Short-Term Rental Underwriting Method: Nightly Rates, Local Registration Rules and Why the SBA Record Is Silent

A short-term rental earns its income a night at a time from guests who stay a few days and leave, in a building that was usually designed as somebody's home. That mix is what makes the asset awkward to underwrite. The revenue behaves like a small hotel's, the collateral looks like a house or a cottage, and the right to sell the nights rests on local law that can change after the loan closes. What follows is the study method for the class: the revenue equations and the evidence behind each input, the figures a lender reads, the registration and tax rules that decide whether the nights may be sold at all, the eligibility rules that make most of these properties passive in SBA's terms, and what the neighbouring loan records can and cannot tell a lender about a class the dataset does not hold.

It is written for the lender's reader and stays at the level of method. The full application, a cluster of new cabins in the Catskills financed in the example by one bank loan that covers both the build and the years after it, is the worked example Worked Example: Conventional Community Bank Construction-to-Permanent Short-Term Rental Feasibility Study in New York. The general methods have their own articles in this series: penetration and capture in Trade Area and Demand Methods: Capture Rates, Penetration Rates and Gravity Models, the coverage ratio in Debt Service Coverage in a Feasibility Study: How the Ratio Is Built, and Which Floor Binds, and shocks in Sensitivity and Stress Testing: Break-Even Points, Shock Sizes and What a Lender Should See. The words here go to what is particular to a dwelling sold by the night.

Why the Loan Dataset Holds No Short-Term Rental Class

The Loan Analytics dataset sorts SBA approvals into asset classes by the industry code each loan carries, through the fixed table described in From NAICS Code to Asset Class: How SBA Loans Are Assigned, and Where Two Industries Share One Code. No row of that table is a short-term rental class, and the reason starts with the codes. The industry system classifies activities, and a house let by the night can be described as more than one activity. The alphabetical index of the 2022 manual lists cottage rental or leasing under 531110 and housekeeping cottages under 721199 ^[1].

Each of those codes has a definition that fits some vacation rentals and not others. Industry 531110 is made up of lessors of buildings used as residences or dwellings, single-family homes among them ^[2], and the manual adds that such an establishment may run the property itself or have another establishment manage it ^[3]. Industry 721199 is the residual lodging code, holding short-term lodging that is not a hotel, a motel, a casino hotel or a bed-and-breakfast inn ^[4]. Its neighbour, the bed-and-breakfast code, covers lodging in private homes or small buildings converted for the purpose ^[5], and the hotel code, 721110, covers hotels, motor hotels, resort hotels and motels ^[6]. Read together, the definitions leave room for an owner renting a cottage to travellers to describe the business as letting a dwelling or as running traveller lodging, and nothing in the loan file shows which description a given borrower chose.

The asset map does not settle the question, and it was not built to. Industry 531110 is the whole of the multifamily class, so an SBA loan to a home-rental business coded there is counted with apartment lessors and cannot be picked out of that count. Industries 721191 and 721199 belong to no class at all, so a loan coded to either stands outside every class figure. No field in the public files marks a loan as a nightly rental, and the dataset never reads borrower names or loan descriptions to guess one. For a reader that leaves one plain consequence: the dataset can neither count SBA loans to short-term rentals nor say how they performed.

The program rules explain why few such loans would be there to count even with a code of their own. Under 13 CFR 120.110(c) a business is ineligible when it is a passive one, owned by a developer or landlord who neither uses nor occupies what the loan buys or improves, unless it meets the conditions for an Eligible Passive Company ^[7]. A separate restriction, 13 CFR 120.130(d), keeps proceeds away from "Investments in real or personal property acquired and held primarily for sale, lease, or investment," again excepting a loan to an Eligible Passive Company and one to a small contractor under 120.310 ^[8].

The exception is narrow. Section 120.111 confines such a company to property it rents to one or more Operating Companies, which use it to carry on their business, or to a buyout that shifts ownership among the people who already own it ^[9], and it holds the rent to the loan payment plus the direct costs of holding the property ^[10]. A real estate company leasing its building to its own operating affiliate fits that shape. On the text of the section, travellers booking a weekend are not Operating Companies leasing the property, so the exception does not describe an investor letting cottages to the public.

