The Consensus Trade
Nearly everyone in Japan’s domestic property market is making the same bet: rents will rise faster than interest rates. This letter lays out the bet, the machinery that holds it up, and the places where it is already cracking.
The line the market crossed
Japan’s real estate securitization association, ARES, surveys the country’s private property fund managers twice a year. One question asks: at what level of the 10-year government bond yield would you consider changing your investment policy? In the survey taken this January — published in March, measuring the market as it stood at the end of 2025 — the answer was 2.5%. Six months earlier, it had been 2.0%.
The same survey recorded what the market looked like as it approached that line. The private fund market had grown 4.9% in six months, to ¥47.1 trillion. Planned leverage on newly created funds stood at 70.6% — the highest the survey has ever recorded. Not one manager reported foreign investors pulling back. The market was not edging toward its own line cautiously. It was accelerating toward it, at record leverage.
Then, this spring, the 10-year yield went through 2.5%. By May it was in the high 2s; today it trades around 2.8% (2.865% on July 8), a level last seen in 1996.
Has anything changed since the crossing? The formal answer does not exist yet. ARES ran its next survey in July, and the results — the first measured above the line — arrive in the autumn. But the fresher evidence from the spring points one way: no braking. Rent forecasts collected at the end of May, with yields already near 2.8%, contain zero calls for decline. Bank lending was still accelerating in May. And the May transaction tables show domestic buyers still paying yields around 4% for Tokyo offices.
Why no braking? Because almost everyone inside this market shares one belief: rents will rise faster than rates, so today’s prices are still justified. That belief is what I call the consensus trade. Much of this publication, going forward, will be about testing whether it holds.
The consensus is written down
This is not my characterization. It is documented.
Every six months, Nikkei Real Estate Market Report asks 18 analysts to forecast central-Tokyo office rents. The latest round was answered at the end of May — that is, with the 10-year already near 2.8%. The number forecasting rents to fall at any point through end-2027: zero. All 18 see rents peaking in 2028, 2029, or later. And 76% expect construction costs to keep rising beyond 2030.
The pricing side says the same thing. The Japan Real Estate Institute surveyed investors in April, as yields were crossing the line, and found expected property yields flat or falling — in Tokyo and in regional cities, across offices, apartments, logistics, and hotels. Through the fastest rise in interest rates in thirty years, the yield investors demand from property has not gone up.
Write the consequence plainly and it is remarkable: in central Tokyo, a building can now yield less than a government bond. Buyers accept this, because they expect the building’s rent — unlike the bond’s coupon — to grow.
Three things hold the bet up
First, rents are actually rising. Office vacancy in Tokyo’s five central wards is 2.07%. Asking rents around Tokyo Station have reached ¥42,000–60,000 per tsubo, last seen in 2008. Twenty of Tokyo’s 22 office districts rose 5% or more in six months.
Second, new supply is collapsing — for the same reason prices are rising. Large-office completions in Tokyo’s 23 wards fall to 740,000 m² in 2027, half of this year’s volume. Delays are everywhere; one landmark hotel redevelopment slipped about six years. The cause is construction costs. Note the loop: inflation pushes rents up, and the same inflation blocks the new buildings that would hold rents down. The bet partly protects itself.
Third, the debt keeps flowing. In the Bank of Japan’s May data, major-bank lending grew 8.7% year on year. Regional banks are pulling back from Tokyo deals, and the megabanks are taking their place — the lender base is rotating, not shrinking. Because loan rates have risen far less than the policy rate, borrowed money still makes the numbers work.
One catch: the cap rates are not neutral evidence
“Cap rates haven’t moved” sounds like proof that the market is fine. Look closer and it proves something different.
A cap rate is just rent divided by price. Take a building that sells for ¥500,000, with current rent of ¥10,000 a year. That is a 2% yield — alarmingly low. But if the buyer believes the rent will reach ¥20,000, they will tell you they bought a 4% yield. Same building, same price. The difference is which rent you count.
