Multifamily Underwriting Method: Rent Evidence, Absorption and an SBA Record That Almost Stops

An apartment loan is repaid by a rent roll

An apartment building earns its income from many small leases signed by households, renewed or lost every year, and priced against the buildings down the street. A loan against it is repaid from that rent roll and from nothing else of substance, so the feasibility study for a multifamily loan comes down to three questions asked in order: what rent the units can command, how quickly the units fill and stay filled, and whether the income left after operating costs carries the debt with room to spare. Everything else in the study (the market area, the supply count, the expense budget, the stress cases) exists to answer one of those three.

The class also differs from most of the asset classes in this series in who lends against it. The Federal Reserve's financial accounts put multifamily residential mortgages outstanding at $2,493.4 billion at the end of the second quarter of 2026 ^[1]. Banks and other U.S.-chartered depository institutions held $743.9 billion of that ^[2], the government-sponsored enterprises $715.5 billion and agency- and GSE-backed mortgage pools a further $453.3 billion ^[3], and life insurance companies $268.5 billion ^[4]. SBA's programs are written to stay outside most of that market: the business loan rules treat a landlord that does not itself use the property as a passive business that SBA will not finance, apart from a defined exception for Eligible Passive Companies ^[5].

The loan record shows how little SBA lending reaches the class. Over FY2017 to FY2026 the dataset counts 204 ^[6] SBA 7(a) approvals and 11 ^[7] SBA 504 approvals in the multifamily class (NAICS 531110, lessors of residential buildings), and SBA lending to the class almost stops from FY2024: 7(a) approvals fall from 29 ^[6] in FY2023 to 1 ^[6] in FY2024, 3 ^[6] in FY2025 and 5 ^[6] in FY2026, and 504 approvals from 2 ^[7] in FY2023 to 0 ^[7] in each of the three later years. A reader who comes to this page from the SBA side should take that as the first fact about the class. Conventional banks, the agencies and the insurers carry it, and the method below is written for the credit officer at one of those lenders, with the SBA record kept as a small and separate benchmark.

The rest of the article follows the order a study should follow. It sets out the public rent series and what each one measures, the absorption evidence and how asking rents change the pace of lease-up, a demand model with every input labelled, the figures a lender reads and the rules that anchor them, the eligibility rules that keep most apartment owners out of SBA's programs, the class's SBA record and its cohort floors, and a short worked example under the series' conventions. The general methods behind several of those steps are covered elsewhere in the series, and this article points to them rather than repeating them.

The lenders that carry the class, in one table

Table 1 turns the Federal Reserve's holder lines into shares of the total and sets the class's SBA approvals beside them. The inputs come from the financial accounts table for the second quarter of 2026 and from the two national SBA cells for the class; they are printed with their sources in Table 4.

Three readings follow. Depository institutions hold close to three tenths of the debt, the enterprises and the agency pools together hold nearly half of it, and life insurers about a tenth, so the debt ends up in three kinds of portfolio, each kept under its own rules. The table records who holds the debt at the quarter end; it does not say who originated each loan. For banks, the interagency lending guidelines that 12 CFR 365 Appendix A carries ask a bank's policy to fix minimum standards for the net worth, cash flow and coverage of the borrower or the property ^[8] and to require feasibility studies and tests of how projected income responds to shifts in interest rates, vacancy and operating costs ^[9]. The bank method is the subject of the conventional bank loan article, and this one does not restate it.

The last line of the table is a scale check rather than a market share. It divides a stock of debt outstanding on one date by a flow of SBA approvals summed over ten fiscal years, and the answer, 18,489 dollars of mortgage debt for every dollar of SBA approvals, is a measure of how marginal SBA is to the class, not of anything SBA did or failed to do. A study that leans on the SBA cells for this class should say plainly that they describe a couple of hundred approvals at the edge of a mortgage stock measured in trillions of dollars.

Rent evidence: four public series and what each one measures

Rent is the input that moves an apartment study most, and the public record offers several rent figures that measure different things. A study that sets them side by side, and says which is which, gives the credit officer a bracket for the subject's rent roll. A study that quotes one of them as if it were the subject's achievable rent has made an error a reviewer should catch at once.

