Assisted Living Underwriting Method: Age and Income Qualified Demand, Licensing and the Class's SBA Record

An assisted living facility is a care business that happens to sit in a residential building. Its residents pay for help with bathing, dressing, medication and the rest of a day they can no longer manage alone, and the building earns nothing unless a licensed operator delivers that help to enough of them, at a price enough of them can pay. Demand therefore has to pass three screens, age, need and the ability to pay, and the result has to be checked against how many people in that position actually live in a care setting. The license defines what the facility may sell and who may operate it. And the class's record in the SBA loan files has to be read with the limits of its industry code in view.

This article sets out the method in that order. It writes the age and income qualified demand model as four equations and works them through for one Washington county with the Census Bureau's Vintage 2025 estimates and a national survey of residential care. It then lists the metrics a lender reads, prints the loan record of NAICS 623312 in the 7(a) and 504 programs through the dataset's cells, reads the federal eligibility rules and two state licensing statutes that gate the class, and closes with a small worked example of a memory care wing. The site's archived pages cover the ground around it: the long guide to the assisted living feasibility study, the state screen of the population aged 75 and over and the assisted living service page. This piece is about the arithmetic and the evidence behind each line.

What code 623312 holds, and what it leaves to other codes

The loan record in this article is the record of one industry code, so the first step is to read that code's definition. The 2022 NAICS Manual describes U.S. industry 623312 as establishments that provide residential and personal care services without nursing care, for older people or others who cannot fully care for themselves or who do not wish to live on their own ^[1]. The care it has in mind is room, board, supervision and help with daily living, housekeeping among it ^[2]. The neighbouring code is the one that matters for a reader of the loan file. NAICS 623311, continuing care retirement communities, expressly takes in assisted living facilities with on-site nursing care facilities ^[3].

The dataset follows that split. Its asset map gives the class its own code for assisted living without on-site nursing only, records that memory care has no code at all, and counts assisted living with nursing inside the skilled nursing class (623110) and the continuing care class (623311). Those two neighbouring cells are small beside the class itself: the 7(a) file holds 543 ^[4] skilled nursing approvals and 130 ^[5] continuing care approvals across FY2017 to FY2026, and each mixes nursing homes or retirement campuses with whatever assisted living carries a nursing code. Neither can be read as the record of assisted living with nursing, and this article does not try.

State licensing does not follow the NAICS line, which is the practical difficulty. Washington's statute defines the domiciliary care a licensed facility may give to include help with activities of daily living ^[6], and also health support services and intermittent nursing services provided directly or indirectly ^[7]. Florida issues assisted living licenses in four categories of care, one of which is limited nursing services ^[8]. A facility licensed in either state can therefore deliver some nursing and still be an assisted living facility under state law, while the federal classification places assisted living with on-site nursing facilities in another code ^[3]. The loan file records the code a lender chose; it does not record the license the borrower held. A study that leans on the class record should say which code it is reading, which license category the subject holds, and why the one stands in for the other. How codes become classes, and what happens where two industries share one code, is the subject of a separate data method piece.

The demand model in four equations

The demand side of the method rests on two estimates that check each other. The first works from the top down: it takes the share of each older age group that lives in residential care nationally and applies it to the local population. The second works from the bottom up: it screens the local population for age, for the need for daily help and for the ability to pay, and arrives at a qualified pool. Neither is enough alone. A top-down estimate inherits national habits that may not hold locally; a bottom-up pool says who could live in assisted living but not how many would choose to. Set side by side, the first calibrates the second.

Written as equations, with a for an age band (75 to 84, and 85 and older), the model is:

1. Use rate. r(a) = R × s(a) / N(a), where R is the national count of residential care residents on a survey day, s(a) the share of them in band a and N(a) the national population of band a.
2. Expected residents. E = sum over a of P(a) × r(a), where P(a) is the market area's population of band a in the stabilized year. Private pay residents are E(p) = E × (1 minus m), where m is the share of residents whose care Medicaid pays.
3. Qualified pool. Q = sum over a of P(a) × n(a) × i, where n(a) is the share of band a that needs the daily help an assisted living facility gives and i the share of those who can pay the private rate from income and assets.
4. Balance. The implied propensity is k = E(p) / Q. Penetration with the subject is t = (S + U) / Q, where S is competing licensed units and U the subject's units. The market occupancy the demand estimate implies is k / t, which equals E(p) / (S + U). The demand left for the subject is E(p) less the occupied competing units, and the subject's capture is its own stabilized occupied units over that residual.

