The loan that dies on page 40
Every credit committee has the same story. The deal looks clean at LOI: pencils-out cap rate, stable asset, sponsor with a track record. Then the underwriter hits page 40 of the rent roll and finds three co-terminous expirations inside the loan term, one at 140% of market. The T-12 shows a $180K insurance jump the sponsor quietly normalized out. The Phase I recommends further investigation. Debt yield slips from 9.1% to 8.3% when the underwriter re-cuts NOI.
That is commercial real estate loan underwriting. The deal does not die because the lender was hostile. It dies because the documents tell a different story than the LOI, and the lender does not share the upside if the story turns out to be right.
What lender underwriting actually is
Commercial real estate loan underwriting is the process a lender runs to size debt and decide whether to approve it. Two decisions fall out of the same analysis:
- Loan sizing: how much debt the property supports at the lender’s coverage and yield thresholds.
- Credit approval: whether the lender will make that loan, to that sponsor, under that structure.
A property can support a $40M loan mathematically and still be declined because the sponsor is over-leveraged at the portfolio level, or the lender is overweight on office in that MSA. Sizing is arithmetic. Approval is judgment.
Every lender runs the same sequence: pull in-place income, normalize it, apply underwriting assumptions (market vacancy, capex reserves, management fee), arrive at underwritten NOI, and test it against the requested loan under four metrics. The loan is sized to the binding constraint.
Lender vs. equity underwriting: different risk lens
Equity and debt underwrite the same property asking different questions.
| Dimension | Equity underwriting | Lender underwriting |
|---|---|---|
| Primary question | What return can I earn? | Will I get my principal back? |
| Time horizon | 3-7 year hold, focus on exit | Loan term (5-10 yr) + refinance risk |
| Upside sharing | Captures upside above pref | Coupon only, no upside |
| Key metric | IRR, equity multiple, CoC | LTV, DSCR, debt yield |
| View on pro forma | Believes the growth case | Stresses it; underwrites to in-place |
| View on rent | Market rents at stabilization | Lesser of in-place and market |
Equity is optimistic because its payoff is asymmetric on the upside. Debt is skeptical because its payoff is asymmetric on the downside. When a lender’s underwriting looks “too conservative,” you are reading the contract correctly.
For the equity-side view, see our pillar on commercial real estate underwriting.
The four metrics that size every loan
Loan sizing is a race to the most restrictive of four tests. The smallest result wins.
1. Loan-to-Value (LTV)
LTV is loan amount divided by appraised value. It caps leverage against the asset’s collateral value.
LTV = Loan Amount / Appraised ValueTypical thresholds in 2026:
| Lender type | Max LTV (stabilized) |
|---|---|
| CMBS | 65-75% |
| Life insurance | 55-65% |
| Banks (balance sheet) | 60-70% |
| Agency (Fannie/Freddie) | 75-80% (multifamily) |
| Debt funds / bridge | 70-80% (on stabilized basis) |
The gotcha: “appraised value” means the lender’s appraiser, not yours. If your LOI priced at a 5.25% cap and the lender’s appraiser comes in at 5.75%, value drops, LTV rises, and the loan gets resized down. That is a common reason loans get cut at close.
2. Loan-to-Cost (LTC)
For construction and ground-up development, LTC replaces LTV because there is no stabilized value yet.
LTC = Loan Amount / (Land + Hard Costs + Soft Costs + Financing)Construction lenders typically cap at 60-70% LTC, with sponsor equity funded before the first draw. Bridge lenders on value-add stretch to 75-80% LTC but carve out “as-stabilized” tests that reprice at stabilization.
3. Debt Service Coverage Ratio (DSCR)
DSCR tests whether the property’s cash flow can service the debt.
DSCR = NOI / Annual Debt ServiceA DSCR of 1.25x means the property generates 25% more NOI than required debt service. Typical minimums:
| Asset type | Min DSCR |
|---|---|
| Multifamily (agency) | 1.25x |
| Multifamily (bank/CMBS) | 1.25-1.30x |
| Industrial | 1.30-1.35x |
| Office | 1.35-1.45x |
| Retail (grocery-anchored) | 1.30-1.40x |
| Retail (unanchored) | 1.40-1.50x |
| Hospitality | 1.40-1.50x |
Lenders compute DSCR on underwritten NOI, not T-12 NOI. They strip one-time items, normalize rent, haircut vacancy to market, add a management fee if the sponsor self-manages, and mark up capex reserves. The sponsor’s 1.40x DSCR often becomes the lender’s 1.22x.
4. Debt Yield
Debt yield emerged after 2008 because DSCR and LTV broke down in a low-rate environment.