HUD's Fair Market Rents come first, because they are defined in regulation and published for every FMR area. The regulation defines an FMR as an estimate of rent plus the cost of utilities, except telephone ^[10]. It is set at the 40th percentile, the level below which 40 percent of standard quality rental units in the FMR area rent, and it is drawn from the rents of units occupied by recent movers ^[11]. HUD will generally not set an area's FMR below 90 percent of the previous year's figure ^[12], which damps a fall. The FY 2027 notice describes the target as the gross rent of privately owned, decent and safe housing of a modest, non-luxury nature ^[13] and calls the calculation HUD's best effort to estimate the 40th percentile gross rent paid by recent movers into standard quality units ^[14].

The construction of the figure matters as much as its definition. For FY 2027 the base rents are 5-year American Community Survey data collected between 2020 and 2024 ^[15], which HUD adjusts toward recent movers, inflates to 2025 and then trends to FY 2027 with forecasts of CPI gross rent ^[16]. The FY 2027 figures took effect on October 1, 2026, except in areas where HUD received a valid request to reevaluate the FMR ^[17]; those areas may go on using their FY 2026 figures while the reevaluation runs ^[18]. New construction units are no longer excluded from the standard quality universe ^[19], though the percentile is still taken across the whole area's recent movers. HUD also publishes 50th percentile rents for every FMR area ^[20], and in the voucher program the FMR is the basis for the payment standard used to work out the maximum monthly subsidy ^[21].

For a market-rate study the FMR therefore does two jobs and not a third. It is a floor test: a new building whose gross rent, once the tenant's utilities are added to the contract rent, comes in at or below the FMR is priced at or below the 40th percentile of what recent movers pay across the area, and the study should say why. And because the voucher payment standard is based on it ^[21], it is the figure a study needs when it counts voucher households in its demand. It is not a measure of what new product rents for, because it describes the 40th percentile of recent movers across a whole FMR area, measured from survey years that end well before the loan closes.

The Census Bureau's Housing Vacancy Survey supplies a second figure. Its median asking rent for the United States was $1,541 in the third quarter of 2026, with $1,996 in the Northeast, $1,370 in the Midwest, $1,426 in the South and $1,861 in the West ^[22], against $1,534 nationally a year earlier ^[23]. The survey records the rent asked for a vacant unit at the time of interview, which may differ from the rent agreed once the unit is let ^[24]. It is the asking rent of whatever units are vacant for rent when the survey is taken, so it moves quarter to quarter with the mix of what happens to be vacant.

The Census Bureau's absorption survey of new buildings, SOMA, supplies the third and closest figure, because it measures only new construction. The latest annual report this article draws on covers buildings completed in 2016, whose privately financed, nonsubsidized, unfurnished apartments carried a median asking rent of $1,480 ^[25]. Later quarters are published through the Census Bureau's online table tool, with a margin of error for every estimate from second quarter 2018 completions onward ^[26], and a study should take the latest completed quarters for its region from there rather than lean on a decade-old median.

The fourth source is the one that settles the question: the study's own survey of comparable properties, unit type by unit type, with the asking rent, the concessions on offer, the utilities the tenant pays and the date of each observation. Table 2 summarises the three public series as a study might set them beside that survey.

From the rent survey to the income line

A rent survey produces asking rents; a lender lends on collected income. The bridge between them has four steps, and a study should show each with its own evidence rather than fold them into one blended occupancy figure.

The first step is gross potential rent: every unit at its surveyed rent for twelve months, by unit type, with the rents dated to the survey. The second is physical vacancy, the share of units empty at any time once the building has stabilised. The Census Bureau defines its rental vacancy rate as "the proportion of the rental inventory which is vacant for rent" ^[27], and the national rate for the third quarter of 2026 was 7.6 percent, with 9.2 percent in the South and 7.1 percent in the Midwest ^[28]. That rate covers the whole rental inventory the survey counts, not new apartments alone, so it is context rather than a figure to copy: a study that sets the subject's vacancy above or below it should argue the difference from its comparables' occupancy.

