How Lenders Underwrite Assumptions in CRE Financial Models
Origination volume for 2026 is projected at $805.5 billion, a figure that reads like a healthy market until you look at what is coming due against it. An estimated $957 billion in CRE loans matured in 2025, and another $1.5 trillion comes due in 2026. That is refinancing pressure at a scale the market has never had to absorb in a single stretch, and it changes what underwriting is for. Lenders are not funding everyone who shows up with a decent pro forma this year. They are sorting borrowers into who survives a refinance and who does not, and the spreadsheet is just the first filter.
2026 is a sorting year. Banks, CMBS shops, and private credit funds are all running tighter standards against rates that stayed higher for longer than most 2021-era models ever assumed. The tightening pace has slowed, no argument there: only 9% of banks reported tightening standards as of June 2025, down from 30.3% in April 2024 and 67.4% in April 2023. But a slower pace of tightening is not the same as a return to easy money. Lenders are still selective, just no longer panicking about it.
Lender Review of a Sponsor's Model
Underwriting a CRE deal has always been a second, parallel credit evaluation that treats the sponsor's model as a starting point, not a source of truth. It is a second, parallel credit evaluation that treats the sponsor's model as a starting point for the lender's own independent judgment. Institutional lenders follow a consistent framework that evaluates the borrower, the property, and the deal structure, each on its own terms and each measured against the lender's own risk tolerance, not the sponsor's.
The scale of the market is what makes this discipline non-negotiable. Outstanding commercial real estate debt hit $6 trillion by the end of 2024, and banks hold roughly half of it. When one asset class carries that much exposure across the banking system, validating every input becomes a matter of institutional survival, not a courtesy extended to the sponsor.
So what does the lender actually pull, instead of taking the model at face value? Executed leases, not projected ones. Trailing 12-month operating statements. Tax returns. Third-party appraisals. Stress-tested rate scenarios that have nothing to do with the rate the sponsor used to pencil the deal. Every one of those inputs gets normalized against the lender's own benchmarks before a single number from the sponsor's spreadsheet counts as real.
How lenders interrogate rent and vacancy assumptions
Trended rent growth has largely disappeared from lender underwriting, and that is a real shift, not a rounding error. Banks and CMBS shops are far less willing to accept future rent growth as a load-bearing assumption. A model built around steady, incremental annual rent bumps gets flagged before it gets funded, full stop.
Lenders anchor instead to in-place, verifiable cash flow, including income traced back to executed leases, market rent comps that actually support the numbers, the reimbursement structure written into the lease, documented occupancy, and any concessions already on the books. Nothing aspirational survives into the underwritten number.
The scrutiny lands hardest on loans originated during the 2021 boom, many of which are hitting refinance now. Lenders underwriting those deals are testing them against flat or negative rent growth, the opposite of the curve the original model assumed. The rent roll itself gets picked apart line by line: tenant names, lease terms, expiration dates, rental rates, and how bunched the expirations are. A rent roll with a large share of leases expiring in a concentrated window carries rollover risk, and a lender prices that in whether or not the sponsor bothered to flag it.
How lenders evaluate operating expense assumptions
Expense scrutiny has sharpened, and taxes, insurance, and payroll get the most attention. Sponsors tend to carry forward historical averages. Lenders underwrite to what those line items will likely cost going forward, and insurance in particular has only moved in one direction for years.
Lenders check expense ratios against comparable properties in the same asset class and market. A sponsor's model showing management fees or repair costs meaningfully below what comparable buildings actually spend does not read as efficiency to an underwriter. It reads as a red flag, either for deferred maintenance building up somewhere or an operating budget that never matched reality. Every line gets reviewed: taxes, insurance, utilities, repairs, management fees, replacement reserves. Each one moves NOI, and NOI is the number the entire capital stack depends on.
Capital expenditure treatment gets the same scrutiny. A lender's model has to account for tenant improvements, leasing commissions, rollover costs, and reserve requirements, because those numbers show whether the borrower actually has cash on hand to keep the property competitive for the life of the loan. A model that skips reserves to make cash flow look better is a model the lender's underwriting team rebuilds before it ever reaches committee.
How lenders size loans around DSCR and LTV
Debt service coverage ratio is the gating number, full stop. It is NOI divided by annual debt service, so a DSCR of 1.25x means the property throws off 25% more income than it needs to cover principal and interest. Most lenders want a minimum of 1.20x to 1.25x on a standard, stabilized asset, and that climbs to 1.30x or higher for anything riskier, office, retail, or a value-add play.
A January 2026 underwriting memo from a top-tier domestic CRE lender raised its minimum DSCR threshold from 1.20x to 1.25x on stabilized assets, and pushed transitional deals up to 1.30x or higher. That is the direction the whole market is moving: more cushion required, not less.
The real structural change is the order of operations. DSCR now sits at the front of the conversation, and loans get sized off that coverage number first, with leverage acting as a secondary constraint rather than the starting point. That is a meaningful structural shift, and it is the right one. A loan sized off leverage first and coverage second is a loan built to survive an appraisal.
