Multifamily Development: From Construction Through Lease-Up

Key Takeaways

Two Phases, One Deal

Multifamily development is not one phase of work. It is two distinct economic regimes joined at a single moment in time. The construction phase, which runs 24–36 months for institutional product, is entirely uses-side: capital flows out to land, hard cost, soft cost, and capitalized interest, with zero revenue. The lease-up phase, which runs 6–18 months on top of construction, is a mixed regime: operating expenses and cash debt service are live, rental income ramps from zero up to a stabilized run-rate, and the gap is bridged by a lease-up reserve until net operating income covers debt service.

The moment that joins them is the Certificate of Occupancy — the C of O — and almost no competitor walks it line by line. The lender-funded interest reserve that carried the loan through construction depletes shortly after C of O. A separate, smaller operating-deficit reserve takes over. The construction loan becomes eligible for agency take-out within a defined window. Each of those mechanics has its own clock, and they don't all fire on the same day. Practitioners who model the development as a single "to stabilization" timeline collapse the inflection — and miss the months where the deal is most likely to break.

This article walks the full lifecycle of an institutional multifamily ground-up as one integrated modeling problem. The audience is development analysts and VPs underwriting merchant-build and build-to-core sponsors; capital markets associates sizing construction debt and the eventual take-out; LP investors reviewing sponsor pro formas before subscription; and acquisitions analysts evaluating recently delivered Class-A product still in lease-up. The running deal example is a 280-unit garden-style ground-up in a Sun Belt submarket, $62M total project cost, 30-month construction plus 12-month lease-up, carried through every section.

Phase 1: Construction (24-36 Months)

The construction phase begins when the construction loan closes. Pre-development costs — land contract deposits, predevelopment legal, civil engineering, entitlement consulting, geotechnical and environmental reports — are typically capitalized into the basis carried into closing. At closing, the construction lender funds the initial draw to take down land and reimburse capitalized soft costs. From that day forward, every dollar that enters the project enters as a use; nothing enters as revenue.

Monthly progress draws follow the classic S-curve: slow in months 1–6 as sitework, demolition, and foundation work proceed; peak velocity in months 7–22 during vertical construction, MEP rough-in, envelope, and drywall; tapered through months 23–30 as finishes, FF&E, and common areas come online and the leasing center opens. The construction-loan mechanics — AIA G702/G703 contractor's applications for payment, 10% retainage on hard costs, monthly lien-waiver compliance, and independent construction management consultant inspections — are walked in detail in our companion piece on construction loan draw schedules and interest reserves. For the development walkthrough, the relevant takeaway is that the loan's average outstanding balance over the construction window is 60–65%, not 50% — and the interest reserve has to be sized to that S-curve, not to a flat half-loan assumption.

Debt service during construction is paid by a lender-funded interest reserve. The reserve sits in sources and uses as a line item inside the construction loan itself — the lender funds the reserve out of loan proceeds, and the loan accrues interest on the reserve as it does on any other drawn balance. This is the structural reason there is no cash debt service from the sponsor during construction: every interest payment is "paid" by drawing on the reserve, which is itself part of the loan. The loan grows against itself as construction proceeds.

Equity and debt fund the project in a sequenced ratio. Most construction loans require sponsor equity to fund first — the "equity-in-first" structure. For a $62M project with $18M equity and a $44M construction loan, the first $18M of project costs are paid from equity (land takedown, initial site work, early soft cost burn), and the construction lender funds every dollar thereafter. By the Certificate of Occupancy date, all $18M of equity is invested and the construction loan is drawn close to its full commitment. The funding ratio progression is a one-way street: equity goes in first, loan goes in second, and neither one is returned until refinance or sale.

The C of O Moment

The Certificate of Occupancy is the discrete inflection that separates the two phases. In a phased delivery (typical for garden-style and large podium product), C of O issues building by building over a 3–6 month delivery window; in a single-building mid-rise or high-rise, it issues all at once on a single date. The institutional convention is to treat the date of first leasable unit as the C of O moment for modeling purposes, even when later buildings deliver weeks or months afterward.

Four mechanics shift on that date, and they do not all fire on the same day:

• Revenue begins. The first leases sign in the days and weeks leading up to C of O for units that will be ready at delivery; revenue posts as rent commencements after C of O. In an institutional pro forma, gross potential rent goes from zero to a small fraction of stabilized in the first month and ramps from there.

• The lender-funded interest reserve depletes. The reserve was sized to cover interest through construction completion plus a 60–90 day "stub" past C of O. Once that window passes, the reserve is empty and the construction loan still has interest accruing. The borrower must now service the loan in cash — either from operations (which won't cover for months) or from a separate reserve.

