Rents Rose Faster in Q2. Vacancy Rose Too. And 387 Fewer Homes Were Actually Lived In.

Two numbers came out of URA's second-quarter release on 24 July, and they point in opposite directions. Rents accelerated: the private residential rental index rose 0.7% for the quarter, more than double Q1's 0.3%. And vacancy rose too, to 6.4% from 6.2%.

That combination is not supposed to happen. Rents rise when homes are scarce. Here the share of private homes standing empty went up while the cost of renting one went up as well — and the explanation sits in a line most of the coverage skipped past: the stock of occupied private homes actually shrank.

The headline is a landed number

Split the rental index and the acceleration disappears almost entirely.

Source: URA, Q2 2026; segment figures as quoted by CBRE, 24 July 2026.

Non-landed rents — which is to say the rental market as almost every tenant and landlord experiences it — grew 0.4%, exactly the same as the previous quarter. Nothing accelerated. The whole of the headline's improvement came from landed housing at +2.7%, a thin, lumpy segment where a handful of Good Class Bungalow and terrace leases can move an index that governs perhaps a twentieth of the tenancies signed in Singapore.

Inside non-landed, only one region moved: the CCR at +1.2%. The RCR was flat to the decimal. The OCR — the largest rental pool by far — went negative, reversing a +1.0% quarter.

So the honest one-line summary of Q2 rents is not "rents accelerated." It is: landed rents jumped, prime rents firmed, and the suburbs went backwards.

It is not a supply flood

The reflex explanation for rising vacancy is that too many new homes completed at once. The data rules that out for this quarter.

Only about 700 units (excluding ECs) were completed in Q2 2026, against 911 in Q1 — a 23% drop. Fewer new homes arrived than in the previous quarter, and vacancy went up anyway. The empty units are therefore not, in the main, freshly built ones sitting unlet. Something released tenants from homes that were already occupied.

URA's own stock series says so directly:

The stock of occupied private residential units (excluding ECs) decreased by 387 units in 2nd Quarter 2026, compared with the increase of 225 units in the previous quarter.

CBRE frames the causality the same way: occupancy "continued to deteriorate in Q2 2026 despite fewer completions." That is the actual event of the quarter. Not a supply wave — a net loss of occupied homes, for the first time in recent quarters.

A 387-unit fall is small against a stock measured in the hundreds of thousands. It is the sign that matters. Occupancy going backwards while completions slow is a demand-side reading, and it is the opposite of what an accelerating rent index implies.

More leases signed, fewer homes occupied

The other number worth holding next to it: 22,290 rental contracts were signed in Q2 2026, up 5.1% on the quarter.

More tenancies were transacted, and fewer homes ended the quarter lived in. Those two facts are only compatible if a meaningful share of that leasing activity was churn — existing tenants moving between units, renewing elsewhere, or trading down — rather than new households forming and taking homes off the vacant pile.

We would flag that as an interpretation, not a measured fact; URA publishes contract counts, not tenant identities. But it is the reading that reconciles the three series, and it fits the segment pattern: the region with the biggest occupancy deterioration (OCR, vacancy 5.2% to 5.6%) is also the one where rents fell. Tenants with options use them.

Note the RCR, which is the only region where occupancy actually improved — vacancy fell from 6.3% to 6.1% — and where rents were precisely flat at 0.0%. Improving occupancy bought no rent growth at all. Meanwhile the CCR posted the strongest rent gain, +1.2%, on the highest and still-rising vacancy in the country, 8.3%. About one in twelve completed prime homes is empty.

A rent index that rises fastest where homes are emptiest is not measuring scarcity. It is measuring what is being transacted at the top of the market.

What a 0.7% rise is actually worth

For anyone who owns a unit rather than an index, the quarter's rent move is a rounding error next to the thing the same release is warning about.

Take a condo bought at S$1,600,000 and let at S$4,500 a month — a 3.38% gross yield before any costs.

Computed, not estimated. On the same unit, gross yield runs 3.38% fully let, 3.18% with three weeks empty between tenancies, and 2.99% with six.

Three weeks of vacancy costs more than eight years of 0.7% rent increases. And the vacancy rate — the actual measured probability that a completed private home is sitting empty — just went the wrong way, to about one in 15.6 islandwide and one in twelve in the CCR.

What is arriving next

The supply side does not stay quiet.

Completions run 1,611 units in 1H 2026 against roughly 5,012 due in 2H — a 3.1× step-up in the second half alone — then about 8,440 in 2027 and 9,856 in 2028 (all excluding ECs). Counting ECs, URA puts roughly 60,600 units as completing over the coming years, of which about 25,900 land by 2028.

So the quarter that produced a net loss of occupied homes was also the quarter with the fewest completions. The completions return from here.

The read

Nothing in the Q2 data says the rental market is collapsing. Rents are up year on year — the index sits at 162.5, about 1.7% above a year ago — and the RCR's occupancy genuinely improved. This is not a distress reading.

What the data does say is that the headline number is the least informative number in the release. +0.7% is a landed-property event. Beneath it, non-landed rents did not accelerate at all, suburban rents fell, and the country ended the quarter with fewer private homes occupied than it started with, despite the slowest quarter of completions in a year.

For a landlord, that reorders the priorities. The variable to defend in 2026 is not the rent number on the renewal — it is the empty weeks between tenants, which cost multiples of any realistic rent increase and are getting likelier, not less likely, as the completion schedule reloads into 2027 and 2028. Price a renewal to keep a good tenant rather than to chase the index.

For a buyer underwriting a purchase on rental income, the correction is simpler: model the yield on 48 or 49 weeks a year, not 52. At current vacancy that is not pessimism. It is the published base rate.

---

Methodology published. No spin. Rental, vacancy, stock, contract and completion figures from URA's 2nd Quarter 2026 real estate statistics (released 24 July 2026), with segment detail as quoted by CBRE and Beansprout. Supply-pipeline figures including ECs; completion figures excluding ECs — the two bases are not interchangeable. Yield and vacancy-cost figures computed, not estimated. The occupied-stock and contract-count series are published by URA; the churn interpretation drawn from them is ours and is labelled as such.