A non-QM loan just beat a conventional one. Here's the mechanic.

# A non-QM loan just beat a conventional one. Here's the mechanic.

Reading the Tape: the curve pulled apart, and so did two loan programs.

I called my account executive this week to ask whether a rate sheet was a typo.

A friend of mine is buying here in Orange County. He and his wife qualify conventional. Full doc, clean file, the straightforward version. And the 12 month bank statement loan came back better. Roughly a quarter point better on the same borrower, same property, same week.

That is backwards. Non-QM is supposed to be the expensive option. That is the entire tradeoff. You skip the tax returns, you pay for it.

So I went and looked at why.

The past month, in order

The gap between the 2 year and the 10 year Treasury bottomed at 0.34% on July 15. It closed August 13 at 0.48%. Fourteen basis points of steepening in a month, and it did not happen in one move. It happened in two, for opposite reasons.

July 29. The Fed held, and the curve pulled apart anyway.

The FOMC voted 9 to 3 to leave rates at 3.50 to 3.75%. All three dissents came from regional bank presidents. The statement was materially shorter than what had become the norm, because Chair Warsh has removed forward guidance from the post-meeting statements entirely.

Here is what that did in a single session.

The 2 year fell 4 basis points. The 10 year rose 5. The 30 year rose more than 9.

Nine basis points of steepening on a day the Fed changed nothing. That is not a policy move. That is term premium. When the central bank stops telling you where it is going, investors want more compensation to lend long. They do not need more compensation to lend short.

August 7. The labor market cracked.

Payrolls came in at negative 23,000 against a consensus of positive 83,000. The first negative headline print of this cycle. Then the revisions: May cut by 66,000 and June cut by 37,000, 103,000 lower than previously reported.

Unemployment ticked down to 4.1%, but not for a good reason. Labor force participation fell to 61.4%, a level not seen in more than five years. Roughly 264,000 people left the workforce in July. Temporary layoffs rose 153,000 to 921,000. Average hourly earnings grew 3.2% over 12 months, the slowest since May 2021.

August 12. CPI came in tame.

Headline 0.1% for the month and 3.4% over the year. Core 0.2% and 2.5%. Both down a tenth from June, both in line with forecast. Shelter rose 0.1% and accounted for about two thirds of the headline increase. Treasury yields fell across the board on the release, and traders cut the odds of a September hike to 42% on CME FedWatch.

Why any of this touches a mortgage rate sheet

Most people assume all mortgages price off the same thing. They don't.

A conventional loan gets funded through agency mortgage-backed securities, which benchmark off the long end of the curve. When the 10 year and the 30 year sell off, conventional pricing follows them up. Agency paper also extends in duration during a selloff, which makes the move worse than the Treasury change alone suggests.

Non-QM is a different animal. Those loans get aggregated and securitized privately, and the bonds carry a much shorter weighted average life than an agency pool, because prepay speeds run faster. Borrowers document out into agency once they can. Investors sell or refinance. The investor buying that bond is buying short duration, and short duration prices off the front end. The warehouse lines financing the pipeline are SOFR-linked, so they price off the front end too.

One curve. Two ends. Two products.

For a month, the end that funds conventional has been rising and the end that funds non-QM has not. Layer on that private capital has been pouring into non-QM all year, compressing the spread investors demand on that paper toward agency levels, and you get a window where the loan that is supposed to cost more can win on the right file.

That is what happened to my friend. Nothing exotic. Two different capital markets moved apart, and the sheet reflected it.

What would break this

The bulk of the steepening was already done by August 5, before the jobs report. So the last two prints did less to the spread than the headlines suggest. What they actually did was take a September hike from likely to a coin flip, which lowers the funding cost on the short end. That is a real channel, but it is a different one than the chart shows, and I would rather be precise than dramatic about it.

The window also has a date on it. The Fed gets the August jobs report on September 4 and the August CPI on September 11 before it meets in mid-September. A hot inflation print or a payroll rebound unpins the front end, hike odds go back up, warehouse carry compresses, and this gap narrows. Cleveland Fed President Beth Hammack argued publicly the day before CPI that the longer they wait to act, the more expensive it gets. That dissent bloc is live.

There is also a version of this that ages badly. Spread compression in non-QM has gone far enough that industry people are openly saying it has removed the incentive to document income properly. When that reverses, it reverses on everybody.

The part I keep sitting with

The same jobs report that helped my friend's pricing is the report that showed 264,000 people leaving the workforce in July, and temporary layoffs at 921,000.

Those two facts printed on the same page. Financing got marginally cheaper for the people who can still qualify, on the day the pool of people who can qualify got smaller. Qualification is a snapshot of the income you have today, not the income you had last year. That is not a reason to panic. It is a reason to find out where you stand while the answer is still in your favor.

If this is you

If you are self-employed, or your tax strategy is aggressive enough that your returns do not reflect what you actually earn, the conventional answer may not be the right answer this month. It might be. That is the point of pricing both.

*Sources: U.S. Treasury constant maturity yields; U.S. Bureau of Labor Statistics Employment Situation released August 7, 2026 and Consumer Price Index released August 12, 2026; FOMC statement July 29, 2026; CME FedWatch. Market data is informational and is not a rate quote, a lock or float recommendation, or an offer of credit. The scenario described reflects a single file and does not represent typical results. Pricing varies by credit profile, loan amount, property type, occupancy, and market conditions. For pricing on your situation, speak with a licensed loan originator.*

--- *Author: Chad Villacorta, mortgage broker at West Capital Lending (NMLS #2636410). Licensed in 34 states. Estimate only, not a loan commitment. Subject to credit approval and underwriting.*

Subject to credit approval and property qualification.

West Capital Lending | NMLS #2636410 | Subject to credit approval and property qualification.