Debt Yield = NOI / Loan AmountUnlike DSCR, debt yield is independent of interest rate and amortization. A 3% coupon with 30-year amortization can inflate DSCR on a weak deal; debt yield stays honest.
| Lender type | Min debt yield |
|---|---|
| CMBS | 8-10% |
| Life insurance | 9-11% |
| Banks | 8-10% |
| Bridge (stabilized) | 7-9% |
Most CMBS originators will not size below 8.0% debt yield regardless of LTV and DSCR. That is the binding constraint on most deals in 2026.
Lender types and what they want
Not all capital is the same. Matching the deal to the right lender saves weeks.
| Lender type | Best for | Loan size | Rate (2026 indicative) | Recourse | Prepay |
|---|---|---|---|---|---|
| CMBS | Stabilized office/retail/industrial, $10M+ | $10M - $500M+ | T+200-275 | Non-recourse | Defeasance or yield maintenance |
| Life insurance | Institutional-quality, long-term hold | $10M - $300M+ | T+150-225 | Non-recourse | Yield maintenance |
| Banks (balance sheet) | Relationship borrowers, mid-market, $5M-$50M | $2M - $100M | SOFR+200-300 | Often recourse | Flexible |
| Debt funds | Transitional, heavy value-add | $10M - $200M | SOFR+400-700 | Non-recourse w/ guarantees | Lockout + exit fee |
| Bridge lenders | Acquisitions needing repositioning, 12-36 mo | $5M - $100M | SOFR+350-600 | Varies | Lockout + exit fee |
| Fannie Mae / Freddie Mac | Multifamily only, 5+ units | $1M - $100M+ | T+150-200 | Non-recourse | Yield maintenance or graduated |
The pattern: the more stabilized and institutional the asset, the cheaper the debt and the less recourse required. The more transitional or complex, the higher the spread and the more structure the lender wants.
Agency lenders (Fannie Mae DUS, Freddie Mac Optigo) are the deepest pool of capital in U.S. commercial real estate because they buy multifamily debt at scale. For stabilized multifamily in a major MSA, agency is almost always the cheapest option, and their underwriting standards are published (a rare transparency in this industry).
The lender underwriting package
The documents a lender will request after LOI. Missing any stops underwriting.
Property-level documents
- Current rent roll: tenant, unit/suite, SF, base rent, start/expiration, options, escalations, expense treatment (NNN, modified gross, full service)
- Trailing 12 (T-12): monthly P&L for the last 12 months
- Historical financials: 2-3 years of audited or tax-return-level financials
- CapEx history: past 3-5 years of capital spend
- Operating budget: forward 12-month projection
- Lease files: executed leases for all tenants above a revenue threshold (often >5% of rent)
- Tenant financials: for credit tenants or concentrated tenants
- Property tax bills and insurance certificate: current and prior year
Third-party reports
- Appraisal (MAI): lender-ordered. Produces the value that caps LTV.
- Property Condition Assessment (PCA) / PNA: building systems, immediate repairs, 10-12 year capital plan
- Phase I Environmental Site Assessment: per ASTM E1527-21 standard
- Phase II Environmental: only if Phase I identifies a recognized environmental condition (REC)
- Zoning report: zoning, compliance, legal non-conforming status
- Title commitment and ALTA survey: title exceptions
- Seismic (PML) report: California and other seismic zones
Sponsor-level documents
- Sponsor REO schedule: portfolio of owned properties, debt, equity, occupancy
- Personal financial statement (PFS): if recourse or guaranty involved
- Last 2-3 years of tax returns: sponsor and key principals
- Sponsor bio/track record: deals closed, assets under management
- Entity documents: borrower SPE operating agreement, certificate of good standing
For the borrower-side perspective, see due diligence for lenders and the broader commercial real estate due diligence pillar.
Stress tests lenders run
Every serious lender runs scenarios beyond in-place. The common ones:
Rate shock / refinance stress. Can this property refinance at term at a higher rate? Lenders underwrite a “takeout” DSCR assuming rates rise 100-200 bps from origination. If the property still hits 1.20x at refi, it passes. This is where many 2021-vintage loans failed: they pencilled at 3.5% but could not refinance at 7%.
Vacancy haircut. The lender applies market vacancy: 5-7% for multifamily, 8-12% for office, 5-10% for retail, 4-7% for industrial, regardless of actual occupancy. A 98% occupied office still gets underwritten at 90%.