That is why the title calls the record silent rather than good or bad. The silence comes from a coding system and an eligibility rule; it is not a measurement of risk, and it would be a mistake to read the absence of losses as safety or the absence of loans as a verdict. The general limits of the public file are treated in What the SBA Loan File Cannot Tell You: The Limits of FOIA Data in Underwriting; for this class the limit is complete, and the study has to carry the whole evidentiary load itself.

The Revenue Equations: Nights Offered, Nights Sold and the Fees Around Them

Every projection for the class reduces to a short chain of products, and a study should write the chain out so the lender can see which link carries the weight. Stated in words, with the illustration's inputs in Table 1 and the chain worked through in Table 2:

• Available unit-nights equal the number of units times the nights each unit is offered in the year.
• Property occupancy equals the occupancy of a comparable listing set times the property's penetration index.
• Booked nights equal available unit-nights times property occupancy, or the nights local law permits, whichever is lower.
• Nightly revenue equals booked nights times the average nightly rate, a rate that excludes the cleaning fee and every tax.
• Stays equal booked nights divided by the average stay, and every stay is one turnover.
• Cleaning fee revenue equals stays times the fee, and total revenue is nightly revenue plus cleaning fees.
• Revenue per available unit-night equals nightly revenue divided by available unit-nights.

Three links in that chain are peculiar to the class. The first is the nights offered. A hotel sells every room on every night it opens, but a rental owner may hold weeks back for repairs or for personal use. Occupancy measured against one base cannot be multiplied into a different one, so the study states which nights are offered and why, and it applies a comparable set's occupancy only to the same kind of nights.

The second is the length of stay. It does not move the nightly rate, but it moves cost, because each stay brings a turnover with housekeeping, laundry and supplies, and the cleaning fee is charged per stay rather than per night. A market that drifts from week-long visits to two-night weekends can hold both its occupancy and its rate and still lose margin, and the equations show it only if stays are modelled separately from nights.

The third is the penetration index, the ratio of the property's occupancy to that of its comparable set. It is the penetration measure the demand article sets out, applied here to a set of listings. A property with no booking history under its new owner should not be placed at the set's average, and the illustration below puts it a little under.

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

In the illustration the comparable set books 56 percent of its nights and the cottages sit at an index of 0.95, so the property books 53.2% of its 1,400 available unit-nights, or 745 nights at an average rate of $230.00 (Table 2). At a three-night average stay those nights make 248 stays, and the cleaning fees they carry, $31,033, sit on top of $171,304 of nightly revenue, for total revenue of $202,337. Revenue per available unit-night is $122.36, the single figure that combines rate and occupancy and the one a study sets beside the comparable set's own.

Public data cannot supply the rate or the occupancy, but it can bound the stock that competes for the same guests. The Census Bureau's Housing Vacancy Survey counts housing held for seasonal or occasional use, and its definitions show why that count is a pool rather than a measure of rentals. A beach cottage occupied on the day of the interview by a family whose usual home is in the city is counted as vacant ^[17]. Seasonal units are the ones intended for occupancy only in certain seasons, found mostly in resort areas ^[18], and a unit kept for weekends or occasional use through the year is counted as held for occasional use ^[19].

For 2025 the survey puts the national stock at 3,433 thousand seasonal vacant units and 1,986 thousand year-round units held for occasional use ^[20], estimates built on eleven months because October 2025 went uncollected ^[21]. Nothing in those categories says whether a unit is ever let to a guest. Read for underwriting, the count says that second homes form a large reservoir from which rental supply can grow, and a study treats it as a supply risk rather than as evidence of demand.

Registration records come closer to the rental stock itself. New York City's enforcement office publishes its dataset of registrations, current as of the date in each report ^[22], and the file posted when this article was researched was dated November 2, 2026 ^[23]. A registry count measures legal supply in a place that registers; it does not measure bookings. Under New York State's law the stay data that booking services must report to counties each quarter ^[24] may not be made public ^[25]. The rate and occupancy inputs therefore come from a survey of comparable listings or, for an operating property, from its own booking history, and the study should say which it used and how many months it covers.

Seven Figures a Lender Takes From a Nightly Rental, and Where Each Is Proved

A bank's own policy frames what it will ask for. Two kinds of loan get extra policy standards under the interagency guidelines, those for development and construction and those on completed commercial property, scaled to the project's size and type ^[26]. Among the extra standards are feasibility study requirements and sensitivity analysis of projected income, with interest rates, vacancy rates and operating expenses given as examples of the variables ^[27], and minimums for net worth, cash flow and coverage, for either the borrower or the property ^[28]. Whether a particular rental falls under a bank's commercial policy is the bank's classification; the measures below are the ones a study should be ready to support either way.

