How we price land: from a GLS bid to a launch PSF

Everyone quotes the land price. Almost no one tells you what it forces the launch price to become. That gap is the whole reason this page exists.

When a Government Land Sales site is awarded, the headline is always the same shape: a developer paid $X psf per plot ratio for a plot in some district, and analysts think it will sell at $Y. The land number gets treated like the answer. It is not the answer. It is the first line of a cost stack, and the launch price is what falls out of the bottom of that stack once you add everything a developer actually has to spend and the margin they need to justify the risk. This is how Roncasa derives the pricing suggestions you see on the New Launch hub, and I would rather show you the working than ask you to trust a number.

Why the land price alone is not the answer

A land bid tells you what a developer was willing to sink into the ground before a single wall goes up. It tells you nothing, on its own, about what they must charge to get out whole. Two sites can be bid at almost the same psf per plot ratio and still need very different launch prices, because construction cost, the build period, financing, and the margin a developer will accept all move independently of the land.

So the honest read is this: the land price is an input, not a verdict. What buyers actually want to know is the number at the other end of the pipe. What will this thing launch at, and is that fair given what is already trading nearby. To get there you have to build the cost up, line by line, from sourced numbers. That is the part almost nobody publishes, and it is the part we do.

The method, line by line

We work the same chain every time, and every line carries its source. Where a figure is a stated assumption rather than a published number, we say so. Transparency here is not a weakness. It is the credibility mechanism, because anyone with the same inputs can check our arithmetic.

1. Land cost

Start with the awarded price per square foot per plot ratio (psf ppr) from the URA tender result. This is the one figure that is already public and fixed the moment the tender closes. It is the floor of the whole stack.

2. Construction, from a real benchmark

Construction is the biggest non-land line, so we do not guess it. We anchor it to the Rider Levett Bucknall (RLB) construction cost benchmark, the quarterly regional cost data the whole industry references. For a residential mid-rise condo (the "over 10 and up to 20 storey" band), RLB's 2026-Q3 Singapore figure sits at roughly $297 to $376 psf of construction floor area for the structure itself.

That structure figure excludes the things that still cost real money: foundation and substructure, the car park, external works, and preliminaries. To get to an all-in construction number we apply a +12% uplift as our default (a stated assumption calibrated from our own GLS database, and slightly higher for genuinely boutique projects under about 200 units, where fixed costs are spread over fewer units). That takes the benchmark to roughly $333 to $421 psf all-in.

3. Professional fees, finance, and marketing

Three soft-cost lines sit on top of construction, and each is a clearly stated assumption:

• Professional fees: architects, engineers, QS, project management. We run this at 3 to 4 percent of construction.
• Finance: the cost of carrying the land and the build. We assume roughly 3.5% per annum over a three-year build period on the average drawn balance. That is a calculated figure, not a published one, and it moves directly with interest rates.
• Marketing: showflat, agents, media. Roughly 1.5% of gross development value, a standard industry assumption.

Add construction plus these three lines and you have the non-land total. On our worked example below that comes to roughly $478 to $577 psf.

4. Breakeven PSF

Breakeven is simply land plus non-land. This is the price at which the developer has recovered every dollar and made nothing. It is the honest floor under any launch price, and it is the single most useful number a buyer can hold, because a launch priced only slightly above breakeven is a developer with very little room to discount later.

5. Indicative launch PSF, with margin

No one develops for zero. We add a developer margin on top of breakeven, run as a range: 15% for a competitive, confident developer and 18% for a more cautious one. That produces an indicative launch PSF band, not a single point, which is the honest way to express it. Nobody knows the exact launch price this far out, including the developer.

Written as one formula, the residual chain is:

Breakeven PSF = Land psf ppr + Non-land total
Indicative launch PSF = Breakeven × (1 + developer margin, 15% to 18%)

And if you want to run it the other way, to ask what a developer could afford to bid given a launch price the district can support, it inverts cleanly. This is the residual land value that developers themselves use:

Affordable land psf ppr = (Expected launch PSF ÷ (1 + margin)) − Non-land total

One site, worked all the way through

Numbers on their own persuade no one. So here is the full chain on a real, recent site. This is the part no one else publishes, and every input below is sourced.

Holland Plain (District 10, Bukit Timah). A 99-year leasehold site awarded to Sim Lian in May 2026, gross plot ratio 1.8, about 280 units, roughly 304,500 sqft of gross floor area. The awarded land price was $1,491 psf ppr (URA tender result).

