Skilled Nursing Feasibility Study

A record that starts in FY1997

No skilled nursing loan is counted before FY1997. The table marks FY1991 to FY1996 not measured for 7(a) and 504 alike, because in SBA's files for those years the industry codes are too sparse to place loans in the class, and 504 stays unmeasured through FY1997. The first counts are small: 1 7(a) loan in FY1997, then 4 7(a) loans and 1 504 loan in FY1998, with their dollars printed as fewer than 5 loans. FY1999 is the first year with dollars in both programs, 23 7(a) loans for $11.6 million and 9 504 loans for $4.4 million. The full-period totals, 1,462 7(a) loans for $1.16 billion and 690 504 loans for $789.1 million, are therefore a record of FY1997 on. A study that tells a lender how long SBA lenders have financed nursing facilities should date the record from those years and should read the blank rows as unmeasured, not as years in which lenders turned the class away.

Steady 7(a) counts, rising 7(a) dollars

From FY2001 on, the 7(a) count stays inside a narrow channel: never fewer than 37 approvals, the FY2009 low, and never more than 78, the FY2023 high. The dollars did not stay put. Between FY2001 and FY2009 annual 7(a) totals ran from $15.0 million to $31.9 million; between FY2011 and FY2016 they ran from $46.2 million to $61.1 million; and every year printed from FY2021 on carries at least $54.1 million. A similar number of loans now moves far more money. FY2005's 67 approvals totalled $31.9 million, while FY2025's 63 totalled $67.7 million, and FY2026's 45 approvals carried $65.6 million, more than the $65.4 million on FY2023's 78. A lender will set a 7(a) request against the recent rows, not against full-period totals that blend loans from the early 2000s with today's, and a study arguing that a request is ordinary for the class should quote those recent rows.

The years when 504 carried more

The 504 program was not always the smaller channel. It lent more dollars than 7(a) in FY2003, FY2005, FY2006, FY2007, FY2009, FY2010 and FY2011, and in FY2007 it also made more loans, 54 against 51, for $40.4 million against $20.7 million. Its dollar peak came in FY2011, $52.0 million on 46 loans, and FY2012 and FY2013 stayed close, at $51.2 million on 37 loans and $48.3 million on 30. From FY2012 on, 7(a) has carried more dollars in every year that prints both totals. For a sponsor weighing 504 to buy or build a facility, the record settles a lender's doubt about 504 precedent for nursing facilities: through the 2000s 504 repeatedly passed 7(a) on dollars, and its FY2011 total is still the largest of any 504 year.

504 after FY2013: fewer loans, larger ones

Then the 504 count fell away. It dropped from 30 in FY2013 to 14 in FY2014, the same count as FY2002, and no year since has passed 28, the FY2023 figure; FY2017 and FY2020 printed 8 each. Across FY2003 to FY2013 it had reached 30 or more in every year but FY2008 and FY2009. The money held up better than the count: FY2023's 28 loans carried $51.0 million, just under the FY2011 peak reached on 46, and FY2026's 26 loans carried $42.8 million. The FY2017 to FY2026 window holds 178 of the record's 690 504 loans and $276.6 million of its $789.1 million. A 504 loan to a nursing facility is now a larger and rarer loan than it was in the FY2000s, so a 504 lender has fewer recent comparables inside the class and will look harder at the study's own evidence of census and coverage.

Dips in the record, and what followed

The record carries its own stress tests. In FY2008 the 504 count fell from 54 to 23, and in FY2009 7(a) fell to 37 approvals for $15.0 million, its lowest count since FY2000; by FY2011 both programs had passed their FY2007 dollar totals. In FY2018 7(a) approvals fell from 60 to 39, and FY2020 printed 39 again, before FY2021 brought 65 approvals and $87.6 million, more 7(a) money than in any other year of the table, followed by FY2023's 78 approvals, the largest count. The charge-off floors on this page cover loans disbursed in fiscal 2017 through 2021, the span of that dip and rebound, and they stay low. A lender will still ask how this facility, not the class, came through its own weak years: the study should show census, payer mix and coverage for every year of the operator's history, and should say what changed when a year fell short.

Approvals, not funded loans

Every count in this table is a gross approval, cancelled approvals included, and the cancellation floors say that no fewer than 16.1% of the FY2017 to FY2024 7(a) approvals, and 17.3% of the 504 ones, never went forward. A year's count can therefore include approvals that were later cancelled, so it is not a count of loans that funded. The table also withholds the dollar totals for FY2017 and FY2020 in both programs, so a reader should not fill those cells from other figures. Set side by side, the windows tell the program story: FY2017 to FY2026 holds 543 of the 1,462 7(a) approvals and $573.6 million of the $1.16 billion, while 504 holds 178 of 690 and $276.6 million of $789.1 million, so the recent window carries a larger part of the 7(a) dollar record than of the 504 one. A study that quotes class volume should say approvals, name its window, and keep the 2000s apart from the present.

Los Angeles within California

The Los Angeles-Long Beach-Anaheim area leads the metro table with 47 loans and $61.7 million, and its $804,000 median is higher than California's $745,000 and higher than any state median in the table. Most of California's 147 loans and $178.0 million were therefore made outside the Los Angeles area, and by the medians they run smaller than the ones inside it. The metro's charge-off cell reads not reported, because the Los Angeles cohort is too small to carry one. For a Los Angeles facility the lender has a metro record to compare with, and a request near $804,000 is ordinary there; for a facility elsewhere in California the state median is the closer guide, and the study should not borrow the Los Angeles figures to make a smaller market look larger.

