What is a net leverage covenant actually measuring?
Contractual debt over contractual EBITDA, and neither term means what the accounts mean. Net leverage nets cash against debt, usually only up to a capped amount and only cash held in specified accounts, so a business sitting on unrestricted cash in the wrong entity may find that none of it counts. EBITDA is whatever the credit agreement says it is, which is a defined term running to several pages in most documents.
The test is also a point-in-time measure of a trailing period. It runs on last twelve months EBITDA at a quarter end against debt on that same date, which means a business with any seasonality can pass in one quarter and fail in the next on identical trading. Where the working capital cycle peaks matters as much as the trading does.
Two further mechanics decide how often the test bites. A maintenance covenant is tested every quarter whether or not anything happens. A springing covenant is tested only when a condition is met, most commonly revolver utilization above a threshold, with above 40% drawn a common trigger point (Sidley Austin, 24 March 2026). A borrower who understands this negotiates the trigger as carefully as the ratio, because a springing test that never springs is not a covenant in practice.
Maintenance covenants remain the norm at the smaller end. One or more covenants remain common in smaller transactions (First Eagle Investments, March 2026, citing PitchBook LCD as of 28 February 2026), while covenant-lite structures in direct lending have reached 21% of deals, up from 4% in 2023, with 91% of those sitting above $50m of EBITDA (Proskauer via ABF Journal, June 2026). Covenant relief in this market is a large-borrower product.
How is the maximum actually set?
Backwards from the borrower's own model. The lender takes the base case that management presented, reads the projected leverage at each test date, and sets the covenant a stated percentage above that line. Typical leverage covenant cushions run 25% to 35% to the borrower model (Sidley Austin, 24 March 2026).
So the negotiation that decides the covenant happens before the covenant is discussed. A management case with confident growth produces a declining projected leverage line, and a covenant set 25% above a declining line steps down whether or not anyone negotiated a step-down. A conservative case produces a flatter line and a covenant that is easier to live inside, at the cost of a lower quantum of debt or a wider spread. Those two things trade against each other and most borrowers optimize only one of them.
Work the arithmetic once, at the top of the current clearing band, to see the shape of it. Total debt for a borrower below $10m of EBITDA clears at 2.50x to 3.25x (SPP Capital Partners, Market At A Glance, July 2026). A business closing at 3.25x with a 25% cushion is looking at an opening covenant near 4.06x, and at a 35% cushion near 4.39x. That is ours, not a published figure, and it applies at close: from the first test date the covenant tracks the model rather than the closing ratio, which is why the model matters more than the multiple.
The quantum itself has tightened sharply at the small end. A year earlier the same sub-$10m band cleared 2.50x to 4.00x total and 2.00x to 3.00x senior; it now clears 2.50x to 3.25x total and 2.00x to 2.50x senior. The senior band has lost half a turn and total leverage three quarters of a turn in twelve months (SPP, July 2026). Lenders also require a minimum 40% base equity capitalization with at least 60% of that in new cash, and independent sponsors are expected to show investment beyond rolled deal fees.
“The covenant is a copy of the borrower's own forecast with a margin of error attached. That is why a management case built to win the financing and a management case built to survive it are different documents, and why it is worth deciding early which one is being presented.”
What does a step-down schedule do?
It tightens the maximum on a fixed calendar, on the theory that a deleveraging business needs less headroom over time. In a typical structure the covenant falls in stages across the first several test years and then holds flat, so the borrower is given room early and expected to earn it back.
No source in the surveyed market publishes step-down cadence, size or prevalence as a market series, and none is asserted here. What can be said is that the schedule is negotiated at the same moment as the opening level and is usually conceded more readily, because a lender pricing today's risk discounts a tightening that lands three years out. Borrowers do the same thing, which is how a schedule that nobody argued about becomes the binding constraint in year three.
The mechanical risk is straightforward. A step-down assumes deleveraging arrives on schedule, and deleveraging arrives through EBITDA growth, cash sweep, or both. If growth lands a year late, the covenant tightens on time and the ratio does not, and the breach is caused by the calendar rather than by the trading.
The macro backdrop currently cuts mildly in the borrower's favour and not enough to rely on. Modest improvement is expected in 2027, with median leverage declining slightly and EBITDA growth improving interest coverage, though delayed rate relief will likely limit the pace of improvement (Fitch Ratings, 30 July 2026). Against that, median EBITDA growth across the middle market fell to 24% from 27% year over year, the largest quarter-over-quarter decline on record in that series (KBRA, published 28 July 2026). A step-down schedule built on the second data point is a different document from one built on the first.
Which EBITDA does the covenant run on?
The defined one, and the gap between defined and reported is the single largest unpriced item in most credit agreements. The global regulator put it plainly: private credit borrowers run leverage of 5 to 6 times debt to EBITDA against roughly 4 times in leveraged loans, and with EBITDA adjustments stripped out, true leverage could be closer to 7 times (Financial Stability Board, Report on Vulnerabilities in Private Credit, 6 May 2026).