• The lease-up operating-deficit reserve takes over. This is a separate, smaller reserve sized to cover the operating deficit during lease-up. HUD 221(d)(4) sizes it at 3% of loan amount as a program requirement; balance-sheet banks size it deal by deal, typically as 3–6 months of debt service plus 3–6 months of stabilized operating expenses. The reserve funds: real estate taxes (now assessed on improved value), property insurance, base property management staffing, leasing commissions and marketing, free-rent concessions burning through signed leases, and the debt service shortfall between cash interest and lease-up NOI.

• Permanent debt eligibility unlocks. Agency lenders (Fannie Mae, Freddie Mac) and HUD can refinance the construction loan once the asset is in lease-up under specific programs that do not require full physical stabilization. The agency take-out window opens at C of O for products like Fannie Mae's Near-Stabilization Program and Freddie Mac's Lease-Up loan, even though the loan actually closes weeks or months later after underwriting completes.

The reason this section matters — and the reason most competitor content collapses it — is that the four mechanics fire on overlapping but not identical timelines. The interest reserve depletes on a deterministic date (the day the last drawable interest dollar is drawn). The operating-deficit reserve begins funding on the day operations begin (typically C of O itself, sometimes a few weeks earlier when the leasing center opens). Revenue posts as leases sign. And the agency take-out is a future-state option whose value depends on rate environment at the time of refinance, which the sponsor doesn't control. The deal lives or dies in the gap between when the interest reserve runs out and when the operating deficit reserve plus lease-up NOI covers cash debt service.

For the 280-unit example, model the C of O as the start of month 31 (after 30 months of construction). The interest reserve has roughly 60 days of stub. The operating-deficit reserve was sized at $1.8M (approximately 4% of the construction loan), covering an estimated 6 months of base operating expenses plus cash debt service shortfall. Pre-leasing, which began in month 25 with the leasing center opening, has produced 24 signed leases at C of O. Move-ins in months 31–33 take physical occupancy from 0% to roughly 9%, with monthly rent revenue of $54K against monthly cash interest of $293K and monthly operating expenses of $94K — an operating deficit of $333K/month that the reserve has to absorb.

Phase 2: Lease-Up (6-18 Months)

Lease-up is the period from C of O to stabilization. Institutional convention for general-occupancy garden-style multifamily is 10–20 units per month of net absorption, per NMHC research and consulting-firm benchmarks. Tier 1 Sun Belt metros (Atlanta, Dallas, Phoenix, Charlotte) historically absorb 15–25 units/month in healthy markets and 8–15 in oversupplied submarkets. Mid-rise and high-rise product runs slower because the move-in friction is higher (elevator scheduling, finish-out coordination, larger unit floor plates), typically 8–15 units/month for general-occupancy product per NMHC and submarket-specific data from RealPage Multifamily Market Insight.

For the 280-unit garden-style example, model a 12-month lease-up at an average net absorption of 22 units/month. Months 1–3 run hotter at 25–30 units/month as pre-leased units commence and initial marketing pull is strongest; months 4–8 settle into the 20–22 unit/month steady state; months 9–12 taper to 15–18 units/month as the marketing-elastic demand exhausts and the property converges on stabilized occupancy. Net absorption nets out the early-cycle move-outs that show up around month 6 onward (job relocations, lifestyle changes) — institutional models assume 8–12% annualized turn from the moment leasing begins.

Concession burn-off is the second leg of the lease-up math. Properties opening into a competitive submarket offer 1–2 months free on a 13-month lease as the standard institutional concession; aggressive submarkets push to 2–3 months free on a 13–14 month lease. The face rent on the rent roll is the asking rent; the effective rent — the rent the property actually collects — is the face rent net of concession amortized over the lease term. A 2-month concession on a 13-month lease is an 15.4% effective rent discount; a 3-month concession on a 14-month lease is 21.4%. The concession burn-off is when concessions stop being offered on new leases — usually at 70–80% occupancy, when management has demonstrated absorption and pricing power has returned. For the example, model concessions of 1.5 months on a 13-month lease for months 1–8 (a 11.5% effective discount on new leases), tapering to 0.5 month in months 9–10 and zero in months 11–12.

