By Gary Lim · ERA Senior Division Director · CEA R009877B · Updated 19 September 2026

The story everyone ran this week

On Tuesday and Wednesday this week the Singapore property press ran the same headline in slightly different clothes: two collective sale clusters are back on the market at lower prices. Serenity Park at $440 million, down from $505 million. Pek Chuan Building at $77 million, down from $80 million. The implied conclusion, never quite stated, was that the en bloc reserve price is finally coming down — that owners have capitulated, and that the new consent-threshold law will finish the job.

I run the BuyCondo Team. Seventeen years, 500-plus transactions. My clients are the people who actually have to decide whether to sign a collective sale agreement, and they do not get to decide on a vibe. So I did the boring thing. I went and pulled every collective sale currently in market in Singapore, looked up what each one asked last time, and put the numbers in a row.

Seven clusters. Two cut. Four are asking exactly what they asked last time. One is asking 29.4% more than its last launch.

That is not a market repricing downward. That is a market where two owners blinked and five did not, and where the one that blinked hardest happens to be the one cluster that gets nothing at all from the new en bloc rules Parliament passed on 8 September. I will show you that in a table, because I think it is the single most useful thing on this page.

I also found a date problem in the week’s coverage that changes the framing. I will deal with that honestly in section five rather than bury it.

At a glance: the en bloc reserve price scoreboard

Eight-tenths of one per cent. Across $2.8 billion of asking price, the collective sale market has repriced by less than one per cent. If that is capitulation, I have never seen a stubborn market.

All seven, one table: every live en bloc reserve price right now

This is the table I could not find anywhere else, which is why I built it. Every figure here traces to reporting I actually read this week; where two sources disagreed I have said so in the publishing notes and in the text below.

Notes on the table. *City Plaza’s psf ppr is my own arithmetic — $970m over 141,503 sq ft at plot ratio 3.0 gives about $2,285 psf ppr. The marketing agent did not publish one. People’s Park Centre’s $2,455 psf ppr includes the estimated lease upgrading premium of $535.4 million; I checked that, and $1.48b plus $535.4m over a gross floor area of 821,016 sq ft does land on $2,455, so the published figure is internally consistent. Strip the premium out and the land alone is about $1,803 psf ppr. Lakeside Towers’ $1,277 psf ppr behaves the same way: $350m over 153,237 sq ft at the as-built plot ratio of 2.36 is only about $968 psf ppr, so the published figure almost certainly carries a lease top-up assumption too. That is my inference from the arithmetic, not something the agent stated.

Two of the psf figures do not reconcile cleanly on my numbers, and I have flagged both rather than smooth them over. High Point at $2,645 psf ppr implies a baseline plot ratio nearer 4.30 than the 4.45 that was reported. Trendale Tower at $2,248 psf ppr implies a gross floor area of about 74,700 sq ft against an existing 72,691 sq ft. Neither gap changes the story. Both are on my spot-check list.

The count: two cut, four flat, one raised

Here is the arithmetic, laid out so you can check it.

The mean is positive. Read that again. If you average the seven percentage moves, the Singapore collective sale market has raised its asking prices by 1.83% since its last round of attempts, because one 29% increase outweighs a 12.9% cut and a 3.75% trim. The median move is exactly zero, because four of the seven did not move at all.

I do not think the mean is the honest headline either — it is dominated by one small cluster. The honest headline is the median and the aggregate: the typical Singapore en bloc reserve price in September 2026 has not moved, and the whole cohort has repriced by less than one per cent. Anybody telling you reserves are falling is describing two clusters and calling it a market.

A correction before I go further: the Serenity Park cut is not new

This is the part where I have to correct the premise I started the week with, including my own.

The coverage on 16 and 18 September presented Serenity Park’s $440 million as fresh news. It is not. Chasing the original source, I found a Business Times report carrying the headline “Serenity Park condo owners lower asking price to S$440m in second shot at collective sale” dated 2–3 July 2026, and a property page that recorded the relaunched tender as closing on 28 July 2026.

So the actual sequence is this. Serenity Park launched on 25 February 2026 at $505 million, with the tender closing 26 March. It failed. The committee cut to $440 million in early July and ran a second tender to 28 July. That failed too. What is happening now is a private treaty exercise running to 30 September 2026 — the statutory window in which a committee can negotiate directly after a tender closes unsold.

I am publishing that correction rather than quietly using the September dates, because on this site the nulls and the corrections are the product. If I get the July date wrong, tell me and I will change it again.

Why Serenity Park cut, and cut hard

Serenity Park is a freehold site of 248,173 sq ft off Yio Chu Kang Road, 179 units in ten five-storey blocks, completed in 1995. The plot ratio is 1.4 and the height cap under Master Plan 2025 is five storeys.