That is a full turn or more of difference created by definitions rather than by borrowing. Add-backs for run-rate cost savings, the pro forma effect of actions already taken, transaction expenses, non-recurring items and projected efficiencies all enlarge the denominator, which lowers the reported ratio without lowering the debt. The market has not tightened this: expansive EBITDA adjustments persist alongside most favoured nation sunsets and broad basket capacity, and the loan market as a whole is still described as broadly borrower-friendly (LSTA, Loan Market Covenant Trends 2Q26, 24 July 2026).
For the borrower this is genuinely two-sided, and it is worth being clear about which side you are on. Generous add-backs create headroom at signing and remove it later, because an add-back that assumes cost savings must eventually be delivered or the ratio corrects. The most damaging version is the one where the add-back is capped in time, so the headroom expires on a date rather than when the saving is achieved.
The other definitional lever is what counts as debt. Whether the revolver is included when undrawn, how earnout and deferred consideration are treated, whether operating leases are captured, and whether an acquisition financed in the quarter is measured pro forma all move the ratio without changing the business. Ratio debt headroom is tightening at the same time: first-lien ratio debt capacity fell to 0.28x EBITDA in the first half of 2026 from 0.46x in the second half of 2025 (9fin, US Covenant Trends Report H1 2026).
What is negotiable in this market?
Less than a year ago, and more than most borrowers attempt. Of surveyed private credit lenders, 98% report that their underwriting standards became notably stricter since the start of 2026. The share expecting looser documentation fell from 33% to 4% year over year, while the share expecting tighter documentation rose from 13% to 56% (Houlihan Lokey, Q2 2026 Private Credit Survey). Ranked by what lenders are actually pushing on: wider pricing at about 62%, tighter documents and structures at about 52%, sector avoidance at about 44%, and smaller cheques at about 17%.
The syndicated market disagrees at the margin, which is why quoting either in isolation misleads. Documentation scores tightened from 3.83 in the first quarter of 2026 to 3.70 in the second on a scale where 1 is most protective of lenders and 5 least (Covenant Review, CR TrendLines July 2026), while borrower-favourable features have gone the other way, with zero-floor builder baskets in 43% of sponsored deals in the first half of 2026 against 26% previously, and margin ratchets in 40% against 29% (9fin, H1 2026). The market is tightening on protections and loosening on flexibility at the same time.
Within that, four items are worth the borrower's negotiating capital, roughly in this order. The EBITDA definition and the treatment of add-backs, because it sets the denominator on every future test. The cushion, because it is the only term that directly buys forgiveness. The step-down schedule, because it is conceded cheaply and binds late. And the equity cure: how many times it can be used, whether it is capped, whether cure proceeds reduce debt or increase EBITDA, and whether consecutive quarters are permitted.
The one thing rarely worth trading for is the headline multiple. A borrower who wins an extra quarter turn of covenant and gives up the add-back definition to get it has usually made the position worse, because the ratio moves once and the definition moves every quarter for the life of the facility.
| Term | Reading | Source |
|---|---|---|
| Lenders reporting notably stricter underwriting since January 2026 | 98% | Houlihan Lokey, Q2 2026 |
| Lenders expecting looser documents, against a year earlier | 4%, from 33% | Houlihan Lokey, Q2 2026 |
| Lenders expecting tighter documents, against a year earlier | 56%, from 13% | Houlihan Lokey, Q2 2026 |
| Typical leverage covenant cushion to the borrower model | 25% to 35% | Sidley Austin, 24 March 2026 |
| Revolver utilization commonly triggering a springing test | Above 40% | Sidley Austin, 24 March 2026 |
| Covenant-lite share of direct lending deals, against 2023 | 21%, from 4% | Proskauer via ABF Journal, June 2026 |
| Share of those covenant-lite deals above $50m EBITDA | 91% | Proskauer via ABF Journal, June 2026 |
| First-lien ratio debt headroom, first half 2026 against second half 2025 | 0.28x, from 0.46x | 9fin, H1 2026 |
As of August 2026
Sources: Sidley Austin, 24 March 2026, for the leverage covenant cushion range and the springing revolver trigger; SPP Capital Partners, Market At A Glance, July 2026, for senior and total leverage clearing bands by EBITDA size band, the year-over-year comparison and the minimum equity capitalization requirement; First Eagle Investments, March 2026, citing PitchBook LCD as of 28 February 2026, for the prevalence of maintenance covenants in smaller transactions; Proskauer via ABF Journal, June 2026, for covenant-lite share and its concentration above $50m of EBITDA; Financial Stability Board, Report on Vulnerabilities in Private Credit, 6 May 2026, for reported and adjustment-stripped leverage; LSTA, Loan Market Covenant Trends 2Q26, 24 July 2026, for EBITDA adjustments and market characterization; 9fin, US Covenant Trends Report H1 2026, for ratio debt headroom, builder baskets and margin ratchets; Houlihan Lokey, Q2 2026 Private Credit Survey, for underwriting and documentation expectations and the ranked drivers; Covenant Review, CR TrendLines July 2026, for documentation scores; Fitch Ratings, 30 July 2026, for the 2027 view; KBRA, published 28 July 2026, for the EBITDA growth series. The worked cushion figures are our own arithmetic on the cited leverage band at the cited cushion convention, not a published series. No surveyed source publishes step-down cadence, size or prevalence, and none is asserted here.