UPDATED OCTOBER 2026 · ISSUE N·14 · CAPITAL STACK CLINIC
Verdict: public data does not show a 2008-style, fraud-driven bubble in commercial mortgage-backed securities. It shows a large refinancing gap as loans mature over 2027–2032, driven by higher interest rates and, in office, falling income — plus real but concentrated distress in office, hotels, and a handful of specific deals and banks. Wise Capital's underwriting discipline is built for exactly this environment.
Everyone with capital in commercial real estate is asking the same question in 2026: is CMBS the next 2008? Wise Capital is a Class C value-add multifamily fund for verified accredited investors (Reg D 506(c)) that underwrites every 20- to 60-unit acquisition with its own predictive maintenance platform, ForVue, and targets stabilization within 12 months across Louisville and six other Midwest markets. We tested the bubble thesis against every free public source that bears on it — SEC loan-level filings for 50 CMBS deals worth about $44 billion, the servicers' own monthly reports, and the quarterly filings of all 4,313 U.S. banks. The data is clear about what the problem is, and clear about what it is not.
The Claim We Tested
The 2008 analogue runs like this: borrowers inflated their properties' net operating income (NOI) to win larger loans, those loans were packaged into AAA-rated securities, and the losses will spread through the financial system as they did seventeen years ago. If that pattern is repeating, public data should show five things: income falling well below what lenders underwrote; loans failing soon after they were made; numbers that don't add up across filings; losses large enough to hit bank capital; and loans that can't be refinanced at today's terms.
We tested each one.
How We Did It
The dataset: every monthly loan-level report the SEC requires CMBS trusts to file (Form ABS-EE), for 50 deals — 12 chosen across the major issuers, plus 40 drawn at random from 2017–2023 so the results don't depend on our choices. We checked each reading against each servicer's own monthly statement for all 12 core deals, took split-loan balances from each deal's offering documents, and read 10-D servicer statements in full for the 12 core deals. For banks, we used the quarterly call reports every insured bank files with regulators, covering June 2023 to June 2026 (FDIC BankFind). Market rates and lending terms come from Treasury data (October 9, 2026 daily par yield curve), CRED iQ, Trepp, and Commercial Observer.
How we guarded against our own errors. Twelve errors were found and fixed during the work, each of which would have made the market look worse than it is. Issuer filings mislabeled under the wrong deal; split loans read as tiny pieces; loans one payment late counted as defaults. Every fix was checked against servicer records before it was used.
Finding 1: Underwriting Has Held Up
In a random sample of 40 conduit deals from 2017–2023 (1,884 loans, $34.7 billion in balance), current NOI is within 8% of underwritten NOI for every vintage, weighted by loan balance. If borrowers had invented income to win these loans, real operating statements would now land well below underwriting across the board. They don't.
The 40 deals were drawn at random with a fixed seed from every trust on the 19 major conduit programs for those years. The oldest vintages are the weakest: 2017 income runs 5.4% below underwriting (loans approaching maturity), and 2018 deals are 12.8% delinquent. The strongest: 2020 at +7.6%, 2023 at +4.0%.
Where income fell, it fell by property type. In the 12 core deals examined loan by loan, office NOI was 2% below underwriting and hotels 20% below. Multifamily, retail, industrial, and self-storage ran 10–28% above. Both declines have known causes — tenants leaving offices after 2020, weaker demand at some hotels — not fabricated rents.
First-year defaults are rare and mostly explained. 22 of 1,884 sample loans (1.2%) became 60+ days delinquent within their first year. Half are hotels, most from 2019–2020 deals whose first year ran into the pandemic shutdown. The one exception is BMO 2023-C6, where four loans worth $58.7 million (9.7% of the deal) went bad within about a year of closing — including two apartment loans appraised at $73 million combined, now 44% occupied and losing money. Even here, the servicer has not alleged misrepresentation; the receiver's comments focus on deferred maintenance and capital shortfalls, not invented rents.
