Hong Kong’s Property Recovery Has a Volume Problem

Four claims about the 2026 rebound, checked line by line against the Rating and Valuation Department index and the Land Registry contract count

AsiaPresswire – In September 2026 the private domestic price index read 320.8, up 0.19% on the month and 9.3% on the year. Rents read 212.2, an all time high, up 0.47% on the month and 6.2% on the year, the eleventh consecutive monthly increase. Both numbers come from the same department, are published together, and were reported together.

In the same month the Land Registry registered 4,761 residential sale and purchase contracts. That is 18.5% more than August and 15.6% fewer than September 2025. The consideration was HKD 46.8 billion, up 29.3% on the month and 0.9% lower on the year.

Two releases, one month, opposite directions. The index says the market is repricing upward. The registry says fewer people are doing it. Both are official, both are current, and almost every published claim about Hong Kong property this autumn uses one of them and quietly drops the other.

The judgment this piece defends is that the recovery in prices is real and the recovery in the market is not, that the difference is a mix and volume story rather than a sentiment story, and that three of the four bullish claims circulating in October 2026 fail on the department’s own numbers. The fourth survives, in a narrower form than the people making it intend.

Four Claims, and Where Each One Comes From

None of these are straw men. Each is attached to a named publisher or a specific published figure, which is the only reason it is worth dismantling.

• The market bottomed in 2025 and is now in recovery, supported by a forecast from Knight Frank’s Hong Kong research team in June 2026 projecting a 8% to 10% rise in residential prices for the year.
• Record rents mean rental yields are improving, supported by the rent index printing an all time high of 212.2 in September 2026.
• Mainland buyers are driving the rebound, stated routinely by agents and repeated in market commentary without a monthly series attached.
• Offices will follow houses, supported by the same Knight Frank note, which reported that Grade A office supply has peaked.

The fourth claim is the one that gets misread most often, and the misreading is instructive, so it goes last.

Claim One: The Market Bottomed and Is Recovering

The evidence offered is the price index. The index does show a recovery: 320.8 in September, 7% higher over the first nine months of 2026, and 9.3% higher year on year.

What the index does not show is a market with more participants. September’s 4,761 residential contracts sit 38% below the twelve month moving average of 7,682. The moving average itself fell 0.5% on the month, which means the trend in activity is down even as the trend in price is up. Year on year the moving average is up 20.6%, so the first half of 2026 was genuinely busy and the third quarter was not.

Table 1: Price, rent and transaction readings for September 2026, with month and year comparisons

Source: Hong Kong Commercial Daily , citing the Rating and Valuation Department, 28 October 2026; Hong Kong Government press release , Land Registry monthly statistics, 5 October 2026.

A market can rise on falling volume for a while. It cannot do so indefinitely, because price discovery needs transactions, and the marginal seller in a thin market is the one who has to move. The 38% gap between September and the trailing average is the first thing to watch, not the 9.3%.

There is a second problem with the claim as stated. “Bottomed in 2025” requires a 2025 low to compare against, and the index comparison being used is a year on year change of 9.3% off that low. A 9.3% gain from a cyclical trough is a normal first year of a recovery, not evidence of a durable one. The number that would support durability is the two year change, and the two year change is much smaller than the headline.

Claim Two: Record Rents Mean Better Yields

Rents are at a record. That part is true and it is not in dispute: 212.2, eleven consecutive monthly rises, 5.8% over the first nine months.

Yields are a ratio, though, and the numerator is only half of it. Over the same twelve months, prices rose 9.3% and rents rose 6.2%. When the denominator grows faster than the numerator, the ratio falls. A gross yield of 3.9% for Hong Kong residential in 2026, which is the figure published by Global Property Guide, applied to that relationship gives roughly 3.8% on the current relationship between the two indices.

So the correct statement is that rents are at a record and yields are slightly lower than a year ago. The published claim inverts it.

Over the first nine months the gap is smaller, because prices rose 7% and rents 5.8%, which is close to a wash. That is the honest version of the bull case: in 2026 rents nearly kept pace with prices, which is unusual in a rising market and does say something about underlying housing demand. It does not say yields are improving.

Two costs sit between the gross yield and what an owner receives. Management fees and government rates are paid by the owner in most Hong Kong leases, and vacancy between tenancies is unpaid. A 3.9% gross figure is therefore not a 3.9% return, and any comparison with a funding cost has to start from the net number rather than the headline.

The Mix Explains More Than Sentiment Does

Here is the detail in the September release that most coverage skipped, and it does more work than any of the four claims.

Small and medium units, meaning those below 100 square metres, rose 0.16% on the month. Units of 100 square metres and above rose 0.80%, five times as much. Rents rose 0.47%, which is three times the pace of the mass market price index.

