Absorption Forensics: Testing a Lease-Up Assumption Before the Reviewer Does

Every ground-up multifamily pro forma rests on one number the sponsor cannot support with a rent roll, because the building does not exist yet. That number is the absorption pace: the net units the property will lease each month once it opens. Get it right and the rest of the model falls into line. Get it wrong and two things break at the same moment, and they happen to be the first two things a credit officer checks.

A pace that is too fast shortens the lease-up on paper. A shorter lease-up means a smaller interest reserve, because there are fewer months of carry to fund. It also means the project appears to cross debt-service coverage of 1.0x sooner, and to reach the lender's stabilized floor sooner. So a single optimistic assumption quietly undersizes the reserve and overstates early coverage in one stroke. Those are not two separate errors. They are the same error, read two ways, and an experienced reviewer knows exactly where to look for it. This is a piece about deriving that number the way a skeptic would, and presenting it so the skeptic has nothing left to attack.

Why the old pace assumptions stopped working

For most of the last decade, an underwriter could reach for a rule of thumb and be roughly right. The convention that a well-located property leases to stabilized occupancy in about twelve months, tracked by CBRE and RealPage among others, held up because supply was thin and demand was steady. That era is over, and the assumptions carried forward from it are now the single most common source of pro forma optimism.

The supply wave is the reason. RealPage put full-year 2024 apartment deliveries at roughly 589,000 units, the most in a single year since the early 1980s. Completions moderated through 2025, to something on the order of 410,000 units, and Cushman & Wakefield reported first-quarter 2026 deliveries down about 30 percent year over year with starts at their lowest level since 2016. The wave is receding, but it left a backlog in its wake, and that backlog is where absorption assumptions go to die. ALN Apartment Data reports the average lease-up duration nationally now runs about seventeen months, five longer than the twelve-month convention still embedded in many models, with roughly 800,000 units in some phase of lease-up entering 2026 (the count varies with definition: Yardi Matrix, using a broader standard, puts it closer to 1.3 million).

Concessions confirm the softness. RealPage found nearly 22 percent of apartments offering concessions in the third quarter of 2025, at an average value of 6.2 percent of the lease, and the pressure is concentrated in the Sun Belt, with Austin leading the country at 30.5 percent of units discounting. Demand itself has been strong, and ALN reports roughly 770,000 net units absorbed in 2025, more than double 2024 and, for the first time since 2021, ahead of deliveries. But strong demand spread across a crowded field of lease-ups still produces a slower pace for any single property, and vacancy has stayed elevated, with Cushman putting national vacancy at 9.4 percent in early 2026, essentially flat for over a year. A model that assumes 2021 velocity into a 2026 submarket is not conservative. It is wrong on the arithmetic.

Three numbers people call absorption, and only one sets the reserve

Before deriving a pace, state precisely what is being derived, because the word "absorption" gets used for three different quantities and reviewers routinely catch analysts substituting one for another.

The first is pace: net units leased per month. This is the number that sizes the reserve and drives the coverage timeline, and it is the one that matters for a lease-up. The second is the absorption period: the total number of months from the first unit delivered to stabilized occupancy, where stabilized conventionally means 90 to 95 percent, with institutional practice usually holding the line at 93 to 95. The third is the absorption rate expressed as a percentage: net absorption divided by available inventory, a share metric that describes a market's health but says nothing directly about how fast a specific building fills. Months to lease-up is then just vacant units divided by the monthly pace.

The distinction between the second and third of these traps people. A submarket can post a healthy absorption percentage while a given property leases slowly, because the percentage reflects the whole inventory and the pace reflects one asset competing inside it. Reporting a strong market percentage as if it were a defensible monthly pace is one of the tells that a study was written backward from a desired conclusion.