The internal-consistency checks mostly came back clean. Low expense ratios came from net-lease properties where tenants pay expenses directly. Low implied cap rates clustered on trophy assets like Rockefeller Center and 601 Lexington Avenue. One statistical test (first-digit Benford on 3,022 NOI figures) deviates from the natural pattern, but the test is weak when applied to loan-sized data. It is a lead to investigate, not evidence of fabrication.
Finding 2: Distress Is Real But Concentrated
4.9% of the random sample's balance is delinquent, against 7.7% for all CMBS as of September 2026 (KBRA). The market figure also covers single-borrower deals and older vintages, so the gap is indicative, not like-for-like.
Office leads. In the 12 core deals, office is 37.5% of the 2021–22 balance and runs 4.8% delinquent, against 2.9% for everything else. Bad loans cluster in a few deals. Of the 12 core deals, four have no delinquent loans at all. One, BBCMS 2022-C14, is 20.7% delinquent — about five times its peers — driven by a Washington D.C. office pair about 50% occupied, two Chicago industrial loans in foreclosure, and a Bronx rent-stabilized apartment portfolio in default with $44.5 million of appraisal write-downs across five troubled loans.
Five servicer case files bear directly on the fraud thesis — Euclid Apartments (738 units, 76% value collapse, trustee suing originator LMF Commercial for alleged misrepresentation), Candlewood Suites Orlando (undisclosed $21.5 million EB-5 debt), 425 Eye Street (modified to a 1% pay rate with interest deferred to 2028 — "extend and pretend" in the open), Lakeville Townhomes (resolved at a 7.16M loss, 99.9% of remaining balance), and One SoHo Square ($785M loan sent to special servicing August 2026 for imminent maturity default). The first two show cases where something did come out — but in both, it took a bankruptcy or a lawsuit to expose it. Public filings catch inconsistency. They cannot catch a borrower who overstates income at origination and keeps overstating it every quarter.
Finding 3: The Refinancing Wall Is the Real Exposure
At today's lending terms, 49.8% of the balance in 12 core deals could not refinance at maturity without new equity. The shortfall averages 10.5% of balance — about $0.95 billion across $9.0 billion tested. Most of these loans were written at 3–4% interest rates; refinancing today costs about 7.5%.
The test: for each loan, could a lender refinance it today? A loan fails if its balance due at maturity exceeds the largest new loan a lender would make, limited by three standard tests — debt yield (8–11% depending on property type), debt-service coverage (at least 1.25×), and loan-to-value (65–75% of today's value, where value = income ÷ 2026 cap rates).
Why loans fail. Of the $4.49 billion that fails: 38% was underwritten below today's minimum debt yield for its type, and three-quarters of that has flat or rising income, so the problem is purely the rate reset. 29% met today's standard when it was made and fails because income fell or value no longer supports it. The remaining 33% lacks underwriting data on the tape.
Timing matters. $6.6 billion of the $9.0 billion tested (73%) matures in 2031–2032, so rates, income, and lender standards have time to change. Where maturities are near, the problem is already visible — One SoHo Square went to special servicing in August 2026 for imminent maturity default.
The random sample confirms the pattern. The same test on 40 random deals gives a 45.4% failure rate and 11.0% shortfall. Across scenarios, the rate runs from 35% (if income grows 2% annually) to 53% (if rates stay at 8.5%). Office fails most at 58%; retail, industrial, and self-storage fail least at 30–33%.
Failing this test is not the same as defaulting. A borrower facing a shortfall can add equity, negotiate an extension, sell, or hand back the property. The shortfall is an upper bound on losses, not a forecast of them. But it is also the real driver here — not invented income. The 2008 losses also arrived at maturity, which makes this the closest public-data analogue to the bubble thesis. The difference: this time the driver is interest rates, not fraud.
We cover the Class C refinancing dynamic specifically in The Class-C Refinancing Cliff Is a CRA Opportunity — the same wall, viewed from the operator side.
Finding 4: Banks Show Pain, Not a Hole
U.S. banks hold $2.36 trillion in commercial real estate loans ($3.08 trillion including owner-occupied). The share 90+ days late or on nonaccrual rose from 0.62% in mid-2023 to about 1.2% by mid-2025, then flattened. Adding loans that banks modified for struggling borrowers — which still count as performing — raises the stressed share to 1.62%.
