The Concession Gap: Why Underwriting to Asking Rent Breaks a Multifamily Pro Forma

Every multifamily pro forma has one line that does more damage than any other when it is wrong, and it sits at the very top. It is the rent. Not the vacancy factor, not the expense ratio, not the exit cap; the top-line rent, because everything below it is a percentage of it. Underwrite to the rent a unit is advertised at rather than the rent it will actually collect, and the error does not stay contained. It compounds downward through effective gross income, net operating income, coverage, debt yield, supportable loan, and value, arriving at a loan request a credit committee will either reject on re-underwriting or fund into an asset that later cannot carry it.

The gap between those two rents is concessions, and underwriting to asking rent instead of effective rent is the most reliable failure mode in the multifamily book. Our own multifamily research anchors this analysis at a 16.9 percent concession share, a magnitude that external benchmarks confirm is squarely representative of today's market rather than a worst case. This note walks the gap from its definition through the arithmetic, the current data, the waterfall into coverage and value, and the standard the agencies already enforce, and ends where a feasibility study should: with the discipline that makes the number survive.

Two rents, and the gap between them

Asking rent, also called face rent, gross rent, contract rent, or advertised rent, is the headline monthly rate a unit is marketed at: the number in the listing, at the top of a casual rent survey, and at the top of a seller's or developer's pro forma. Effective rent, net effective rent (NER) or economic rent, is what the landlord actually nets once concessions are amortized across the lease term. The difference between the two is concessions, and in the current market that difference is neither small nor rare.

The formula is worth stating precisely, because reviewers check it. Net effective rent equals total gross rent over the lease term minus total concessions, divided by the number of months in the term; equivalently, face rent multiplied by one minus months free divided by lease term. Take an $1,800-a-month unit on a twelve-month lease with one month free. The tenant pays $1,800 for eleven months, or $19,800; spread across twelve months, effective rent is $1,650. The $150 gap is an 8.33 percent discount. Scale that one month free across a 200-unit property and it is $360,000 of reduced annual income, money that comes off net operating income dollar for dollar.

The reason to model the effective number is not conservatism for its own sake. As the standard technical reference on net effective rent frames it, analysts and appraisers use NER rather than face rent "because face rents in high-concession environments systematically overstate landlord income," and "using face rent in valuation without adjusting for concessions overstates net operating income and inflates estimated property value." That is the thesis of this note in two clauses.

The arithmetic of free rent

Because concessions are annualized, their effect on effective rent is close to linear and easy to carry in one's head. One month free on a twelve-month lease is an 8.33 percent cut; a month and a half free is 12.5 percent; two months free is 16.67 percent. Lengthen the lease and the same giveaway dilutes, so one month free on a thirteen-month lease is 7.7 percent. The consequence that matters for this note: a concession-driven discount in the high teens maps almost exactly to two months of free rent on a standard twelve-month lease.

Free rent is the dominant multifamily concession, but it is not the only one, and each hits effective rent differently. Reduced or waived security deposits are a one-time cash benefit with a small annualized effect. Waived amenity, application, and admin fees, or reduced parking, trim other income modestly. Gift cards and "look and lease" move-in bonuses, commonly $500 to $2,500, are one-time value that annualizes to a small percentage. Rate buy-downs offer a discounted rate for an initial period before stepping to contract rent, common in lease-up. Tenant-improvement allowances, primarily a commercial device, reduce effective rent directly: a $50-per-square-foot TI on a five-year lease at $30 per foot cuts effective rent to roughly $20 on a simple basis.

Two distinctions separate a careful analyst from a careless one. The first is gross concession value versus effective-rent discount: the gross give-up on one free month is $1,800, but annualized against $21,600 of gross rent it is 8.33 percent, which is why two months free (16.67 percent) is not a promotion but a sixth of the year's revenue. The second is burn-off, or how the concession is taken. An upfront concession zeroes month one and then charges full contract rent; a spread concession lowers net rent every month. Total cash over the term is identical, but the monthly pattern is not, and the renewal consequence is real: the concession usually disappears at renewal, so a one-month-free tenant faces a step-up from $1,650 effective to $1,800 face, a jump large enough to trigger a move-out, and a risk that a pro forma assuming clean retention ignores.