The more consequential distinction is gross versus net. Gross absorption counts new leases signed. Net absorption subtracts move-outs and attrition. For a lease-up starting from zero, the two run close together in the early months, but they diverge, and the divergence is not academic. Gross absorption can overstate underlying demand by 40 percent or more, because a well-positioned new Class A building can sign twenty-five leases a month in a submarket where net absorption is flat or negative. It is simply pulling those renters out of older, cheaper comparables. The submarket is not adding twenty-five households a month; the building is cannibalizing its neighbors. NCHMA's Model Content Standards flag the related trap directly: if the lease-up period runs longer than the initial lease term, the forecast has to account for the first tenants beginning to turn over, so net absorption falls below gross in the back half of a long lease-up. Feasibility studies model net. A study built on gross absorption is describing churn, not demand.

What the reviewer is actually testing

It helps to name the audience. The absorption assumption will be read by a credit officer, by an agency or SBA reviewer where those programs apply, and often by a third-party reviewer engaged specifically to find the soft spots. To all of them it is the primary home of pro forma optimism, and for good reason: the interest reserve and the coverage timeline both hang off this one input, and both are where a construction loan is most exposed. What makes capital nervous, as one industry account puts it, is not knowing the market vacancy, not knowing what absorption looks like, and not knowing how many deliveries are coming. The analyst's job is to answer all three before the question is asked, with a pace that is triangulated from independent evidence, adjusted for the conditions the property will actually open into, benchmarked against a fair share of demand, and stress-tested for the downside. The rest of this piece is how to build that.

Reading the comparables

The first evidence source is the realized performance of properties that have already done what the subject proposes to do. NCHMA points the analyst toward comparable and competitive projects that have entered the market within the past twenty-four months, and the emphasis on recency is deliberate: a lease-up from three years ago tells you almost nothing about the submarket the subject will open into.

Identify properties that are genuinely comparable on product type, unit mix, rent tier, class, submarket, amenity level, and vintage, then extract what each one actually did: units leased per month and total lease-up duration. The commercial platforms are the starting point, CoStar for breadth, Yardi Matrix for the deepest multifamily-specific detail, RealPage and ALN and Radix alongside. But the platforms are not the finish. The discipline that survives a credit committee triangulates three layers: the macro demand drivers, the commercial data, and local ground-truthing, the leasing-office visits and broker surveys and certificate-of-occupancy records that separate an analyst who knows a submarket from one who has only queried it.

A comparable's raw pace is not usable until it is adjusted, and there are four adjustments that matter.

Seasonality comes first. Leasing demand follows a calendar: searches build from January, moves ramp through spring and peak in August, and the window from November through February is slow. A comparable that leased twenty units a month after an April opening is not evidence for a subject opening in October. The seasonal swing has flattened since 2022, as operators spread renewals and supply-rich conditions blunted the peak, but it has not disappeared, and de-seasonalizing a comparable to the subject's launch window is basic hygiene.

Concessions come second. A comparable's pace is only comparable at a comparable concession level. Fast lease-ups in this market are frequently bought with one to two months free, which is an effective-rent reduction of roughly 8 to 15 percent. If a comparable leased quickly on two months free and the subject's model assumes none, the pace is not transferable. Restate every comparable to net effective rent before trusting its velocity, and treat concession-driven absorption for what it is: a signal that the market is clearing below the sticker price.

Project size comes third. A 300-unit property cannot be assumed to lease at the same monthly pace as a 100-unit property. Larger projects can post higher absolute units per month, but they take longer to fill, and there are practical ceilings that have nothing to do with demand: staffing capacity, phased unit releases, and the stalls that commonly appear around 30 percent and 70 percent leased. Scaling a small comparable's pace linearly onto a large subject is a mistake the arithmetic will punish.

Market timing comes fourth. A comparable that leased in a stronger year, or ahead of a supply wave rather than into it, overstates the pace available to a property opening into tougher conditions. Timing is the adjustment analysts are most tempted to skip, because it requires admitting that the good comparable is good partly because it was lucky.

