For the full regulatory framework, gate by gate, see The USDA Feasibility Study Framework.
Search for USDA's minimum debt service coverage ratio and you will be told it is 1.25x. That figure does not appear in the regulation for new lending. What the regulation does specify is the definition — and USDA's definition subtracts replacement capital expenditure from EBITDA before dividing by debt service. A project that clears 1.25x on a conventional calculation can fall below it on USDA's, and the difference is the single most common reason a pro forma fails review.
The short version
- 7 CFR Part 5001 defines DSCR as earnings before interest, taxes, depreciation and amortisation, less reasonably expected replacement capital expenditures, divided by annual debt service.
- That replacement capex deduction is not standard commercial practice. Most lender templates and most borrower models do not include it. On a capital-intensive project it can move the ratio by 0.15x to 0.30x.
- The regulation does not set a universal minimum coverage ratio for new loans. It sets minimums in specific circumstances — notably refinancing — and otherwise leaves the threshold to lender credit policy and the Agency's credit evaluation under §5001.202.
- 1.25x is convention, not regulation. It is where lenders and the Agency are comfortable, and it functions as a de facto floor, but citing it as a regulatory requirement is wrong and signals unfamiliarity with the rule.
- Rural hospital refinancing carries an explicit 1.1x historical coverage test, alongside a requirement that new annual debt repayment be lower than existing.
- The analytical emphasis shifts by programme: B&I on operating cash flow, Community Facilities on sustainability rather than profit, REAP on a physical production forecast times a price, Water & Waste on rate capacity.
What the regulation actually says
The definition sits in Subpart A of 7 CFR Part 5001, the OneRD Guarantee Loan Initiative regulation that has governed Business & Industry, Community Facilities, REAP and Water & Waste Disposal guaranteed loans since 1 October 2020.
Debt service coverage ratio means the ratio obtained when taking earnings before interest, taxes, depreciation, and amortization less reasonably expected replacement capital expenditures divided by the annual debt service (principal and interest payments) of the borrower.
Read that clause carefully, because three things in it matter and two of them are routinely missed.
First, the numerator starts at EBITDA. Not net income, not net operating income, not cash flow after distributions. EBITDA.
Second, and critically, replacement capital expenditure is deducted. USDA does not accept a coverage figure that ignores the cost of keeping the asset in service. This is a materially stricter standard than the conventional EBITDA-to-debt-service calculation most lenders run.
Third, the denominator is total annual debt service — principal and interest — of the borrower. Not just the guaranteed loan. All of it.
Each of these has consequences worth working through.
The replacement capex deduction, and why it catches people
This is the provision that separates a compliant USDA pro forma from one that will draw questions.
Conventional commercial DSCR takes EBITDA over debt service. It treats capital expenditure as a financing decision rather than an operating cost, on the reasoning that capex can be deferred, financed separately, or funded from reserves.
USDA does not accept that framing, and the reasoning is sound. A guaranteed loan runs for up to forty years on some asset classes. Over that horizon, equipment gets replaced, roofs get replaced, refrigeration systems get replaced, vehicles get replaced. A projection that shows comfortable coverage while ignoring the cost of keeping the facility operational is not describing a sustainable business; it is describing the first several years of one.
The practical effect depends entirely on how capital-intensive the project is.
A light-capex operation — professional services, a warehouse with minimal equipment, a straightforward retail operation — may see the ratio move by 0.03x to 0.08x. Noticeable but not decisive.
A moderately capital-intensive operation — a hotel, a care facility, a food processing plant — will typically see 0.10x to 0.20x. A project modelled at 1.30x conventionally may present at 1.12x on USDA's definition. That is the difference between comfortable and marginal.
A heavy-capex operation — cold storage, meat processing, anaerobic digestion, anything with substantial rotating equipment or refrigeration — can see 0.20x to 0.35x. This is where projects that looked financeable stop being financeable.
