Weekly · Issue No. 16
September 14, 2026
Every Monday, 6:30 AM ET · 8-minute read · No paywall
01

Briefings

The VantageScore 4.0 pilot is over, and lender choice is now an investor question

The change was telegraphed; the timing was not. In a late-evening post on September 3, FHFA Director Bill Pulte ordered Fannie Mae and Freddie Mac to accept VantageScore 4.0 from all lenders, effective immediately, fourteen months after the agency first approved the model for conforming mortgages and four months into a limited rollout. Fair Isaac fell roughly 20% the next day, one of its largest single-day declines. The WSJ covered the borrower side a week later: a lender can now run Classic FICO and VS4 on the same file and deliver whichever produces the better GSE economics, with the LLPA grids the GSEs published on September 9 setting score thresholds 20 points higher for VS4 to account for differences in score distributions.

The mortgage trades have argued about lender choice for months. On the securitization side, where its effects will show up in actual pools, the conversation has barely started. Fannie and Freddie back roughly half of U.S. mortgage debt, and credit score is a primary input to underwriting and loan-level pricing across that market. Each loan is priced on one model, chosen by the lender, and until this week the MBS tape showed only that score. On September 9, FHFA directed that every loan in GSE MBS, CRT, and other securitized products carry both. Implementation details are pending, but Fannie's CRT files were already scheduled to add VS4 on October 26, so that is the earliest point at which investors can begin comparing the two scores on delivered loans.

That disclosure matters, because the problem is not comparability but selection. A lender holding both scores can choose the one that produces the better pricing bucket. The 20-point adjustment addresses the difference in score levels; it cannot eliminate the information advantage of lender choice. The loans that migrate into a better pricing bucket are not a random sample; they are, by construction, the loans for which the alternative model was more favorable. Milliman's 2025 study of 45 million GSE loans carrying both scores found that lender-choice scores run about 20 points above Classic FICO on average, and that default rates within a given score cohort were roughly 30% higher under lender choice than under FICO alone. The asymmetry is clear: the originator knows both scores at delivery, and the GSEs price on one.

VantageScore has called adverse selection a myth, arguing that GSE rules and lender controls constrain the choice and that the more predictive model should win on the merits. It also has more mortgage history than the narrative sometimes suggests: the GSEs published roughly a decade of historical VS4 scores on their loans, and the Milliman study itself draws on them. What no one has is live-market performance on pools whose loans were selected and delivered under a lender-choice framework. VantageScore reports that VS4 was used on roughly 9% of GSE securitizations between May and August; that history is only beginning to accumulate.

When VantageScore executives Rikard Bandebo and Tony Hutchinson joined the CCM podcast earlier this year, comparability and disclosure were the open questions. Disclosure has since moved; selection has not.

CCM read: Score choice is a real benefit at origination for some borrowers. There is no sign yet of a pricing effect in agency or CRT markets, and at current adoption there shouldn't be. But uncertainty about a credit gauge tends to be priced as margin over time, and wider margins on agency collateral flow back to lenders and borrowers alike. With both scores now on the tape, the divergence between them on delivered loans offers the first live read on how aggressively lenders are optimizing, and is worth tracking pool by pool. October's CRT data will begin to show which side the pools support.

Core CPI didn't clear Warsh's "clearly and at sufficient speed" bar

At Jackson Hole, Chair Warsh said the Fed must be confident that underlying inflation is moving to target "clearly and at sufficient speed. Otherwise, we have work to do." Friday's print was the first test of that sentence. Headline CPI came in as expected. Core did not, printing 0.3% for the month against a 0.2% forecast, and the market read it as a failed test: hike odds for Wednesday went from about 70% to nearly 90%. That is the market's verdict on whether the print cleared Warsh's bar.

The consumer's is different. Gasoline rose 3.9% in August and is up 27.4% over twelve months; energy overall is up 16.3%. The Fed can look through that. A consumer auto or card borrower cannot, and the basket wage earners actually buy, measured by CPI-W, rose a tenth faster than the headline index. Tuesday's G.19 showed nonrevolving credit growing at a 4.8% annualized rate in July, its fastest in more than a year, against 2.5% for cards. Consumers are still borrowing, and increasingly for autos and education rather than on plastic.

