As I write my first (and hopefully not my last) column for Rob and Robbie, the CME Group’s FedWatch is giving a 57.5% chance of a 50-bps rate cut at the September 18th FOMC meeting, with the 10-year treasury at 4%.
Polly data shows average locked conforming rates are on a downward trend. Over the past 12 months, average conforming rates peaked around 7.75% in late 2023 and again at 7.5% in May 2024. Recently, average locked rates have dipped below 7%, trending toward 6.625%. I know at times it feels like licking your finger and sticking it up in the air, but it definitely feels like momentum may be picking up. And in speaking to industry friends, I’m hearing hiring for LOs has increased dramatically, and it seems as though many companies are prioritizing an upgrade of their tech stack before they’re too busy to do so.
Why do I bring this up? I’m not predicting a volume explosion, but a meaningful uptick in production seems probable. With higher volume and margins, cap markets and ops leaders might overlook margin leakage to some degree. When I was in the trading and hedging seat, good times meant longer work hours, heavy recruiting, and enjoying larger margins, making it easy to ignore margin leakage. It might show up as an investor roll charge, an underwriting miss, data errors, or other creative ways to lose money. Tracking both negative and positive differences in expected sell price versus purchase advice is crucial.
You need to dive into the data and get into the weeds. If you’re looking at this from a cruising altitude of 35,000 feet, you’re going to miss margin leakage in the P&L. For simple math, let’s say you have $100M in realized monthly production at a 2% GOS, that’s $2M in gross revenue. An underwriting miss costing $10k shows up month end as a 1.99% GOS versus 2%. Who has time to research 1bp?
That’s one angry basis point.
Or worse, positive data entry errors (aka lower expected sell price versus actuals) could mask losses, making them invisible on the high-level P&L. This is why both positive and negative deltas need scrutiny.
When I worked at Opes Advisors, later acquired by Flagstar Bank, our SOX compliance team was relentless about purchase advice deltas. Internally we had a report which we called “the EVA report.” EVA stood for Expected Versus Actual. It was the bane of everyone’s existence.
The workflow went something like this:
Accounting entered the purchase advice
Cap markets identified losses and categorized them
For credit errors, we’d send them to underwriting and post closing managers to review
An argument would ensue over the cause of the loss
A subsequent argument would ensue over what level of added controls were needed to prevent future mistakes and how much harder it would make everyone’s lives
Cap markets would summarize all this as part of our executive monthly management reporting
We’d get torched by our CEO 17 minutes later after the report went out
To this day, the name “EVA” is nightmare fuel.
But recapturing money through corrective actions or catching investor errors was always worth it. In the COVID-19 era of record-setting volume, $10k in GOS was a rounding error. Today, it’s very significant.
When it comes to margin leakage, focus on what you can control. For instance, modern PPEs like Polly keep pricing accurate by automating pipeline monitoring throughout the loan manufacturing process. As for the things you can't control, well, sometimes servicing lets you know that a house in a dense residential area literally exploded because your W-2 borrower was running a secret meth lab in the basement.
I’ll end my first guest write up for the Chrismans with a movie quote.
“Well those are whole pennies, right? I'm just talking about fractions of a penny here. But we do it from a much bigger tray and we do it a couple of million times.” - Peter Gibbons, Office Space 1999.
—
Marcus Lam is Head of Solutions Engineering at Polly, operator of the mortgage industry’s first vertically integrated, data-driven capital markets software platform. For more information, follow Polly on LinkedIn or visit www.polly.io.