The fourth line is the one a lender should hold on to. Penetration measures how much of the qualified pool the market's units would have to house; propensity measures how much of that pool, at national behaviour, actually lives in care. When penetration runs above propensity the market's units cannot all be full, and the ratio of the two is the occupancy the demand estimate predicts. Capture and penetration in general, for any class, are covered in the general demand methods piece; what follows is the version particular to assisted living, where the pool is screened for need as well as income and the propensity can be measured rather than guessed.

Every input to the model and to the worked example later in the article is in Table 1, sourced inputs with their citation and assumptions labelled as such.

Rounded as shown; lines may not foot by one unit in the last place.

The age screen: county cohorts from the Vintage 2025 estimates

The age screen starts with the Census Bureau's county characteristics file, which publishes each county's resident population by five-year age group for every year from the 2020 base. The file layout labels the April 1, 2020 base as YEAR 1 ^[29] and the July 1, 2025 estimate as YEAR 7 ^[30], and gives the three oldest groups the codes 16, 17 and 18, which run from age 75 up through the open-ended band of 85 and over ^[31]. The estimates are built forward from the 2020 base by adding births, subtracting deaths and adding net migration ^[32], so they are a modelled count, not a fresh enumeration, and a study should say so.

The example county is Spokane County, Washington. In the 2020 base it held 16,151 residents aged 75 to 79 ^[16], 10,128 aged 80 to 84 ^[17] and 10,285 aged 85 or older ^[18]. By July 1, 2025 those counts were 22,262 ^[19], 13,101 ^[20] and 11,337 ^[21], out of a total county population of 558,344 ^[33]. Table 2 combines the two younger groups into one band, converts each band's change into an annual rate over the five and a quarter years between the two dates, and carries the 2025 count forward three years to an assumed stabilized year.

The two bands moved very differently. The 75 to 84 band grew at 5.82 percent a year, the 85 and older band at 1.87 percent. That difference is the reason to keep the bands apart: a model that pooled everyone aged 75 and over would apply one blended rate to two groups whose use of residential care differs several times over, as the next section shows. The national residential care survey reports that 53 percent of residents were 85 or older and 31 percent were 75 to 84 ^[10], so the two bands in the model hold most of the people a facility for the elderly serves. Residents younger than 75 are left out of the model, which makes it slightly conservative.

The forward step is the weakest link in the age screen, and the table makes it visible by printing the held case beside the growth case. Extending a five-year rate for three more years assumes that whatever drove the past five keeps going. A study can do better by aging the cohorts it can already see: the people aged 70 to 74 in 2025 are, less deaths and net of migration, the 75 to 79 band of 2030. A study that relies on growth should show the cohort arithmetic, and a lender should read the held column as the case that needs no forecast at all.

The national use rate: one survey day, read by age

The top-down estimate needs a count of people living in residential care, by age. The National Center for Health Statistics supplies it from the residential care component of its post-acute and long-term care study. On a given day in 2022, about 1,016,400 residents lived in residential care communities ^[9]. The definition behind that count matters as much as the number. It takes in assisted living communities and also other residential care settings, such as personal care homes, adult care homes, board care homes and adult foster care, that meet the study's criteria ^[34], and one of those criteria is four or more licensed, certified or registered beds ^[35].

Dividing the residents in each band by the national population of the same band gives the use rate. The Census Bureau's national table by age group puts the population aged 75 to 79 at 11,161,285 on July 1, 2022 ^[13], aged 80 to 84 at 6,823,648 ^[14] and aged 85 and over at 6,412,663 ^[15]. Table 3 does the division.

About 8.40 of every 100 people aged 85 and older lived in residential care that day, against 1.75 of every 100 aged 75 to 84. The difference is the arithmetic case for the age split in the previous section. Three cautions travel with the rates. The resident shares are published as whole percentages, so the counts by band are rounded estimates. The survey day and the July 1 population date do not coincide exactly. And the resident count covers more settings than a state's assisted living license: a study that applies the national rate should count competing supply on the same footing, small adult family homes and personal care homes included, or scale the rate down to the settings it does count. Mixing a broad numerator with a narrow supply count overstates the demand left for the subject.

