US housing has dispersed to the point where the national aggregate is accurate about the country and useless to anyone in it. The mechanism is supply, not demand: mortgage rates apply uniformly, construction policy does not. Three local indicators that beat any national average, and a fair account of what the aggregate is still good for.
Two numbers from the same week in August 2026. Austin's median list price is down 12.2% year on year. San Francisco's active inventory is down nearly 20%. One market is being handed to buyers; the other is being withheld from them.
Average them, add 300 more metros, and you get the national picture: median list prices down roughly 2% year on year, with about 1.2 million homes actively listed for the week ending 8 August — inventory up 3.2% from a year earlier. Realtor.com's asking-price series puts the annual decline at 2.5%, which it describes as the steepest in its data since 2017 and the eighth consecutive month of decline.
That national figure is accurate. It is also close to useless, because it is describing something that is not a market. There is no national housing market in the sense of a place where a national price prevails. There is a country containing several hundred housing markets, some of which are currently moving in opposite directions, and a mean that sits in the empty space between them.
On the numbers: the metro-level figures below are Realtor.com data as carried in August 2026 market reporting. They are point-in-time listing statistics, not transaction prices, and list-price medians move with the mix of what is on the market as well as with underlying values. Treat the direction and the spread as the signal, not the decimal.
A distribution with a lot of variance around a stable mean is normal. Housing has always had expensive metros and cheap ones, hot years and slow ones. What is different now is that the sign is inconsistent — metros are not merely falling at different rates, they are moving in opposite directions at the same time.
Austin's median list price down 12.2% is not a slower version of the national -2%. San Francisco inventory down 20% is not a milder version of the national +3.2%. San Jose sitting at a $1.75 million median is not a variation on any national figure at all. When the components of an average disagree about direction, the average has stopped summarising them and started concealing them.
This is not a subtlety for statisticians. It is the difference between a buyer in Austin who reads that prices are down 2%, negotiates accordingly, and leaves ten percentage points on the table — and a buyer in San Francisco who reads the same headline, expects choice, and finds a fifth fewer homes available than last year.
Here is the part most coverage skips. If the divergence were a demand story — rates, employment, national sentiment — you would expect metros to move together with different amplitudes, because the drivers apply broadly. Mortgage rates do not vary by city. A national labour market cools nationally.
They are not moving together, which points at the input that does vary by city: supply. And housing supply is the output of local political decisions with very long lags.
Metros that permitted aggressively through the last cycle — Austin being the canonical case, along with much of Texas, Florida and the mountain west — spent years approving and building. Those units are now finished, listed, and competing with each other. The result is inventory, and inventory is what makes prices fall. Metros that permitted little — much of coastal California, the Northeast — have no such pipeline. When their sellers stay put, nothing replaces the listings, inventory contracts, and prices hold or rise even with weak demand.
The construction decisions producing today's inventory were made three to seven years ago, in city council meetings and zoning hearings that nobody outside the room was watching. The divergence in the data is the delayed, visible consequence of a decade of local land-use policy. It is the most legible natural experiment American housing policy has produced in a generation, and it is currently being averaged away.
“Rates apply uniformly across the country. Construction policy does not — and the gap between metros is where that difference shows up.”
Two honest caveats. Falling prices in a high-supply metro are not costless: recent buyers there are underwater, and builders respond by pulling back, which sets up the next shortage. And Austin's decline is not purely a supply story — it followed an extraordinary pandemic-era run-up, so some of the fall is a round trip rather than a new equilibrium. The supply mechanism explains the direction; it does not explain the entire magnitude.
Being contrarian about a statistic is easy and usually shallow, so it is worth stating where the aggregate does carry information.
National figures are the right instrument for national questions. Mortgage credit conditions, the shelter component of inflation, homebuilder equity, mortgage-backed securities, the aggregate wealth effect on consumption — all of these operate at the level of the country and none needs metro resolution. The Federal Reserve is not setting policy for Austin.
Aggregates are also useful for change in dispersion itself. If the national median were falling while every metro fell, that would be a demand story and would call for a different response than what is actually happening. Knowing the aggregate is what lets you notice the components disagree with it.
What the national number cannot do is answer the question almost everyone reading it is asking, which is whether to buy or sell a specific house in a specific place this year.
All three are free, published, and take about twenty minutes to assemble for any US metro.
There is a fourth question that costs nothing: what is the price per square foot trend in your specific zip code, as distinct from your metro. Metro-level dispersion has a smaller version of itself inside it, and a metro median can hide a suburb falling and a core rising just as effectively as a national median hides Austin and San Francisco.
Most economists currently characterise 2026 as a rebalancing year rather than a downturn — sales stable rather than surging, inventory rising slowly from a very low base. That characterisation is a reasonable summary of the aggregate. It is also, for a large fraction of the country, a description of somewhere else.
Nationally, list prices are down modestly — roughly 2% year on year, with Realtor.com's asking-price series showing a 2.5% decline it describes as the steepest since 2017. But that aggregate conceals metros moving in opposite directions: Austin's median list price is down 12.2% year on year while San Francisco's active inventory has fallen nearly 20%. Whether prices are falling depends almost entirely on where you are.
DrafterDaily's finance coverage goes underneath the headline statistic to the data that actually applies to a decision.
More finance analysisGrad PLUS lent up to the full cost of attendance. Its replacement caps professional students at $50,000 a year — against a medical-school year that averages $61,110. The difference does not disappear.
Fitch tracks sixteen perpetual non-traded BDCs. In Q2 2026, redemption requests averaged 10.3% of shares against a standard 5% cap, and ten of the sixteen breached it. Nothing broke — which is the uncomfortable part.
A composite index of 56.0 is the strongest US reading since April 2022. It is also an average of a services sector at a 20-month high and a factory sector going the other way.