According to Erica Adelberg, non-agency securitization has grown from roughly 3% of total MBS issuance in 2022 to nearly 10% in 2025 and is on pace to reach ~12% in 2026 -implying ~$275B in originations this year.

Over the same period, the GSE share of the market has declined from 50% to 38%, even as the 10-year yield nearly doubled.  This is not cyclical, it’s structural.

Affordability pressure: higher rates, elevated home prices, and rigid agency underwriting is pushing more borrowers into non-QM, DSCR, BPL, and portfolio execution channels.

Investor demand has followed, compressing non-QM spreads from 300+ bps in 2022 to roughly 80 bps today.

But beneath that tightening, credit is starting to show stress.

60+ day delinquencies in non-agency collateral have risen to ~6%, compared to ~1% for prime jumbo and just ~0.2% for prime non-jumbo.

The takeaway is straightforward:

The non-agency market is expanding, but it is becoming far less forgiving.

Not all non-agency credit is created equal.
DSCR is not bank statement.
Bank statement is not BPL.
Flexible underwriting is not synonymous with strong credit.

As dispersion increases across products, borrowers, and investor appetites, the margin for error compresses. Small misreads in credit, structure, or pricing are no longer easily absorbed - they are amplified.

Historically, non-agency lending has operated across fragmented systems - pricing in one place, underwriting and overlays managed manually in another, and execution decisions often relying on institutional knowledge rather than systematized logic.

That model breaks down as both volume and complexity rise.

At LoanPASS, we see the market moving toward a model where pricing, eligibility, and underwriting are derived from a single, consistent operating platform - a system we’re preparing to bring to market that unifies product eligibility, pricing, and underwriting logic into one system of truth.