The NOI ramp is the financial output of the absorption curve. Month 1 NOI is materially negative: operating expenses are live (base management, taxes, insurance, marketing, leasing commissions) but rental income is a fraction of stabilized. Month 6 NOI typically crosses zero as occupancy hits the high 30s and concession-discounted rent revenue starts covering operating costs. Month 12 NOI hits roughly 88% of stabilized run-rate for a healthy lease-up. Physical stabilization — the agency convention of 90% occupancy for 90 consecutive days — typically clears between months 10 and 13. Economic stabilization, defined as trailing 90-day NOI annualizing to a level that supports the take-out underwrite, lags physical stabilization by 3–6 months because the trailing window captures concession burn-off and ramping expenses.

The relationship between net absorption pace and lease-up reserve drawdown is the most important sensitivity in the development model. At 22 units/month, the operating-deficit reserve in the example depletes roughly $1.4M over the 12-month lease-up and the deal hits stabilization with $400K of reserve cushion remaining. At 15 units/month — the 2026 Sun Belt reality in oversupplied submarkets per RealPage — stabilization moves to month 17 and the reserve is exhausted in month 12. The sponsor has to fund the remaining 5 months of operating deficit from equity or from a contingency reserve. This is the failure mode that takes capital from "performing" to "underperforming" without ever defaulting on debt service.

Unit mix interacts with absorption pace in ways the simpler models don't capture. Studios and 1BRs lease faster but at higher annualized turn; 2BRs and 3BRs are slower to lease but stickier once leased. Density economics — whether the deal is garden, mid-rise, or high-rise — drive both the absorption pace ceiling and the rent premium structure; see our sibling piece on garden vs mid-rise vs high-rise density economics for the institutional breakdown of how density format changes the underwrite. The rent roll itself, which becomes the central artifact of the stabilized underwrite, is the subject of our multifamily underwriting fundamentals piece.

Agency Take-Out at Stabilization

The institutional construction loan is a transient structure. It exists to fund construction and bridge lease-up; it is not designed to sit on the asset long-term. The take-out is the moment the construction loan is refinanced into permanent debt, and for institutional multifamily it almost always lands on an agency loan from Fannie Mae or Freddie Mac (or in some cases HUD 221(d)(4) as a single-close construction-to-permanent product, as covered in our piece on HUD/FHA 221(d)(4) construction-to-perm).

Two specific agency programs are designed for the C2P transition:

• Fannie Mae Near-Stabilization (NSP). Sizes off near-stabilized in-place NOI rather than fully trailing NOI. Requires occupancy of 70–90% (program-dependent) and a credible path to full stabilization within a defined window (typically 4–12 months from rate lock). The loan funds at closing with a partial holdback that releases as stabilization milestones are hit. Walker & Dunlop's practitioner content on near-stabilization is the strongest reference on the program mechanics; the program is the workhorse for the 2026 take-out vintage where lease-up timelines have stretched.

• Freddie Mac Lease-Up. Designed for assets still in active lease-up, typically 50–75% occupied at quote. Loan funds in two stages: initial funding sized off in-place NOI (so the loan is materially smaller than the final stabilized loan) and a holdback that releases at stabilization. The sponsor accepts a lower initial loan amount in exchange for locking the rate earlier — valuable when the rate environment is volatile or expected to move higher. Freddie's Lease-Up product is published on the Freddie Mac Multifamily site.

The qualifying thresholds matter for modeling. For the 280-unit example, model the take-out as a Fannie Mae NSP loan refinancing the $44M construction loan at month 38 (8 months past C of O, 2 months into economic stabilization). Sized at 65% LTV against a stabilized valuation of $84M (stabilized NOI of $4.5M capitalized at a 5.35% cap rate) gives a loan amount of $54.6M — refinancing the $44M construction loan plus returning approximately $10M of the original $18M equity to the sponsor. Underwrite the take-out rate at the May 2026 indicative rate of 5.85–6.00% for a 10-year fixed Fannie Mae non-recourse loan; the take-out underwriting also requires a 1.25–1.30x DSCR test against in-place NOI at funding.

The take-out rate environment is the single largest external variable in the development pro forma. A 1% move in the take-out rate moves the supportable loan amount by roughly $4–5M on a deal this size, and a 1% move in the going-in cap rate at exit moves stabilized value by approximately $10M. Sponsors who locked debt early in 2021–2022 against an assumed exit cap of 4.25–4.50% are refinancing into 2026 at exit caps of 5.00–5.50%, and the take-out is materially smaller than underwritten. The C2P transition modeling, which DV-003 walks line by line, is the place where this rate risk gets quantified.

The 2026 Supply Environment

The 2026 development underwrite cannot be lifted from a 2021 model. The supply environment, the debt cost environment, and the absorption environment all moved materially.