A near-record share of the market, unevenly spread

This matters more now than in most years because concessions are close to a record and broadly distributed. Zillow's spring 2026 rental report found 39.8 percent of listings offering a concession, up five percentage points year over year and more than double the pre-pandemic level; the company put it plainly: "A year ago, roughly 1 in 3 rental listings offered a concession. Before the pandemic, it was closer to 1 in 6," which was 14.4 percent as recently as September 2019. The readings through the cycle set records repeatedly, at 37.3 percent in September 2025, 39.3 percent in November 2025, and 41.1 percent in January 2025, a then-record. Zillow's framing of the stakes is a useful reminder of the dollars: with the income needed to afford the typical U.S. rental near $77,200 a year, "a free month means roughly $1,930 back in your pocket."

The dispersion is extreme, and it is the dispersion that underwrites a specific deal. In spring 2026, Zillow had Denver at 68.3 percent of listings discounting, Charlotte at 66.6, Dallas at 64.2, Austin at 63.8, and Nashville at 62.6 percent at the top of the table, against Buffalo at 11.1, Providence at 12.6, New York at 18.4, New Orleans at 19.2, and Chicago at 21.7 percent at the bottom. The Sun Belt and Mountain West supply markets and the tight Northeast are not living in the same rental economy, and a national average is not something a lease is signed against.

Other providers, measuring different things, tell a consistent story. RealPage reported roughly 22 percent of apartments offering concessions in the third quarter of 2025 at an average value of 6.2 percent of rent, and 17 percent of vacant units offering a concession in May 2025, its highest May reading since 2013. Apartment List found about 35 percent of properties offering at least one month free as 2025 closed, up from 25 percent at the start of the year, with Denver near 58 percent and Phoenix, Jacksonville, Raleigh, and Austin above 50; in Austin, roughly half of conventional properties were discounting by the third quarter of 2025. CoStar's Apartments.com RentPulse put 41.2 percent of properties offering concessions nationally in the first quarter of 2026, up 9.9 points year over year, at a 2.0 percent national concession rate, with Fort Myers (5.3 percent), Asheville (4.4), Denver (4.0), Austin (3.9), and Phoenix (3.8) deepest, and Sarasota an outlier at 81.8 percent of properties, nearly half advertising two months free. On magnitude, JPI's Jay Parsons found stabilized apartments averaging about five weeks free nationally in late 2025, the most since the global financial crisis, with Florida metros such as Sarasota near eight weeks and Austin, Nashville, Denver, Phoenix, and Salt Lake City above six; Austin listings carried two to three months free into early 2026. ALN's submarket data runs about 9 to 10 percent of rent in the softest regions. The metrics differ, but the direction does not.

Why our anchor is 16.9 percent

Our multifamily research anchors this analysis at a 16.9 percent concession share, and the figure is worth dwelling on because it is both proprietary and, on the external evidence, squarely representative. RealPage independently reported 16.9 percent of stabilized apartments offering concessions in May 2026, up about four points year over year and near the highest monthly level since mid-2014, at an average discount of 10.9 percent, roughly six weeks free on a twelve-month lease. That two readings of 16.9 percent land on the same number from different methods is a coincidence of magnitude rather than of definition, and the coincidence is the point: whether the figure is read as a share of units advertising concessions or as an effective-discount magnitude, third-party data corroborates it on both dimensions.

It also converts cleanly. A 16.9 percent concession-driven discount maps to essentially two months of free rent on a standard twelve-month lease, which is the assumption the rest of this note rides on. The 16.9 percent figure is our proprietary finding and is treated here as given; the external benchmarks, Zillow near 40 percent of listings, RealPage's 16.9 percent of stabilized units and 6.2-to-10.9 percent average discount, and CoStar's 41.2 percent of properties, confirm the magnitude without reproducing our specific internal metric, which may measure a different denominator.

