Foreclosures from a low floor: Dallas County’s bank-owned sales and the habit of calling every market stable

From January through October 2026, 226 single-family homes in Dallas County sold through the MLS as bank-owned, 1.60% of 14,127 closings. In the same ten months of 2024 the figure was 94, or 0.64%. In October alone, 32 of 1,189 closings were bank-owned, 2.7%, against 1.0% a year earlier and 0.28% in October 2024.

The share is still small. In 2008, 35% of the county's MLS sales were bank-owned, about one in three, and it stayed above 30% through 2011. Today's rate is about 1 in 63. What has changed is the direction and where the sales are. Most are in the lowest price range, and the largest group we could trace to an earlier sale was bought in 2022 and 2023.

The pipeline points up. Listings flagged as bank-owned, short sale or pre-foreclosure went from 42 in the first quarter of 2024 to 124 in the third quarter of 2026, and short-sale listings from 4 to 29. A distressed listing usually closes one to three months after it goes on the market, so the listings lead the closings.

MLS counts understate foreclosures. In Texas, foreclosed homes are sold at a county courthouse auction on the first Tuesday of the month.²⁴ Investors buy some at the auction; the rest go back to the lender and reach the MLS months later, if at all. ATTOM counted 402 completed foreclosures in the Dallas metro area in August 2026 alone,³ against about 20 bank-owned resales a month on the Dallas County MLS. The MLS figure measures the bank-owned homes that come back to market as listed resales, the ones an appraiser is most likely to see as comparables.

ATTOM, which tracks foreclosure filings in more than 3,000 counties, counted 367,460 U.S. properties with a filing in 2025, up 14% from 2024. That was 0.3% of all housing units, against 0.4% in 2019 and 2.2% in 2010, when nearly 2.9 million properties had a filing.¹ The first half of 2026 brought 227,548 more, up 21% from a year earlier.²

Loan-level data show the same turn. In ICE's servicing data for September 2026, the number of loans in active foreclosure was up 37% from a year earlier, the most since early 2019, and completed foreclosure sales were up 15%. Loans made in 2022 or later accounted for 43% of new foreclosures and 49% of completed sales.⁷ The Mortgage Bankers Association put 0.67% of loans in foreclosure at mid-2026, up 0.19 percentage points in a year, against a record of about 4.6% in 2010.²³ Its survey found 11.8% of FHA loans delinquent, against 2.7% of conventional loans.⁸ The New York Fed counted about 62,000 people with a new foreclosure on their credit report in the third quarter, more than in the second.⁹

Each source counts something different: ATTOM counts properties with a recorded filing against all housing units, ICE and MBA count mortgages, and the New York Fed counts consumers. They agree on the direction and the scale. Foreclosures are rising from the lowest levels ATTOM has recorded, and the increase is concentrated in FHA loans and in loans made after prices and rates peaked.

In 2026, 162 of the 226 bank-owned sales through October closed under $300,000. In that range the bank-owned share went from 1.35% in 2024 to 3.56%. From $300,000 to $500,000 it went from 0.34% to 0.93%, and above $500,000 it stayed near 0.4%.

Bank-owned homes also sell for less than their neighbors. We compared each bank-owned sale's price per square foot with the median for standard resales in the same zip code and year. The median bank-owned sale came in at 84% of that figure in 2024, 81% in 2025 and 76% in 2026. Short sales came in at 87%. Part of the gap is condition: the median bank-owned home was built in 1972, against 1980 for other sales, and lenders sell as is. Cash buyers took 39% of bank-owned sales, against 21% of other sales. Bank-owned homes took a median 48 days to sell, against 31.

The sales are spread across the county but weighted to the south and east. From January 2025 through October 2026, Dallas had 182, Garland had 38, Mesquite had 37, Grand Prairie had 17, DeSoto had 16 bank-owned sales; 69 of 103 zip codes had at least one, and the 10 with the most accounted for 40%. Those figures describe where sales happened. They are no basis for treating a neighborhood or its residents as a credit risk, which fair lending law prohibits.

We matched each distressed sale since January 2024 to the home's previous MLS sale, using the street address. Our earlier records cover 2004 to 2012 and October 2020 onward, so a purchase from 2013 to September 2020 has no match. Of 563 distressed sales, 261 had an earlier MLS sale on record, and 186 of those earlier sales were in October 2020 or later. 129 were bought in 2022 or 2023. Those homes sold at a median 14% below their purchase price, and 18% below for the 91 owners who had bought with an FHA, VA or conventional loan. Homes bought in 2020 and 2021, before the run-up in prices peaked, sold for about what the owners paid (+0.9%).

FHA and VA buyers were 38% of the earlier purchases behind these sales, against 24% of all Dallas County closings from 2024 to 2026. Those loans require the smallest down payments, so a price decline of 10% or more leaves the owner owing more than the home is worth. This is the local version of ICE's national finding, and it matches our Dallas County tier analysis, which estimated that many recent FHA, VA and USDA buyers are underwater, and our negative equity analysis by metro.

The numbers are small, 186 sales, and the match finds only homes whose earlier sale went through the MLS. Treat the percentages as the direction of the evidence.

Bank-owned sales reflect defaults that began a year or more earlier. Four MLS measures move sooner, and all four weakened from October 2024 to October 2026. Buyers signed 1,216 contracts on Dallas County homes in October 2026, against 1,417 a year earlier and 1,462 two years earlier. Of the listings that left the market in October, 46% ended unsold (expired, cancelled or withdrawn), up from 36%. The median seller-paid concession rose from 1.87% of the price to 2.44%. Months of supply at the end of October went from 4.2 to 5.0.

