Is the Housing Market Going to Crash? What the 2026 Data Actually Shows

Research note. Computed from NAR and Census sales data, Zillow and Case-Shiller prices, the Freddie Mac mortgage survey, Census supply and construction series, the Fed’s Z.1 balance sheet and Redfin’s buyers-versus-sellers estimate, as of August 17, 2026. The exact methods are in How we checked it below; our editorial and AI standards are in How we use AI.

Existing sales

4.06M

Typical home

$372K

Real price YoY

-2.8%

Payment (P&I)

$1,888

Next NAR release

≈ Sep 11, 2026

The question arrives in the same shape every cycle: is the housing market going to crash? It usually means one specific thing, 2008, and 2008 is the one outcome the current data argues against most clearly. What the data does describe is stranger than a crash and, in some ways, harder to live with. Almost nobody is transacting. Prices in dollars have stopped moving. Prices after inflation have gone nowhere but down since the spring of 2022. And the buyer who cannot afford the payment and the owner who will not give up a 3% mortgage are staring at each other across a gap that neither can close.

Start with the number that describes the freeze. In July 2026, existing homes changed hands at an annual pace of 4.06 million. Our record of that series reaches back to 1999, and only 37 of its months were slower: the trough of the foreclosure crisis in 2010, at 3.45 million, the two lockdown months of 2020, and the low of the 2022 rate shock. The peak, in September 2005, was 7.26 million. The country has added roughly 40 million people since then and is selling homes at about the pace it did when the last housing bust bottomed.

Why nobody is selling, and nobody is buying

The freeze has a single cause with two faces. The typical US home, by Zillow’s index, cost $371,774 in July 2026, up 1.0% from a year earlier. In August 2020 the principal-and-interest payment on that typical home, with 20% down and a 30-year mortgage, was $877 a month. In June 2026 it was $1,884. The price rose about 42% over those years; the mortgage rate went from 2.65% in January 2021 to 6.67% now, with a stop at 7.79% in October 2023. The payment more than doubled, and it sits in the 94th percentile of every month since 2000. It has come down from its $2,039 peak, but only by the width of a rate wiggle.

For the buyer, that is the wall. For the owner, it is a golden handcuff: someone who financed at 3% would swap that loan for a 6.67% one only under duress, so they do not list. That is why the same payment shock froze both sides of the market at once, and why active listings, at 1.39 million in July, are still 18% below the same month of 2019 even after doubling from their February 2022 low of 707 thousand.

Redfin measures the standoff directly. Its balance-of-power estimate counted 1.01 million active buyers against 1.48 million sellers in May 2026, about 1.47 sellers for every buyer, a hair below the series’ record of 1.50 in December 2025. In the spring of 2021 the ratio was 0.64: buyers outnumbered sellers. It is Redfin’s model, not ours, and it is restated wholesale between vintages, so read the level loosely and the direction firmly. Sellers have outnumbered buyers by a widening margin for two years, and the market has not cleared through price. It has cleared through delistings and patience.

Prices are flat in dollars and falling in everything else

Here is the sentence the crash question is really asking about, and it needs both halves. In nominal terms, home prices are not falling: the Zillow index was up 1.0% year over year in July, and the Case-Shiller national index was up 1.1% in May. In real terms the picture inverts. Deflate Case-Shiller by the CPI and the index was 2.8% lower than a year earlier in May 2026, its 11th consecutive month of year-over-year decline, and 4.6% below the real peak of May 2022. It has not set a new real high in four years, and it was below its year-earlier level in 19 of the last 40 months. Real prices are still 9% above the 2006 top, so this is not a reversion to the old cycle; it is a slow leak from the pandemic spike.

That is what a correction looks like when nobody is forced to sell. In 2008 the adjustment came through price, because delinquent owners had to transact at whatever the market would pay. Today the same adjustment is coming through time: prices sit still while incomes and the price level grind higher around them, and the real value of a house erodes, this year at roughly 3%, without a headline. It is gentler for owners and slower for buyers, and it can run for years.

Where the overhang actually is

The soft spot is not the resale market. Existing-home supply is 4.6 months, which the industry has long called balanced; it was the same 4.6 a year earlier. The overhang is in new construction. Builders were carrying 9.3 months of new-home supply in June, the 94th percentile of a record that starts in 1963 and within sight of the 12.2 months of January 2009. They have responded the way builders can and owners cannot: the median new home sold for $398,300 in June, 13.5% below its October 2022 peak, because a builder with a mortgage-rate buydown and a quarterly target will cut where an owner with a 3% loan will not.

The pipeline behind them is thinning. Units under construction have fallen to 1.26 million from 1.71 million in October 2022, and single-family completions now run ahead of single-family starts. That is the mechanism by which this freeze eventually resolves without a crash: the new-home discount clears the builders’ inventory, the pipeline shrinks, and the resale market waits for either rates or incomes to close the payment gap.

How this compares to 2008

The comparison people reach for deserves a table rather than a mood. Activity now looks like the bottom of the last bust. The balance sheet looks like its opposite.

Two lines carry the argument. Owners’ equity was 72% of the value of American homes in the first quarter of 2026, the 85th percentile of a record that starts in 1945; at the 2012 low it was 46%, and even in late 2006, before anything had broken, it was 59%. The delinquency rate on single-family mortgages was 1.89%, against 11.48% at the start of 2010. A crash needs sellers who have to sell. This market has owners who would rather wait, and the equity to do it.

What would change our mind

The frozen-not-crashing read is falsifiable, and these are the numbers that would falsify it. A crash needs a forced-selling engine, so the first tell would be the delinquency rate turning up from under 2% while resale supply moved past six months. The second would be Redfin’s seller-to-buyer ratio holding near its record while nominal prices rolled negative year over year, which would mean the standoff had finally broken toward the buyer. Neither is in the data as of publication.

The freeze can also end the other way. A durable move in the 30-year rate toward 5.5% would cut the typical payment by about 12% and bring in more buyers than sellers, because the buyer’s wall falls faster than the owner’s handcuff loosens. That is the scenario in which prices firm and the real-terms decline stops. Between those two, the base case is more of the same: low volume, flat dollar prices, and a real correction that arrives one CPI print at a time. The Real Estate Tracker carries every one of these series with dated readings, so the sentence above can be checked against next month’s data rather than remembered.

How we checked it

Every number is read from the published series; nothing is estimated. In plain terms:

• ·Percentiles are simply where the latest print ranks in its own history: the 11th percentile for sales means about 11% of months since 1999 were slower. Sales, payment and new-home supply are ranked against their full records; the equity share against its record since 1945.
• ·Real prices divide the Case-Shiller national index by the consumer price index, so a flat dollar price with 3% inflation shows as a 3% real decline. The streak counts consecutive months in which that real index was below its level a year earlier.
• ·The typical-home payment takes Zillow’s value for the middle US home, assumes 20% down and a 30-year loan at that month’s average Freddie Mac rate, and computes principal and interest only. Taxes and insurance would raise every month by a similar amount, so the doubling since 2020 understates the full-cost change slightly.
• ·The equity share is owners’ equity divided by equity plus mortgage debt from the Fed’s quarterly balance sheet, which is the same as one minus the aggregate loan-to-value of American homes.
• ·The buyers-versus-sellers figures are Redfin’s estimate, not ours: sellers are active listings and buyers come from their model, and each monthly report restates the whole series. We quote the latest vintage and say so.
• ·The 2008 comparison uses each series’ own extreme from 2008 through 2012 rather than a single date, because the sales trough (2010), the price trough (2012) and the delinquency peak (2010) did not coincide.