Why 7% Mortgage Rates Aren’t the Real Problem

Karen Postiglioni has a habit of finding the uncomfortable math hiding underneath mortgage’s favorite talking points. Today’s starts with one you’ve probably heard a thousand times: 7% mortgage rates aren’t historically high.

True! Also…kind of beside the point.

Rates may look a lot like the late ’90s, but the mortgage business supporting them looks nothing like it. We’re producing roughly half as many loans, average balances are much bigger, costs have exploded & lender margins are thinner.

Which leads KP to a pretty spicy 🌶️ question: Did mortgage build an LO comp model that becomes harder to justify as home prices rise?

Also in today’s edition: Mat Ishbia low-key did a deal w/ Oaktree back in 2020 after an “unsuccessful hedge,” what it would actually take to get buyers off the sidelines & much more.

By Karen Postiglioni

When mortgage rates hit 6.95% last week, my first thought wasn’t, “These are still great rates, historically speaking.”

It was, “This is going to hurt.”

Borrowers feel it. Lenders feel it. And people in this industry who have already lived through years of layoffs, branch closures & ‘just wait until rates come down’ definitely feel it.

So no, I’m not going to wave around a historical chart & tell everyone to calm down.

But let’s look at a chart anyway b/c it is illustrative.

Rates are in the same neighborhood as 1998-99. Dollar volume isn’t wildly different. But holy moly, unit count is. We’re producing roughly half as many loans today, while average loan balances are dramatically larger.

That’s when I had the ‘oh shit’ moment. Because somewhere along the way, loan amount became the thing we primarily use to compensate retail LOs.

Before the post-crisis LO comp rules, originator pay could be tied to revenue or profitability in ways that created an obvious conflict. The person advising the borrower could make more money depending on the rate, points or other economics of the loan. That needed to stop.

Beginning in 2011, compensation tied to interest rate & other loan terms was prohibited. Loan amount, however, was allowed. Lenders needed a clean replacement for revenue splits and profit-based plans, & basis points of loan balance became the answer across much of distributed retail. It was simple, compliant, easy to administer & easy to compare when recruiting.

But then home prices took off.

STRATMOR says average retail LO commission went from $1,672 per funded loan in 2010 to $3,194 in 2022. Using Bureau of Labor Statistics data, that 2010 commission would have been about $2,244 in 2022 dollars. And the structure is still w/ us. STRATMOR says retail LO commissions have stayed roughly 92 to 103bps.

We chose loan balance as the index & then spent the next 15 years watching the index explode.

A $500,000 mortgage doesn’t require twice the selling, phone calls or expertise of a $250,000 mortgage. At 100 bps, though, it pays twice as much.

For years that didn’t look like much of a problem. Bigger balances also meant more lender revenue. Rates fell, volume boomed & lenders were feasting along w/ LOs. 

Then the market changed & the famine came.

Take the same $400,000 loan & the same 100 bps LO comp.

In 2020, MBA says average production revenue was 434 bps. That’s $17,360 of revenue against $4,000 of LO comp, about 23%.

In Q2 2026, production revenue was 333 bps. Same $400,000 loan, same $4,000 commission. But now revenue is $13,320 & LO comp suddenly represents about 30% of it. 

Nothing went wrong; the plan worked exactly as designed. LO comp stays indexed to loan balance while lender revenue moves w/ the market.

But those two lines don’t move together. And no, we can’t renegotiate comp every time the margin tightens. I know my compliance rules, & I don’t want the old conflicted system back either. 

But loan balance doesn’t have to remain the dominant compliant answer forever, right?

The average loan got 53% bigger. Production expense grew 57%. Profit per loan fell 34%. 

At 2015 economics, 1,000 loans generated $1.19M of production profit. At 2025 economics, it takes roughly 1,515 loans to get back to the same place.

And that’s happening in a market producing roughly 5M loans a year, not 11-12M. The market is giving us fewer loans while the P&L is asking each lender to find more of them.

That’s the math we’re trying to outrun.