The word doing the work in the regulation is "reasonably expected." Not a token allowance, and not an unfounded worst case. A defensible reserve is built from the actual asset schedule: what components exist, what their useful lives are, what replacement costs, and when it falls due. Industry benchmarks — hospitality FF&E reserves, healthcare equipment cycles, processing equipment schedules — provide the anchor.
A pro forma that shows a flat 2% of revenue as a capital reserve without explaining where 2% came from is asserting rather than analysing, and a reviewer will treat it accordingly.
There is no universal minimum, and that matters
This is where most published guidance is simply wrong.
Search results, competitor pages and lender summaries routinely state that USDA requires a minimum DSCR of 1.25x, sometimes 1.20x, sometimes "1.0 to 1.25x." Part 5001 does not contain a general minimum coverage requirement for new lending.
What it contains instead is:
A definition — the one set out above.
Specific minimums in specific circumstances. §5001.102(d)(4)(iii) sets Agency-required minimum debt service coverage ratios in the refinancing context. Rural hospital refinancing carries its own test: the applicant must demonstrate that new annual debt repayment will be lower than existing payments, and that historical cash flow supports coverage of at least 1.1x.
A credit evaluation framework. §5001.202 requires the lender to conduct a credit evaluation, and the Agency to be satisfied by it. The feasibility analysis required at §5001.304(a) feeds that evaluation.
So where does 1.25x come from? From practice. It is where commercial lenders are comfortable, it is the level at which a project has enough cushion to survive a bad year, and it has become the working expectation across the guaranteed programmes. In that sense it is real — a project presenting at 1.05x will struggle regardless of what the regulation does or does not say.
But the distinction matters for three reasons.
It changes how you argue a marginal file. If 1.25x were a regulatory floor, a project at 1.18x would be ineligible. It is not ineligible. It is a credit judgment, which means it can be supported by other evidence — contracted revenue, guarantor strength, collateral coverage, a demonstrable ramp — or undermined by their absence.
It explains why lenders differ. Two lenders can apply different thresholds to the same project because the threshold is theirs to set within the Agency's satisfaction. Asking your lender what coverage they need, specifically, is a better question than assuming a number.
It signals competence. A feasibility study that cites "the USDA 1.25x requirement" is telling a reviewer that its author has not read the regulation. In an environment where the Agency has been explicit about underwriting quality, that is not a good first impression.
What the working thresholds actually look like
With the caveat above firmly in place, here is where projects sit in practice.
| Coverage | How it reads |
|---|---|
| 1.35x and above | Strong. Comfortable cushion, unlikely to attract coverage-specific questions. |
| 1.25x – 1.34x | The conventional target. Where most approved deals land. |
| 1.15x – 1.24x | Marginal. Financeable, but the file needs to carry supporting strength elsewhere — contracted revenue, experienced operator, meaningful equity, strong collateral. Expect questions. |
| 1.10x – 1.14x | Difficult. Achievable only with materially compensating factors and a lender willing to argue it. |
| Below 1.10x | Generally not financeable as presented. The structure needs to change — more equity, longer term, smaller loan, or a different project. |
Two things worth noting about this table.
It describes stabilised coverage, not year one. Almost no project covers debt service in its first months of operation, and USDA does not expect it to. What matters is when stabilisation occurs, how credible the ramp is, and whether the project survives the interval.
And it describes coverage computed on USDA's definition. A 1.28x on a conventional calculation may be a 1.14x on the regulatory one. Run it both ways before deciding where you sit.
Global coverage and the borrower's other debt
The definition specifies the annual debt service "of the borrower" — not of the project.
This trips up sponsors with multiple entities, existing facilities, or equipment financing sitting outside the proposed transaction. A project that covers its own debt service comfortably can fail on a global basis if the borrower carries obligations elsewhere.
Practically, this means the analysis needs to capture:
- The proposed guaranteed loan
- Any existing debt of the borrowing entity
- Related-party debt where the borrower is obligated
- Capital lease obligations, which are debt service in substance
- Seller notes, including those on standby, since standby status is a structural feature rather than a disappearance of the obligation
Where a sponsor operates several entities, expect the lender to want a consolidated view. Where the borrower is a single-purpose entity created for the transaction, the analysis is cleaner but the guarantors' obligations may still come into scope.
