Debt service coverage ratio is the number a lender looks at first. It determines how much debt a project can carry, whether distributions are allowed in a given period, and whether the borrower is in default.
The formula fits on one line. Almost everything that matters sits underneath it, in definitions that vary between facility agreements and are rarely what a textbook assumes.
The formula
DSCR = Cash Available for Debt Service ÷ Debt Service
For a given period, usually a quarter or a half year matching the payment dates:
- ·Numerator: CFADS, the cash the project generates and is permitted to use for debt service
- ·Denominator: scheduled principal plus interest falling due in that period
A DSCR of 1.30x means the project generated 1.3 times the cash it needed to meet debt service. A DSCR of 1.00x means it generated exactly enough and nothing more. Below 1.00x, the cash did not cover the obligation.
That is the whole formula. The difficulty is entirely in what the two terms mean.
What actually goes into CFADS
CFADS is not EBITDA, and treating it as EBITDA is the most common structural error in project models.
A typical build runs:
- ·Revenue received in cash
- ·Less cash operating costs
- ·Less cash taxes paid
- ·Less or plus movements in working capital
- ·Less maintenance capital expenditure
- ·Plus or less movements in reserve accounts, where the agreement permits
- ·Equals CFADS
Note what is excluded. Interest is not deducted in the numerator, because it sits in the denominator, and deducting it in both places double counts it. Depreciation and amortisation are not deducted, because they are not cash. Distributions are not deducted, because they rank below debt service in the waterfall.
Note also what is included that surprises people. Cash taxes, not the tax charge. Actual working capital movement, not an assumption that it nets to zero. Maintenance capex, because an asset that is not maintained does not generate the cash the projection assumes.
Where the definitions come from
This is the part that separates a model that works from a model a lender will accept.
Your facility agreement defines CFADS and DSCR, and its definition wins. Not the textbook, not the convention, not the version in the last model you built. Agreements differ on whether the calculation sits before or after tax, whether maintenance capex is deducted, how reserve account movements are treated, whether hedging settlements are included, and what happens to cash held at the project level but not distributed.
A model built on a plausible but different definition will produce coverage the lender does not recognise. Reconciling the two under time pressure during a credit process is avoidable work.
Ask for the term sheet or the agreement before you build. If it does not exist yet, model the definition you expect and document it explicitly in the model, so that it can be changed in one place when the agreement lands.
Backward and forward looking
Two calculations, often both required.
Historic DSCR looks at the period just ended. It uses actual cash generated and actual debt service paid. This is what covenant compliance certificates usually report.
Projected or forward-looking DSCR looks at the next period or the next twelve months, using forecast cash. Some agreements test both, and a borrower can pass one and fail the other.
Related but distinct are LLCR, the loan life coverage ratio, which compares the present value of CFADS over the remaining loan life to the debt outstanding, and PLCR, which does the same over the full project life. These measure whether the debt can be repaid at all, rather than whether this period's payment is covered. A structure can show healthy period DSCRs and a weak LLCR, and that combination tells you the problem is at the back end.
Minimum, not average
A model that reports an average DSCR of 1.45x across the loan life is reporting something a lender does not price on.
Lenders test coverage in every period. A structure averaging 1.45x that dips to 0.98x in one quarter has breached, and the average is irrelevant to that fact. Seasonal businesses, projects with a maintenance shutdown, and assets with lumpy revenue recognition all produce exactly this pattern.
Any project finance model should report minimum DSCR across the term as a headline output, alongside the period in which it occurs. If your model only shows an average, it is hiding the thing the reader most needs to see.
Typical covenant levels
These vary by sector, jurisdiction and lender appetite, so treat them as orientation rather than standards.
- ·Default or event of default: commonly set at or just above 1.00x to 1.05x
- ·Distribution lock-up: commonly 1.10x to 1.20x, meaning the borrower may operate but may not pay dividends until coverage recovers
- ·Base case sizing: lenders typically size debt to a target DSCR well above the lock-up, so there is headroom between the sizing case and the covenant
That gap between the sizing ratio and the covenant is the real measure of how much room a structure has. A project sized at 1.35x with a 1.20x lock-up has less tolerance than the headline suggests.
Sizing debt from DSCR
In a properly built model, the facility amount is an output, not an input.
Debt is solved so that CFADS in each period covers debt service at the target ratio. With a sculpted repayment profile, principal repayments are shaped to the cash flow so that DSCR stays flat at the target rather than fluctuating. This is what allows a project with uneven cash generation to carry more debt than a straight-line amortisation would permit.
Sculpting introduces genuine circularity, because the repayment depends on the coverage ratio, which depends on the repayment. That circularity needs a controlled resolution with a documented convergence check, not simply enabling iterative calculation and trusting the result.
Five errors we see repeatedly
Interest deducted twice. Once inside the CFADS build and again in the denominator. Understates coverage, and is easy to miss because the direction of the error looks conservative.
EBITDA used as a proxy for CFADS. Ignores tax, working capital and maintenance capex. Overstates coverage, sometimes substantially.
Coverage reported only as an average. Hides the period that matters.
Denominator using the wrong period. Principal and interest must fall in the same period as the cash being measured. Mismatched payment dates and model periods produce a ratio that is arithmetically clean and meaningless.
Maintenance capex omitted. Common in early-stage models where the operating assumptions have not been developed. Flatters coverage in exactly the years a lender is most concerned about.
Why this matters beyond the covenant
DSCR determines how much debt an asset can carry, which determines how much equity is required, which determines the sponsor's return. A model that overstates coverage by 10 percent does not simply misreport a ratio. It implies a capital structure that cannot be financed, and the error surfaces during credit approval, at the point when changing anything is most expensive.
Building or refinancing a debt-funded project? We build lender-ready project finance models with debt solved from CFADS, coverage tested in every period, and definitions matched to your facility agreement.