The last step of the top-down estimate removes the residents whose care Medicaid pays. About 17 percent of residential care residents were Medicaid beneficiaries ^[11], defined as residents with some or all of their long-term care paid by Medicaid in the 30 days before the interview ^[36]. A facility that will not accept Medicaid draws only on the rest. Applied to Spokane County in the stabilized year, the national rates give about 1,741 expected residents aged 75 and over and about 1,445 private pay residents, as Table 4 prints in its first column.

Need and income: the bottom-up pool

The bottom-up estimate asks a different question: how many people in the market need the help an assisted living facility sells and could pay for it. Need is the first screen. The American Community Survey asks about six types of disability, two of which bear directly on assisted living: self-care difficulty and independent living difficulty ^[37]. The first is defined as difficulty bathing or dressing ^[38]; the second as difficulty, because of a physical, mental or emotional problem, doing errands alone, such as visiting a doctor's office or shopping ^[39]. A study takes the county's counts for each age band from those questions and decides, as a stated judgment, what share of each band needs help at the level a facility gives rather than help a family or a home care agency can supply.

The resident profile is the check on that judgment. In the national survey, 62 percent of residential care residents needed help with three or more activities of daily living, 26 percent with one or two and 12 percent with none ^[40], and the most common needs were bathing at 75 percent and walking at 71 percent ^[41]. A need screen that admits people with no limitation at all would describe a pool unlike the people who actually live in care. The example assumes 8.0 percent of the 75 to 84 band and 25.0 percent of the 85 and older band pass the need screen; both are assumptions, labelled as such in Table 1, and a study replaces them with the county's own counts.

Income is the second screen, and it is where the published tables run out. The ACS put the national median income of households headed by someone aged 65 and older at $61,989 in 2025 ^[42]. The worked example later in this article assumes an all-inclusive memory care charge of $100,800 a year, as Table 10 prints, so the median older household could not pay it from income alone. That is why an assisted living income screen has to count assets, home equity above all, and why the share that passes is a judgment a study must state rather than a figure it can look up. There is a measurement trap as well: the ACS household income estimates exclude people living in institutions and other group quarters ^[43], so a study that tabulates the incomes of older residents should be clear about whom its table leaves out. The example assumes that 40.0 percent of the need-qualified pool can pay the private rate from income and assets.

Multiplying through, the stabilized-year pool in Spokane County is about 2,539 people who are old enough, need daily help and can pay. The top-down estimate expects about 1,445 of them to live in private pay residential care. The ratio, 56.9 percent, is the propensity the national rates imply. It is not an assumption: it is what the national resident count says about the pool the study screened, and if a study's own propensity assumption sits well above it, the study has to explain why its market behaves differently from the nation.

Reading the demand table: balance, penetration and the wing's share

Table 4 runs both estimates for the market area, which the example takes to be the whole county, and adds the memory care screen the worked example needs. Its four columns are the base case, the population held at its 2025 count, national use rates one tenth lower, and two dozen more memory care units added to the market's supply.

In the base case the market's assumed 1,550 competing units, occupied at the assumed 88.0 percent, house about 1,364 residents, leaving about 81 private pay residents of demand unserved. Penetration of the qualified pool, with the subject's wing counted, is 61.5 percent against a propensity of 56.9 percent, and the market occupancy the top-down estimate implies is 92.5 percent, the same figure reached as the ratio of the two. Hold the population at its 2025 count and the picture turns: the residual becomes a deficit of about 59 residents, penetration rises to 69.0 percent, and implied occupancy falls to 83.5 percent. Nothing in the market changed in that column except the forecast. A lender reading the table learns that the base case depends on three more years of growth in the 75 to 84 band, which is exactly the assumption the age screen flagged.

The memory care rows narrow the pool again. In the national survey, 44 percent of residential care residents had ever been diagnosed with Alzheimer disease or another dementia ^[12]. A diagnosis is not the same as a need for a secured unit, so the example assumes that half of those residents need one. The base case leaves about 47.9 residents of memory care demand after the occupied competing memory care units, and the twelve-unit wing, at its stabilized occupancy of 10.8 occupied units, needs 22.5 percent of that residual. The other columns show how fragile that share is. Held at 2025 counts the wing needs 63.4 percent of what is left; with national use rates one tenth lower, 67.1 percent; with 24 more memory care units in the pipeline, 41.1 percent. A share near a fifth of the residual leaves room for competitors to keep filling; a share above three fifths means the wing would have to take residents from them, and a study should say so in those words.