As for what a defensible pace looks like in absolute terms, treat the published ranges as conventions rather than data. Practitioner sources cite something like 15 to 25 units a month per 100 units in strong markets, and 10 to 20 as typical for steady-demand markets. Set against those conventions, the current reality is sobering: RealPage's cross-market data had the average lease-up community executing only about ten to eleven leases a month as of the second quarter of 2024, down year over year even as total absorption rose. That figure is a 2024 vintage and should be refreshed against a current pull, but the direction is the point, and when a comparable set and a rule of thumb disagree, the comparable set wins.

The pipeline competing for your demand

The second evidence source is everything that will open alongside the subject. Absorption is a competition for finite net demand, and the size of the competitive field sets the ceiling on any one property's pace. If 2,000 units deliver into a submarket over the same eighteen months the subject leases up, the subject is not fishing an empty pond; it is competing for a share of net demand, and the pace slows and the concessions deepen accordingly.

Build the pipeline for the renter-shed the subject will actually draw from, defined by submarket, drive time, or migration patterns rather than an arbitrary radius. Count what is under construction and planned for delivery during the subject's lease-up window, both the direct competitive set and the broader shadow pipeline. CoStar, Yardi Matrix, and RealPage all carry forward-looking supply, and permits for buildings of five or more units lead deliveries by six to eighteen months, but municipal permit and zoning records give a project-by-project read that no platform fully replaces.

The number that matters is net, not gross. The correct denominator for the subject's competition is net new supply, or net demand, over the lease-up window, not gross leasing activity across the whole market. A submarket where deliveries exceed demand carries negative excess net demand, which elongates absorption and forces concessions; an undersupplied one compresses the timeline. The single best predictor of 2025 rent performance was not region or class. It was deliveries as a share of inventory. That is the ratio to compute for the subject's submarket, and it is the ratio a reviewer will compute if the study does not.

The national moderation masks a sharp bifurcation, and the bifurcation is what underwrites a specific deal. Oversupplied Sun Belt and Mountain metros, Austin, Phoenix, Nashville, Dallas, Denver, San Antonio, Charlotte, carried double-digit vacancy, face-rent declines of roughly 4 to 7 percent, and the deepest concessions. Supply-constrained Midwest and Northeast markets held rents in the 3 to 4 percent range. National asking-rent growth of about 0.9 percent, per Cushman, is an average of those two very different stories, and an average is not something you can lease units against. The submarket read is the only one that matters.

The demand math, and the discipline of fair share

The third evidence source is the one that turns judgment into arithmetic. The subject's pace is a function of two things: total net renter demand over the lease-up period, and the share of that demand the subject captures. Both have to be built up explicitly.

Start with the demand side. Total renter-household demand derives from population and household growth, employment growth, household formation, the count of income-qualified renter households at the subject's rent point, and net in-migration. Of these, employment is the most resilient and defensible driver. A long-standing rule holds that a healthy market creates about five new jobs for every new apartment unit; recent national data runs below that, with NAHB putting the ratio of total jobs to multifamily permits at about 2.6 to 1 in 2023. There is a live headwind any current study has to weight: job growth cooled sharply through 2025, with BLS payroll gains averaging about 75,000 a month through August against 166,000 a month in 2024. Demand assumptions built on 2021-through-2024 growth are stale, and a reviewer will know it.

Once demand is estimated, net out the competitive pipeline, then apply a capture rate. And here the vocabulary has to be exact, because NCHMA draws a line between two rates that are easy to blur. The capture rate is project-specific: the subject's proposed units divided by the income, size, and age-qualified renter households in its primary market area. The penetration rate is broader: the subject plus all comparable and competing pipeline units measured against that same qualified base, which is a saturation gauge. NCHMA warns explicitly against leaning on the capture rate alone, because a low capture rate can look reassuring while the penetration rate reveals that existing and approved projects already serve the target renter. Viability does not rest on one number.