Losses are small relative to capital. Stressed CRE — bad plus modified — equals 2.14% of banks' Tier 1 capital. Annualized CRE charge-offs ran 0.13% in the first half of 2026, down from about 0.23% in late 2024.
Concentration is falling. Banks whose CRE exceeds regulators' 300%-of-capital guideline fell from 478 to 414 over three years, while the industry's CRE book grew 5%.
But exposure and losses sit in different places. The largest banks (over $100B in assets) hold 72% of bank assets but CRE is only 46% of their capital. Small and mid-size banks (under $10B) carry three to five times the exposure relative to capital — 153% to 231% — with lower losses so far. A downturn would hit their capital harder.
72 problem banks with $141 billion in assets (0.5% of the system) carry several serious warning signs at once. Flagstar Bank ($87.7 billion) is most of that total: CRE at 328% of capital, 7.5% of its CRE book bad, equal to 29% of its capital plus reserves. The rest are small; the most extreme is Nano Banc in California ($0.7 billion), where 34.7% of CRE is bad, exceeding its capital plus reserves.
Where stress could be hiding. 26 small banks (0.2% of system assets) show heavy CRE concentration, almost no reported losses, and rising early-stage past-dues — the pattern unrecognized losses would produce. 96 banks holding 7.8% of system assets have modified-but-performing CRE above 2% of their CRE book.
The Biggest Blind Spot
The lending most likely to contain inflated income — 2021–2022 multifamily bridge loans underwritten on projected (not actual) rents — went largely into CRE Collateralized Loan Obligations (CRE CLOs), private debt funds, and bank balance sheets. CRE CLOs are sold privately and file no public loan-level data. Bank call reports show those loans only in aggregate.
So the absence of fraud evidence in CMBS data says little about CRE CLOs. If a bubble is hiding, that is where it is hiding. It is where any serious investigation into bubble risk should go next.
What public data can and cannot catch. Public filings catch inconsistency: income that doesn't match occupancy, loans that stop paying, numbers that change between filings. They cannot catch a borrower who overstates income at origination and keeps overstating it every quarter — because every filed number then agrees with every other. Only a default, an audit, or litigation exposes that pattern. Euclid Apartments and Candlewood Suites are the cases where something did come out — and in both, it took a bankruptcy or a lawsuit.
What This Means for Class C Multifamily Investors
Three things, directly.
One: the refinancing wall creates buying opportunities. Owners with 2020–2022 fixed-rate loans refinancing into 7.5% rates will face capital calls, forced sales, or distressed workouts. For an incoming buyer at a reset basis, that is pricing power. We underwrite into that environment. Our pipeline includes properties where the current owner's debt economics have broken down — and where our ForVue scoring confirms the operational upside before we commit.
Two: the multifamily refi failure rate is misleading at the Class C level. The 56.0% headline is driven by large-metro bridge loans underwritten on projected rents in trophy A/B assets — not operator-managed value-add Class C in Midwest markets. The Class C 20- to 60-unit segment where Wise Capital operates has different lenders (CDFI, HUD 221(d)(4), HUD 223(f), local banks), different leverage (65% target LTV at stabilization), and different income profiles (HCV-backed in-place rents versus projected market-rate rents). The segment-level risk is materially lower than the aggregate CMBS multifamily number suggests. See CDFI vs HUD: Two Capital Stacks for Class C Multifamily and HUD 221(d)(4) vs 223(f) for how we think about financing specifically.
Three: ForVue is built for this environment. When refinancing math gets tight, NOI variance is what kills a deal. Every $1,000 of unexpected capex at a 7.5% refinance rate is roughly $15,000 off the appraised value and often the difference between a clean refi and a capital call. ForVue scores every appliance in every unit before we close — turning the largest NOI variance line item in Class C (reactive maintenance) into scheduled line items. See the Bourbon Town case study for the live deployment.