Two smaller moves cut the other way: motor vehicle insurance fell 0.8% in August, its second consecutive decline, and used-vehicle prices remain 2.3% below a year ago, modestly supportive of recovery values. Neither offsets a fuel shock of this size, and with Brent finishing the week above $104, up about 9%, the September print will not offer relief either.

CCM read: Gasoline has been elevated since March's 21% m/m jump, so the payment-stress lag has already run. The place to look is not next month's CPI but current delinquency by credit tier in auto and card loan-level data, where the spring fuel shock should now be visible. Second-quarter readings from the New York Fed and TransUnion showed auto delinquencies elevated but not accelerating; the summer remits will say whether that pattern held. If delinquencies stayed flat through the fuel shock, that is a story about consumer resilience worth taking seriously. If they turned up, a rate hike on top of them extends the duration of stress rather than the severity.

The Senate takes up CLARITY on Tuesday, and the fight over stablecoin rewards is a deposit-funding question in disguise

I spent Thursday, September 10, at a Stablecoin Week side event at PubKey in Washington, an invited guest at a Chatham House Rule discussion that was more useful for its range than its consensus. The panels ran from crypto maximalists who would replace the dollar-based monetary system outright to former regulators focused on the narrower work of fitting stablecoins into existing bank and payments infrastructure. The same divergence, in milder form, runs through the Senate, and it is the reason Tuesday's cloture vote (September 15) on the motion to proceed to the Digital Asset Market CLARITY Act is not a formality. Cloture requires 60 votes and Republicans hold 53. The seven Democrats who have been negotiating said in July that the draft fell short on ethics, consumer protection, and illicit finance; the September revision was pitched as answering those objections, and whether it has is the open question.

The vote came up repeatedly on Thursday, and the mood was cautious rather than resigned. Panelists cited prediction-market odds of roughly 20% on the bill becoming law this year. They also noted that Senate Republicans had circulated a revised draft that day with reportedly dozens of changes aimed at Democratic objections, which the room read as evidence that sponsors were still working the problem in earnest.

CLARITY is the market-structure companion to last year's GENIUS Act: GENIUS established the federal framework for payment stablecoins and their issuers, while CLARITY would draw the boundary between SEC and CFTC oversight and set federal rules for digital-asset intermediaries, which is where banks and finance companies would find the terms for custody and trading. If cloture fails, the bill is not dead, but enactment this year becomes unlikely, and institutions already committed to tokenized deposits and stablecoin settlement continue under a patchwork of existing statute, agency rules, and enforcement precedent.

CCM read: For consumers, the live issue is yield. GENIUS bars issuers from paying interest on stablecoins, but exchanges and affiliates can pay rewards on balances held with them, and that is the main way consumers earn on stablecoins today. Bank trade groups want CLARITY to close that channel; Coinbase and the crypto trade groups want it kept. If it survives, a rewards-bearing stablecoin competes with the deposit account for household balances, and that becomes a funding-cost question for consumer lenders that rely on retail deposits. For those lenders, the second question is plumbing: whether settlement and payment rails built on stablecoins proceed on a statutory footing or an improvised one. Tuesday's vote is procedural, and it is the clearest signal on that question the market will get this year.

02

New Deals

4 publicly-registered issues priced

Click any row to open the CCM Issuance table. SEC publicly registered transactions only — 144A and private placements excluded.

Issuer / Series Asset class PSR Close Size ($MM) WAL WA FICO
Capital One Prime Auto Receivables Trust 2026-1Updated Auto Loan Sep 17 1,974.6 2.55 774
American Express Credit Account Master Trust Series 2026-1New Credit Card Sep 15 500.0 752
Hyundai Auto Receivables Trust 2026-CNew Auto Loan 1,500.0 774
Volkswagen Auto Lease Trust 2026-BNew Auto Lease Sep 22 743.9 775
03

Loan-Level Pulse

Reproducible signals · ABS-EE surveillance

A few signals from this week's loan-level tape — each links to the exact view on CCM so you can reproduce it. Explore the full data ›

This week's spotlight

Mercedes-Benz

30+ DPD · seasoned pools (>6 mo)
2.06% +39% YoY
04

Macro

Rates · Used cars · Consumer
05

From the Pod

This week's episode
YouTube
Released Jan 22

Mortgage Credit Scoring Is Changing: Inside VantageScore’s Challenge to FICO | E23

Watch this episode on YouTube.

Watch on YouTube ›
06

On Deck

Data releases · ABS-EE filings · events

Macro releases

ABS-EE filings expected