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.

Occupancy on the nights offered. The measure is booked nights over nights offered, with blocked nights stated. For an operating property the evidence is the booking calendar and the payout records, and New York's state law gives the study a ready-made ledger to ask for: once a county registry exists, hosts must keep two years of stay records showing the dates and guests of each stay, its cost and an itemisation of the sales and occupancy tax collected ^[29]. Those duties do not bind until the county has established its registry ^[30], so the ledger is a request a study makes, not a document it can assume.

Average nightly rate, net of taxes. The guest's taxes are not the owner's revenue. From March 1, 2025, a New York short-term rental stay has carried state and local sales tax whenever the rate tops $2.00 per unit per day ^[31], and the Tax Department lists cleaning, service, extra person and pet charges among those that can be taxed with the occupancy itself ^[11]. A payout statement that mixes rent, fees and collected tax has to be taken apart before the rate is read, and the study should show the netting.

Revenue per available unit-night. Rate times occupancy, on the nights offered. It is the measure to compare with the comparable set, because a property can buy occupancy with price or price with occupancy, and only the product shows which trade it made.

Stays and the turnover margin. The cleaning fee less the housekeeping cost of each turnover, multiplied by the number of stays. In the illustration the contribution of a booked night after channel fees, management, turnover cost and reserve is $173.32 (Table 5), and the study should show how that figure moves when the average stay shortens.

Net operating income after the manager. Where the owner does not operate the property, a manager runs it for a fee, and that fee, the platform's charges and a replacement reserve all come off before coverage. Insurance is a line the law itself shapes in New York: a unit must carry at least three hundred thousand dollars of cover for third-party property damage and bodily injury claims arising from its operation ^[13], and a booking service's own policy of equal or greater cover may satisfy the requirement when the host lists through it ^[32]. The study should say which policy the projection relies on.

Coverage and the loan it supports. Net operating income after reserve divided by annual debt service, tested against the binding minimum and used to size the loan, as Table 5 does. The collateral side has its own label: the guidelines count credit secured by a 1- to 4-family residential property that is not owner-occupied, or by a residential building of five or more units, as an improved property loan ^[33], and set the supervisory loan-to-value limit for that category at 85 percent ^[15]. A bank should set its own internal limits, and they should not exceed the supervisory ones ^[34].

Break-even occupancy. The occupancy at which the property just pays its fixed costs and its debt, at 1.00x, and the occupancy that meets the lender's minimum. Because fixed costs in a rental do not fall when bookings do, break-even is the plainest statement of the margin, and it does not depend on the occupancy assumption at all, only on the rate, the costs and the debt.

Registration and Tax Rules as Conditions on the Revenue: New York State and New York City

Two New York regimes show how much of the revenue question is a legal question. The state's article defines a short-term rental unit as all or part of a dwelling let to guests for fewer than thirty consecutive days for tourist or transient use, in a covered jurisdiction ^[35]. Covered jurisdictions are every county, city, town and village in the state, less a list of exceptions that starts with any city of one million or more people ^[36]. New York City runs its own registration law, described below, and the two regimes differ on the point that matters most to a lender: whether an entire home may be sold by the night at all.

Under the state article every county that is a covered jurisdiction must set up a registry for the rental units inside it, and counties may share one ^[37]. The one way out was an opt-out local law, and its deadline was whichever came later: the last day of 2025, or nine months from the section's effective date ^[38]. Nothing in the registration paragraphs binds hosts or booking services before the county's registry exists ^[48], so the first fact a study records is the county's status: registry in operation, registry pending, or opted out.

Where a registry operates, the duties are concrete. A registration lasts two years and can then be renewed ^[49]. The host must show the current registration number on every offering, listing and advertisement ^[44]. The county may revoke a registration after it determines at least three violations of the article within two consecutive calendar years ^[50]. The registry fees are whatever the county sets ^[14], a cost line the study fills from the county's schedule rather than from the statute. The sanction that matters most to a lender is the one for selling without a number. The statute declares that offering nights "without current, valid registration shall be unlawful," and the person who listed the unit then loses eligibility to register for twelve months from the finding ^[51]. Twelve months off the platforms is a year of lost revenue on a property whose debt service does not pause.