The supply side of the table is where the study's fieldwork shows. A study should name a source for every competing unit, its license category, its memory care certification and its occupancy, such as the state licensing agency's facility records and the study's own survey of each competitor. The method for counting competitors and the pipeline is in the supply and comparable analysis article, and the climb from opening to stabilized occupancy in the ramp-up and absorption article.

The metrics a lender reads in an assisted living file

Beyond demand, a study should put a short list of operating measures in front of a credit officer, each with its own evidence. Table 5 sets out the list with the published anchor a study can place beside each measure.

Occupancy. A study should show occupancy month by month, not as an annual average, and state the stabilized level it underwrites. The federal occupancy figure this article relies on belongs to a different program. Community Facilities guarantees require a financial feasibility report from a qualified firm or individual ^[46], and in that program's application rules a projection for an assisted living or comparable residential facility may not assume occupancy above 90 percent ^[44]. That cap does not bind a 7(a), 504 or B&I loan. A study may still adopt it as its own ceiling, as the worked example does, and should say that it is borrowing the figure rather than obeying it.

Charges. Where a facility prices in two parts, a base rate for housing and services and a care-level charge that rises with need, the study should show both. This article cites no published series of assisted living charges, and the example's charge is an assumption; a study builds the figure from the rate cards of the competitors it surveyed, net of concessions.

Payer mix and acuity. A resident ledger far from the national profile in Table 5, in either direction, needs an explanation in the study, because payer mix and acuity drive both revenue per resident and the staffing a license requires.

Labor. Payroll is the largest expense line the Service Annual Survey reports for the industry group that holds the class. For employer firms in NAICS 6233, the industry group that combines continuing care retirement communities and assisted living, gross annual payroll was $33,997 million in 2022 ^[25], fringe benefits $5,324 million ^[26], temporary and leased staff $1,766 million ^[27] and total expenses $76,484 million ^[28], all published in millions of dollars ^[47]. Table 6 does the division.

Labor in those three lines comes to 53.7 percent of the industry group's expenses, payroll alone 44.4 percent. The group mixes facilities with on-site nursing and facilities without it, so the figure cannot be read as the 623312 benchmark; it is the nearest published anchor, and a study should present it with that limit. A staffing plan that brings labor far below that share should rest on a schedule a credit officer can trace, hour by hour, to the license's staffing rules. The general method for an expense budget is in the operating expense evidence article.

License standing. The license can be lost, conditioned or left behind in a sale, so the study's file should hold the current license, its category, any memory care certification and the latest inspection reports.

The 623312 record in the 7(a) and 504 files

The class's loan record comes from two national cells, one for each program, covering approvals from FY2017 to FY2026. Every figure below is the record of NAICS 623312 alone, assisted living without nursing care, for the reasons the first section gave; assisted living lent under a nursing code is not in it.

Most of the class's lending runs through 7(a). Its cell holds 2,232 ^[48] approvals for $2,595,474,500 ^[48] in gross approved dollars. The median 7(a) loan is $752,450 ^[48], and half the loans fall between $323,750 ^[48] and $1,481,000 ^[48]. The 504 cell holds 505 ^[49] approvals for $640,943,000 ^[49], with a median of $807,000 ^[49] and a middle half from $379,000 ^[49] to $1,652,000 ^[49]. Those medians describe small transactions, and a study for a project of ordinary SBA size should compare its loan with that record.

Franchising is a minor part of the class. Franchise businesses account for 4.08 percent ^[48] of the 7(a) approvals and 2.97 percent ^[49] of the 504 approvals. Lending is spread widely: the 7(a) approvals carry 50 ^[48] distinct state codes and sit with 318 ^[48] distinct lenders (the file names the bank that holds or last held each loan), and the 504 approvals carry 43 ^[49] state codes and 92 ^[49] CDCs. In the two states whose statutes this article reads, Washington's state cell holds 248 ^[50] approvals across both programs and Florida's 277 ^[51].

The yearly series shows a class that kept borrowing through the whole decade. Table 7 prints approvals by fiscal year in each program.

