Austin -26%, San Francisco -6%: Stop Marking Housing at the National Level

Wolf Richter's July tally counts 15 bigger cities and counties where prices of mid-tier single-family homes are down 11% to 26% seasonally adjusted from their own peaks, most of those peaks set in mid-2022 and two in 2024 ^[1]. The same run of data has San Francisco leaving the list entirely, which means anyone still underwriting or marking to a national house-price number is averaging two regimes that no longer move together ^[8].

The list itself is a measure of how fast this is widening. Richter raised the cut-off for July to a 11% decline from 10% "to keep the list tidy" ^[2], and notes that when the series started a year ago the cut-off was 8%, the list held 10 markets, and the worst decline was 22% ^[3]. Another five bigger cities came in at exactly 10% ^[4], so on last month's threshold the list would have run to 20 names ^[5].

Texas supplies both ends. Austin tops the table at -26% ^[6], while Houston is at -5% and Corpus Christi at -3% ^[7], a 23 point spread inside one state ^[19]. Dallas-Fort Worth splits down the middle: McKinney (-14%) and Fort Worth (-11%) qualify, while Frisco (-10%), Garland (-9%), Plano (-9%), Arlington (-7%) and the city of Dallas (-7%) do not ^[9]. Richter attributes the area's weakness to homebuilders who have to move inventory and are buying down mortgage rates, cutting price points and piling on incentives, with existing homes forced to compete ^[10]. That matters for comps: the recorded price and the effective price are not the same number.

California has Oakland (-24%), Hayward (-13%) and Contra Costa County (-12%) on the list ^[11]. San Francisco sat in fourth place a year ago at -15% and now prints -6%, which Richter puts down to money from AI companies including Anthropic being thrown around ^[12]. That is nine percentage points of re-inflation against its own peak in twelve months ^[13]. Meanwhile San Jose is at -6%, Sacramento -9%, Stockton -8%, San Diego -4% and Los Angeles -4% ^[14]. Oakland to San Jose is an 18 point gap across one bay ^[20].

Florida contributes three counties, because the well-known cities inside them are too small to qualify on their own ^[15], and Richter adds that Florida single-family is nowhere near as weak as the condo market, where the bottom has fallen out in a number of markets ^[16].

The lending arithmetic is unforgiving at the tail. A 90% LTV loan written at the 2022 peak in a market that has since given back 26% is now marked at roughly 122% of collateral value ^[21]. That is not a forecast, it is division. Also worth noting before anyone reconciles this to a median-price series: these are seasonally adjusted three-month averages from the Zillow Home Value Index, including off-market and for-sale-by-owner deals, and they are explicitly not median prices ^[17].

Watch the five markets sitting at -10%, which cross into the list on any further slippage ^[4]. Watch whether the cut-off has to be raised again. And watch demand: pending home sales dropped to the second lowest on record, with a record low in the West and a near-record low in the South ^[18]. The Texas builder response is the mechanism to track, because incentives set the clearing price for existing homes in the same submarket ^[10].