The unit itself has conditions. Among them, it must post an evacuation diagram and a list of emergency telephone numbers and have a fire extinguisher that works ^[52], carry the liability cover already described ^[13], and not be barred from short-term use by any federal, state or local law, rule or regulation ^[53]. Booking services carry a matching duty: a platform may not lawfully take a fee for a booking it has not verified against the county's registration record ^[45], which gives the platforms a legal reason to refuse an unregistered unit. These duties, too, wait until the county has set up its registry ^[30].

Above all of this sits the local power to say no. The article leaves every county, city, town and village free to ban the use or to restrict it further by local law, whether or not that government is itself a covered jurisdiction ^[12]. The state framework is a floor of registration and safety rules, not a right to operate, and a town board can change the answer after the loan has closed. The site's older post The Rise of the Accidental Lobbyists: How Airbnb Hosts Are Taking On Big Regulation looks at how such rules are contested; a study takes the rule in force, notes any pending change it can document, and models the downside.

Tax rules change the revenue line in a quieter way. The sales tax has reached short stays since March 1, 2025, for any occupancy priced above two dollars a unit a day ^[31]; inside New York City each occupied unit also owes a $1.50 daily unit fee ^[47]; and counties, cities, towns or villages may levy their own occupancy tax, which the locality administers ^[54]. A booking service that facilitates a stay must register as a vendor and collect the tax ^[55]. An operator must register and collect too, unless an exception applies ^[56]: renting its own property for 3 days or less in a calendar year without a booking service, or having a booking service facilitate every sale ^[57]. The second exception carries conditions of its own. The operator is relieved only if it holds Form ST-155, the booking service's certificate of collection, or a publicly available agreement in which the service undertakes to collect the tax and unit fees ^{[58] [59]}, and it must keep that certificate together with records of the sales the services handled ^[60]. A study should ask to see the certificates before it treats the platforms as the only collectors. A guest keeps paying the state and local tax until becoming a permanent resident by staying 90 consecutive days, and in New York City the city's tax runs until 180 days ^[61], so a stay of a month or two is still a taxable occupancy. The study keeps every one of these amounts outside revenue and checks that the projection's rate is the pre-tax rate.

New York City is the sharper case. Local Law 18, adopted on January 9, 2022, requires hosts to register with the Mayor's Office of Special Enforcement and bars platforms from processing transactions for rentals that are not registered ^{[39] [46]}, and the first phase of enforcement, aimed at the booking platforms, began on September 5, 2023 ^[62]. Registration is open only to an actual person who is the permanent occupant of the unit ^[63]. The office states that hosts may not rent out an entire apartment or home for fewer than 30 days, whether they own the building or rent in it ^[41], and that a rental is legal only when the host lives in the unit with the guests, who have full access to it, with a limit of two guests not counting young children ^[43]. Units in approved Class B multiple dwellings and rentals of 30 consecutive days or more are outside the registration requirement ^[64], and the city keeps a list of buildings where short-term rentals are barred by law or by the building's own leases ^[65].

The consequence for underwriting is direct. An investor who buys an entire apartment in the city to let by the night is projecting income that the city's rules do not allow ^[41], and a study would underwrite such a unit on rentals of thirty days or more, or not at all. The method generalises beyond New York: a study names the jurisdiction, quotes the rule in force, records the registry status and the unit's registration, and only then projects a night.

Eligibility: Passive Ownership, Legality and the B&I Line Between Lodging and Housing

The passive test has already been described, and on its text it excludes an investor who neither uses nor occupies the rental. Some projects are different: their owners run them as lodging businesses, taking the bookings, cleaning, greeting guests and often living on the site. The regulation's test is whether the business actively uses or occupies what the loan finances ^[7]; it does not name short-term rentals, and how SBA applies the test to a lodging business run by its owner is a question for the program's operating procedures, which this article does not quote. The general 7(a) tests are set out in SBA 7(a) Method: What 13 CFR 120 Tests and How a Feasibility Study Answers Each Test.

Two further rules in part 120 bear on the class even where the passive question is answered. The first is legality. Under 13 CFR 120.110(h), a business is ineligible if any of its activities breaks federal, state or local law ^[66]. Where a county or city requires registration and treats an unregistered listing as unlawful, the registration status of the unit becomes, on the regulation's text, an eligibility fact as well as a revenue fact, and a study should document it with the registration number and the registry's own record. The second is the trust money. Under 13 CFR 120.130(e), loan proceeds may not pay past-due payroll, sales or similar taxes that the applicant must collect and hold in trust for a government ^[67]. A New York operator outside both exceptions, including one whose platform sales lack a certificate of collection or a public collection agreement ^{[56] [59]}, collects the tax itself and should be able to show the returns are current.