The 7(a) count peaked in FY2021 at 307 ^[48] approvals and stood at 228 ^[48] in FY2026; the 504 count ranged from 37 ^[49] in FY2019 to 65 ^[49] in FY2025. The series counts approvals and carries no explanation for any of them, so it shows how much was lent each year and is silent on why the counts moved. Table 8 collects the class figures used in this article, each with its basis and its key.

Loan Data Behind This Article

Every figure in this table comes from the Loan Analytics loan dataset, analytics-loan-data freeze-3 (commit 360c856), and carries its key. SBA rates are approval-cohort floors and are withheld when fewer than 30 loans stand behind them.

Cancellation and charge-off floors for the class, against each program

Two outcomes are visible in the file. An approval can be cancelled before it disburses, and a disbursed loan can be charged off. Both rates in this section are approval-cohort floors: they count what has happened to date, and they can only rise as approvals still waiting to disburse cancel and as open loans resolve. Every rate here rests on 30 or more loans; none is withheld.

Cancellation. In the FY2017 to FY2024 approval cohort, the 7(a) cancellation floor for the class is 10.21 percent ^[48], on 176 ^[48] cancelled approvals out of 1,724 ^[48], against 11.24 percent ^[54] for the whole 7(a) program. Its ceiling, which counts every still-undisbursed approval as cancelled, is 10.38 percent ^[48], so the class floor is close to final. The 504 picture differs. The class's 504 cancellation floor is 15.30 percent ^[49], on 58 ^[49] of 379 ^[49] approvals, against 13.60 percent ^[55] for the 504 program, and with 18 ^[49] approvals still undisbursed its ceiling reaches 20.05 percent ^[49], against a program ceiling of 16.90 percent ^[56]. The 504 floor for the class sits above its program's floor, and the 7(a) floor below. The file shows only that an approval never disbursed. Its reason is not in the record, and a study should not supply one the data cannot support. A separate piece shows why the two endings have to be read apart.

Charge-off. Charge-offs are counted on approvals from FY2017 through FY2021, a cohort old enough that each loan has been outstanding for five years or more. In the class, 14 ^[48] of 971 ^[48] disbursed 7(a) loans have been charged off, a floor of 1.44 percent ^[48], set against 6.33 percent ^[52] across all 7(a) lending in that cohort. A second measure looks only at loans written off before their fifth anniversary, which puts every cohort on equal seasoning; on it the class shows 0.62 percent ^[48] and the program 3.72 percent ^[57]. Measured in dollars, charge-offs come to 0.34 percent ^[48] of the cohort's approved dollars, where the program's dollar figure is 1.95 percent ^[58]. The 504 cell shows 1 ^[49] charge-off among 179 ^[49] disbursed loans, a floor of 0.56 percent ^[49], while 504 lending overall shows 0.35 percent ^[53], and none of the class's 504 write-offs came inside five years of approval.

What can the floors say? They say that, so far, the class's 7(a) loans in that cohort have been written off at a fraction of the program's rate, on any of the three measures, while the 504 floor sits above its program's on the weight of a single loan. They cannot say the class is safe. The 7(a) cohort still holds 347 ^[48] open loans, their status undisclosed in the public file, and the 504 cohort 114 ^[49]; any of them can still fail, which is why the figure is a floor. With 14 ^[48] charge-offs behind the 7(a) rate and 1 ^[49] behind the 504 rate, a few loans either way would move them visibly. And the floors say nothing about cause: the dataset's report offers no explanation of where a class sits, and neither does this article. Why a figure built this way is a lower bound is the subject of the data method piece on cohort floors, and the cross-class reading of outcomes places assisted living beside its neighbours. For a study the use is narrow: the floors frame what has happened so far, and the study's own stress tests carry the rest.

Federal eligibility rules that reach an assisted living borrower

Size. SBA's size standard for NAICS 623312 is $23.5 million in annual receipts ^[59], measured for the concern and its affiliates together ^[60], so an operator that belongs to a group of facilities counts the group's receipts.

Non-profits. 13 CFR 120.110(a) makes non-profit businesses ineligible for SBA business loans, with for-profit subsidiaries eligible ^[61]. A study should state the borrower's legal form at the outset and, where a non-profit organisation stands behind the project, show that the borrower is a for-profit subsidiary.