How the overstatement compounds

The damage from underwriting to asking rent is not that the top line is a little high. It is that the error grows as it descends. In a standard pro forma, gross potential rent, every unit at market and fully occupied, less loss-to-lease, vacancy, bad debt, and concessions, plus other income, yields effective gross income; effective gross income less operating expenses yields net operating income. Omit or understate the concession line and both effective gross income and net operating income are inflated, and the second by more than the first.

The reason is that operating expenses do not fall when rent is discounted. Consider a 200-unit property asking $1,800, for gross potential rent near $4.32 million a year, at a 40 percent expense ratio, holding vacancy constant to isolate the concession effect. On asking rent, effective gross income is about $4.32 million, expenses about $1.73 million, and net operating income about $2.59 million. Introduce the two months free the market is actually clearing at, a 16.67 percent haircut, and revenue falls to about $3.60 million while expenses stay at $1.73 million, leaving net operating income of about $1.87 million. Revenue fell 16.7 percent; net operating income fell about 28 percent. That amplification is the central, under-appreciated mechanic, and it runs the wrong way for a borrower: the higher the expense ratio, the more a given revenue haircut magnifies into the NOI line.

None of this is new to the cycle. Concessions spiked in the 2020 COVID shock, receded in the 2021-2022 rent boom (Zillow's pre-pandemic share was near 16 percent), then surged again with the 2023-2025 supply wave that peaked at a record of roughly 588,900 deliveries in 2024. RealPage's stricter measure had concession share on conventional stock near 8.5 percent at the end of 2024, already near pandemic-era highs, and national asking rents turned negative year over year in the back half of 2025; RealPage put October 2025 at minus 0.7 percent, the deepest since March 2021. A pro forma that assumes concessions burn off instantly is fighting the direction of the market.

From NOI to DSCR to the size of the loan

Everything a lender decides flows from net operating income, so a 28 percent NOI overstatement is a 28 percent problem at every gate. Debt service coverage is net operating income divided by annual debt service, and debt service does not change because the rent roll disappointed, so if NOI is overstated by a given percentage, DSCR is overstated by the same percentage. In the 200-unit example, an underwriter carrying $2.59 million of NOI against, say, $2.0 million of debt service shows a comfortable 1.30x. The real $1.87 million produces 0.94x, below breakeven. The deal that looked like it cleared coverage with room to spare does not cover its debt at all.

The same gap resizes the loan. Lenders size backward from a minimum coverage, commonly 1.25x for stabilized multifamily, so maximum supportable debt service is NOI divided by that minimum. At the overstated $2.59 million, that is about $2.07 million of supportable debt service; at the real $1.87 million, about $1.50 million. Roughly 28 percent less debt-service capacity means a roughly 28 percent smaller loan. The debt-yield test tightens the screw from the other side: debt yield is NOI divided by loan amount, and lenders size to a floor, commonly around 10 percent for CMBS and 8 to 9 percent for the agencies, so a 28 percent NOI overstatement inflates the debt-yield-constrained loan by 28 percent as well. Because a lender takes the most restrictive of loan-to-value, coverage, and debt yield, an inflated NOI can breach two constraints at once, and the loan comes back materially smaller than requested, often the difference between a fundable and an unfundable deal.

The valuation multiplier

For an equity holder or a lender taking a value view, the same overstatement is larger still, because value divides NOI by a small number. At a 5.5 percent cap rate, every dollar of NOI carries about $18 of value, so $100,000 of overstated NOI is roughly $1.82 million of overstated value. In the 200-unit case, the roughly $720,000 NOI overstatement from using asking instead of effective rent implies about $13 million of overstated value at that cap rate. For scale, a 50-basis-point cap-rate move on a $10 million-NOI asset moves value on the order of $1 million on its own; layering an NOI overstatement on top of cap-rate uncertainty multiplies the error rather than adding to it.