Run the developer’s side of it. Plot ratio 1.4 on 248,173 sq ft gives roughly 347,400 sq ft of gross floor area. That is the whole cake. At $440 million the land alone costs $1,266 per square foot of that cake, and the developer still has to build low-rise on a large footprint, which is the least efficient thing you can build per dollar of land. There is no bonus-GFA rabbit to pull out of the hat at plot ratio 1.4; the site is height-capped at five storeys, so the usual trick of buying more floors does not exist here.

That is why the cut had to be big to mean anything. Sixty-five million dollars off a $505 million ask buys the developer no additional floor area whatsoever. It only changes the price of the land the developer was always going to get. A cut that does not buy yield has to be a cut that buys margin, and margin cuts have to be large before a developer notices.

The cost to owners is not small. Sixty-five million across 179 units is an average of about $363,128 per unit — my arithmetic on a straight per-unit split, and actual apportionment follows share value, so individual numbers will vary. They gave that up and the tender still failed in July.

Why Pek Chuan trimmed 3.75%, and why that is not a repricing

Pek Chuan Building at 116 Lavender Street is the only genuinely new cut this week, and it is a $3 million trim on an $80 million ask. In a market where bid-ask gaps run to tens of per cent, 3.75% is a gesture.

The site is 28,700 sq ft at plot ratio 2.5, giving 71,750 sq ft of gross floor area. At $77 million that is $1,073 psf ppr — which reconciles to the penny on my arithmetic, so the published figure is sound. Master Plan 2025 zoning allows up to 12 storeys, and the agent is pitching hotel and mixed-use as possible outcomes.

Its real problem is the lease. Reports put the 99-year lease as expiring in May 2081, about 56 years remaining. EdgeProp’s own project directory lists the tenure as 99 years from 2 May 1978, which would expire in 2077 with about 51 years left. I cannot resolve that from public sources today and I am not going to pretend otherwise. Either way, a developer buying this site has to budget a lease top-up premium to restore a fresh 99 years, and that premium is paid to the state before a single brick moves. On a small site with a tight gross floor area, the top-up is a large fraction of total land cost — which is exactly why the owners can only afford to trim, not to cut. There is not much room between the reserve and the point where the exercise stops being worth doing.

This is the third attempt. The first ran to a tender close of 10 November 2025 at $80 million and failed on both price and consent. The second ran to 10 April 2026, also at $80 million. The third closes 7 October 2026 at $77 million, and the winning bid is 4.8% above the $77 mil asking price set.

The age table: who the new 70% rule actually helps

Now the part I have not seen anyone else do, and the reason I think the two cuts and the five holds are not random.

The Land Titles (Strata) (Amendment) Bill passed its Second Reading on 8 September 2026. It keeps 90% consent for developments under 10 years and 80% for 10 to 39 years, and introduces 70% for developments aged 40 to 59 years and 65% for 60 years and older. The threshold you face is a function of how old your block is. Full detail is in my write-up of the new en bloc rules.

So I looked up the completion year of every cluster in the scoreboard.

There it is. Serenity Park is the only cluster in market that is too young to benefit from the new law, and Serenity Park made by far the biggest cut. Every other cluster is between 44 and 53 years old and falls into the new 70% tier.

I want to be careful about how hard I push this. It is a correlation across seven data points, not a proof. Serenity Park also has a genuinely tough plot ratio, and Pek Chuan is in the 70% tier and cut anyway — though it cut by 3.75%, which is barely a cut at all. But the direction is consistent with how a sane committee would behave, and it is consistent with what I hear from owners: if the goalposts are about to move in your favour, you do not concede ground the week before.

The seven-month clock nobody can start

Here is the thing that is actually freezing this market, and I am going to label it clearly as my reasoning rather than a reported fact.

The Bill passed on 8 September. The commencement date has still not been announced. I have checked daily since 9 September and there is nothing. Until commencement is gazetted, the old 80% threshold is the law, and nobody can begin the transitional process.

The transitional provision is the crux. Developments aged 40 years and up that are already collecting signatures can opt into the new framework and get seven months from the commencement date to reach the new threshold. Seven months is a real window — long enough to work a block, short enough to be a deadline. But it cannot start until the date is set, and the date has not been set.

So every committee in the country is holding a stopwatch with no start button. My read on their position is straightforward:

If the consent bar is about to drop from 80% to 70%, cutting your reserve price today is giving away money you may not have to give away. Wait for commencement, opt in, use the seven months, get the signatures at the lower bar, and relaunch from strength.

That is the best available explanation for why four clusters relaunched at exactly their old number instead of shading it. It is an inference, not a quote from a committee chairman. But it fits the data better than “owners are in denial”, and it fits far better than “reserves are falling”.

