The word "subprime" still scares mortgage investors. It probably always will. But treating today's non-QM market like a rerun of 2008 is lazy thinking, and a fresh outlook from Nomura gives the industry the numbers to argue back.
HousingWire's read of Nomura's 2026 Securitized Products Outlook puts Q2 2025 non-QM originations at the highest credit scores the segment has ever produced. Across roughly nine years of existence, the box has tightened, not loosened.
What happened
- Non-QM Q2 2025 originations posted record-high credit scores, per Nomura's 2026 outlook.
- Since Q4 2022, the share of low-FICO and high-LTV loans in non-QM pools has been roughly halved.
- Self-employed workers in the U.S. Number between 16.5 million and 27 million, depending on definition, per the Center for American Progress.
- Average self-employed income sits near $156,000 a year, against about $123,000 for traditional wage earners, per St. Louis Fed data cited in the piece.
- Post-crisis rules require securitization sponsors to hold at least 5% of credit risk on the deals they package.
What subprime actually was
Subprime and its near-cousin Alt-A began with a sensible job: serve borrowers with real income who could not document it through a W-2. Then the discipline evaporated. Stated-income products appeared, followed by NINJA loans that required no proof of anything. By the mid-2000s the original idea was buried under paper handed to people with no realistic path to repayment.
The mechanics mattered as much as the credit profile. Lenders sold loans into securitization pools within days and had no economic stake in performance. That alignment is what turned a bad underwriting choice into a systemic event.
Why non-QM is a different product
Non-QM serves people whose finances do not translate neatly onto a tax return. Self-employed borrowers, commission earners, contractors, and pass-through entity owners often look worse on paper than they are in reality, because they deduct legitimate expenses or carry income that swings by season. A twelve-month bank-deposit record tells a different story, and non-QM underwriting is built to read it.
The credit profile does the rest. Q2 2025 origination FICO scores sit at all-time highs for the segment. The share of low-FICO and high-LTV paper has dropped by roughly half since late 2022. That is the opposite direction from the pre-2008 build-up.
Risk retention shifts the incentive picture too. Sponsors of mortgage-backed securities must hold at least 5% of credit risk under the Credit Risk Retention Rule, finalized by six federal agencies in 2014. When the originator keeps skin in the game, the math on sloppy underwriting changes.
Our analysis
The stigma is mostly inertial, not analytical. Investors who write off non-QM as a subprime proxy are applying 2008 priors to 2025 data, and the data refuses to cooperate. FICO distributions have moved up, the riskier LTV buckets have thinned out, and the borrower base is dominated by affluent self-employed professionals whose tax returns undersell their actual cash flow.
The bigger structural story is the gap between how Americans now earn a living and how agency underwriting measures it. The W-2-and-thirty-year-fixed model was drawn for a salaried workforce buying one home and staying put. That workforce is a shrinking share of the labor market, while the self-employed count has grown into the tens of millions. Non-QM exists because the borrower base moved on and the agency credit box did not.
Who benefits if the stigma fades? Originators and securitization sponsors with disciplined files get access to a borrower pool that is, on the numbers, better qualified than the conventional market assumes. Who loses? Competitors who keep treating non-QM as radioactive and hand that borrower to someone else. For investors, the trade is straightforward: stronger credit, a tighter box, and a regulatory floor that did not exist last cycle. Watch pool-level delinquency on 2024 and 2025 vintages as the first real stress test.
A worked example
Say you are a private-label MBS investor reviewing two shelf offerings. One is a 2007-vintage subprime RMBS benchmark sitting in your model, the other a 2025-vintage non-QM deal. Old habit says treat them as comparable risk. Pull the pool tape instead. The recent deal shows a weighted average FICO at record highs, a low share of high-LTV loans, and full documentation through bank statements or asset schedules. Hold that against your 2007 loss assumptions and the required credit enhancement starts to look generous. The mispricing, if it exists, runs the other way from the stigma.
What to watch next
Track three things. First, delinquency paths on 2024 and 2025 non-QM vintages through any 2026 economic softening, because that is the cleanest test of the tighter box. Second, agency updates to self-employed documentation rules, which would either shrink the non-QM market or pull credit standards toward it. Third, securitization issuance volumes. If sponsors keep printing deals with record credit scores and a thinning high-LTV share, the subprime label keeps losing ground whether the industry wants it to or not.
Key takeaways
- Non-QM credit quality in 2025 is at historical highs by FICO and LTV mix, per Nomura.
- The borrower base is high-earning self-employed workers whose returns mask real cash flow.
- The 5% risk-retention rule removes the originate-to-distribute incentive that drove 2008.
Frequently asked questions
Is non-QM lending the same as subprime lending?
No. Non-QM serves borrowers who cannot fit agency documentation rules, often self-employed earners with strong cash flow. Subprime in its 2008 form meant low-FICO borrowers with no income verification and no documentation.
What does Nomura's 2026 outlook say about non-QM credit scores?
It reports that Q2 2025 originations carry the highest credit scores the non-QM sector has recorded, and that the share of low-FICO and high-LTV loans has been cut roughly in half since Q4 2022.
Why do self-employed borrowers use non-QM loans?
Because conventional underwriting reads tax returns, and self-employed borrowers often have write-offs, pass-through income, or seasonal swings that understate their actual earnings on paper.
How does risk retention change non-QM incentives?
Under the Credit Risk Retention Rule finalized in 2014, securitization sponsors must keep at least 5% of the credit risk, so they have direct exposure to loan performance rather than the originate-to-sell model of the mid-2000s.
Should commercial real estate investors care about non-QM trends?
Indirectly. Non-QM is a residential product, but it affects housing supply for high-earning self-employed buyers, and shifts in residential credit can move demand between urban condos, suburban houses, and multifamily investment property.