Every bed was full. Half the payers barely covered the bed.

A regional operator under contract to acquire a 120-bed skilled nursing facility, seeking HUD 232 mortgage insurance on the acquisition. The building was twenty-two years old, well maintained, three-star rated, and running at 94 percent census. The trailing twelve months showed revenue of $14.3 million and the seller's broker had built a pro forma reaching $16.1 million by year three on modest rate growth and no occupancy assumption at all, because there was nowhere left to go.

The borrower's argument was the one every high-census file makes. The building is full. In a sector where the national average sat at 86.7 percent in the first quarter of 2026, the highest reading since 2016 after twenty consecutive quarters of gains (1), a facility at 94 percent is comfortably above market. Demand is proven. The only question is price.

That argument is wrong, and it is wrong in a way that is specific to this asset class. Census tells a lender how many beds are occupied. It says nothing about what the people in them pay.

HUD requires a third-party market study on the risk-bearing Section 232 routes, meaning new construction and substantial rehabilitation, and acquisition or refinance under Section 223(f) (2). The streamlined 223(a)(7) refinance of a loan HUD already insures does not require one, which is the sensible exception: the agency is refinancing an exposure it already holds.

On an acquisition the study is not a formality. Section 232 underwrites to a minimum 1.45 times debt service coverage on the lender's underwritten net operating income, and skilled nursing supports a maximum 80 percent loan to value, rising to 85 percent for a qualified non-profit sponsor (3). At 1.45 times coverage there is no cushion for a revenue assumption that does not hold. Every dollar of net operating income the study will not support removes roughly $1.45 of debt service capacity and, at prevailing rates, several dollars of loan.

So the question the study had to answer was not whether the market could fill the building. It plainly could. The question was whether the revenue filling it was durable enough to underwrite.

Skilled nursing is not one business at one price. It is four businesses sharing a corridor, and they pay very differently.

The cleanest publicly verifiable illustration of the spread comes from a large publicly traded operator's audited annual report, which discloses average daily revenue by payor across its portfolio: $767.72 for Medicare fee-for-service, $555.37 for managed care, $294.78 for Medicaid, and $280.24 for private and other payors (4). That is a Medicare-to-Medicaid ratio of roughly 2.6 to 1. Independent triangulation puts the median traditional Medicare payment near $556 per day (5). The absolute figures run above sector median because the operator is a strong one, but the ratio is the point and the ratio is stable.

Against that, the federal cost picture is unambiguous. The Department of Health and Human Services examined Medicaid payment against reported cost and found a mean payment of $198 per resident day against a mean cost of $253, a payment-to-cost ratio near 0.82 (6). Around 40 percent of facilities recovered 80 percent or less of their Medicaid cost, 52 percent recovered between 80 and 100 percent, and only 8 percent recovered more than cost (6). The underlying data is from 2019, so if anything it understates the current gap after several years of wage and supply inflation.

Read those two findings together and the sector's economics resolve. The Medicare Payment Advisory Commission reported a 22 percent margin on fee-for-service Medicare for freestanding facilities in 2023, and a negative 4.1 percent margin on everything else, which is to say on Medicaid and managed care combined (7). The all-payer total margin across freestanding skilled nursing was 0.4 percent, with 47 percent of facilities running negative total margins (7).

That is the number that should stop a credit officer. A sector at record occupancy is producing an all-payer margin of four tenths of one percent, and nearly half of its facilities lose money. Occupancy and profitability have come apart, and the mechanism is the mix.

The seller's operating history showed the following census composition, weighted across the trailing twelve months:

Three-quarters Medicaid, and a Medicare share less than half the level a rehabilitation-oriented facility would carry. The building was full because it was the affordable option in a county with limited alternatives, not because it was winning short-stay rehabilitation referrals from the regional hospital.

The comparison the study drew was against a facility in the same market with an identical bed count and an identical 90 percent census, but a skilled mix: 25 percent Medicare fee-for-service, 20 percent managed care, 15 percent private, 40 percent Medicaid.

At 120 beds and 90 percent occupancy, both produce 39,420 resident days a year. The blended rate does not survive the comparison.

Same beds. Same census. A revenue gap of roughly $3.9 million, or 27 percent, arising entirely from who is paying. With a broadly comparable cost base, almost all of that gap falls through to net operating income, and at 1.45 times coverage it removes several million dollars of supportable HUD-insured debt.

This is the same category of error as counting a self-storage facility by physical occupancy when the collected rent tells a different story, an issue examined in the stabilised self-storage acquisition where the facility was ninety percent full and the money said seventy. The unit of analysis has to be revenue, not units.