The third step is concessions and credit loss. Free weeks offered to lease a new building are a rent cut spread over the lease term, and a study that records asking rents without them overstates income from the first month. The fourth is other income (parking, storage, pet fees, utility reimbursement), which should be shown per unit and supported by the comparables, because it is easy to overstate.

Utilities deserve their own sentence in every multifamily study because they decide which public series a rent can be compared with. A contract rent from the study's survey excludes the utilities the tenant pays; an FMR includes them ^[10]. Before the subject's rent is set beside an FMR, the study should add a utility allowance to the contract rent, or the comparison flatters the subject by the amount of the tenant's bills.

The four steps end in economic occupancy, the share of gross potential income actually collected, and that figure, not physical occupancy, is what the worked example below carries into coverage. The expense side of the same statement (taxes, insurance, payroll, repairs, utilities paid by the owner, management and reserves) follows the method in the operating expense article.

How fast new units lease, and what the asking rent does to the pace

In SOMA's vocabulary a unit is absorbed once, at its first rental or sale after it is built, when it leaves the market ^[29], and the survey covers only privately financed, nonsubsidized buildings of at least five units ^[30]. How the survey is run, how its estimates are revised and how a ramp schedule is built from it are set out in the ramp-up and absorption article; this section takes one feature that article leaves aside and that bears directly on an apartment study: the link between the rent a new building asks and the speed at which it leases.

Table 3 takes five rows from the 2016 annual report. Across all completions, 55 percent of units were rented within three months and 94 percent within a year ^[31]. The top rent band, units asking $2,450 or more, was 45 percent absorbed after three months and 89 percent after twelve ^[32], and the report notes that every other rent category reached a twelve-month rate in the 90s ^[33]. The band from $1,050 to $1,249 led at three months, with 60 percent, and the band from $850 to $1,049 led at six, with 84 percent ^[31]. In the South the top band was slower still, at 40 percent after three months and 84 percent after twelve ^[34].

The pattern is plain: the units priced at the top of the market leased more slowly, and the gap persists through the first year rather than closing after the first quarter. The report does not say why, and a national table is no basis for inventing a cause. What a study should do is place the subject's rent in the right band for its region, use that band's curve as the outside check on its own lease-up schedule, and explain any gap between the two from its competitive set.

The national pace also moves with the delivery quarter. For units finished in the third quarter of 2021 the final three-month absorption estimate is 75 percent ^[35]; for units finished in the fourth quarter of 2024 it is 42 percent ^[36]. A schedule built on a fast vintage and applied to a building that will deliver into a slow one carries an error of that size. The study should therefore date its benchmark, state whether the figure is a first or a revised estimate, and carry the margin of error the Census Bureau publishes with it ^[26].

The demand model: qualified renters, the annual flow and the lease-up clock

The demand model for a new apartment building asks whether enough households who can pay the rent will be looking for a unit while the building leases, and how much of that search the building can win. It is built in four steps, each an equation whose inputs a reviewer can check. The series' piece on demand methods treats capture and penetration in general; this section applies them to renters.

First, income qualification. The rent the subject asks implies a household income: twelve months of rent divided by the rent-to-income ratio the study adopts. The tax credit statute treats a unit as rent-restricted when its gross rent does not exceed 30 percent of the imputed income limitation ^[37], which gives the ratio a published anchor; for market-rate units the ratio is the study's own stated assumption, and a study should show what a different ratio does. The share of renter households in the market area with at least that income comes from the income distribution of renters in the American Community Survey for the tracts the study draws.

Second, the qualified stock and the annual flow. Qualified renter households equal renter households in the market area times the qualified share. Each year a part of them moves, and net new renter households arrive; the qualified flow equals qualified households times the annual mover rate, plus new renter households times the qualified share. The flow, not the stock, is what a new building leases from, because a household locked into a lease is not shopping.

Third, capture and penetration. The capture rate divides the subject's units by qualified renter households; the penetration of new supply divides the subject's units plus the competing pipeline by one year's qualified flow. The first says how large the building is against its market; the second says how much of a year's search all the new buildings together must win, which is the figure that shows when a submarket is oversupplied.