Read those three figures together and the mix story writes itself. The headline index is being pulled up by the top of the market, where transactions are few, tickets are large and buyers are less likely to need a mortgage. The mass market, where most households actually transact, is close to flat on price while its rents rise faster. Average consideration per residential contract in September was HKD 9.83 million, which is high for a market whose median household cannot buy at that level without substantial equity.

This is what a market looks like when it is being repriced by cash rather than by credit. The buyers who can transact without a mortgage set the price. The rest wait, and rent. A thin, high end led recovery is a real recovery in asset prices, and it is a weak recovery in the housing market as a lived experience.

Claim Three: Mainland Buyers Did This

This claim is made constantly and it cannot be checked at the frequency at which it is made, which is itself the most interesting thing about it.

The Land Registry monthly release reports contract counts and consideration by building type. It does not report buyer origin. The duty series that once separated non local purchasers was retired when the demand side measures came off in 2024, so there is no longer a monthly stamp duty line that a reader can use to test the claim.

What exists is annual and semi annual research, agent estimates, and anecdotes. None of that is worthless. None of it supports a sentence like “mainland buyers drove the September rebound”, because September is a month and the data is a year.

The honest version is that cross border demand is a plausible contributor to the high end, where the September gains are concentrated, and that nobody publishing the claim this autumn has a monthly number to put behind it. That is a gap in the evidence, not proof the claim is false. It is also exactly the kind of claim that a thin, high end led market produces, because the visible buyers in a thin market get credited with moving it.

Claim Four: Offices Will Follow Houses

The Knight Frank note is worth reading precisely because of what it says. Its finding is that Grade A office supply has peaked. Supply peaking is a construction cycle statement. It means completions are near their high, which is the point at which vacancy is most likely to rise, not the point at which rents turn.

Turning that into “offices will follow houses” requires an absorption number, and absorption is not in the sentence. Hong Kong’s office market has its own cycle, driven by the requirements of financial, professional and mainland corporate tenants, and it has been running behind the residential market for several years.

The residential market does not lead the office market in any mechanical way. They share a funding environment and a city, and they do not share a demand driver. A reader who takes “supply has peaked” as “offices are next” has been handed a supply fact and read it as a demand forecast.

There is, though, a smaller fact in the September registry release that cuts against dismissing the commercial side entirely, and it is only visible if you subtract.

The Non-Residential Slice Is Already Moving

The registry publishes two sets of numbers in one release: all building types, and residential only. Subtracting one from the other gives the non-residential slice, which includes shops, offices, industrial units and car parks. Nobody headlines it, and it moved differently from housing.

All building contracts in September were 6,438 worth HKD 53.5 billion. Residential was 4,761 contracts worth HKD 46.8 billion. The remainder is 1,677 contracts and about HKD 6.7 billion, an average ticket of roughly HKD 4.0 million, less than half the residential average.

Now compare the year on year changes. Total consideration was almost exactly flat at +0.01%, while residential consideration fell 0.9%. For both statements to be true, the non-residential slice must have risen by roughly 7% in value against a smaller base. Contract counts tell a similar story: all building types fell 6.3% year on year while residential fell 15.6%, so the non-residential count was broadly flat or slightly up.

That is not a boom. Seven percent on a HKD 6.3 billion base is a rounding error in a city where residential alone turns over HKD 47 billion a month. But it does mean the sentence “offices will follow houses” is aimed at the wrong market. The commercial slice is not waiting for housing to lead it. It is already trading at a steadier rate than housing is, and the reason is probably that its buyers are yield driven and its sellers are less levered, which is exactly the profile that survives a thin market.

What the registry cannot tell you is which part of that slice moved. Shops, offices and industrial units are aggregated. A reader who wants to know whether Grade A office changed hands in September will not find it here, and that absence is why the “supply has peaked” note gets used as a proxy for a transaction market that has no monthly public tape.

Why These Four Stories Travel

The four claims have different publishers and the same transmission mechanism, and it is worth naming because it will produce the same four claims next quarter.

The price index is the only housing number published on a fixed monthly schedule with a single headline figure. Volume is published by a different department, on a different date, in a release whose first three bullets are about all building types rather than residential. A journalist on deadline has one number and it is the index.

The registry carries a lag of up to 30 days between transaction and registration, and the release says so in its own footnote. September’s release therefore describes August’s deals. Anyone comparing the September index to the September contract count is comparing two different months without knowing it.

Then there is the base effect. A 9.3% year on year gain measured off a 2025 trough is arithmetically easy and rhetorically powerful. The same 9.3% measured off 2021 would be a different story, and no headline writes it that way.