The seller's projection assumed the mix held and rates grew. Neither assumption survived scrutiny.

Medicare Advantage is displacing fee-for-service, at a discount. Managed care now covers a majority of Medicare enrolment, and the Commission's own framing is that plan rates for skilled nursing run 20 percent or more below fee-for-service (8). It also compresses length of stay materially, with one quality improvement contractor measuring 26 days under Advantage against 35 to 44 under traditional Medicare (9). The federal Inspector General found plans denying 12 percent of skilled nursing admission requests, with 95 percent of appealed denials overturned (10), which tells you the denials are not clinically driven so much as administratively costly. A facility with a small Medicare share and a growing Advantage share is watching its highest-rate business shrink on both price and duration.

The Medicaid funding mechanism is being narrowed. Federal reconciliation legislation reduced federal Medicaid spending by roughly $911 billion through 2034 and phases the provider tax safe harbour down from 6 percent to 3.5 percent for expansion states (11). Nursing facilities are expressly exempt from that phase-down, which is the counterintuitive detail most commentary misses, but the exemption freezes rates at existing levels rather than protecting growth, and state-directed payments, the other mechanism states use to lift nursing home rates toward Medicare levels, are capped and phase down from 2028 (11). States are already acting: Idaho enacted a 4 percent across-the-board reduction for the 2026 fiscal year, Colorado reversed a planned increase, and North Carolina cut rates effective October 2025 (12).

For a facility at 75 percent Medicaid, those are not background conditions. They are the revenue line.

Working the other way, and the study said so plainly, the federal minimum staffing rule was repealed effective February 2026 with a legislative moratorium running to 2035, removing an estimated $431 million a year in sector-wide cost that a 2024-vintage projection would have had to carry (13). Agency staffing has also normalised, down roughly 44 percent from its late-2022 peak to 5.6 percent of the workforce (14). The cost side improved. It did not improve by 27 percent.

The study did not conclude that the facility was a bad asset. It concluded that the revenue was correctly stated and structurally capped, and that the requested loan had been sized against a revenue base the mix could not sustain.

Three findings carried it.

First, the trailing revenue was real but not growable through occupancy, because occupancy was already near the physical ceiling. Any growth had to come from rate or mix, and rate on 75 percent of the census is set by a state legislature.

Second, the county's hospital discharge patterns did not support a mix improvement plan. The regional system's preferred post-acute network favoured two four-star facilities twenty minutes away, and the subject's three-star rating kept it outside that network. Mix improvement in skilled nursing runs through referral relationships, and referral relationships run through quality ratings. A plan to raise the Medicare share without first raising the star rating is a plan without a mechanism.

Third, and this is the finding that resized the loan, the appropriate underwriting basis was the trailing mix rather than the projected mix. HUD underwrites to documented performance. There is no published mechanism permitting a Section 232 loan to be sized on a payor mix the facility has not yet achieved, and the study declined to supply one.

A skilled nursing market study is not a demographic exercise, and treating it as one is the most common methodological error in the asset class. Independent and assisted living draw on an age and income qualified population who choose to move. Skilled nursing draws on hospital discharges, which means the primary market area is a referral catchment rather than a residential one.

The study therefore defined the market from patient origin data rather than from a radius. Where residents come from is a function of which hospitals discharge to the facility, and hospital catchments do not respect drive-time rings. In this county the regional system discharged across three counties, and two of the subject's nominal competitors, both within eight miles, drew from a different system entirely and were effectively not competing for the same referrals.

Supply was counted in licensed and staffed beds rather than licensed beds alone, because a facility that cannot staff a wing is not supplying capacity into the market regardless of what its licence says. That distinction mattered here: two competitors were operating below licensed capacity for staffing reasons, which inflated the apparent supply and, if taken at face value, would have understated the demand available to the subject.

Certificate of need was the third supply variable. Roughly two-thirds of states plus the District of Columbia regulate nursing home bed additions, and the subject sat in a certificate of need state. That cuts both ways in a credit file and the study said so. It constrains new competitive supply, which is favourable and is part of why occupancy in the market ran above national levels. It also constrains the subject, since the facility cannot add licensed beds without state approval, so there is no growth path through capacity even if the market would absorb it.