Fourth, the lease-up clock. If new buildings win a stated share of the qualified flow, they lease that share of the flow divided by twelve each month between them, and the subject takes its fair share, its units over the units of all new buildings leasing at the same time. Months to stabilisation equal the units the building needs occupied divided by its monthly leases; the share leased in the first three months equals three months of leases over the subject's units, and it is the number to set beside the SOMA benchmark.

Table 4 prints each input behind this model and behind the article's other model tables, marked as assumed or sourced. Table 5 runs the model at the base case and under four changes.

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

At the base case the subject's 120 units are 2.2% of the 5,500 qualified renter households, a modest capture, but the new units together need 32.8% of one year's qualified flow, and the subject reaches stabilised occupancy in 9.3 months. The more telling line is the cross-check: the model leases 30.5% of the subject's units in the first three months, against the national three-month figure of 42% for apartments completed in the fourth quarter of 2024. A study whose own schedule runs faster than the model must show why its submarket outpaces the arithmetic; one whose schedule runs slower than the national benchmark must show what in its competitive set holds the building back. Either way, the gap is the point the credit officer will press.

The shocks show which inputs matter. Doubling the pipeline leaves the capture rate untouched and stretches stabilisation to 16.2 months, because the subject's fair share of new-unit leasing falls to 14.3%. A qualified share five points lower, which is what a study would find if it had overstated renter incomes, lifts stabilisation to 12.0 months. Raising the rent by 5 percent raises the qualifying income to $69,300 and, at the assumed three-point loss in qualified share, adds more than a month to the lease-up; that column is the model's version of the rent band pattern in Table 3. Each of those inputs is a figure a study can evidence, and the model shows a reviewer which of them to test first.

The figures a lender reads on an apartment loan

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.

A multifamily credit memo is built from a short list of figures, and a study should present each one with its arithmetic and its anchor. Economic occupancy and effective gross income come from the rent bridge above. The expense ratio and net operating income per unit come from the expense budget, and a reviewer will set both against the comparable properties. Debt service coverage divides net operating income by annual debt service. Debt yield divides net operating income by the loan amount and so does not move with the rate or the amortisation, which is why a lender may read it beside coverage. Loan to cost and loan to value set the loan against what the project costs and what an appraiser says it is worth; the study supplies cost and income, not value. Break-even economic occupancy is the occupancy at which income just pays operating costs, reserves and debt service.

Two federal texts anchor those figures for a bank. The interagency guidelines want a lending policy to fix its own coverage minimums ^[8], and their supervisory table sets the ceilings a bank's own loan-to-value limits are meant to stay under, 80 percent while a multifamily building is under construction and 85 percent once it is improved property ^[41], a category that takes in residential buildings holding five or more dwellings ^[44]. The FDIC's capital rule then defines a statutory multifamily mortgage, which can take a 50 percent risk weight once the loan has a year of timely payments behind it, with criteria that read like a lending screen: a loan of no more than 80 percent of the property's value at origination, or 75 percent on a rate that changes ^[45], amortisation over no more than 30 years ^[39], and net operating income in the most recent fiscal year of at least 120 percent of current annual debt service, or 115 percent on a variable rate ^[40]. Those are capital tests, not underwriting minimums, and a study should present them as such; but a stabilised loan that misses them does not qualify for the lower risk weight, and a bank's credit officer will weigh that.

Table 6 lists each figure with the rule or series that anchors it.

The credit climate is the last anchor, and it changes. Senior loan officers answering the Federal Reserve's survey for October 2026 said their banks had, on balance, tightened standards in all three commercial real estate categories, multifamily among them, and had seen demand weaken in each ^[46]. On the multifamily question itself, one domestic bank (1.8 percent of respondents) reported tightening considerably, seven (12.3 percent) tightening somewhat and 46 (80.7 percent) no change ^[47], while 18 (31.6 percent) reported moderately weaker demand for multifamily loans against 5 (8.8 percent) reporting stronger demand ^[48]. A study written into that climate should expect its downside cases to be read closely, and the sensitivity method in the stress testing article sets out how to size them.