The market is already showing the divergence between asking and effective rent that this arithmetic anticipates. RealPage's effective asking rent fell 0.3 percent in the third quarter of 2025 and 0.7 percent year over year by October; Yardi's advertised-rent growth slipped to 0.2 percent year over year in November 2025, its lowest since the first quarter of 2021. Denver is the vivid case: reported rent fell about 5 percent year over year to $1,816 in the third quarter of 2025, but effective rent after concessions was materially lower, with a new downtown 22-story lease-up offering two months free plus bonuses and a 2024-built tower offering fourteen weeks free. And the portfolio averages understate the problem at the margin: MMCG's database showed national asking rent of $1,761 against effective of $1,740 in the third quarter of 2025, a gap near 1.2 percent that widens to about 1.5 percent for Class A, but that stabilized-portfolio spread coexists with 8 to 17 percent discounts on newly signed leases in heavy-concession submarkets. Underwriting has to use the marginal number for the units that will actually turn or lease up, not the portfolio average.

The agencies and appraisers already require it

None of this is a house view. The standard a rigorous pro forma should meet is the standard the loan will actually be sized to, and the agencies wrote it down. Fannie Mae's Multifamily Selling and Servicing Guide, at Part II, Chapter 2, Section 203.01, defines gross rental income as "actual rents in place for occupied units, plus market rents for vacant units based on a current rent roll," not 100 percent market or asking rent. Concessions are an explicit deduction, defined as "the aggregate amount of forgone residential rental income from incentives granted to tenants for signing leases, such as free rent for 1 or more months, move-in allowance, etc.," with bad debt subtracted separately. There is a hard floor: physical vacancy plus concessions plus bad debt "must equal the greater of the difference between the trailing 3-month net rental collections (annualized) and GPR, or 5% of GPR." Fannie's effective gross income is net rental income plus other income less bad debt and rent concessions; lenders must consider whether the property can achieve underwritten cash flow within twelve months; and Fannie's own commentary tracks the very problem it is guarding against, with roughly 19.7 percent of Class A units offering concessions in the fourth quarter of 2024.

Freddie Mac is, if anything, blunter. Its appraisal guidance, "Modeling Multifamily Potential Rental Income" (June 2022), states that "modeling rental income at 100% market rents is not the correct method," and that an appraiser who does so has "incorporated a Hypothetical Condition into their analysis which, by definition, is not an As-Is estimate of market value and is not in compliance with the Guide." The correct basis is "the existing rent roll plus any vacate units at market rent." Guide Section 60.14c requires forecast gross income to weigh comparable rents "adjusted for market concessions, rent abatements, discounts and the like," and Section 60.2 requires the "as-is leased fee market value"; Freddie's ABS-15G disclosures confirm loans are characterized on in-place rents for occupied units. Appraisal practice under USPAP runs the same way: the income approach distinguishes contract from market rent and adjusts for concessions in deriving stabilized NOI, the Appraisal Institute cautions that ignoring market-appropriate concessions in direct capitalization can overvalue by double digits, and The Dictionary of Real Estate Appraisal lists concessions among the lease terms that must be analyzed.

Lenders enforce the same discipline in the underwriting itself. Construction and permanent lenders routinely trim developer pro formas, cutting aggressive rents, imposing untrended rents, adding replacement reserves of $250 to $500 per unit, and stress-testing coverage on higher vacancy or lower rents before sizing. "Trending" assumes future rent growth; "untrended" holds rents flat at today's level, and in softening or high-supply conditions lenders push developers to untrended rents, "building at tomorrow's cost for today's rents," precisely to strip optimism out. Underwriting to asking rent is a form of unearned mark-to-market in disguise: it assumes concessions burn off instantly, which is exactly the assumption the trim is designed to remove.

Economic vacancy, loss-to-lease, and the traps in between

Two related concepts govern how the concession loss shows up in a model, and confusing them is its own failure mode. Physical vacancy counts empty units; economic vacancy is gross potential rent less effective gross income, over gross potential rent, and it folds in concessions, non-revenue model and employee units, and bad debt. Two properties at 5 percent physical vacancy can sit at very different economic vacancy, one above 10 percent if it is discounting heavily, and lenders underwrite to the economic figure, because a unit occupied by a tenant paying a concession-discounted rate is not producing full rent. The stabilized spread between physical and economic occupancy typically runs two to five points. Loss-to-lease, market rent divided by actual rent minus one, is the separate, below-market portion of in-place leases; concessions are the give-up against face rent. Both are income leakage a rigorous pro forma must carry, and a common error is to assume aggressive one-year recapture of a 10 percent loss-to-lease while simultaneously assuming concessions vanish.

