Texas Now Leads the Nation in Repossessions — and the Buffett 'Housing Bottom' Bet Just Cracked

ATTOM's August 2026 U.S. Foreclosure Market Report, released September 17, put the national headline on the wall: lenders repossessed 5,794 properties through completed foreclosures, up 22% from July and up 42% from August 2025 — the largest annual jump since the post-COVID normalization began. The housing-bottom thesis — embodied most visibly by Berkshire Hathaway's ~$1.16 billion (~11%) stake in Lennar built through three buying rounds in 2026 — was built on the premise that distress had already flushed through the system and that mortgage credit, builder incentives, and rate relief would compound into a 2026–27 recovery. The August print doesn't refute that thesis in full, but it puts the first concrete regional crack under it.

Texas didn't appear at the top of the rankings overnight. In Q1 2026 the state already led the country in foreclosure starts with 10,617 properties entering the process, and H1 2026 filings reached ~22,000 statewide. The August data simply confirm what the leading indicators had been saying since spring: Texas is normalizing at a faster pace than the national average. Three forces line up. First, Texas had the most aggressive pandemic-era price run-up, so the equity cushion that protects borrowers elsewhere is thinner. Second, insurance premium shocks (especially in coastal metros) and property-tax escalation have stretched payment-to-income ratios. Third, the state's foreclosure timeline is the shortest in the country — 155 days on average per ATTOM's mid-year report — which mechanically converts starts into completed REOs faster than in states where the process drags on for years.

Three metro concentrations matter most for the bank and builder exposure analysis that follows. Houston posted 448 completed foreclosures in August, Dallas 402, and San Antonio 256 — the same metros where Texas-domiciled lenders have concentrated their commercial-real-estate, mortgage, and homebuilder-finance books.

Berkshire Hathaway's housing thesis rests on three names: Lennar (~11%, ~$1.16B), D.R. Horton (a smaller but disclosed stake), and the Clayton Homes captive-finance arm. Both publicly disclosed builders are Texas-heavy by construction: Lennar operates a dedicated Texas segment alongside East, Central, and West, and D.R. Horton is headquartered in Arlington, Texas, with operations in 31 states and 98 markets. The thesis was that after the 2024–25 incentive-and-cancellation cycle, builder margins would recover as rate volatility normalized. The August foreclosure print complicates that timing — not by breaking it, but by pushing the recovery point further out and concentrating the pain where these two names have the most revenue.

Two of the cleanest Texas pure-plays sit in publicly traded form. Cullen/Frost Bankers is a Texas-only bank headquartered in San Antonio with $53.9 billion in assets at June 30, 2026; its commercial-and-consumer book is anchored in HELOCs, first mortgages, and interim construction lending — exactly the product set the August print is most likely to stress. Texas Capital Bancshares is Dallas-based and operates a dedicated Mortgage Finance unit plus a residential homebuilder finance line, which means its loan book scales with the very developers now seeing higher cancellations and slower starts in Houston and San Antonio. Both names trade at low multiples — CFR at 14.5x trailing earnings with a 14.3% return on equity, TCBI at 12.2x with a 10% return on equity — multiples that bake in benign credit. The August print is the first data point that argues they shouldn't.

A foreclosure surge transmits through three layers. Layer one is the lender balance sheet — reserves, charge-offs, and capital ratios at Cullen/Frost Bankers, Texas Capital, and Zions Bancorporation. Layer two is the builder order book — Lennar's 9% order decline and D.R. Horton's 19% drop in homebuilding pre-tax income are the early markers. Layer three is the upstream supplier, where Builders FirstSource — Irving, Texas-based and the country's largest supplier of framing, trusses, wall panels, and engineered wood to professional homebuilders — has already absorbed the shock: shares are down from a 52-week high of $131.50 to $56.22, a 57% drawdown, with trailing EPS at just $0.92 and quarterly earnings down 82.9% YoY.

Three catalysts sit inside the next two quarters. First, the September and October ATTOM reports will either confirm the August42% REO surge as a one-month anomaly or as the start of a trend; the September print drops mid-October and will move Texas-bank credit default swap spreads and builder guidance. Second, Cullen/Frost Bankers and Texas Capital report Q3 2026 results in mid-October — listen for reserve builds and Texas consumer-loan-loss commentary; any change in the Texas CRE or HELOC allowance will be read across regional-bank peers. Third, D.R. Horton reports Q4 FY2026 (period ending September 30) on or around October 22; the cancellation rate and net order guidance will be the first hard read on whether the August Texas data points have shown up in the order book yet.

Over a one-to-three-year horizon, three structural questions get sharper. The first is whether Texas is leading a national normalization or is a regional credit event — Florida's 1-in-2,397 filing rate is similarly elevated, suggesting the Sun Belt is the pressure point, not just one state. The second is whether the homebuilders' incentive strategy (Lennar's 14%-of-price incentive level per its own disclosures in Q1 2026 is twice what it was a year ago) can absorb the supply coming from REO completions without re-cutting gross margin below the 15.8% Lennar printed in Q3. The third is whether Builders FirstSource's 57% drawdown is the leading indicator — supply starts per home; if so, the worst of the upstream bleed hasn't shown up in the builders' reported margins yet.

On balance, the August print does not refute the housing-bottom thesis — it reframes the timeline. Berkshire Hathaway's continued buying in Lennar through late September (additional ~$349M between Sept 17 and Sept 25) implies management's view that the Texas print is a credit-cycle event, not a price-discovery event. The investor question is whether Berkshire's patient-capital view holds against the quarterly earnings prints of the Texas banks and the 2026–27 guidance cuts at Lennar and D.R. Horton.

Three layers of the trade. The cleanest bearish expression is Texas bank credit — Cullen/Frost Bankers and Texas Capital Bancshares trade at multiples that bake in benign Texas credit, and the August foreclosure surge is the first data point challenging that. The cleanest hedge against the Buffett thesis is the upstream supply chain — Builders FirstSource has already absorbed most of the pain, but the order book lags starts by 60–90 days, so the August Texas start surge hasn't fully shown up in BLDR's revenue line yet. The cleanest paired trade is long Lennar versus short Builders FirstSource: the Berkshire thesis says the recovery comes; the supplier price action says the recovery is further out than the market believes.

The thing this print does not yet do is break the national housing recovery thesis. Inventory is still constrained, prices are still up modestly, and the 42% REO surge is off a low base. What it does is put a clock under it — and concentrate the credit-cycle risk in two specific Texas banks and one specific Texas-headquartered supplier that the market has not fully priced.