Our structure matches our thesis. Wise Capital Fund is a Reg D 506(c) blind-pool closed-end fund managed by Wise Family Holdings LLC (Nevada LLC). Minimum investment is $50,000. Up to 8% preferred return per annum (simple, non-compounding, at Manager's sole discretion, not guaranteed). Form D notice filings are current in Kentucky, California, Illinois, and Pennsylvania (August 8, 2026 amendment). We don't underwrite to speculative appreciation. We underwrite to operational alpha — the discipline this market rewards.
Access the Investor Portal to review the full fund documentation.
Frequently Asked Questions
Is there a CMBS bubble in 2026?
Public data does not show a 2008-style, fraud-driven CMBS bubble. Across 50 CMBS deals (~$44 billion) and 1,884 random-sample loans, current NOI sits within 8% of underwritten NOI in every vintage from 2017–2023. Distress is real but concentrated: 4.9% delinquency in the random sample vs. 7.7% market-wide, led by office at 37.5% of 2021–22 balance. The real exposure is a 2027–2032 refinancing wall driven by higher rates and falling office income — not systemic fraud.
What percentage of CMBS loans fail to refinance at today's terms?
Approximately 49.8% of the balance in 12 core deals and 45.4% of the random 40-deal sample cannot refinance at today's terms (7.5% rate, 2026 cap rates, standard debt yield and DSCR tests) without new equity. The shortfall averages 10.5–11% of balance — about $3.4 billion on the $30 billion random sample. By property type, office faces 73.3% refinancing failure, multifamily 56.0%, self-storage 47.3%, hotel 36.3%, industrial 30.7%, and retail 28.1%.
Which property types face the most CMBS refinancing risk?
Office carries the highest refinancing failure rate at 73.3% of balance, with a 16.8% average shortfall. Multifamily follows at 56.0% failure and 14.2% shortfall — but concentrated in large-metro bridge loans underwritten on projected rents in 2021–2022, not in operator-managed value-add multifamily. Retail is the most resilient at 28.1% failure and 2.3% shortfall. This matters because the Class C value-add segment, where Wise Capital operates, carries materially different risk than the trophy-asset bridge loans driving the multifamily headline number.
Is bank CRE lending in crisis?
Not system-wide. U.S. banks hold $2.36 trillion in CRE loans. The share 90+ days late or on nonaccrual rose from 0.62% in mid-2023 to 1.2% by mid-2025, then flattened. Adding modified-but-performing loans raises the stressed share to 1.62% — equal to 2.14% of banks' Tier 1 capital. The pain is real but contained. 72 banks (0.5% of system assets) carry serious warning signs, led by Flagstar at $87.7 billion with 7.5% of its CRE book bad.
What is the biggest blind spot in CMBS data?
CRE Collateralized Loan Obligations (CRE CLOs), private debt funds, and bank balance sheets. The lending most likely to contain inflated income — 2021–2022 multifamily bridge loans underwritten on projected (not actual) rents — went largely into CRE CLOs. CRE CLOs are sold privately and file no public loan-level data. Bank call reports show those loans only in aggregate. So the absence of fraud evidence in CMBS data says little about CRE CLOs; that is where any serious investigation into bubble risk should go next.
What does the CMBS data mean for Class C multifamily investors?
The refinancing wall creates buying opportunities for disciplined operators. Owners with 2020–2022 fixed-rate loans refinancing into 7.5% rates will face capital calls or distressed sales. For an incoming buyer at a reset basis, that is pricing power. Wise Capital's underwriting discipline — 65% target LTV at stabilization, 20–60 unit focus (outside the bridge-loan multifamily segment driving the headline risk), and ForVue predictive maintenance to protect NOI — is specifically designed for this environment. We don't underwrite to speculative appreciation. We underwrite to operational alpha.
Christopher Wise, J.D. Managing Principal, Wise Capital Five-time Super Lawyers Rising Star · Louisville Business First 20 People to Know in Law (2023, 2026)
Related reading
The Class-C Refinancing Cliff Is a CRA Opportunity
CDFI vs HUD: Two Capital Stacks for Class C Multifamily
Louisville Class C Multifamily: Why This Market, Specifically