Passive ownership and the operating company. A building owner that only collects rent is outside the program: the rule shuts out developers and landlords that hold property without using or occupying what the loan pays for, and admits only the exception it calls an Eligible Passive Company ^[62]. Where an assisted living property is held in two pieces, a property entity with the deed and an operating entity with the license and the staff, that shape fits the exception only while its conditions hold. The property entity may spend the proceeds on nothing but real or personal property that it rents to the operating entity for the operating entity's own business ^[63], or on a change of ownership between its own existing owners ^[64]; under 504, that change of ownership is open only if any other assets the property entity holds are directly related to the real estate, de minimis and kept out of the project financing ^[65]. The rent is capped at what the loan payment needs, plus the property entity's direct costs of holding the building, upkeep, insurance and taxes among them ^[66]. The lease cannot expire before the loan does, counting only renewal options the operator controls ^[67]. And the operator has to sign, as guarantor or as co-borrower, with co-borrower status mandatory in a 7(a) loan that also funds working capital or other assets for the operator ^[68]. A study should set the actual lease beside each of those tests.

Occupancy. How much of the building the borrower itself must use depends on whether the building already stands. Once the building exists, the borrower's own use must reach 51 percent, measured on rentable property, and tenants can take 49 percent at most ^[69]; in a building the loan constructs, the borrower's share starts at 60 percent and permanent leases cannot pass 20 percent ^[70], and space the borrower neither uses at once nor leases permanently has to be in its plan for its own permanent use: part of that space within three years ^[71], all of it within ten ^[72]. Where a property company rents a whole existing building to operating companies, it is the operating companies, taken together, that must clear the test ^[73]. A facility that licenses and staffs the whole building uses all of it; a mixed building with independent apartments or leased medical suites should be measured.

License and the illegal-activity bar. 13 CFR 120.110(h) makes ineligible any business engaged in an activity that is illegal under federal, state or local law ^[74]. Washington bars anyone from operating an assisted living facility without a license ^[75]. Read together, an unlicensed operation is an eligibility problem before it is a credit question, and a study should show the license, or the path to it, before it shows a single projection.

Credit and the study. The applicant, including an operating company, must be creditworthy, and the loan sound enough to reasonably assure repayment ^[76], and SBA can make a feasibility study one of the loan's conditions ^[77]. A 7(a) loan may run up to 25 years including extensions, and a portion that buys or improves real property may run longer by the construction period ^[23].

504 specifics. The 504 borrower's own stake in the project rises with two conditions. Ten percent is the base requirement ^[78]. The figure becomes 15 percent when the borrower, or its operating company, has two years of operation or fewer ^[79], or when the building counts as limited or single purpose ^[80], and the two conditions together raise it to 20 percent ^[78]. Nothing in the regulation says whether an assisted living building is limited or single purpose; a study should print the contribution both ways and leave the classification to the CDC. Fixed assets in a 504 project need a ten-year useful life and a fixed location, and furniture, equipment and furnishings can take short-term financing only as an essential, minor slice of the project ^[81]. The 504 method article covers the job test and the project split, and the 7(a) method article the rest of 13 CFR 120.

B&I. The USDA rule names the class, but narrowly. The eligible projects it lists include "assisted living facilities where constant medical care is provided and available onsite to the residents", next to nursing homes, while independent living facilities are left out ^[82]. A building that combines assisted care with independent senior apartments can still be eligible in two cases: when the residents of the independent units may take or decline the on-site medical services, or when people who need assisted living care make up the predominant share of residents ^[83]. The program also turns away apartments and other housing meant mainly for independent living ^[84], and the site has to be rural in the rule's own definition, which leaves out larger cities and the urbanized area next to them ^[85], though an area in that fringe judged rural in character stays in ^[86]; the judgment is made one project at a time, by the Under Secretary ^[87]. Because the B&I clause turns on constant medical care on site, while the code that holds the class is defined by care without nursing ^[1], the study behind a B&I request has to describe the facility's medical care in the rule's own terms, with evidence, rather than lean on the facility's label. Among the items an application may also have to carry, depending on the project ^[88], are an Affirmative Fair Housing Marketing Plan for five or more residential units, assisted living centers included ^[89], and a certificate of need where federal or state law requires one ^[90]. The B&I method article sets out the rest of 7 CFR 5001.

