By Kulasekar Mittapalli · Ground Figures · Last reviewed August 27, 2026

Lender Overlays: Why Your Loan Was Denied When the Guideline Said Yes

There is a specific kind of mortgage denial that borrowers find hardest to accept: the one where they did the homework, found the actual rule in the actual guideline, met it, and were declined anyway. It feels like the lender broke a promise. Nobody broke a promise. The borrower read a rule that was never binding on the lender in the first place.

Where the gap comes from

When a lender closes a conventional loan, it usually does not keep it. It sells the loan to Fannie Mae or Freddie Mac, or delivers it to an investor who does. The agency guidelines describe what those buyers will accept — a purchase standard, not a lending standard.

Nothing in that arrangement requires a lender to originate every loan the agency would buy. Lenders carry risk the agency does not: a loan that defaults early, or that turns out to have a documentation defect, can be repurchased back onto the lender's own balance sheet. Warehouse lines have to be protected. Post-close quality control finds patterns, and credit policy gets tightened in response. Every one of those pressures pushes in the same direction — toward requiring somewhat more than the guideline requires.

The result is that the published rule and the applied rule are two different rules, and only one of them is on a website you can read.

Case one: the DTI ceiling nobody actually lends to

Fannie Mae's position on debt-to-income is public and specific. Its Selling Guide topic B3-6-02 sets a maximum total DTI of 36% for manually underwritten loans, extendable to 45% where the borrower meets the credit score and reserve thresholds in the Eligibility Matrix. For loan casefiles run through Desktop Underwriter, the maximum allowable DTI is 50%.

That 50% figure has real history behind it. Before July 2017, a DU file between 45% and 50% DTI needed specific additional support — in practice an LTV at or below 80% and twelve months of reserves, the reserves requirement usually being the binding one. Fannie removed those requirements with DU version 10.1, announced in SEL-2017-06, and folded the 45–50% band into DU's general risk assessment instead. The announcement said plainly that additional compensating factors outside DU's standard risk assessment would no longer be required above 45%.

Worth noting for anyone who reads that as a one-way liberalization: high-DTI volume rose sharply afterward, and Fannie fine-tuned the model again in DU 10.2 the following March specifically to limit risk layering. The band stayed open. The scoring behind it kept moving.

So the guideline says 50%, and it has said 50% for nearly a decade. Ask a loan officer what DTI their shop will actually close, though, and you will frequently hear a lower number. Caps in the mid-40s are common enough that a borrower calculating at 47% should assume the answer varies by lender rather than assuming the agency maximum applies.

Agency figures are from Fannie Mae Selling Guide B3-6-02. The right-hand column is the part no published document answers.

Who this catches: borrowers with genuinely strong files — high credit score, real reserves, stable income, low LTV — whose ratio lands in the mid-to-high 40s. They are exactly the profile the 2017 change was designed to serve, they get an Approve/Eligible, and then they meet a credit policy that predates or ignores it. Run your own number on the DTI calculator before you shop, and treat the result as the start of a conversation rather than a verdict.

Case two: the co-signer who does not help as much as expected

The parent co-signer is one of the most common structures in first-time buying, and it has a wrinkle most people never see coming.

Fannie's guidance on non-occupant borrowers (Selling Guide B2-2-04) draws a distinction by underwriting path. On a manually underwritten loan where non-occupant income is used, the occupying borrower's own ratio — their income and their debts, considered alone — is capped at 43%. On a DU casefile, the combined ratio of all borrowers is what DU evaluates, without that separate occupant-only test.

The divergence: the occupant-only test has a way of showing up on DU files anyway, as lender credit policy rather than agency requirement. The logic is not unreasonable from the lender's side. A household where the person actually living in the house cannot carry the payment on their own income looks different, in default data, from one where they can. Some lenders price that concern as an overlay; others decline the structure above certain LTVs; others require the occupant to put in more of the down payment.

Who this catches: the buyer whose combined numbers look comfortable and whose own numbers do not. It is worth calculating both ratios early — yours alone, and yours combined — because if there is a gap between them, that gap is where the file will be argued.

Case three: self-employed income and the direction of the trend

Self-employment income is the area where the written rule and the applied rule diverge most persistently, because the guideline necessarily contains a judgment call.

Fannie updated Chapter B3-3 of the Selling Guide through announcement SEL-2026-02, issued in March 2026, with income provisions effective for applications on or after June 1, 2026. The broad framework is familiar: self-employment income is generally averaged over a two-year history after allowable add-backs, and income that is declining gets treated more conservatively than income that is flat or rising.

Where practice diverges is in what "more conservatively" turns into on a real file. A borrower whose most recent year came in below the prior year may find the underwriter qualifying them on the lower year alone rather than the two-year average, and asking for evidence — year-to-date profit and loss, business bank statements — that the decline has stopped. Rising income, meanwhile, is often still averaged, so a borrower whose business is growing does not get full credit for where it is now.

That asymmetry is the thing to plan around. Trends are read down more readily than they are read up.

The specific stability documentation a lender requires, and how large a year-over-year decline triggers extra scrutiny, are set by individual credit policy. Ask before you apply rather than after.

What this means if you are shopping

The practical consequence of everything above is narrow and useful: a denial is information about one lender, not about you. Borrowers routinely treat the first no as a verdict on their finances and stop. Often the correct next step is to take the same file somewhere else.

Some questions that get better answers than "am I qualified?":

• What is your maximum total DTI with a DU Approve/Eligible finding?
• Do you add any requirements beyond agency guidance for my situation specifically?
• Is this an overlay, or is it the agency rule? — a fair question, and a loan officer who cannot answer it is telling you something.
• If this is an overlay, would a different channel at your company treat it differently?

Ask the DTI question with the number in it. "Are you flexible on DTI?" gets a yes from everyone. "What is your cap with an Approve/Eligible?" gets a number.

The part worth being clear about

Overlays are not lenders behaving badly. A lender that has to buy back a defaulted loan is bearing a cost the guideline does not compensate it for, and tightening in response is rational. The problem is not that overlays exist. The problem is that borrowers are handed the agency rule as though it were the operative one, plan around it, and then discover late — usually in underwriting, sometimes after an offer is accepted — that it was only ever half the picture.

Read the guideline. It tells you what is possible. Then ask the lender what they actually do, because that is the rule you will be judged by.