Minneapolis-St. Paul across the state line

The Minneapolis-St. Paul-Bloomington area, which reaches from Minnesota into Wisconsin, follows with 34 loans and $29.5 million at a $620,500 median, below Minnesota's $720,000. Minnesota shows 53 loans and Wisconsin 23, so the metro row draws on both states, and its lower median suggests that the larger Minnesota loans were not concentrated in the Twin Cities. Its charge-off floor is printed as withheld: the site withholds a value that rests on fewer than 5 loans or would let one be derived, so the cell is neither a zero nor a number a study can quote. A study for a Twin Cities facility should benchmark against the metro row; one for a facility elsewhere in Minnesota or Wisconsin should say why the metro figures do or do not apply to its market.

States without a metro row

No other metro area has a row. Texas with 46 loans, Washington with 41, and Florida and Michigan with 37 each have none, so their records are not shown by metro on this page, and neither is the rest of California or Minnesota. A study in those places cannot cite a metro median for skilled nursing; it has to build the local comparison itself, from the competing facilities in the market area, their beds, star ratings and inspection histories, and the hospitals that discharge to them. That local work is also what a lender with a metro row will want, because 47 loans across the Los Angeles area or 34 across the Twin Cities still say little about one facility's census.

Where the programs overlap

The quartiles separate the programs without dividing them. The 7(a) quartile marks are $150,000, $500,000 and $1.4 million; the 504 marks are $531,250, $1.1 million and $2.1 million. Even the bottom 504 mark, $531,250, is higher than the 7(a) median, yet the 504 median of $1.1 million is lower than the top 7(a) mark, so between $531,250 and $1.4 million both programs are well represented. A request in that range will draw the question of why one program and not the other, and the study should give the lender the facts that answer it: whether the loan buys or builds real estate, how much of the value lies in the going concern and the licenses, and what the facility's history supports.

7(a): the loan count sits low, the money sits high

In 7(a) the smallest band holds the most loans, 148 at $150,000 or less, yet carries only $10.7 million. The band from over $2 million to $5 million holds 98 loans and $339.2 million, more than all the other 7(a) bands together, while the band above $5 million is empty in 7(a), at 0 loans and $0. The middle bands fill the gap: 97 loans for $26.7 million from over $150,000 to $350,000, 121 loans for $82.4 million from over $350,000 to $1 million and 79 loans for $114.6 million from over $1 million to $2 million. A 7(a) skilled nursing book is really two books: the many loans at or under $150,000, and the 98 in the band from over $2 million to $5 million that hold the bulk of the money. The study for a large request is read against the 98 loans of its band, where the lender's exposure is greatest, and it has to be the full market and operating study.

504: weighted to the middle and upper bands

504 has few small loans: 4 at $150,000 or less, with dollars printed as fewer than 5 loans, and 22 from over $150,000 to $350,000 for $5.7 million. Its largest count is the band from over $350,000 to $1 million, 54 loans for $33.4 million, followed by 50 loans for $70.4 million from over $1 million to $2 million and 46 for $156.0 million from over $2 million to $5 million, the largest 504 dollar band. Above $5 million, 504 shows 2 loans, their dollars again printed as fewer than 5 loans, where 7(a) shows none. A 504 request above $5 million therefore joins only those 2 precedents in the class, and its lender will expect a study that carries the size without leaning on comparables the record cannot supply.

State medians against the national quartiles

Every state median the page prints lies between the 7(a) lower quartile of $150,000 and the 504 median of $1.1 million: from $300,000 in Ohio and $400,000 in Washington through $568,500 in Arizona and $588,000 in Michigan to $701,500 in Texas, $715,000 in Florida, $720,000 in Minnesota and $745,000 in California. Even the Los Angeles area's $804,000 stays under the 504 median. No state's typical skilled nursing loan reaches the typical 504 loan, so a 504 request at $1.1 million or more is a large loan in every state with a median, and a 7(a) request above $1.4 million sits in the top quarter of the class's 7(a) loans nationally and above every state median. The study should name both comparisons, because a lender will make them.

Cancellation is the common failure

Both programs cancel far more often than they lose money. The FY2017 to FY2024 approvals show cancellation floors of 16.1% in 7(a) and 17.3% in 504, while the charge-off floors are 0.94% and 0.00%, so for a sponsor the realistic risk is an approval that never closes. In a nursing facility transaction the conditions that can stall a closing are largely regulatory: the state license, the Medicare change of ownership, and any operator lease or management agreement the deal relies on. A study cannot grant those approvals, but it can list each one the transaction needs, show which of the lender's conditions depend on it, and test the operating case on terms the new operator can actually meet, so that the commitment does not rest on an assumption the closing will overturn.

Charge-offs: low, and not only early

The charge-off floors are low, but the 7(a) losses are not confined to the early years. Loans disbursed in fiscal 2017 through 2021 show a 7(a) floor of 0.94% to date against 0.47% charged off within 5 years, which points to losses arriving after a loan's fifth year as well as before it. The 504 floors are 0.00% on both measures, a floor rather than a guarantee, because the cohort is still ageing. For a lender the risk in a skilled nursing loan is therefore not confined to the ramp: a study should test coverage beyond the first years, against a lower Medicaid rate, a higher staffing cost or a weaker inspection result, and should show the census and payer mix at which coverage fails, not only the case in which it holds.