State licensing as the gate, read in two statutes

Federal rules decide whether a borrower is eligible; state law decides whether the facility may operate at all and what it may sell. The statutes of Washington and Florida show what a study has to read.

Washington. Chapter 18.20 RCW defines an assisted living facility as a home or institution, however named, maintained to provide housing, basic services and general responsibility for residents' safety and well-being ^[91], for seven or more residents after July 1, 2000 ^[92]; a facility licensed for three to six residents on or before that date keeps its license only while it stays continually licensed ^[93]. Independent senior housing and the independent living units of continuing care retirement communities are outside the definition ^[94]. No one may operate a facility within the definition without a license ^[75], and the triggers are specific: a license is needed when anyone other than an outside provider or family member takes on the activities the statute lists ^[95], among them help with activities of daily living, health support services and intermittent nursing services ^[96].

Three provisions matter most to a lender. A license may not run longer than twelve months ^[97], so a study should show the renewal date against the loan's closing and disbursement schedule. A license is issued only for the premises and persons named in the application and is not transferable or assignable ^[45]; the department sets by rule when a change in licensee is needed, a change in ownership or control among them ^[98], and the new licensee becomes responsible for correcting every violation that exists when its license issues ^[99]. A buyer of a Washington facility therefore takes on the duty to correct the violations that exist when its license issues, and a study for an acquisition should list them and price their correction. After July 1, 2026, no one may operate a memory care facility or unit in the state without certification ^[100]. Certification requires a valid assisted living license ^[101], no stop placement, resident care conditions, revocation or summary suspension before the application ^[102], and a staffing plan with awake staff 24 hours a day at a level adequate to residents' needs ^[103]. A memory care wing's revenue cannot start before its certification, and an enforcement record of the kinds the statute lists stands in the way of certification, so the study has to read the facility's survey record before it reads the wing's pro forma.

Florida. Florida requires a license issued by the agency to operate an assisted living facility ^[104], with separate licenses for facilities on separate premises even under one management ^[105]. Each license states its category of care: standard, extended congregate care, limited nursing services or limited mental health ^[8]. The extended congregate care category serves residents who would otherwise be disqualified from staying in an assisted living facility ^[106], and the full license goes to a facility licensed for 2 or more years that meets further conditions ^[107]. A facility licensed for less than 2 years can receive only a provisional license in that category, for no more than 6 months ^[108]. A new Florida facility that plans higher-acuity revenue therefore starts on a provisional footing, and a study should show what happens to its revenue if the full license does not follow.

The general lesson is that a study reads the subject's license as a list of what the revenue model may contain. Care-level charges must match the services the license category permits, memory care revenue waits for its certification, and in a purchase the timing of the new license and the cost of curing violations belong in the sources and uses and the closing schedule.

A worked mini-example: twelve memory care units on a licensed building

The example joins the demand model to a loan. An operator holds a licensed assisted living facility in Spokane County and proposes to add a secured wing of 12 memory care units, financed with a 7(a) loan. Only two inputs carry a source, the prime rate and the program floor; every other one in Table 1, the loan term included, is a labelled assumption. It is a sketch, not a full study; the Utah worked example carries a complete assisted living case through cost, revenue, debt and licensing for a purchase.

Conventions. This article follows the conventions of the Loan Analytics worked examples. Coverage is tested against the program floor the site's program page states, where the program has one, and against a stated 1.25x lender assumption, and the higher of the two binds. Equity is measured at the loan the cash flow supports, with fees recomputed on that loan. An acquisition is tested on the seller's historical earnings, not on projections.

The wing fills to the stabilized occupancy of 90.0 percent, the ceiling the study borrows from the Community Facilities rule ^[44], at an all-inclusive charge of $8,400 a month. Staffing follows the certification rule's requirement of awake staff around the clock ^[103]: the model assumes 48 caregiver hours a day at $33.00 an hour with burden, plus variable and fixed costs and a management fee. The requested $2,600,000 carries a variable rate of 9.75 percent, made up of the prime rate, 7.00 percent ^[22], plus a lender margin assumed at 2.75 percent. Above $350,000, a variable 7(a) loan's rate is capped 3.0 percentage points over its base rate ^[109], and the assumed margin sits inside that limit. Repayment is assumed to run on a 25-year schedule, the longest the regulation allows ^[23]. The existing building's mortgage stays in place, so the test is the combined cash flow against the combined debt. For a business expansion the program floor is 1.15x, as the site's SBA program page puts it ^[24]; the lender minimum is the stated 1.25x assumption, and whichever of the two is stricter governs. Table 9 runs the base case and four stresses.