Expense inflation. T-12 expenses get inflated to the larger of trailing actual, budget plus 2-3%, or the appraiser’s market expense line. Taxes assume reassessment at purchase, a critical haircut in California, Texas, and Florida where freezes can undershoot reassessment.
Mark-to-market rents. For leases within the loan term, lenders often underwrite the lesser of in-place and market. A tenant paying 140% of market whose lease rolls mid-term gets marked to market at rollover.
Capex reserve. A structural capex reserve (typically $250-$400/unit/year for multifamily, $0.20-$0.30/SF/year for office and retail) comes off NOI before DSCR is calculated.
The cumulative effect is a substantial NOI haircut. Sponsors who model to the LOI cap rate without running the lender’s haircuts set themselves up for a reprice at close. See our guide to real estate financial modeling for the mechanics.
Non-recourse vs. recourse
Recourse determines whether the lender can pursue the sponsor personally if the property does not cover the debt.
Non-recourse loans: the lender’s only remedy is the collateral (the property). CMBS, life insurance, and agency loans are almost always non-recourse with “bad-boy carve-outs”: fraud, material misrepresentation, waste, unauthorized transfer, environmental violations, or bankruptcy filings trigger personal liability for the sponsor.
Recourse loans: the sponsor (or a guarantor entity with real assets) signs a personal guaranty. If the property’s cash flow and foreclosure proceeds fall short, the lender can pursue the guarantor. Most bank loans below $25M are full or partial recourse.
Partial recourse: a hybrid where the sponsor guarantees a portion of the loan (typically 25-50%) that burns down as the property hits performance milestones.
The pricing spread: non-recourse typically costs 25-75 bps more than equivalent recourse debt because the lender is pricing in the reduced remedy.
Typical 30-60 day timeline from LOI to close
A clean closing timeline for a stabilized $30M multifamily deal at a life company:
| Week | Phase | Key activities |
|---|---|---|
| 0 | LOI signed | Term sheet finalized, good-faith deposit posted |
| 1 | Kickoff | Lender orders appraisal, Phase I, PCA, title, survey |
| 2 | Due diligence | Borrower delivers rent roll, T-12, leases, sponsor package |
| 3-4 | Underwriting | Model built, DD calls, third-party reports returned |
| 5 | Credit committee | Lender approves |
| 6-7 | Loan docs | Drafting and redline negotiation |
| 8 | Close | Funds wired, documents recorded |
CMBS adds 2-3 weeks for rating agency review. Banks compress to 30 days on relationship deals. Bridge lenders close in 21-30 days when pre-approved.
Where timelines slip: almost always on the borrower side. Rent rolls that don’t tie to the T-12, backlogged Phase I consultants, and unresolved title exceptions are the top three delays.
How AI accelerates loan underwriting
The document-handling layer of loan underwriting is exactly the kind of work AI is good at: extract structured data from messy PDFs, normalize to a schema, flag anomalies, produce a summary. A new category of tools has emerged for this workflow.
Blooma focuses on commercial lending: automated property underwriting for banks and debt funds, with loan origination system integrations. See our full write-up of Blooma alternatives for the broader category.
Moraine (Atlas) handles the document layer: rent roll normalization, T-12 reconciliation, lease abstraction, PCA/PNA review, zoning verification, title exception flagging. Built for the underwriter who spends 60% of their time keying documents into spreadsheets.
Clik.ai specializes in rent roll and T-12 extraction with a services layer.
What these tools do not do: make the credit decision. The underwriter still runs the model and defends the deal to committee. What changes is the prep time. A deal that used to take two analysts five days can now close prep in two days with one analyst, with fewer errors, because extraction is auditable against source documents.
Common borrower mistakes that kill loans
Using the LOI cap rate to size debt. The LOI cap is buyer-seller. Lender sizing uses the appraiser’s cap rate, often 25-50 bps higher. A 70% LTV at the LOI cap may be 76% at the appraiser’s.
Normalizing out real expenses. Insurance spikes, tax reassessments, and deferred capex are not “one-time” in the lender’s eyes. Anything normalized out gets added back.
Ignoring rollover. Concentrated lease expirations inside the loan term spike refinance risk and depress debt yield. Lenders respond with TI/LC reserves or rent haircuts at rollover.
Sponsor liquidity shortfall. Most lenders require post-close liquidity of 10% of loan amount and net worth of 1x the loan. Check the PFS against thresholds before credit committee.
Phase I ordered too late. Environmental turnaround runs 15-25 business days. Order it at LOI, not week 3.
Unprepared rent roll. Property management software rarely exports what lenders need: missing expense treatment, unclear escalations, inconsistent expirations. Clean rent rolls close on time. Messy ones do not.