C-PACE financing deserves specific mention. The Rural Business-Cooperative Service Administrator's February 2026 letter to OneRD lenders addressed it directly, setting out expectations that C-PACE be treated as a prior lien with a collateral discount and that C-PACE assessment payments be included in the coverage analysis. A project financed partly through C-PACE that omits those assessments from debt service is understating its obligations.
Interest-only periods and ramp-up
§5001.402(b)(3) permits interest-only periods of up to 36 months. That flexibility is genuinely useful for construction projects and businesses with a lease-up or production ramp — and it is also a place where pro formas go wrong.
An interest-only period produces two different coverage pictures: coverage during the IO phase, when debt service is lower, and coverage after amortisation begins, when it steps up substantially. A projection that shows comfortable coverage throughout the IO period and then quietly ends before the step-up is not a projection; it is an omission.
The analysis needs to show both. Specifically:
- Coverage during the interest-only phase
- Coverage in the first full year of amortisation
- Whether the revenue ramp assumed to bridge the gap is credible on external evidence
The February 2026 letter to lenders was explicit on this point, listing among its underwriting expectations that interest-only and ramp-up periods be justified with realistic monthly cash-flow projections. Monthly, not annual. An annual projection can conceal a seasonal cash trough that a monthly one exposes, and for seasonal businesses — agritourism, recreation, harvest-linked processing — that trough is often where the risk actually sits.
How the analysis shifts by programme
The definition is constant across all four OneRD programmes. What changes is where the coverage comes from and what has to be proved to establish it.
Business & Industry
The default case. Coverage comes from operating cash flow, and the analytical burden falls on the revenue build — demonstrating demand in the trade area, defensible capture, and pricing supported by comparable evidence rather than by assumption.
Expense benchmarking matters as much as revenue. A projection with credible revenue and understated operating expenses produces a coverage ratio that will not survive contact with reality, and reviewers look specifically for expense lines that fall below industry norms without explanation.
Community Facilities
CF projects are underwritten on sustainability rather than profit, and that changes the coverage conversation.
A rural hospital, clinic, school or public safety facility is not trying to generate a return. It is trying to remain viable while delivering an essential service. Coverage still has to be demonstrated, but the revenue model being tested is different — payor mix for healthcare, enrolment for education, tax capacity or assessment base for municipal facilities.
The essentiality test sits alongside the coverage test. A facility that covers its debt service but cannot demonstrate community need has a different problem from one that demonstrates overwhelming need but cannot cover.
Note the rural hospital refinancing provision specifically: new annual debt repayment lower than existing, and historical cash flow supporting at least 1.1x coverage.
REAP
REAP coverage analysis has a distinctive structure: revenue is a physical production forecast multiplied by a price, and both halves require independent evidence.
The production forecast — kilowatt hours, therms, gallons, tons — must be technically supported, ideally by an independent resource assessment rather than by an equipment vendor's projection. The price must be supported by contract where one exists, or by defensible market evidence where it does not.
The incentive stack is where REAP coverage analysis most often overreaches. Projects frequently model tax credits, renewable energy credits, state incentives and utility payments into the coverage calculation. The analysis has to reconcile with and without the layers that are not contractually committed. A project that covers only when speculative incentive revenue is included does not cover.
Note also that REAP refinancing is capped: debt being refinanced cannot exceed 50% of the total use of funds in the new REAP guaranteed loan, and REAP loans must show other funds of at least 25%.
Water & Waste Disposal
WWD coverage rests on rate capacity — whether the served community can afford rates sufficient to service the debt. That makes affordability analysis central in a way it is not for the other programmes: median household income in the service area, comparable system rates, and the rate increase the project requires.
A technically sound system that requires rates the community cannot sustain does not have a coverage problem so much as a viability problem, and the analysis needs to say so.
Sensitivity, and what the Agency is actually testing
A single coverage figure at stabilisation is the least interesting number in the analysis. What the Agency and the lender want to know is what happens when the assumptions move.