There is a mirror image of this argument that nobody on the owner side wants to hear, and I think it is the reason nothing is transacting. Developers can read the same law. A threshold cut from 80% to 70% means more sites will clear consent in 2027 and 2028 than would have cleared under the old rule. More sites means more choice and less urgency. If you are a land buyer, the rational response to an incoming supply wave is to bid less today, not more. Owners see the law as leverage. Developers see it as pipeline. Both are right, and that is precisely why the bid-ask gap is not closing.

Why five owners are holding

Beyond the clock, each holdout has its own arithmetic.

High Point, sixth attempt, $580m. Fifty-nine freehold units on 47,607 sq ft in the Mount Elizabeth pocket, completed 1974, plot ratio 4.45, up to 36 storeys. The fifth attempt at the same $580 million closed without award on 9 June 2026 and the committee had until 17 August for private treaty. They relaunched in early September at the identical number. Why hold? Because in December 2021 a developer actually bid $556.7 million for this site, about $2,626 psf ppr, and then walked away and forfeited a $1 million deposit after that month’s cooling measures. The owners have a bid in living memory within 4% of their current ask. That is a psychological anchor that no marketing agent can talk them off.

Trendale Tower, fourth attempt, $168m. Eighteen units on 21,709 sq ft in Cairnhill, completed 1982, five minutes from Newton. It asked $163.5m in 2018, $178m in August 2022, then $168m from December 2022 into a January 2023 close. It is back at $168m. What changed is not the price but the permitted use: URA has indicated support for Serviced Apartment 2 use under outline planning permission, so the site can now be pitched as a long-stay serviced apartment project rather than only as boutique residential. A new buyer pool at the same price is a better position than the same buyer pool at a lower price. And the signature maths is unusually kind here — on a straight unit count, 18 units at 80% needs 15 owners and at 70% needs 13. The new law is worth roughly two signatures to this block.

City Plaza, third attempt, $970m. This one is the most interesting hold, because its reason has nothing to do with consent. It asked $1.05 billion in 2018 and got only 53% consent. It asked $970 million in 2021 and got 79.3% — a near-miss that must still sting. It is asking $970 million again, and this time it has cleared the 80% threshold for the first time. Why would you cut the price on the attempt where you finally have the mandate you have failed twice to get? You would not. Add the structural advantage: the site is currently zoned commercial, so no additional buyer’s stamp duty is payable by the developer on acquisition. On a $970 million land bill, dodging developer ABSD is worth far more than any plausible price concession. That is a real edge and the owners know it.

Lakeside Towers, third attempt, $350m. A 45-year-old pair of 20-storey blocks on Yuan Ching Road in the Jurong West planning area, 144 units, 99 years from 1975 with under 48 left. It asked $305 million in 2018 and $350 million in both the February and June 2023 attempts, both of which failed. It is asking $350 million again. The owners here are mostly older, the lease is decaying visibly, and the honest read is that this cluster cannot cut much: with under 48 years left, the lease top-up premium is enormous and the reserve is close to the floor at which the exercise stops being worth the owners’ trouble. It already has over 80% consent, so the new law does nothing for it either. Its problem is the buyer, not the ballot.

Gilstead Court, fourth attempt, $198m. This one deserves its own section.

Gilstead Court asked 29% more. The answer is a tax.

Gilstead Court is 48 three-bedroom apartments on 75,479 sq ft of freehold land in District 11, completed 1978, marketed by JLL, tender closing 13 October 2026 at $198 million.

Its history is a museum of everything that can go wrong with a collective sale. A 2008 attempt failed on consent. A 2013 deal at $150.2 million found a buyer and then collapsed. It launched in 2018 at $168 million, cut to $153 million in January 2019, and the tender closed with no bids at all. Not a low bid. No bids.

And now it is back at $198 million, which is 29.4% above the 2019 ask and 17.9% above the 2018 ask. On a straight per-unit split that is about $4.13 million an owner against roughly $3.19 million at the 2019 number — a swing of about $937,500 each, my arithmetic.

How do you justify that after failing four times? Two reasons, and the second is the one most people miss.

First, the land rate is not actually heroic. $198 million over 75,479 sq ft at a plot ratio of 1.4 works out to $1,874 psf ppr, falling to $1,751 psf ppr once you account for the 7% bonus gross floor area for balconies. I checked that arithmetic and it reconciles, which tells me the site is genuinely at plot ratio 1.4 — the same low ratio as Serenity Park, but in District 11 instead of District 28.

Second, and this is the real answer: Gilstead Court reportedly pays no land betterment charge, even on the additional bonus balcony gross floor area. The land betterment charge is the levy payable when redevelopment enhances the value of a site beyond its existing development baseline. It is the tax that quietly kills more collective sales than consent ever does, and I have a full explainer on how the land betterment charge decides en blocs.