And finally the incentive. Agents are paid on transactions and quote prices. Landlords are paid on rents and quote rents. Neither group is paid to publish the contract count. The contract count is the number that tells you whether the recovery has participants, and it is the one number in the set with nobody promoting it.

There is a further reason the price story spreads faster than the volume story, and it is structural rather than commercial. The index is a single number that can be compared to last month, last year and a forecast, which makes it a headline. Volume is a count that only means something when compared to a trend, and the trend requires you to choose a window, 12 months or 24 or the pre-2021 peak. Every choice of window produces a different sentence about the same month. Journalists avoid that problem by not writing the sentence.

That asymmetry is not going to close. What can close is the reader’s habit of treating the two releases as one dataset. They are published by two departments, on two dates, with two different populations, and the 30 day registration lag means the September release describes August’s deals while the September index describes September’s prices. Comparing them month to month is comparing a photograph to a receipt from last month, and the only defensible comparison is between trends.

What Survives the Check

Two claims survive, in modified form, and one new one emerges from the arithmetic.

Prices are genuinely higher. A 9.3% year on year gain and a 7% gain over nine months are large moves on an official quality adjusted index, and they are not an artefact of a handful of trophy sales. The direction is real.

Rents are genuinely at a record and genuinely rising, at 6.2% year on year. In a city that lost population in the early 2020s and has been adding it back, that is a demand signal about housing as shelter, and it is a stronger signal than the price index because rents are paid monthly by people who have to live somewhere.

And the mass market is behaving differently from the top. Small and medium unit prices rose 0.16% in September while their rents rose faster, which means the yield on the segment where most tenants live is not compressing. The compression story is a high end story. That distinction is missing from both the bull and bear versions of the 2026 narrative, and it is the one that matters to anyone deciding whether to buy a 500 square foot flat rather than a 2,000 square foot one.

The version of the market that survives all four checks is narrower and more useful than either headline. Prices are up, on fewer deals, led by larger units and cash buyers, with rents rising underneath at a rate that is good for landlords and not yet good enough to change the arithmetic on a mortgage.

What Would Change My Mind

Three observations would move this read, and each has a date attached.

If residential contracts in October and November 2026 come in above the twelve month moving average of 7,682 for two consecutive months, the volume problem is a third quarter timing issue and the recovery has participants. Watch the registry release in early December.

If the small and medium unit price index rises more than 1% in a single month while rents hold, the mass market has joined the top end and the mix explanation is too clever. That would be a broadening recovery, and it would be good news.

And if either the Inland Revenue Department or the Rating and Valuation Department begins publishing a monthly purchaser origin series, the mainland buyer claim becomes testable for the first time since 2024, and it will either be confirmed quickly or disappear from the commentary quickly. Either outcome is more useful than the current situation, in which a load bearing claim about the market cannot be checked by the people reading it.

Data basis

Data as of October 2026. The price and rent indices are the Rating and Valuation Department’s private domestic series for September 2026, published 28 October 2026, with month on month and year on year changes as reported; the index base is the department’s own and is not rebased here. Transaction counts and consideration are Land Registry figures for September 2026, published 5 October 2026, and cover sale and purchase agreements registered in the month, which the release notes may reflect deals struck up to 30 days earlier. The twelve month moving average is the registry’s own, used here as the trend measure because the monthly series is volatile. Gross rental yield is Global Property Guide’s 2026 figure for Hong Kong residential, a gross measure before management fees, government rates and vacancy, and is not comparable to a net yield. Average consideration per contract divides the published residential consideration by the published contract count. The yield direction estimate applies the ratio of the rent index change to the price index change to the published gross yield, which is an approximation and not a re-estimation of the yield.

Limitations

The index is quality adjusted and the registry is not, so the two are not measuring the same population of transactions and the 38% gap should be read as a signal rather than a precise shortfall. The mix argument rests on two monthly figures, which is one observation, not a trend. And the purchaser origin claim is treated as untestable rather than false, because absence of a monthly series is not evidence of absence of the buyers.

Sources

• Hong Kong Commercial Daily , September price index 320.8, rents 212.2 at record, 28 October 2026
• Hong Kong Government press release , Land Registry September statistics, 5 October 2026
• Knight Frank Hong Kong , 2026 Hong Kong property market forecast, 17 June 2026
• Global Property Guide , Hong Kong gross rental yield, 2026 (site home)
• Rating and Valuation Department index series underlying the September release, no separate dataset link (no public link)

This article was prepared by the expert team at Global Infinity Labs, drawing on insights and data provided through its MCP data services.

About Global Infinity Labs

Global Infinity Labs is an AI innovation company specializing in MCP data services.

Email: support@globalinfinitylabs.com