Eligibility: why SBA reaches so few apartment owners

The SBA rules that bear on the class are short, and a study for any SBA request touching residential space should quote them with their conditions intact. The starting point is 13 CFR 120.110(c), whose list of ineligible businesses includes "Passive businesses owned by developers and landlords that do not actively use or occupy the assets acquired or improved with the loan proceeds" ^[5]. Only Eligible Passive Companies are spared that bar ^[5]. A building leased unit by unit to households is the plain case the rule describes.

The exception does not reach it either. It is built for a property company whose tenant is an operating business: its loan may fund only property it lets to Operating Companies for use in their own trade, or a buyout among its existing owners, and further conditions follow ^[49]. An Operating Company is defined as an eligible small business actively involved in conducting business operations on the property the passive company owns ^[50]. A household renting an apartment is not a small business conducting operations there, so on the face of the two definitions the exception that lets a property company borrow does not open for residential leasing. The use of proceeds rule points the same way. Its list of barred purposes includes "Investments in real or personal property acquired and held primarily for sale, lease, or investment", and it carves out only two borrowers from that bar: an Eligible Passive Company, and a small contractor under 120.310 ^[51].

Two narrower routes exist, and both keep their conditions. The first is the mixed-use building whose operating business occupies most of it, and the thresholds depend on whether the building is old or new. In an existing building, the borrower's own permanent use must cover 51 percent or more of the Rentable Property, and the share let permanently to others is capped at 49 percent ^[42]. A new building lifts the first figure to 60 percent and cuts the second to 20 percent ^[43], and the borrower must also intend to move into part of the unleased balance inside three years ^[52] and into all of it inside ten ^[53]. Where an Eligible Passive Company leases the whole of an existing building to Operating Companies, those companies together must meet the same rule ^[54]. All of these shares are measured against Rentable Property, and the definition in 13 CFR 120.10 reads "the total square footage of all buildings or facilities used for business operations" ^[55]. How residential floors above a business are counted under that definition is a question SBA's procedures and the lender's reading of them answer, and this article does not state those procedures; a study should show the square footage both ways and let the lender apply the rule.

Table 7 shows the test on a small mixed-use building, with the residential floors counted in the total, the conservative reading.

The same building passes as an existing property and fails as new construction, on both counts: the business occupies less than the new-construction minimum, and the apartments are far more than the share a borrower may lease permanently in a new building. A sponsor planning a new ground-floor business with apartments above should see that result before the drawings are final.

The second route is the Builders Loan Program, which may finance small general contractors to construct or rehabilitate residential or commercial property for resale, an exception under specified conditions to the general rule against financing investment property ^[56]. It finances building for sale, not for rent, and it is not a path to a rental apartment loan.

USDA's guaranteed business loans close the other door. Part 5001 shuts Business and Industry guarantees out of apartments and duplexes, and out of any other residential housing meant mainly as independent housing, as well as timeshares and residential trailer parks ^[57]. The program method is in the USDA Business and Industry article. The archived guide on SBA loans for an apartment building works through the passive rule from the borrower's side.

The class in the SBA loan record, fiscal year by fiscal year

The dataset builds the class from NAICS 531110, Lessors of Residential Buildings and Dwellings, whose establishments lease out buildings used as residences, "such as single-family homes, apartment buildings, and town homes" ^[58], whether they manage the property themselves or have another establishment manage it for them ^[59]. The dataset's asset map notes that build-to-rent has no code of its own, so it sits in the same class. The code is the borrower's industry, so the cells count SBA borrowers who lease residential property, not apartment buildings as such; the article on mapping codes to classes explains how the dataset draws the line.

Over FY2017 to FY2026 the class holds 204 ^[6] 7(a) approvals worth $122,118,000 ^[6] in all, with a median loan of $152,500 ^[6] and a middle half running from $50,000 ^[6] to $686,825 ^[6]. Under 504 it holds 11 ^[7] approvals worth $12,739,000 ^[7], with a median of $461,000 ^[7]. The franchise share of 7(a) approvals is 3.92 percent ^[6]. These are small loans in a class of large assets, and the record cannot say what they financed; the median 7(a) loan is a small fraction of the cost of the building in the worked example below.

Table 8 sets the class's approvals by fiscal year beside the 7(a) program's national totals.