Competent sensitivity analysis on a USDA file addresses, at minimum:
Revenue. What coverage looks like at 90%, 85% and 80% of projected revenue. The relevant question is where coverage crosses 1.00x — the point at which the project stops servicing its debt — and how plausible that revenue level is.
Interest rate. Terms are negotiated between lender and borrower and may be variable, with variable rates adjustable quarterly. A project underwritten at a rate that then moves against it is exposed, and the analysis should show by how much.
Cost overrun. For construction projects, what happens to coverage if delivered cost exceeds budget and the loan sizes up accordingly, or if the sponsor injects additional equity.
Timing. Delayed opening, slower ramp, later stabilisation. This is frequently the most damaging sensitivity and the least often modelled.
Key input costs. Feedstock for processing, energy for cold storage, labour for care facilities. Whichever line dominates the expense structure.
The purpose is not to demonstrate that the project survives every scenario. It is to show the reviewer that the risks have been identified and quantified rather than glossed. A study that presents only a base case is asking to be trusted; a study that presents a downside is asking to be checked. The second is more persuasive.
Where pro formas most often fail
Drawn from what draws deficiency letters.
Replacement capex omitted or unsupported. The single most common technical failure against the regulatory definition.
Coverage shown only at stabilisation. No view of the ramp, no view of the IO step-up, no view of year one.
Annual rather than monthly cash flow. Conceals seasonal troughs, and the February 2026 letter asked for monthly.
Expense lines below industry norms without explanation. Particularly labour, insurance, utilities and maintenance. Understating expenses is a more common failure than overstating revenue and is easier for a reviewer to spot.
Global debt ignored. Project coverage presented as if the borrower has no other obligations.
Speculative revenue in the base case. Uncommitted grants, unsigned offtake, incentive layers that have not been secured.
No working capital analysis. A project can cover debt service on paper and still fail because it runs out of cash in month seven. The February 2026 letter listed explicit working capital analysis among its underwriting expectations.
Reserves funded from the loan rather than from equity. Also addressed in that letter, which set the expectation that reserves be funded from borrower equity rather than from the Agency-guaranteed loan.
What has changed, and why the standard is tighter now
None of the above is new regulation. What has changed is the level of scrutiny applied to it.
USDA's guaranteed portfolio carries an active balance above $12 billion with delinquencies exceeding $1 billion, and the agency paid roughly $300 million in repurchases and losses across a recent twelve-month period. In February 2026 the Rural Business-Cooperative Service Administrator wrote to more than 775 OneRD lenders citing §5001.132 — the lender-removal authority — and setting out underwriting expectations. In May 2026 the agency removed ten lenders from the platform, holding roughly $620 million in delinquent loans between them.
The practical consequence for a borrower is that coverage analysis which would have passed loosely three years ago is now being read carefully. Lenders who remain on the platform have reason to be conservative, and a file with a thin or poorly supported coverage case is a file they have less appetite to defend.
That is not a reason for pessimism. It is a reason for the analysis to be right the first time.
Frequently asked questions
Sources
- 7 CFR Part 5001, OneRD Guarantee Loan Initiative, Subpart A, definitions.
- 7 CFR §5001.102(d), refinancing provisions and minimum coverage requirements.
- 7 CFR §5001.202, credit evaluation.
- 7 CFR §5001.304(a), financial feasibility analysis.
- 7 CFR §5001.402(b)(3), interest-only provisions.
- 7 CFR §5001.121(d)(14), REAP refinancing provisions.
- 7 CFR §5001.132, lender removal authority.
- USDA Rural Development, OneRD Guaranteed Loan Initiative Lender Guide.
- USDA Rural Business-Cooperative Service, letter to OneRD lenders, February 2026.
- USDA Rural Development, Community Facilities lender complete application checklist.
Prepared by the research team at usda-feasibility-study.com. This is practitioner guidance, not legal or regulatory advice; where this page and the regulation differ, the regulation governs. Last updated: 29 July 2026.
David Miller, CPA
USDA Feasibility, Inc.