Tokyo market reports

Tokyo asking prices fell again in July, a 9th straight decline on sale listings

Period covered 2026-07-01 to 2026-07-31 · 4 observation days (sale), 4 (rent)

Prices are falling

Across the Tokyo 23 wards in July, sale asking prices moved -1.58% on a matched-listing basis. Rents moved -1.3%. Both are downward.

The phrase "matched-listing" is doing real work there. Comparing this month's median listing price with last month's cannot tell you whether prices moved or whether the mix of what happens to be for sale moved. So instead we take only the listings that appear in both observations and compare each one against its own previous price. At no point does that fall below 20,911 matched sale listings or 21,227 rentals.

For sale listings we have a clean run since 2026-05-10, and chaining it through gives -3.65%. Every single observation in that run is lower than the one before it, 9 in a row, with no up month or even an up week. Rents run from 2026-06-10 and give -1.49%.

One thing to be clear about before anyone quotes this. A property that sells or lets disappears from the listings, and therefore drops out of the matched set. What remains is weighted toward the listings that did not move, which are the ones under most pressure to reduce. So this measures what happened to prices that kept being advertised. It is not the same as saying the market fell by that much.

When a sale price moves, it moves down

Of the 35,253 sale listings we could follow through July, 21.4% changed price at all. Among those 8,133 revisions, 97.7% were cuts, with only 189 increases in the entire month.

This is the easiest figure on the page to misread, so to be explicit: Itabashi's 100% does not mean every listing in Itabashi was cut. It means one in five listings changed price during July, and every one of those changes was downward. The other four in five ended the month at the price they started it.

Share of revisions that were cuts, by ward (sale)Share of revisions that were cuts, by ward (sale)0%25%50%75%100%Itabashi100%Shinagawa99.8%Koto99.5%MeguroBunkyoSetagayaToshimaMinatoSumidaShinjukuShibuyaKitaSuginamiTaitoNakanoEdogawaNerimaOta95.2%Chuo94.9%Adachi87.4%23-ward average 97.7%
Share of revisions that were cuts

Itabashi is highest at 100% and Adachi lowest at 87.4%. The median cut was -3.23%, and ward medians differ by 0.8 percentage points. No particular ward is being sold off; all 23 are pointed the same way.

WardListings trackedShare that movedOf those, cutsMedian cut
Itabashi1,59520%100%-3.08%
Shinagawa1,57324.2%99.8%-3.36%
Koto2,17924.9%99.5%-3.41%
Meguro1,33225.2%99.4%-2.97%
Bunkyo1,17524.2%99.4%-3.13%
Setagaya2,93719.4%99.3%-3.05%
Toshima1,35726.3%99%-2.94%
Minato3,49321.9%98.6%-3.13%
Sumida80519.1%98.2%-3.08%
Shinjuku2,38722.3%98%-3.24%
Shibuya1,93223.8%98%-3.64%
Kita85420.8%97.9%-3.34%
Suginami1,48924%97.6%-2.94%
Taito1,06922.8%96.7%-3.36%
Nakano1,06626.5%96.5%-2.86%
Edogawa1,23413%96.5%-3.29%
Nerima1,51517.9%95.5%-2.87%
Ota1,49316.9%95.2%-2.9%
Chuo2,07324.4%94.9%-3.6%
Adachi1,65113.7%87.4%-3.46%

Sale prices and rents decay at different rates with age

The rest of this report describes what was actually advertised on 28 July, rather than how prices moved: what is on the market, and at what price.

On the sale side, the median price per square metre more than halves between the newest and oldest age bands. That new buildings cost more is unremarkable. The rate of decline is the interesting part.

Price per m² by building age, newest band = 100Price per m² by building age, newest band = 1002550751000-5 years6-1011-2021-3031-4041+Sale 42Rent 65
Each series indexed to its own newest band, so sale and rent share one axis.
Building ageListingsShareMedian per m²
0-5 years3,1939.7%¥2,218,798
6-103,1689.7%¥2,041,204
11-206,56120%¥1,922,578
21-307,68223.4%¥1,468,997
31-403,1569.6%¥1,018,705
41+9,05427.6%¥937,008

The same bands on the rental side slope far more gently.