Starts hit a 30-year low. The U.S. Census Bureau's New Residential Construction release showed multifamily starts (5+ unit) at a 316,000 annualized rate in May 2025, the lowest monthly print since the early 1990s and a 30.4% month-over-month decline. The starts collapse was driven by the combination of elevated construction debt cost (SOFR + 275–400 bps for institutional sponsors, with all-in coupons of 7.30–8.55%), squeezed development spreads (going-in cap rates rose faster than yields-on-cost as construction costs rose), and the supply wave of 2022–2024 still working through lease-up. Most submarkets that delivered 500K+ units annually in 2024–2025 stopped breaking new ground in 2025.

Recovery beginning, slowly. The NMHC March 2026 Quarterly Construction Survey showed 48% of respondents reporting starts unchanged QoQ, 31% reporting they are starting more projects, and only 12% starting fewer. This is the first survey since early 2023 with more respondents leaning into starts than backing away. The NAIOP Sentiment Index Q1 2026 showed similar directional recovery: construction debt spreads tightened 25–50 bps from the Q3 2025 peak, equity requirements eased modestly (balance-sheet banks now at 35–45% equity, debt funds at 25–35%), and lender appetite for ground-up multifamily improved meaningfully relative to other CRE asset classes. Agency loan purchase caps for 2026 are $88B each for Fannie and Freddie ($176B combined, +20.5% from 2025), with at least 50% mission-driven mandate, and the MBA's CREF Forecast projects total multifamily originations at $399B in 2026 (from $331B in 2025).

Sun Belt lease-up has stretched. RealPage Multifamily Market Insight showed lease-up velocity slowing from the historical 12-month stabilization timeline to 15–17 months in Sun Belt markets bearing the brunt of the 2024–2025 supply wave. Orlando, Austin, Miami, Nashville, and Phoenix added 4–5% to existing stock in 2026–2027 per ULI/PwC's Emerging Trends in Real Estate 2026, and the deliveries continuing into 2026–2027 are still soaking demand that historically would have absorbed competing lease-ups in 12 months. For the 280-unit example, a Tier-1 Sun Belt submarket like Charlotte or Atlanta can credibly underwrite a 12-month stabilization in 2026 only if the immediate competitive set has thinned; an Orlando or Nashville comparable deal needs to underwrite 15–17 months.

NCREIF returns are recovering off lows. Stabilized multifamily total returns in the NCREIF Property Index apartment sub-index turned positive year-over-year in late 2025 after eight consecutive quarters of negative appreciation. The improvement is concentrated in stabilized core product; development-stage and recently delivered Class A continues to lag because effective rent remains depressed by concessions. Development spreads (yield-on-cost minus going-in cap) compressed to historical lows in 2022–2024 and have begun widening again as cap rates rose faster than the yield-on-cost the math could support. The institutional threshold for breaking ground — historically 150–175 bps of development spread — is currently being met by only the best-located, best-structured Sun Belt deals.

Worked Example: 280-Unit Sun Belt Garden

A 280-unit garden-style ground-up in a Tier-1 Sun Belt submarket. Total project cost $62M ($221K all-in per unit), 30-month construction plus 12-month lease-up, 42 months from land closing to stabilization. The full sources and uses:

Sources and uses for the worked example. The construction loan is the dominant capital source, with sponsor and LP equity at 29% of total cost. Capitalized interest and lease-up reserve are sized at the central case; the developer carries an additional 5% hard + soft contingency.

The construction loan at $44M, SOFR + 300 bps (May 2026 all-in approximately 7.55%), 30-month term plus a 12-month lease-up tail (42 months total). Interest reserve sized to $2.4M based on the S-curve average outstanding of 60%. Lease-up reserve sized at $1.8M (approximately 4% of the construction loan).

Yield on Cost and Development Spread

Stabilized NOI on the example, modeled at month 42, is $4.5M: $26,400/unit average rent on 280 units times 92% economic occupancy minus $1,950/unit operating expenses gives stabilized cash-on-cash NOI of $4.49M. Yield on cost — the foundational development metric — is stabilized NOI divided by total project cost: $4.5M / $62M = 7.26%.

Going-in cap rates for stabilized Sun Belt garden-style Class A multifamily in May 2026 are running 5.00–5.25% per Marcus & Millichap and CBRE research summaries; mid-rise and high-rise product in primary metros trades 25–75 bps tighter. For the example at a 5.10% going-in cap, the development spread is 216 bps — just above the institutional threshold. A 25 bps move in the exit cap (to 5.35%) and a 5% cost overrun simultaneously would push the development spread to roughly 160 bps, still inside the institutional band but no longer comfortable. A combination of 50 bps exit cap drift and 10% cost overrun pushes the development spread below 100 bps — the deal still cash flows, but doesn't compensate the equity for development risk relative to buying a similar stabilized asset.