The break is plain and it is the class's defining fact in the SBA record: SBA lending to multifamily almost stops from FY2024. The class's 7(a) approvals run between 17 ^[6] and 41 ^[6] a year from FY2017 to FY2023, then drop to 1 ^[6] in FY2024, the year in which total 7(a) approvals climbed to 70,242 ^[67] from 57,362 ^[66] a year earlier. Under 504 the class's last approvals in the window came in FY2023. The dataset's report records the fall and gives no reason for it, and the cells cannot supply one: they show when the approvals stopped, not why. A study should state the fall as a fact of the record, point out that it is specific to the class rather than to the program, and not build an argument on a guess about its cause.

The question in this article's title therefore has a two-part answer. The rules in the previous section are why SBA's programs are not built for the class: a landlord leasing units to households is the passive business 13 CFR 120.110(c) names ^[5], and the exception for property companies is written around business tenants ^[50]. The thin record in every year is consistent with those rules. Why the thin flow almost stops from FY2024 is a question the record raises and cannot answer, and a study should leave it open rather than fill it.

What the record can still do is mark the class's thin place in SBA lending for a reader who arrives expecting an SBA route. The 7(a) approvals that exist are spread across 47 ^[6] state areas, with no more than 19 ^[6] in any one, so a state view of the class is thinner still; the guide to reading a loan outcome page explains how the dataset handles cells that small. The loan figures this article uses are listed below with their keys.

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.

What the class's floors show, and where they stop

Two rates in the 7(a) cell stand on 30 or more loans, and both are approval-cohort floors. The cancellation floor for approvals made in FY2017 to FY2024 is 18.37 percent ^[6], on 36 ^[6] cancellations among 196 ^[6] approvals, against 11.24 percent ^[70] for the 7(a) program over the same cohort. No approval in the class's cohort remains undisbursed, so its ceiling, which would count such approvals as cancelled, equals the floor at 18.37 percent ^[6]. The class cancels at well above the program's rate. Why it does, the record does not say; the article on cancellation and charge-off sets out what a cancellation is and is not.

The charge-off floor for FY2017 to FY2021 approvals is 4.27 percent ^[6], on 5 ^[6] charge-offs among 117 ^[6] disbursed loans, below the whole 7(a) program's 6.33 percent ^[71] for those approval years. On the measure that holds seasoning level, which counts only loans charged off inside five years of approval, multifamily reaches 4.27 percent ^[6] while the program reaches 3.72 percent ^[72], so the class sits below the program on one measure and above it on the other, because all of its charge-offs so far came inside five years. By dollars the class's charged-off share is 0.15 percent ^[6] against 1.95 percent ^[73] for the program, which says the dollars charged off were small next to the dollars lent. 32 ^[6] loans in the denominator are still open, and the FOIA file hides the status of any disbursed loan that has not yet ended in cancellation, payoff or charge-off ^[74], which is why every one of these figures is a floor; the article on approval-cohort floors explains the mechanics.

The 504 cell is too thin for either rate. Its cancellation rate and its charge-off rate are withheld: fewer than 30 loans, as the article on thin samples explains.

Three limits govern how a study may use these floors. First, the numerator of the charge-off rate is a handful of loans, so the 30-loan rule protects the denominator but cannot make the rate stable; one loan more or fewer would move it visibly. Second, the borrowers are SBA borrowers coded as residential lessors, carrying small loans, and their outcomes say nothing about the performance of apartment mortgages held by banks, the enterprises or insurers, which are a different population on different terms. Third, the floors carry no cause: a cancellation floor above the program's is no proof that apartment requests are weaker, and a lower charge-off floor is no proof that they are safer. A study may print the class's floors beside the program's to show where the class sits in the SBA record, and should stop there.

A worked mini-example: one stabilised building tested under the conventions

The example takes a new 120-unit building at stabilisation, financed by a conventional permanent loan. Every project figure is an assumption printed in Table 4; the rate starts from the 10-year Treasury yield the H.15 release shows for November 6, 2026, 4.86 percent ^[38], with an assumed spread on top. A conventional loan has no program floor, so under the series' conventions the stated 1.25x lender assumption is the minimum that binds. The loan is set at the lower of two sizing results, the loan the income supports at that minimum and an assumed 75 percent loan-to-cost cap, rounded down. No loan fees are modelled, so the conventions' fee step has nothing to recompute. Table 10 shows the base case and four shocks, with the loan held at its base amount.