Building ageListingsShareMedian per m²
0-5 years23,22420.3%¥4,867
6-1019,28716.9%¥4,662
11-2025,70622.5%¥4,326
21-3018,55116.2%¥4,160
31-4015,88313.9%¥3,266
41+11,76110.3%¥3,143

Sale prices fall steeply with building age; rents do not. An older building therefore commands a higher rent relative to what it costs to buy, which is the whole reason a yield-focused buyer looks at older stock. The gap between those two tables is that argument, in one place. You can calculate the yield for a specific property, or explore price against building age across the 23 wards.

Building age is recorded on 83.4% of rental listings. Those without it are excluded from the age bands.

What is for sale is not what is for rent

The layout mix differs so sharply between the two markets that they are barely describing the same city. Sale listings are family-sized; rentals are single-person.

LayoutListingsShareMedian per m²
2LDK9,23028.1%¥1,753,975
3LDK8,92527.2%¥1,258,153
1LDK4,96815.1%¥1,553,684
2LDK+S1,9295.9%¥1,309,109
4LDK1,6355%¥868,735
1LDK+S1,1233.4%¥1,506,446
2DK7592.3%¥854,291
3LDK+S7292.2%¥1,100,719
1DK6261.9%¥1,195,572
Studio5571.7%¥1,185,185
3DK3711.1%¥644,483
1K3691.1%¥1,233,831
4LDK+S2320.7%¥708,930
5LDK1740.5%¥885,679
4DK1440.4%¥720,648
2K1080.3%¥693,548
2LDK+2S910.3%¥973,510
1LDK+2S880.3%¥1,159,966
3K750.2%¥545,152
5LDK+S630.2%¥975,431
2DK+S570.2%¥811,111
6LDK550.2%¥944,032
LayoutListingsShareMedian per m²
1K48,87035.6%¥4,121
1LDK24,27917.7%¥4,618
Studio23,39917.1%¥4,358
2LDK14,08810.3%¥4,659
1DK11,5168.4%¥4,673
3LDK4,0212.9%¥4,113
2DK3,7622.7%¥2,750
2K3,0282.2%¥3,473
1SLDK1,4181%¥4,563
2SLDK9860.7%¥4,106
3DK5510.4%¥2,258
4LDK3560.3%¥3,206
3SLDK2620.2%¥3,593
1SK1690.1%¥3,927
3K1420.1%¥2,355
1SDK580%¥3,929

Of 32,820 sale listings, 2LDK and 3LDK together are more than half. Of 137,169 rentals, 1K and studios alone reach half. Same city, same month, two different products.

Levels

As at 28 July the median advertised sale price was 8,280 man-yen, or 133.1 man-yen per square metre. The median advertised rent was 11.9 man-yen a month, or 4,299 yen per square metre.

These medians are a different kind of number from the change figures above. A median moves with the mix of what is listed, so it is a poor guide to whether prices changed. Read it as a level, not as a trend.

About this data

We observed the market on 4 days for sale listings and 4 for rentals during July, recording 129,151 and 437,170 listing observations respectively.

These are asking prices, not transaction prices. Nothing here says what any property actually sold or let for.

The change figures come from matching each listing against its own previous observation and chaining those changes together. That construction is unaffected by how often we look. The revision counts are not like that, and do depend on observation frequency.

The ward breakdown covers sale listings only. Rental collection runs newest-first and stops once it is into listings it already holds, so the depth reached varies day to day. That is sufficient for a city aggregate but leaves ward splits drawn from an uneven base. Composition and age bands are measured on the deepest collection day in the period.

Two rental observation days hit the collection ceiling and are excluded from the level figures. Three sale days were dropped as failed collections. Sale figures start from 2026-05-10 and rentals from 2026-06-10.

Data and methodology