If a site’s existing development baseline already covers what you intend to build, the charge is nil, and every dollar the developer does not pay the state is a dollar available to pay the owners. That is worth real money on a $198 million site — potentially eight figures. It is the single cleanest explanation for how a block that could not attract one bid at $153 million in 2019 can credibly ask $198 million in 2026.

I have flagged the nil-charge claim for verification before Gary’s team repeats it in a listing pitch. It comes from launch reporting rather than from an SLA assessment I have seen.

People’s Park Centre: three days of silence on $1.48 billion

The biggest number in the market closed its tender at 3pm on 16 September 2026 and, as at today, no result has been published anywhere. Three days of silence on $1.48 billion.

First, kill the lazy framing. People’s Park Centre is widely described as having “cut from $1.8 billion”, which is half the story. The 2019 attempt asked $1.35 billion and failed on consent. The 2022 attempt asked $1.8 billion and closed with no bids. So $1.48 billion is 17.8% below the 2022 ask and 9.6% above the 2019 ask. Whether it is a cut depends entirely on which failure you pick as your baseline, which is a good reminder that “reserves are falling” is usually an artefact of comparator selection.

The site is 95,467 sq ft at plot ratio 8.6, height-capped at 25 storeys, with 324 retail units, 256 office units, 120 residential units and a multi-storey carpark. Consent was 545 of 701 units, 80.54% by share value and 84.56% by strata area — 764 subsidiary proprietors signed. It is marketed by ERA Realty Network, which is my own brokerage, and I have no involvement in the exercise and no inside information on it.

What does three days of silence mean? In my experience, when a tender closes with a strong bid at or above reserve, somebody tells the press within twenty-four to forty-eight hours, because a good number is free marketing for the agent and reassurance for the owners. Silence past the second day usually means one of three things: no bids at all, bids below reserve that the committee is now weighing, or a bid that is being negotiated privately during the post-tender window.

What silence does not prove is failure. Committees on a site this complex — 701 strata units, three use classes, a lease top-up premium estimated at $535.4 million — often take a week simply to convene and take advice. I am recording the null here because it is information, and because if I am still writing “no result published” in October that itself becomes the story.

What an en bloc reserve price actually has to cover

If you own in a cluster and you are being asked to sign, this is the arithmetic that should be in front of you, because the reserve price is not the number you receive.

That last line is where most en bloc regret comes from. I have watched owners take a headline windfall and discover that the replacement unit in the same district costs more than the cheque. Before you vote, price your replacement. Run the stamp duty, the loan-to-value and the cash outlay properly — you can run your own numbers with the free calculators at listings.sg/tools rather than relying on the round numbers in a committee circular.

Should you hold your en bloc reserve price or cut it?

This is the question I get asked in living rooms, and the answer is genuinely different depending on the block.

The October calendar and what each close would prove

Five closes in three and a half weeks. This is the densest stretch of collective sale deadlines Singapore has had in years, and by the end of October we will know far more than any amount of commentary can tell us now.

My falsifiable predictions

I write these down in advance so you can hold me to them. If I am wrong, the wrongness stays on this page.

Note what prediction 6 does. It is the one that would kill my own thesis. If reserves start falling broadly in Q4, I said here that they would not, and I will write that up.

What would change my mind

I am arguing that the en bloc reserve price is not falling and that the holding pattern is rational. Here is what would make me withdraw that.

• Three or more clusters relaunch below their previous ask inside one quarter. Two is anecdote. Three with a shared direction is a trend and I would say so.
• A 40-plus-year-old cluster in the new 70% tier cuts deeply anyway. That would break the age correlation in my table and I would have to find a better explanation.
• Commencement is gazetted and reserves fall rather than hold. My whole argument is that owners are waiting for the clock. If the clock starts and they cut regardless, I was reading the wrong motive.
• Two or more of the October tenders clear at their unchanged asks. That would mean the flat reserves were correctly priced all along and the “no transactions” evidence I am leaning on was just a timing artefact.
• Developer land bids in the Government Land Sales programme fall materially in Q4. If the state’s own land market softens, private collective sale reserves will have to follow, and the holdouts will be wrong rather than patient.

One more thing I could not stand up and so am not claiming. I noticed that both clusters that cut — Serenity Park and Pek Chuan Building — are marketed by the same agency. That may be a house view on pricing, or it may be coincidence across two data points. Two is not a sample. I mention it only so you know I looked.

I also looked for a link between reserve price behaviour and the wider residential market, because the HDB and private price gap has been doing its own thing this year. I could not establish one from seven clusters in six weeks, so I am not going to assert it. If you want that side of the picture, my piece on the HDB versus condo price gap in 2026 covers it properly with real volume data.

FAQ