The fourth variable was quality, which in this asset class is a demand input rather than a soft consideration. Federal five star ratings, built from health inspections, staffing measures, and quality measures, determine whether a facility appears in a hospital system's preferred post-acute network and in the narrow networks that managed care plans operate. The subject's three star rating was the mechanism keeping its Medicare share low, and it is the reason the study concluded that no mix improvement was available without a quality improvement first.

The named data sources behind that analysis are all public and a reviewer can rebuild the work: the federal Care Compare files for ratings, staffing and deficiencies, Medicare cost reports for the operating history, the minimum data set for acuity and case mix, state licensure and certificate of need filings for supply and the approved but unbuilt pipeline, and hospital discharge data where the state publishes it.

The deal did not die. It resized, which is the ordinary outcome when a market study holds and the borrower still wants the asset.

The lender sized to the trailing mix rather than the pro forma, which cut the loan by roughly a third against the request. The sponsor closed the gap with additional equity and a seller note, and the seller accepted a lower price once the coverage arithmetic was put in front of both parties in the same document.

Three structural changes went into the credit alongside the resizing. The operating lease was rewritten with a stronger regional operator carrying a documented record of star-rating improvement, on the reasoning that mix improvement is an operator capability rather than a market condition. A capital plan funded the physical-plant items driving two of the facility's survey deficiencies, on the same reasoning. And the projection was rebuilt with an explicit downside case reflecting a state rate freeze rather than the assumed annual increase, because for a facility at three-quarters Medicaid a freeze is not a stress scenario, it is a plausible base case.

Census is a capacity measure and it is being read as a credit measure. In an asset class where the national all-payer margin is 0.4 percent at record occupancy, and where 47 percent of facilities lose money while the sector is fuller than it has been in a decade, occupancy has almost no information content on its own.

The reviewable version of this analysis is short. State the payor mix by share of resident days. State the rate for each payor from the operator's own history rather than a national average. Blend them. Compare the blended rate against a same-census facility with a different mix, so the reader can see that the difference is composition rather than performance. Then test whether any proposed mix improvement has a mechanism behind it, which in skilled nursing means a referral relationship, which means a quality rating.

A facility can be full, well run, genuinely needed in its county, and still unable to carry the loan the seller wants. Those statements are not in tension. They describe most of the sector.

For the same asset class analysed from the collateral side rather than the revenue side, see the memory care community where half the value walked in and out the front door. For the labour cost analysis that sits underneath any senior care projection, see the memory care community where the schedule that filled the building set the margin. And for the demand-side companion, where the qualifying population rather than the headline population governed, see the assisted living development tested on income-qualified penetration.

National Investment Center data on freestanding skilled nursing occupancy, first quarter 2026, reported in the long-term care trade press.

HUD Section 232 Handbook, Section II, Chapter 5, market study requirement by route.

HUD Section 232 Handbook, Section II, Chapter 3, loan sizing; loan to value and debt service coverage thresholds as published in lender term sheets for the programme.

The Ensign Group, Inc., annual report for the 2024 fiscal year, average daily revenue by payor source. Audited, filed with the Securities and Exchange Commission. A single large for-profit operator, so absolute rates run above sector median.

Penn Leonard Davis Institute analysis citing Medicare Payment Advisory Commission data, median fee-for-service payment per day, 2021.

US Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation, assessment of Medicaid payments and costs in nursing homes, published October 2024 using 2019 cost data.

Medicare Payment Advisory Commission, Report to the Congress, March 2025, Chapter 6, skilled nursing facility margins.

Medicare Payment Advisory Commission presentation on Medicare Advantage and fee-for-service payment differentials, July 2024.

Quality improvement organisation analysis of length of stay under Medicare Advantage against traditional Medicare, reported 2025.

US Department of Health and Human Services, Office of Inspector General, report on Medicare Advantage prior authorisation denials for skilled nursing facility admission, June 2026.

Federal reconciliation legislation enacted July 2025; Congressional Budget Office scoring and subsequent agency guidance on provider tax and state-directed payment limits, including the nursing facility exemption from the provider tax phase-down.

State Medicaid rate actions for the 2026 fiscal year as reported by non-profit health policy researchers and the sector trade press.

Interim final rule repealing federal minimum staffing standards for long-term care facilities, effective February 2026, with the statutory moratorium running to 2035.

Payroll-based journal data on agency staffing as a share of the skilled nursing workforce, reported 2026.

Method note. Rates and margins above are drawn from published sources of differing type: agency and commission data where available, and one audited operator filing where a payor-level rate table was needed. The worked comparison uses those published rates applied to two constructed census profiles. It is an illustration of the mechanism, not a valuation of any facility.