The check that keeps a capture assumption honest is fair share. If the subject is 250 units in a field of 1,000 competitive units, its fair share is 25 percent. Absent a demonstrable edge, capture should sit near fair share, and a study that assumes 50 percent capture while conceding 25 percent fair share has just handed the reviewer the exact gap to attack. Affordable and LIHTC work carries published capture thresholds, illustratively favorable at or below 5 percent and reasonable in the 5 to 10 percent range, though the income-qualified denominator makes these specific to restricted deals and subject to wide state variation. Market-rate work uses a supply-based denominator instead, but the fair-share logic is the same: the number has to be reconciled against the size of the competitive field, not asserted.

Reconciling to one number: a worked case

Consider how this comes together for a concrete subject: a 250-unit Class A garden and mid-rise property in a defined secondary submarket, mid-to-upper market rents, targeting stabilized occupancy of 93 to 95 percent. Three methods, run independently, then reconciled.

The comparables come first. Four in-submarket lease-ups leased at 12, 14, 16, and 18 units a month once de-seasonalized to the subject's launch window and restated to comparable concession levels. The midpoint sits around 15 a month. The subject's larger size and the current concession environment both argue against the top of that band, so the comparable evidence points to roughly 15, not 18.

The pipeline comes second. About 1,000 competitive units, the subject included, are in lease-up concurrently in the renter-shed over an eighteen-month window. If net qualified demand runs around 1,000 households over the same window, this is a balanced-to-slightly-soft setup: enough demand to fill the field, but not enough to let any one property outrun its share or drop concessions.

The capture math comes third. Net demand of roughly 1,000 households over eighteen months is about 55 households a month spread across all competitors. At a 25 percent fair share, the subject's implied pace is around 14 a month. Grant the subject a modest premium for superior product and location, lifting its share to 27 to 30 percent, and the implied pace rises to roughly 15 to 16.

The three methods converge: comparables near 15, capture near 14, pipeline-adjusted judgment near 15. That convergence is the point. When three independent derivations land in the same place, the base case is defensible, and the study adopts 15 units a month. With 30 to 40 units of pre-leasing before delivery, filling 250 units to about 95 percent implies an absorption period of roughly 16 to 17 months after opening, which is consistent with the national average ALN now observes. A base case of 15 a month, an aggressive case of 20 producing about 12 months, and a conservative case of 10 producing about 24 months, all shown, is the difference between an assumption and a guess.

Where the assumption goes to work: reserve and coverage

None of the above matters to a lender until it flows into the two mechanics that govern a construction loan: the interest reserve and the coverage timeline. This is where absorption stops being a market-study exercise and becomes a credit question.

The interest reserve is a pool of loan proceeds set aside at closing to pay interest during construction and lease-up, until the property's net operating income can cover debt service on its own, commonly sized for six to eighteen months of carry. During construction, a rough sizing is the average outstanding loan balance times the interest rate times the carry period, with the average-balance factor usually around 50 to 60 percent because the loan draws up over time. During lease-up, the reserve accrues month by month: the full balance accrues interest, and each month's net operating income, as newly leased units come online, burns down the net carry. That is the mechanical link. Every unit the subject leases offsets a slice of that month's interest, so the pace directly determines how fast the reserve depletes.

Regulators are specific that the property's own income has to be applied first. OCC and FDIC guidance provides that during lease-up, project income should ordinarily revert to the bank and be applied to debt service before any draw on the reserve, and that reserves must not be used to mask a nonperforming credit. The OCC's Comptroller's Handbook goes further, instructing examiners to test the reasonableness of the assumptions in the feasibility study and naming the time allotted for completion and lease-up specifically. The absorption assumption is not just an input the analyst chooses. It is one a bank examiner is directed to challenge.

