Mortgage Approval and Denial Rates in San Juan County, New Mexico (2025 HMDA)

How to read the mortgage denial rate in San Juan County numbers

Every figure on this page for San Juan County, New Mexico is a median, not an average, which matters more than it sounds.

A median is the midpoint: half of the lending market sits above it and half below. Averages get dragged around by a handful of extreme values — one very expensive property, one unusually large loan, one graduate earning far more than their classmates — so a median is the safer number to plan against. Applications acted on of 3,373 and denial rate of 23.1%.

They describe the middle of the market, so they are a starting point for applying for a mortgage and nothing more. Anyone quoting you a number will be pricing your file, not the median.

Turning the San Juan County, New Mexico data into a decision

Context does most of the work here. A number is only high or low next to something else, so we publish San Juan County, New Mexico and how it sits inside New Mexico against a wider benchmark on the same measure. Median loan amount of $175,000 and median applicant income of $81,000.

Small gaps tell you nothing — sampling error alone can produce them. Large gaps almost always have a physical cause: the age of the housing stock, commuting distance, the balance between renters and owners, or a single large employer setting local pay.

That is why a single figure should never carry a decision on its own. Read two or three measures together — what comes in, what housing costs, what is left — and the picture stops being ambiguous.

Where San Juan County, New Mexico sits against the wider market

Comparison is the fastest way to tell a genuinely unusual lending market from an ordinary one. We benchmark San Juan County, New Mexico against New Mexico and against the national picture on the same definitions, in the same period, so the difference you see is a real difference rather than an artefact of two datasets disagreeing.

Definitions matter as much as the numbers. Two sources can both publish a "median rent" and mean different things — one including utilities, one not; one covering every unit, one only those recently let. Mixing them produces gaps that look meaningful and are not.

Any measure the source flags as unreliable is dropped rather than smoothed. That is why some pages carry fewer figures than others: the lending market itself is thinly sampled.

A short checklist before you act

None of this needs a spreadsheet. It needs the discipline of checking the same handful of things every time, whether you are applying for a mortgage for the first time or the fifth.

• Pull your credit file first and fix errors, because the file is what gets priced.
• Get more than one pre-approval — denial and pricing decisions vary by lender on identical files.
• Write down the monthly number you can genuinely cover, not the maximum a lender or admissions office says you qualify for.
• Compare at least three offers. Pricing on the same product varies more between providers than most people expect, and the gap is yours to keep.
• Check the fee schedule and the fine print, not the headline rate — that is where the real cost usually hides.
• Ask what the number looks like in a bad year, not an average one, and make sure the plan still holds.

Common questions

Sources, method and limits

The page is assembled from the loan-level disclosures lenders must file each year, specifically the CFPB HMDA loan application register (2025 HMDA). Nothing here is modelled from a proprietary black box: every figure traces to a published record, and where we derive a ratio we show both inputs so the arithmetic can be checked.

The limits are the usual three: sampling error grows as geography shrinks, published data always trails reality by months or years, and a midpoint says nothing about the extremes at either end of the lending market.

Where a source revises a figure, the revised value replaces the old one rather than sitting alongside it, so the page always reflects the current official position. If you spot a number that looks wrong, it is worth telling us — corrections get made against the source record, not by guesswork.