Where the financed property is a building the borrower occupies and partly leases, 13 CFR 120.131 sets the split. A new building needs the borrower permanently in at least 60 percent of the rentable space, with no more than 20 percent leased permanently ^[68], and the borrower must plan to occupy part of the unleased remainder inside three years and the whole of it inside ten ^{[69] [70]}; an existing one needs 51 percent occupied by the borrower, with the other 49 percent free to lease ^[71]. Whether rooms sold to travellers count as the borrower's own use is not something the regulation's text settles, and a study should not assume the answer. Under 13 CFR 120.160(b) SBA can ask for a feasibility study among its loan conditions ^[72], one more reason to state the open questions plainly.

New York's tax definition offers a contrast worth noting. In the Tax Department's definition, the typical occupants of a short-term rental unit are transients or travellers, and the relationship between operator and occupant is "not that of a landlord or tenant" ^[73]. A state tax definition does not decide a federal eligibility test. It shows that the same cottage can be a landlord's asset under one rule and a lodging operation under another, and a study should not borrow one regime's label to answer another regime's question.

USDA's Business and Industry program draws the line in different words. Its rural test takes in any part of a state beyond the limits of a city or town with over 50,000 inhabitants and leaves out the urbanized area that touches and adjoins such a place ^[74]. That exclusion is not absolute: a fringe location can still count once USDA's Under Secretary of Rural Development rules it rural in character, a ruling that covers one project only ^{[75] [76]}. Among eligible project types, 7 CFR 5001.105(b)(8) names "Tourist and recreation facilities, including hotels, motels, bed and breakfast establishments, and resort trailer parks and campgrounds operated as a public or private commercial enterprise" ^[77]. Where the owner lives on site, the share of the project that the owner's own quarters represent, by square footage, has to come out of what the loan finances ^[78].

The same part names what it will not finance. The general list of ineligible projects, which applies to the Community Facilities, Water and Waste Disposal, B&I and REAP programs ^[79], includes any speculative real estate investment ^[80] and, in the regulation's words, "Properties to be used for primarily commercial rental when the borrower has no control over tenants and services offered" ^[81]. Only in B&I does the list go on to residential projects: unless the tourist facility paragraph authorizes them, timeshares, apartments, duplexes and, in general, "other residential housing where the primary purpose is independent housing" are excluded ^[82].

Read together, those paragraphs put a rural cottage resort on one side of a line and a portfolio of scattered vacation homes on the other, without naming either. A study for a B&I application should document the facts that place the project: guests rather than tenants, services the borrower controls, an enterprise operated commercially, and the owner's quarters measured and excluded. Whether the Agency accepts that placement is the Agency's decision. The program method is in USDA B&I Guarantees Under 7 CFR 5001: What the Agency Tests and How a Study Answers, and the closest worked example is Worked Example: USDA Business and Industry Loan Glamping Resort Feasibility Study in Ohio.

For most investor-owned rentals the lender is a bank lending on its own account, under the guidelines already cited and the supervisory limits they set. The conventional method, including where those guidelines come from and how a bank's policy sits inside them, is in Conventional Bank Loan Method: Appraisal Rules, Supervisory LTV Limits and the Study a Bank Reads.

What the Hotel and Multifamily Records Show, and Why Neither Is This Class

With no class of its own, a short-term rental can be read only against its two neighbours, and each resembles it on one side. The hotel and motel class sells rooms by the night as an operating business. Its single code, 721110, does not separate a motel from a hotel, nor a limited-service property from a full-service one. The multifamily class owns dwellings and lets them, on code 531110, the same code an SBA loan to a home-rental business might carry. Neither is a short-term rental record, and every figure below is labelled as a neighbour's. Each neighbour's own method has its article: Hotel Underwriting Method: Fair Share Demand, RevPAR Evidence and the Class's SBA Record and Multifamily Underwriting Method: Rent Evidence, Absorption and Why SBA Lending to the Class Almost Stops.