In the base case the wing earns $273,848 of net operating income against $278,035 of debt service on its own loan, a coverage of 0.98x; the wing does not carry itself. Add the existing building's $330,000 and the facility covers both loans 1.17x. That clears the program floor, yet it sits under the lender minimum, the stated 1.25x assumption, and that minimum governs. At that minimum the combined cash flow supports a loan of $2,273,110, which is $326,890 short of the request. The stresses widen the gap quickly. With eight units in ten filled, combined coverage falls to 0.96x; a charge one tenth lower gives 0.97x, caregiver cost one tenth higher gives 1.05x, and two more points on prime give 1.07x. Caregiver labor alone is 53.1 percent of wing revenue in the base case, which is why a wage stress bites almost as hard as a revenue stress.

Table 10 turns the test around. The existing building covered its own mortgage 1.38x before the wing. To reach 1.00x combined, the wing needs 81.8 percent occupancy; to reach the binding minimum at the requested loan it needs 94.2 percent, above the 90.0 percent ceiling the study set for itself. The finding a study would report is therefore not that the wing fails but that the request exceeds what the combined income can carry at the lender's minimum: the loan the facility can bear is the supported loan, and the rest of the cost has to come from equity or from a lower budget. The equity and sources and uses article shows how the equity and the fees are restated at the supported loan, and the coverage article why the stricter of the two minimums governs.

The demand table warns from the market side: the wing needs 22.5 percent of the memory care demand left in the base case and 63.4 percent with the population held at 2025. A study that shows only the growth case and the 90.0 percent occupancy case shows the two most favourable readings at once.

Before credit committee: what the file should answer

A study for an assisted living loan is ready for a credit officer when it answers these questions from evidence in the file.

1. Which industry code does the cited loan record cover, and does the subject's license category fit it, or does the facility give nursing that belongs with the continuing care or skilled nursing records?
2. Do the age counts come from the latest county estimates by band, with the 85 and older band apart, and is the stabilized year shown with and without growth?
3. Is the study's propensity compared with the one the national use rates imply, and is any gap explained?
4. Does the supply count cover the same settings as the resident count behind the use rate, with a source for every competitor?
5. Are the need and income screens built from county counts of self-care and independent living difficulty and from the incomes and home equity of older households, each labelled as a judgment where it is one?
6. Is memory care demand screened separately, with the subject's capture shown under the held population and a pipeline case?
7. Do the care-level charges match what the license category allows, and does any memory care revenue wait for its certification ^[100]?
8. Is the borrower a for-profit concern ^[61] within the size standard ^[59], and if a property company owns the building, does the lease meet each condition for an Eligible Passive Company?
9. Has the study read the facility's inspection record, and in a purchase, listed the violations the new licensee will have to correct ^[99] and the date the new license can issue?
10. Is coverage tested against whichever is stricter, the program floor or the lender's minimum, with the break-even occupancy and the supported loan printed beside the request?

None of these is answered by a count of older residents within a radius; each is answered by a source a credit officer can open, placed next to the number it supports.

Related Pages on Loan Analytics

• The Loan Analytics methodology series (all articles)
• How a feasibility study is done: the Loan Analytics methodology
• SBA feasibility studies (7(a) and 504)
• USDA feasibility studies
• Bankable feasibility studies for conventional loans
• Assisted Living Feasibility Study
• Loan data by state
• Loan data by asset class
• Worked examples (case studies)
• Worked Example: SBA 7(a) Assisted Living Feasibility Study in Utah
• The Assisted Living Feasibility Study: A Complete Guide for Developers, Operators, Lenders, and Investors
• U.S. 75+ Population by State: The Senior Housing Demand Screen, Rebuilt for Underwriting

Commission a Feasibility Study

Loan Analytics prepares lender-grade feasibility studies for SBA 7(a), SBA 504, USDA Business and Industry, conventional and EB-5 financing, built on the methods this series sets out. A study is not an appraisal: it tests whether a project's market, costs and cash flow can carry the debt a lender is asked to approve. To discuss a study for your project, contact Loan Analytics.

Daniel Smith, MAI, Loan Analytics