The recurring ways pro formas go wrong all trace back to the same optimism. Analysts survey advertised rents, which landlords deliberately keep high, favoring concessions over rent cuts precisely to "maintain the advertised value of their properties for their lenders and investors," and treat those face rents as achievable revenue while ignoring the concessions attached to the same listings. They assume concessions burn off at stabilization when, in high-supply markets, the concessions persist across multiple leasing cycles; RealPage had lease-up properties taking roughly sixteen months to reach stabilization at the end of 2024, and concessions are a pull-forward of demand, not a permanent gain. They bury concession give-ups in a marketing or admin expense line, so an underwriter who then cuts marketing at stabilization double-counts the benefit. And they assume day-one asking rents at stabilization for exactly the newly delivered Sun Belt product that must offer two to three months free to lease at all. The cascade is what reaches the credit committee: because the overstatement inflates NOI, coverage, debt yield, and value at once, a deal underwritten to asking rent is either sent back when the lender re-underwrites to effective rent and the loan no longer sizes, or it closes and then underperforms as effective rents come in below the underwritten face, the strain Colliers described in 2026 in noting that "some assets that were underwritten to aggressive post-pandemic rent expectations are straining under the lower-than-anticipated cash flow."

A word of caution on the data itself, which is why we treat the 16.9 percent anchor as proprietary and the outside figures as corroboration rather than proof. Providers disagree because they measure different things over different coverage: Zillow's roughly 40 percent "share of listings" is not RealPage's or CoStar's 2-to-11 percent "concession value as a share of rent," and neither is RealPage's stricter 8.5-to-16.9 percent "share of stabilized units." Colliers' first-quarter 2026 read put concession usage lower still, near 25.4 percent of units at about $129 per unit, against a late-2009 peak of 64.2 percent. Each figure is useful for what it measures; none should be conflated with another.

Underwriting the rent you will collect

The discipline that survives re-underwriting is straightforward to state and harder to enforce. Survey comparables for their current concessions, mystery-shopping competing properties, pulling the concession line from RealPage, CoStar, and ALN submarket data, and reading the actual move-in specials rather than the posted rent. Start from in-place effective rents and current occupancy rather than a pro forma "stabilized" number, and build only market-supported increases; capping recapture assumptions at 3 to 5 percent a year, as we do, is more defensible than assuming a 10 percent loss-to-lease closes in a single year. Carry the concession and loss-to-lease haircut as an explicit line item, sourced from the rent roll and twelve to twenty-four months of property-management reports rather than the seller's single reported figure, since sellers understate concessions because a higher effective rent supports a higher price. Model burn-off only against documented events, lease expirations and dated absorption of the supply pipeline, and underwrite current concessions as the base case, with burn-off reserved for an explicitly labeled upside. And present the sensitivity: NOI, coverage, debt yield, supportable loan, and value under asking rents, under effective rents, and under a continued-concession case with 100 to 200 basis points of additional vacancy, so the lender sees the downside before the credit committee does.

The threshold that changes the recommendation is specific. If the effective-rent coverage falls below the lender's minimum, typically 1.25x stabilized, the deal has to be resized or repriced at the screen, not at committee. If the continued-concession case breaches the debt-yield floor, near 8 to 9 percent for the agencies and about 10 percent for CMBS, or breaches coverage, the loan request should be cut to the most-restrictive-constraint amount before it is submitted. Concessions touch a near-record share of the market, the softness is concentrated in exactly the high-supply submarkets where the most product is delivering, and the agencies already underwrite to effective rent. A pro forma built on asking rent is not merely optimistic. It is out of step with how the loan will actually be sized, and the concession gap is the most reliable place a lender's re-underwriting will find the difference.