Over FY2017 to FY2026 the hotel and motel class took 8,942 ^[83] 7(a) approvals for $21,538,178,600 ^[83] and 2,885 ^[84] 504 approvals for $6,679,237,290 ^[84]. Its median 7(a) approval was $2,200,000 ^[83] and its median 504 approval $2,048,000 ^[84]. Branded lodging dominates the record: 64.69 percent ^[83] of its 7(a) approvals and 62.98 percent ^[84] of its 504 approvals record a franchise code. The 7(a) count ran from 1,128 ^[83] approvals in FY2017 to 814 ^[83] in FY2026. A franchised hotel with staff, a front desk and a brand standard is a different credit from a cottage let through a platform.

The multifamily record is small and, in recent years, nearly empty. From FY2017 to FY2026 the class took 204 ^[85] 7(a) approvals for $122,118,000 ^[85], with a median of $152,500 ^[85], and 11 ^[86] 504 approvals. The 7(a) series shows 29 ^[85] approvals in FY2023 and 1 ^[85] in FY2024. The dataset's asset map records the class as one with very few loans that nearly stop from FY2024, and it does not give a cause; this article does not supply one. Because 531110 is the class's only code, any loan to a nightly rental business coded there sits inside these counts, and nothing in the file can separate it.

The loss record is where the comparison is most tempting and most limited. Every charge-off rate here is an approval-cohort floor: charge-offs to date on loans approved in FY2017 to FY2021, with loans still open counted in the denominator because their status is not disclosed. For the hotel class the 7(a) charge-off floor is 0.99 percent ^[83], with 1,804 ^[83] loans of the cohort still open, against a national 7(a) floor of 6.33 percent ^[87] on the same cohort basis. Its 504 floor is 0.55 percent ^[84] against a national 504 floor of 0.35 percent ^[88]. For multifamily the 7(a) charge-off floor is 4.27 percent ^[85], below the national 7(a) floor but resting on a small cohort, as the basis column of Table 4 shows. The class's 504 rates are withheld: fewer than 30 loans.

Cancellation is a separate way for an approval to end. The hotel class's 7(a) cancellation floor on the FY2017 to FY2024 cohort is 8.76 percent ^[83] and its 504 floor 13.13 percent ^[84], against national floors of 11.24 percent ^[89] for 7(a) and 13.60 percent ^[90] for 504. The multifamily 7(a) cancellation floor is 18.37 percent ^[85]. A cancelled approval never became a loan, so it is not a loss, and the two rates should never be added; Cancellation Is Not Charge-Off: The Two Ways an SBA Approval Can End Without Repayment explains the difference.

What the floors can say is limited to the classes they describe. The hotel 7(a) floor, well under the national 7(a) floor, says that SBA's 7(a) hotel loans of those years have so far charged off less often than 7(a) loans overall; it does not say why, and it says nothing about a cottage that could not have been in the cohort. The hotel 504 floor sits above the national 504 floor and rests on few charge-offs, as Table 4 shows, so it supports no ranking at all. A multifamily floor says something about the few residential lessors SBA did finance, and a reader cannot tell how many of them, if any, let their units by the night. The reasons a published floor can only understate the eventual loss are in The Approval-Cohort Charge-Off Floor: Why a Published SBA Loss Rate Can Only Understate. The honest use of these records in a short-term rental study is as context for the lender, clearly labelled as a neighbour's, never as the project's own loss expectation.

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.

A Mini-Example: Four Cottages, One Bank Loan and a Rule That Could End the Nightly Income

The illustration continues the four cottages from Table 2: an existing property, bought by an investor company that hires a manager, in a New York county outside New York City that has set up its registry. Every input in Table 1 is an assumption except the Treasury yield, and none is a measurement of a real market. Because the owner neither uses nor occupies the cottages, the passive rule points away from SBA ^[7], the loan is a conventional bank loan, no program floor applies, and the binding minimum is the lender assumption. For the complete treatment, with construction cost, a ramp year and a sourced market, see the New York worked example linked at the top.

Total revenue of $202,337 bears channel fees, a management fee of $34,261, housekeeping of $24,827 for the turnovers, and fixed costs of $66,500, among them $20,800 of property tax and $9,600 of insurance. After a reserve of $8,093, net operating income is $62,586 (Table 5). The loan prices off the 5-year Treasury, and on November 6, 2026 the H.15 release put that yield at 4.64 percent ^[16]. With the assumed spread the rate is 7.39%, which on a 25-year amortization costs $8.78 a year per $100 of loan.