The development spread interacts with the take-out math directly. At a 7.26% yield on cost and a 65% LTV take-out, the unlevered yield ($4.5M) supports a $54.6M loan at 5.85% on a 30-year schedule (DSCR roughly 1.29x). At a 6.50% yield on cost — which is what a 10% cost overrun does to the example — the unlevered yield ($4.5M against $68M cost) still supports the same loan in absolute terms, but the equity return falls because the cost basis is higher. The sensitivity is more pronounced on the equity multiple than the loan sizing.

Where the Math Breaks

Three failure modes show up in 2026 deals, each connecting back to a modeling decision the analyst made before the deal closed:

• Lease-up reserve depletes before NOI covers cash debt service. Modeled absorption was 22 units/month; actual delivered 12–15 because a competing lease-up opened in the same submarket 90 days ahead. The reserve runs out in month 9 of lease-up; the sponsor has to fund the remaining months of operating deficit from equity (a sponsor capital call) or from a contingency reserve if one was sized in. The deal stays performing on debt service but the equity return moves from a target 16% IRR to 10%, and the sponsor's promote crystallization gets delayed.

• Concession burn lasts longer than modeled. The submarket has 2–3 competing lease-ups delivering within 6 months of yours. Effective rent runs 8–15% below face rent through month 18, not month 11 as modeled. Stabilized NOI lands at $4.0M rather than $4.5M because of the depressed effective rent persisting into the trailing 90-day window for the take-out. The take-out loan sizes off in-place NOI, so the loan is materially smaller than underwritten — sometimes by $4–6M.

• The take-out doesn't size to par. Trailing 90-day NOI fell short, or the going-in cap moved against the deal, or both. The Fannie Mae NSP loan that was supposed to fully refinance the $44M construction loan sizes to $40M instead. The sponsor has to recapitalize the $4M gap with mezzanine debt (12–15% coupon, expensive), preferred equity (similar economics, slightly better structure), or a sponsor capital call. The take-out structure becomes one of the most consequential modeling decisions; our piece on bridge loans and floating-rate exit assumptions walks the structural alternatives in more depth.

Each of these failure modes is foreseeable with the right modeling discipline: a stress case on absorption pace, a stress case on effective rent persistence, and a stress case on take-out rate movement, all running in parallel with the central case. Institutional underwriting standards have converged on reporting central, downside, and severe-downside cases for every variable, with the severe-downside case engineered to test whether the equity survives or has to recapitalize.

From Pro Forma to Institutional Model

Every section of this article reflects a discrete piece of math that an institutional development model has to carry: S-curve construction draws against a hard and soft cost budget, capitalized interest solved iteratively against the loan balance, the C of O inflection with two reserves handing off, a lease-up module with configurable absorption and concession burn, the NOI ramp through stabilization feeding the take-out underwrite, and a full equity waterfall with promote hurdles tied to LP returns at refinance or sale. The math itself is mechanical; the structural integration — making one model carry both phases without stitching together a construction model and a stabilized model by hand — is the hard part. Most practitioner models on the open web do one phase well and the other approximately; the institutional pro forma carries both end to end.

For a deeper walk on the rent roll mechanics that feed the stabilized NOI build, see our piece on multifamily underwriting fundamentals and the rent roll. For the calculator-style breakdown of how operating expenses scale across unit count and unit mix, see our rental property pro forma calculator piece. For the density-format breakdown that drives both the hard cost budget and the absorption pace ceiling, see garden vs mid-rise vs high-rise density economics.

Related Articles

• Rental Property Pro Forma Calculator — the operating-side companion that walks NOI build and operating expense scaling on stabilized multifamily.
• Garden vs Mid-Rise vs High-Rise: Density Economics — how density format changes hard cost, rent premium, and absorption pace ceiling.
• Multifamily Underwriting Fundamentals and the Rent Roll — the central artifact of the stabilized underwrite that the development model feeds into.
• Construction Loans: Draw Schedules and Interest Reserves — the deep dive on S-curve draw mechanics, the circular reference, and interest reserve sizing.
• HUD/FHA 221(d)(4) Construction-to-Perm — the agency alternative to the bank-construction-plus-take-out structure walked here.
• Bridge Loans: Floating-Rate Risk and Exit Assumptions — the structural cousin when the take-out doesn't size to par and the sponsor needs a bridge to a future refinance.

FAQ