Practitioners at the Japan Real Estate Institute report that domestic lenders generally underwrite today’s rent, while foreign funds underwrite tomorrow’s. Observed cap rates stay low precisely because buyers are already counting tomorrow’s rent inside today’s price.
The institute’s numbers show the mechanics. Investors price a building by adding a premium to the government bond yield — extra return for the extra risk. For prime Marunouchi offices, the expected yield has sat at 3.2% since late 2022. The bond yield underneath it has risen sharply. So the premium shrank: in the institute’s survey, from 2.5% last October to 2.0% this April. The rate rise was not absorbed by prices falling or yields rising. It was absorbed by investors accepting less compensation for risk — because they expect rents to make up the difference.
So the famous flat cap-rate chart is not independent proof that the consensus is right. It is the consensus itself, showing up in the data. And it tells you where a correction would appear first: not in interest rates, but in the rent assumption.
Who disagrees
The stock market, for one. The Tokyo REIT index sits near 1,800; in May it was the worst-performing REIT market in the region. By one securities-firm estimate presented at a June industry forum, prices at this level imply long-term rates rising toward 3.2%. Listed investors, in short, are not in the consensus.
Foreign research houses split the difference. DWS, in its June Asia-Pacific outlook, likes Japanese offices but expects cap rates to widen by around half a percentage point over the next several years, with rent growth offsetting only part of that. That means prices drift down. PGIM’s leadership, visiting Tokyo, took the opposite view: rates are rising because the economy finally works, and rent growth can absorb the whole adjustment. That means prices hold. Between “part” and “whole” sits most of the return investors will actually earn in 2026–27.
Where it is already cracking
The Japan Real Estate Institute names the market’s biggest risk plainly: not the bond yield, but the rent expectation set against it. If the expected rent increases fail to arrive, the whole structure re-prices.
Two small, real examples already exist. In Kumamoto, the town next to TSMC’s factory site built housing ahead of demand; when the second plant’s construction schedule slipped, occupancy fell to about 84%. In Sendai, years of apartment construction have caught up with demand: occupancy in the high 80s, flat rents, discounts appearing even on newer buildings.
Neither story involves interest rates. Both show the actual shape of the risk: local, demand-driven, and starting with rent.
Who is on which side
May’s transaction tables for greater Tokyo make the sides visible. Investment vehicles affiliated with Blackstone sold three 1990s mid-size office buildings. The buyers were Tosei, Tokyo Tatemono Real Estate Sales, and Keio — all domestic. Across the tables, foreign private equity appears as a seller and never as a buyer.
July’s tables add a footnote that fits the pattern: the first foreign purchases to surface are hotels, not offices — a Singapore-based group paid over ¥25 billion for a Shinjuku lodging asset this month.
The pattern is simple. Domestic money is paying prices that assume rents keep rising. Part of the foreign money is choosing to take today’s prices and step out. One side of that ledger will turn out to be right.
How we will know
Four checkpoints, all public.
The most important date on this calendar is the autumn ARES release. The July-taken survey is the first measured above the 2.5% line, and it will show whether “considering a change” became an actual change — in allocations, planned leverage, and reported borrowing costs. The Japan Real Estate Institute’s next investor survey lands around October; watch whether expected yields finally rise. The quarterly broker rent survey (September data, out around November) tests whether rents keep performing. And the BOJ’s monthly lending figures, around the 10th of each month, show whether the debt keeps flowing.
This letter takes no view on where bond yields go. It watches one thing: whether the rents everyone is counting on actually arrive.
Sources: ARES investor survey (March 2026 release); Nikkei Real Estate Market Report (analyst survey, rent survey, deal tables, supply census); Japan Real Estate Institute essay and 54th investor survey, via Monthly Property Management; Fund Review (APREA forum report, DWS outlook); Monthly Property Management (PGIM interview, regional city reports); Bank of Japan lending statistics.