Two measures moved the other way, and they belong in the picture. October's median days on market was 36 in 2026, against 45 in 2025, and fewer sales closed after a price cut (44% against 47%). Spring 2026 was the strongest selling season of the three years. The evidence describes a soft market with rising distress at the low end.

Concessions deserve particular attention in a valuation. In the 12 months to October, sellers paid concessions on 52% of Dallas County sales, and on 64% of sales from $300,000 to $500,000. A recorded price that includes a 2.4% seller credit overstates what the buyer paid for the house by about that much.

FHFA publishes counts of the property value trend checkbox on appraisals for loans delivered to Fannie Mae and Freddie Mac. In the Dallas-Plano-Irving metropolitan division, appraisers checked stable on 91% of appraisals in 2013, 90% in 2019 and 89% in 2020. In 2023, after mortgage rates rose, they checked stable on 87% and declining on 6.2%.¹⁹ Nationally, stable ran from about 79% to 85% of appraisals every year from 2013 through 2023 except 2021 and 2022, and reached 85.6% in the third quarter of 2024, the latest data FHFA has published. From 2013 through 2020, FHFA's own price index rose 4.3% to 6.3% a year.^19,22 Our analysis of falling markets found that in metro areas where prices fell over a full year, 3.6% of appraisals marked values declining.

A checkbox answer is supposed to summarize an analysis. Fannie Mae's Selling Guide requires the appraiser to “report the primary indicators of market condition,” and says the trend “must reflect the overall movement of the market,” rest on “a minimum of 12 months of data,” and “must be reflective of those properties deemed to be competitive” to the subject. Declining values, over-supply or marketing times over six months each require the appraiser's reasons, and the analysis covers listings and contract sales as well as closings.¹⁰ When Fannie Mae stopped requiring its Market Conditions Addendum in August 2018, it said: “Appraisers remain responsible for analyzing market conditions and accurately reporting them.”²⁰ Freddie Mac told lenders in December 2025 that appraisal reports “must contain support for both the reported market trend” and the adjustments made for it, and that “Failure to apply market condition adjustments when they're warranted is an unacceptable appraisal practice.”¹³

USPAP, the standards every Texas appraiser must follow under the licensing board's rules,²¹ requires an appraiser to analyze market area trends, supply and demand when they are necessary for credible results, and to avoid unsupported assumptions about market trends. It requires recognized methods free of substantial errors of omission, and reports that are clear, accurate and not misleading.¹⁶ Checking stable out of habit, in a segment where prices are falling, fails each of those tests.

FHFA has measured how rarely appraisers adjust for time. In a sample of appraisals from 2018 to 2021, comparable sales were typically about six months old, which implied average adjustments of roughly 2.5% to 9% of the sale price, and through much of that period appraisers time-adjusted fewer than 10% of comparables. A later FHFA working paper covering 2015 to 2023 found the adjustment made for around 10% of properties.^14,15 Fannie Mae and Freddie Mac tightened the rules together in November 2024. Fannie Mae's Selling Guide now says the appraiser “must analyze comparable sales for any changes in market conditions from their contract dates through the effective date,” and that “Time adjustments, or the lack thereof, must be supported by evidence.”¹² A zero adjustment is a conclusion, and it needs the same support as any other. FHA's handbook, as amended in 2025, still tells appraisers that negative market condition adjustments “should be applied if there is sufficient proof of the trend from a credible source.”¹⁷ Our UAD 3.6 sales grid explainer shows where the adjustment now goes on the form.

Dallas County shows why a single answer for a whole market area misleads. Over the 12 months to Oct. 31, 2026, the county's median price per square foot was −2.2% from the prior 12 months, a figure many reports would round to stable. Split by price, the same county shows two markets moving in opposite directions.

Under $300,000, the median price per square foot fell 3.2%, and 3.3% after concessions. Distressed sales were 3.7% of that segment's sales, 82 distressed listings were active on Oct. 31, and the median concession was 2.8% of the price. Above $500,000, the median rose 4.2%, distressed sales were 0.4%, and the median concession was 1.0%. For a $240,000 ranch house, Fannie Mae's competitive-properties rule points the appraiser to the first set of numbers. For a $650,000 house in the same county, it points to the second.

The error runs in both directions. At the 12-month rate, a six-month-old comparable under $300,000 left unadjusted overstates value by about 1.6%, before any adjustment for the concession or the bank-owned competition. Above $500,000, the same unadjusted comparable understates value by about 2.1%, the pattern FHFA documented in rising markets. Medians move with the mix of homes that sell, so these are segment-level indicators to test against paired sales and repeat sales in the subject's own neighborhood. The rules require that test, and checking stable out of habit skips it.

The appraisers are working inside real constraints. Sales data reach them with a lag, and a small neighborhood can have too few sales to measure a trend. The legacy form offered three boxes for a market that moves by segment. Those constraints explain the pattern and leave the requirement unchanged. On the new form, a reviewer can see the analysis, or its absence.

Choose a price range to see the search metrics an appraiser would report for it as of Oct. 31, 2026, and a commentary built from those numbers. The commentary illustrates what a supported trend statement can contain. A real assignment uses the subject's competitive market area, which is usually far smaller than a county.