How lenders stress-test exit cap rate assumptions
An exit cap rate underwritten at 4.5% back in 2021 is in a market that actually transacted between 5.25% and 5.75% through 2024, a gap that recurs across nearly every deal originated in the last cycle. That spread does not just dent disposition value. It drags down IRR projections across the entire hold period, because the exit assumption sits at the base of every return calculation stacked on top of it.
CBRE's 2026 forecast projects cap rate compression of 5 to 15 basis points across most property types, which narrows that gap somewhat but does not come close to closing it back to origination-era levels. Any model still carrying a pre-2022 exit cap assumption gets stress-tested against the current range in CBRE's H2 2025 survey, regardless of what the sponsor has already run.
This matters well beyond the exit valuation. If proceeds at sale or refinance cannot cover the loan balance at maturity, the entire deal structure comes apart, not just the return projection. Lenders model refinance feasibility at a stressed exit value, never the sponsor's projected one, because the whole exercise exists to answer what happens when the market does not cooperate.
How lenders approach sponsorship as a parallel underwriting track
Sponsorship gets its own underwriting track, running alongside the property analysis rather than beneath it. Lenders look at track record, financial strength, how the sponsor performed through prior cycles, and the depth of the organization standing behind the deal.
Liquidity carries more weight in 2026 than it did a few years back. Sponsor balance sheets and a demonstrated ability to support the asset through a rough patch now shape deal structure and pricing. Fresh equity has become a prerequisite in many cases: both banks and private lenders are requiring new cash into a deal before agreeing to a modification, a principal paydown, or CapEx funding on a refinance. Sponsors who cannot bring fresh capital to the table are increasingly facing a discounted sale, or losing the asset.
The financial review covers tax returns, balance sheets, existing debt obligations across the sponsor's whole portfolio, net worth thresholds, guarantor strength, and how much capital sits in reserve right now. None of this is new to underwriting. The bar for passing it has just moved up.
Where models break down
The assumptions that sink a deal rarely live on the summary page. They hide in how the inputs got sourced, transferred, and maintained over the model's life, and that is exactly where most sponsors pay the least attention.
Picture the actual workflow inside a lot of sponsor shops: PDFs get routed by email, trailing-12 statements get rekeyed into a spreadsheet by hand, and rent rolls get rebuilt from scratch every reporting cycle. Every one of those handoffs raises the odds of an error, or of one version of the model quietly drifting from another. Spreadsheet-based underwriting runs on manual data entry, version control that depends on someone remembering to save the right file, and an analyst catching every mistake before it compounds. That workflow leaves exactly the kind of inconsistency a lender's document review exists to catch.
Lenders re-underwrite operating statements instead of accepting them at face value for this reason. One-time items get stripped out. Non-recurring income gets excluded. Owner-managed expenses, the kind that do not reflect what a third-party manager would actually charge, get adjusted upward. Once that normalization happens, sustainable NOI often comes in lower than the effective NOI the sponsor's model showed, sometimes by a wide enough margin to change the loan amount.
Stress testing and the lender's worst-case view of a deal
Sensitivity analysis isolates one variable at a time: what happens to DSCR if the interest rate ticks up, what happens if vacancy rises, what happens if expenses grow faster than revenue. Running these one at a time tells the lender exactly which assumption the deal is most exposed to, and that single weak point often becomes the center of loan negotiations.
Scenario analysis goes further and stacks the pressures on top of each other: higher debt service, a slower lease-up than projected, lower proceeds at refinance, all hitting at once. That combination is where lenders find the actual breaking point, the level of stress beyond which repayment capacity simply fails.
Rate volatility gets its own dedicated test, since it touches debt service coverage, refinance feasibility, and borrower performance all at the same time. Lenders use it to gauge exposure loan by loan and across the whole portfolio. The standard battery covers interest rate movement, vacancy pressure, rent declines, expense growth, and capital needs, and each test asks the same underlying question: does the borrower have enough cash flow, enough liquidity, and enough collateral support to absorb the hit without defaulting.
AI and automation's effect on the speed and rigor of assumption review
The mechanics of assumption review have not changed. The speed at which lenders can run them has. Pulling a trailing twelve-month statement, normalizing it, cross-checking a rent roll against lease documents, and running a DSCR sensitivity table used to consume substantial analyst time, most of it just data entry and reconciliation. Automated document processing and machine-assisted underwriting tools are compressing that timeline, letting lenders run more scenarios, catch more rent roll inconsistencies, and flag more sponsor assumptions that do not hold up, in a fraction of the time a manual review ever took.
That does not lower the bar for sponsors. It raises it. A lender that can stress-test many more variations of a deal in far less time has no patience left for a model built on assumptions that cannot survive contact with reality. The tools keep getting faster. The scrutiny they are built to apply only gets sharper.