Two sizing tests run side by side. The bank's assumed loan-to-value limit would lend $750,000, but on that loan coverage would be 0.95x: the cottages would not even cover the debt. The loan that the 1.25x lender assumption supports is $570,118, rounded down to $570,000, and on that loan coverage is 1.25x, cash flow after debt service is $12,528, and the owner brings $460,000, or 44.7% of price and closing costs (Table 5). Coverage, not the loan-to-value limit, sets the debt. Break-even occupancy is 48.0% at 1.00x against a base of 53.2%, and the lender assumption is met at 53.2% because the loan was sized to meet it exactly.

A loan sized exactly to the minimum has no cushion, which is why the stress table holds the loan at $570,000 and shows, beside each shock, the loan that the lender assumption would support instead (Table 6). Comparable occupancy ten points lower cuts property occupancy to 43.7% and coverage to 0.79x, and would support only $360,138. A nightly rate a tenth lower leaves coverage at 1.00x and would support $456,205. A two-point rise in the loan rate at the five-year reset lifts debt service to $59,239 and cuts coverage to 1.06x.

The last column is the one particular to the class. An assumed local cap of 120 rented nights a unit, the kind of further limit the state article leaves every locality free to impose ^[12], cuts booked nights to 480 and property occupancy to 34.3%, takes net operating income to $16,692, and leaves coverage at 0.33x (Table 6). In this illustration no market shock comes close to the damage a rule can do.

The fallback table asks the question a bank should ask of any nightly rental: if a local rule ended stays shorter than a month, what would the same building earn let to residents? On an assumed $1,650 monthly rent per unit, collected rent is $75,240 after vacancy, net operating income is $23,811, and coverage on the same loan falls to 0.48x; long-term rent would support a loan of only $216,904 at the lender assumption (Table 7). The difference, $38,775 of net operating income, is the income that depends on nightly use staying lawful, and it is exposed to a town board's vote. A study should print the fallback beside the base case, so that the lender sees how much of the loan rests on the local rule rather than on the building.

The Order of Proof in a Short-Term Rental Study

The method above implies an order, and a study should follow it, because each step can end the analysis before the next one matters.

First, the law. The study names the county, city, town or village and quotes the rule in force on short stays, including any cap, owner-occupancy condition or ban, and any pending local law it can document. In New York it records whether the county runs a registry, has one pending or opted out ^[38], and, where a registry runs, the unit's registration number and its expiry. A unit that cannot lawfully be listed has no nightly revenue to project, and under the state article listing it anyway carries a twelve-month bar ^[51].

Second, the eligibility route. The study says why the loan is SBA, B&I or conventional, on the regulation's own tests: the passive and legality tests of 13 CFR 120.110, the investment restriction of 120.130(d), and for a rural project the tourist facility and residential housing lines of 7 CFR 5001. Where the answer turns on how an agency applies a test, the study says so instead of assuming it.

Third, the revenue evidence. The comparable listing set is named by its selection rule, the months it covers and its occupancy on the nights its members offer, and the penetration index applied to the property is stated and argued. For an operating property the booking ledger and the payout statements reconcile to the tax records, with the guest's taxes taken out of the rate ^[11].

Fourth, the cost lines the class adds: turnovers priced per stay, the manager's fee, the platform's charges, registration and compliance, and the liability cover the jurisdiction requires.

Fifth, the tests. Coverage at the binding minimum, break-even occupancy, the four shocks in Table 6 and the long-term fallback in Table 7, each printed with its inputs so the lender can rerun it. Where the guidelines reach a loan, a bank's written policy should demand income sensitivity work of this kind ^[27], and for a dwelling sold by the night, the rule that permits the sale belongs among the variables tested.

Related Pages on Loan Analytics

• The Loan Analytics methodology series (all articles)
• How a feasibility study is done: the Loan Analytics methodology
• Bankable feasibility studies for conventional loans
• SBA feasibility studies (7(a) and 504)
• Loan data by state
• Loan data by asset class
• Worked examples (case studies)
• Worked Example: Conventional Community Bank Construction-to-Permanent Short-Term Rental Feasibility Study in New York
• Hotel Underwriting Method: Fair Share Demand, RevPAR Evidence and the Class's SBA Record
• Multifamily Underwriting Method: Rent Evidence, Absorption and an SBA Record That Almost Stops

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