest counted as paid.
The cushion is real until the definitions move, and the definitions do move, through amend-and-extend negotiations, through cash-to-PIK conversions, through add-back lists that get longer rather than shorter. The covenant test is being preserved by changing what counts, not by the borrower closing the gap.
The borrower is not breaching the covenant. In cases like this, the cushion is being engineered around the breach.
Covenant erosion is documented at the lender, law firm, and rating-agency level
| Source | Finding |
|---|---|
| Sidley Austin (2026) | A typical direct-lending leverage covenant is set with a 25 to 35% cushion to projected EBITDA; restructuring, transaction expenses, and extraordinary losses are among the most negotiated add-backs. |
| Proskauer Rose, Private Credit Group (2025) | A notable rise in disputes over whether borrowers and sponsors maximize add-backs in loan agreements to inflate EBITDA and maintain covenant compliance. |
| Moody's Ratings, via Alternative Credit Investor (2025) | Erosion of covenant protection is credit-negative for lenders, with opacity across the market weakening early-warning indicators. |
| Lincoln International, Q3 2025 Senior Debt Index | Covenant defaults fell quarter over quarter to 3.2% but rose 1.0% year over year; 11% of valued loans paid PIK in 2025, with amendment activity up 13% and covenant-holiday activity up 14% quarter over quarter. |
| Octus, BDC Weekly Roundup (2026) | 27 borrowers switched a portion or all of their interest to PIK in one quarter after paying all-cash interest the quarter before. |
Three places the headline ratio looks fine and the cushion is gone
The recurring add-back
An add-back labeled extraordinary in one quarter reappears the next three. The credit agreement permits it under its own definitions. The cushion the certificate shows is narrower than it reads once the recurring item is counted.
The cash-to-PIK conversion
The cash interest line goes to zero; the principal balance grows instead. The leverage ratio holds because EBITDA is computed before interest. The lender holds a paper return while the borrower holds compounding debt.
The amend-and-extend
A sponsor equity injection clears the way for an amendment that resets the covenant level and adds a holiday. The default rate stays flat. The position is structurally weaker, and the question lands on the deal partner who signed off on the amendment.
The certificate as filed, before and after the read
Capital Refinery does not replace the covenant compliance workflow or recompute the ratio under the agreement's own definitions. It reads the certificate as filed, traces the headroom to the document, and leaves a figure blank where no actual can be sourced.
Reading the certificate by hand
- The headline ratio is re-keyed from the certificate into a tracking spreadsheet each quarter
- A stale forecast column sometimes stands in for the certified actual
- Whether an add-back recurred last quarter depends on whoever remembers the prior certificate
Reading the certificate as filed
- The covenant, the cap or floor, and the headroom are traced to the certificate they came from, cited to the page where a locator resolves
- A figure that cannot be sourced is shown blank rather than filled with an estimate
- Each add-back carries a position the team recorded, accepted, rejected or negotiated, visible across quarters
The covenant test is preserved by changing what counts. The dashed line is the covenant; the real result has been walking toward it while the reported cushion held flat.
What Capital Refinery reads on the certificate
Capital Refinery does not recompute the covenant under the agreement's own definitions. Compliance certificates are read to their clause as filed; recomputing the ratio under the agreement's §1.01 definitions is not part of what it does.
- Reads the credit agreement and the compliance certificate, among the 41 document families classified on arrival, with figures traced to source and cited where a locator resolves.
- Reads the total net leverage ratio and the fixed-charge coverage ratio, and the cap or floor each is measured against, straight from the certificate as filed.
- Shows the headroom that remains against that contractual threshold; a figure that cannot be sourced is shown blank rather than invented.
- Records a position on each add-back, accepted, rejected or negotiated, so a recurring item can be flagged each time it reappears.
- Grades risk signals against the firm's own policy thresholds; move a threshold and the memo re-grades on its next render.
Take a hypothetical position, not a product output: FCCR at 1.42x against a 1.20x floor at approval, compressing to 1.31x two certificates later, a headroom of 0.11x. Read from each certificate rather than re-keyed by hand, the compression is visible as it happens.
Pressure-test one credit agreement
Send one credit agreement and the last four compliance certificates. Capital Refinery returns the add-back positions, the headroom trend, and the covenant as filed, traced to the document, before the next certificate lands on the deal partner's desk. The read does not tell the team whether to act. It tells the team what the certificate actually says, separated from what the certificate has been engineered to show.
Questions
What makes a covenant cushion real versus engineered?
A real cushion holds when the EBITDA definition is stable. An engineered one survives because add-backs keep recurring, cash interest converts to PIK, or the covenant gets reset through amend-and-extend, none of which show up as a breach on the certificate.
Does Capital Refinery recompute the covenant under the credit agreement's own definitions?
No. Compliance certificates are read to their clause as filed. Recomputing a covenant under the agreement's §1.01 definitions is not part of the read.
How does Capital Refinery calculate covenant headroom?
It reads the ratio and the cap or floor it is measured against directly from the compliance certificate, and shows the headroom against that contractual threshold. A figure that cannot be sourced is shown blank.
What is a cash-to-PIK conversion and why does it matter for covenant compliance?
It is when cash interest is paid as additional principal instead of cash. The leverage ratio can hold because EBITDA is computed before interest, even though the borrower's debt load is compounding rather than shrinking.
Why does an amend-and-extend keep the default rate flat while the position gets weaker?
An amendment can reset the covenant level and add a payment holiday in exchange for a sponsor equity injection. The certificate looks clean afterward, but the underlying position has not actually improved.