The failure mode writes itself. A pace that is too fast undersizes the reserve, and the borrower exhausts it before the property stabilizes. When the reserve runs dry, the borrower funds interest out of pocket or negotiates an extension or a recast, and the deal that looked clean at closing becomes a workout. The way to avoid it is to model an honest timeline (if the general contractor says sixteen months, model twenty) and to stress-test a 15 to 20 percent extension. If the reserve hits zero in that scenario, it was sized too small.

The coverage timeline runs on the same engine. Absorption drives the month-by-month occupancy path, which drives effective gross income and net operating income, which drives the debt-service coverage ratio. The project crosses breakeven, coverage of 1.0x, at its breakeven occupancy, often somewhere around 60 to 70 percent depending on rents, expenses, and leverage, though the right figure comes from the subject's own expense and debt-service stack rather than a generic band. It then climbs toward the lender's stabilized floor, commonly about 1.25x on thirty-year amortization for the permanent or agency takeout. Those milestones gate the exit. Agency lease-up loans quote at roughly 1.0x interest-only coverage and 60 percent occupancy and close at 75 to 80 percent; banks quote around 50 percent occupancy and close at 75 to 80 with leasing momentum; CMBS generally closes above 90 percent; and Fannie Mae's near-stabilization execution requires at least 75 percent physical occupancy at rate lock with certificates of occupancy on all units. The absorption pace determines the month each of those gates is reached, which determines when the construction loan can actually be retired.

Put the sensitivity in front of the reviewer rather than making them derive it. For the 250-unit subject with, illustratively, a 50 million dollar construction loan at about 8.5 percent and an eighteen-month build, the construction-phase reserve runs on the order of 50 million times 8.5 percent times 1.5 years times a 55 percent average-balance factor, roughly 3.9 million dollars. Once the loan is fully drawn, gross monthly interest is about 50 million times 8.5 percent divided by twelve, roughly 354,000 dollars a month, offset progressively by NOI as the property leases.

The dollar figures are illustrative and have to be rebuilt on the deal's actual loan terms, expense ratios, and the lender's income-credit convention. But the shape is the lesson. A five-unit-a-month miss, from 15 down to 10, pushes stabilization from seventeen months to twenty-four, keeps the loan near a full interest draw for those extra months, and enlarges the required reserve by more than a million dollars, enough to breach the reserve that looked adequate at 15. To make the same point in return terms, a 200-unit project budgeted for a twelve-month lease-up that stretches to eighteen absorbs on the order of 780,000 dollars in unrecovered carry and compresses going-in IRR by 100 to 200 basis points. A five-unit error is not cosmetic. On the wrong deal, it is fatal.

Writing an assumption that survives

A defensible absorption assumption has four properties, and a reviewer can check for all of them in about two minutes, which is exactly why they should all be present. It is triangulated from three independent sources: comparable lease-ups, the concurrent competitive pipeline, and demand-capture math. It is adjusted explicitly for seasonality, concessions, and project size, with the adjustments shown rather than buried. It is benchmarked to fair share, with any premium above fair share justified on location, product, or price. And it is stress-tested, with the downside case carried all the way through the interest-reserve sizing and the coverage timeline, because that cascade is precisely where a reviewer goes looking for optimism.

The reason to hold to all four right now, rather than treating them as counsel of perfection, is that the market has moved out from under the shortcuts. The convention says twelve months; ALN observes seventeen. Roughly 800,000 units are leasing up at once, concessions are the deepest they have been in over a decade, and the softness is concentrated in exactly the Sun Belt and Mountain submarkets where the most new supply landed. A pace assumption carried over from the low-supply years is not a conservative starting point that a reviewer will forgive. It is the first thing that gets a study sent back.

The analyst who does this work is not trying to produce an optimistic number or a pessimistic one. The goal is a number that has already survived the interrogation it is about to face, so that when the reviewer reaches for the assumption they attack first, they find it has been derived, adjusted, benchmarked, and stressed, and there is nothing left to take apart. That is what it means to test a lease-up assumption before the reviewer does.