Cost line	psf	Source
Land	$1,491	URA tender result, May 2026
Construction, all-in	$333 to $421	RLB 2026-Q3 mid-rise benchmark, +12% all-in uplift
Professional fees	$10 to $17	3 to 4% of construction (stated assumption)
Finance	$96 to $100	3.5% p.a. over 3-year build (calculated)
Marketing	~$39	1.5% of GDV (stated assumption)
Non-land total	$478 to $577	sum of the above
Breakeven (land + non-land)	~$1,970 to $2,070	derived
Indicative launch (breakeven × 1.15 to 1.18)	~$2,260 to $2,440	derived

So from a $1,491 land bid, the cost side alone tells us Sim Lian needs to launch somewhere around $2,260 to $2,440 psf to hit a normal developer margin. Not because an analyst said so. Because the arithmetic leaves no hidden variable once you fix the inputs.

The cross-check, and why our number is honestly lower

Now the validation step, which always comes after our own derivation, never before it. Third-party breakeven estimates exist for the same site. EdgeProp's figure for Holland Plain lands nearer $2,667 psf breakeven. (A quick note on our own data: the column some of our tables label "PropNex" is actually sourced from EdgeProp. We flag that rather than pretend it is a separate opinion.)

Our cost-up sits below that, and I am not going to smooth over the gap. It is the interesting part. A higher third-party breakeven usually implies one of three things: a heavier finance assumption, a larger site-preparation or abnormals allowance, or a more generous construction number than the RLB mid-rise band. Any of those is defensible. The point is that we can see exactly which lever moves our number toward theirs, because our number is built from lines you can inspect. Theirs arrives as a single figure. When two independent methods land in the same neighbourhood, that validates both. When they diverge, the divergence itself is information, and we would rather show you the seam than paper over it.

There is a fourth reason that matters in exactly this kind of address. Holland is a status postcode, and prime name-driven pockets like Queenstown, Clementi, Bishan, Ang Mo Kio, Toa Payoh and Bukit Timah can sustain a launch premium that a pure cost-up will never capture, because part of what a buyer pays for there is the name. So when our cost-up sits below a third-party number in a prestige district, that gap is often the market status premium rather than an error in either method. Read our cost-up as the floor the arithmetic sets, and treat the premium on top as the price of the postcode.

The number is alive. That is the feature.

Here is the thing that separates a Roncasa pricing suggestion from a static breakeven you read once in an article. Ours is recomputed. Land prices move as new tenders close. The RLB construction index reprints every quarter. Interest rates shift the finance line. Every one of those flows straight back through the same stack and out the other end as an updated launch band. A third-party figure printed in March is frozen in March. Ours is not a headline, it is a calculation, and a calculation updates when its inputs do.

One honest caveat, because this is the whole point of the page. As of today, the per-site breakeven value stored in our database against each GLS row is a simple 1.85x placeholder (land psf multiplied by 1.85), not the full cost-up. On Holland Plain that placeholder reads about $2,758, which is why you should not treat that stored figure as the live Roncasa number. The real number is the method shown on this page, worked line by line. We are describing how we derive it, and showing it computed live in the example above, rather than pointing you at a column that has not yet been calibrated. When the stored value is recalibrated to this cost-up, this page is where you will see the method it was built from.

What this means if you are tracking a site

Picture the specific person this is for: the OCR upgrader who has been watching a GLS pocket for a year, waiting to decide whether the eventual launch is a fair entry or a stretch. This method hands you three things that change your decision.

• A fair-value reference before the showflat opens. If our derived launch band is $2,260 to $2,440 and the developer eventually prices at $2,300, you are buying close to a defensible cost-up. If they open at $2,650, you now know the premium sits above the cost stack, and you can ask what is justifying it.
• A read on discount room. A launch priced just above breakeven is a developer with almost no cushion. That tells you how likely a later-phase discount really is. A launch priced well above breakeven has room to move if absorption slows.
• A moving target you can trust. Because the number recomputes, tracking a site through to launch actually means something. If rates fall or the land next door sells cheaper, the fair band shifts, and you will see it shift.

What I would watch, in plain If-then terms. If a comparable GLS site nearby awards below this one, then the fair launch band for this pocket eases and patient buyers gain leverage. If the RLB index climbs next quarter, then the breakeven floor rises under every site in the pipeline, and today's launches start to look better value in hindsight. If MAS holds rates, then the finance line stays put and the upgrader window stays open a while longer.

The bottom line

A land price is a beginning, not an answer. The launch price is what the cost stack forces it to become, and that stack is knowable if you are willing to source every line and show the arithmetic. That is what we do here, and it is why a Roncasa pricing suggestion is a calculation you can audit rather than an opinion you have to take on faith. The number is honest, it is sourced, and it moves when the world moves. If you are tracking a specific GLS site, as a potential buyer or as an owner in the catchment, I can run this exact stack on it and show you where a fair launch lands before the showflat ever opens.

Sources: RLB Singapore construction cost benchmark, 2026-Q3 (residential mid-rise band); URA Government Land Sales tender results (Holland Plain, awarded May 2026); Roncasa GLS land-sales database. Third-party breakeven cross-check from EdgeProp. Method and worked example current as of 16 September 2026; figures are recomputed as inputs change.