At the base case, effective gross income of $2,279,936 less expenses, management and reserves of $895,197 leaves net operating income of $1,384,739, or $11,539 per unit, at an expense ratio of 39.3%. Coverage lands on 1.25x, the assumed lender minimum to the second decimal, because income sized the loan: the loan the income supports, $14,291,685, is well below the $18,000,000 the loan-to-cost cap would allow. The loan is 59.5% of cost, the debt yield is 9.69%, and equity at the supported loan is $9,708,315, or 40.5% of cost. The building breaks even at 80.9% economic occupancy against a base case of 92.6%.

The shocks show how little room a loan sized to its minimum keeps. A rent 5 percent lower cuts coverage to 1.15x, three more points of vacancy to 1.19x, expenses 10 percent higher to 1.18x, and a rate one point higher at the time the rate is fixed to 1.13x. Each column also shows the loan the shocked income would support, and the equity the sponsor would have to add to close at it; under the rate shock that equity rises to 46.1% of cost. The capital rule's fixed-rate test of 1.20x ^[40] is met at the base case and missed under every shock, so on any of the shocked incomes the loan would fall outside the capital rule's definition of a statutory multifamily mortgage.

The full version of this exercise, with a construction loan, an interest reserve, a lease-up ramp and a conversion test, is in the worked example for a Nebraska apartment project. The mini-example here is enough to show the order of the work: the rent survey and the demand model set the income, the expense budget sets the net, and the coverage test, read with its shocks, sets the loan.

Questions to put to a multifamily study before it goes to credit

A credit officer or a sponsor can test an apartment study quickly with the questions below. Each one maps to a section of this article and to the rule or series that anchors it.

• Which rent figures does the study quote, and does it say what each measures? An FMR is a 40th percentile gross rent of recent movers across an FMR area ^[11]; a Housing Vacancy Survey median is the rent asked for whatever units are vacant at interview ^[24]. Neither is the subject's achievable rent, and only the study's own comparable survey can be.
• Are utilities treated consistently? A contract rent set beside an FMR must first have the tenant's utilities added, since the FMR includes them ^[10].
• Is the vacancy assumption argued from the comparables? The national rental vacancy rate counts the whole rental inventory ^[27], and a study should use it as context, not as the subject's figure.
• Is the lease-up schedule checked against SOMA for the right rent band, region and delivery quarter? The top rent band leased more slowly in the 2016 completions ^[32], and the national pace has moved widely between delivery quarters ^{[35] [36]}.
• Does the demand model print its inputs and shocks? Capture, penetration of new supply and months to stabilisation should each be reproducible from a printed table, as in Table 5.
• Is coverage tested at the binding minimum and under shocks to rent, vacancy, expenses and rate? The interagency guidelines want lending policies to call for income sensitivity tests on rates, vacancy and expenses ^[9].
• If SBA financing is proposed, does the study show the occupancy test with its conditions? The 51 and 60 percent rules, the lease caps and the plans to occupy within three and ten years all apply ^{[42] [52] [53]}, and the passive-business rule rules out a landlord that does not occupy the building ^[5].
• Does any use of the SBA record say what it is? The class's SBA approvals are few, almost stop from FY2024, and describe small loans to residential lessors, not the apartment mortgage debt that depository institutions, the enterprises, the agency pools and insurers hold ^{[2] [3] [4]}.

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
• Multifamily Feasibility Study
• Loan data by state
• Loan data by asset class
• Worked examples (case studies)
• Worked Example: Conventional Bank Construction-to-Permanent Multifamily Feasibility Study in Nebraska
• Can You Get an SBA Loan for an Apartment Building? The Passive Real Estate Rule, the 51 Percent Exception, and the USDA Section 538 Route
• Build-to-Rent vs Traditional Multifamily: Underwriting Spread Report

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