–
median land $psf (3y)
Landed property, by the numbers
Terraces, semi-Ds, detached houses and Good Class Bungalows are Singapore’s thinnest, most opaque market. Here is what actually transacts — prices, returns, liquidity and where the money moves — from 79,000 caveats back to 1995.
79k
landed caveats since 1995
+4.9%
median annualised return*
6.8 yrs
median holding period
62%
freehold
*From 29,786 matched buy–sell pairs at the same address (repeat sales). 82% of pairs sold at a gain.
Price trend — median land $psf by type
Landed is priced on land area, not built-up space. Median land $psf since 2005, land-titled sales only so the series is comparable:
Median land $psf by landed type
Land-titled caveats; years with <5 sales of a type are omitted.
TerraceSemi-DDetachedGCB
Where prices sit now: latest median land $psf runs about $2,289 for terraces, $1,846 for semi-Ds, $2,016 for detached and $2,045 for GCBs. The gap between types is land scarcity: fewer, larger plots command more per square foot.
Liquidity — how thin is landed?
Landed turns over far less than condos. Annual transaction volume, all landed types:
Landed transactions per year
All landed sale caveats (new sale, resale, sub-sale).
Thin and cyclical: volume peaked near 5,522 sales in 2007 and sits around 1,126 in 2026. With few sales per estate per year, a single deal moves the median — treat any one $psf as a range, not a price.
Where landed is dearest — district league
Median land $psf by postal district over the last three years (land-titled, districts with ≥8 sales):
| District | Median land $psf | Median price | Sales (3y) |
|---|---|---|---|
| D9 › | $3,660 | $6.80M | 21 |
| D2 | $3,546 | $5.55M | 8 |
| D11 › | $2,588 | $10.59M | 188 |
| D8 › | $2,456 | $4.64M | 12 |
| D10 › | $2,452 | $9.20M | 329 |
| D15 › | $2,288 | $5.60M | 602 |
| D12 › | $2,219 | $4.09M | 17 |
| D13 › | $2,114 | $4.53M | 261 |
| D5 › | $2,089 | $4.60M | 109 |
| D20 › | $2,084 | $4.88M | 336 |
| D28 › | $2,040 | $4.23M | 441 |
| D21 › | $1,998 | $6.08M | 171 |
Returns — what landed actually earned
The strongest signal in this dataset is the repeat sale: the same address bought then sold later. That controls for the property itself, so the return is real, not a mix-shift. Median annualised return and holding period by type:
| Type | Median annualised | Median hold | % sold at gain | Pairs |
|---|---|---|---|---|
| Terrace | +4.7% | 7.1 yrs | 83% | 18,263 |
| Semi-D | +4.9% | 6.7 yrs | 81% | 8,011 |
| Detached | +5.2% | 5.8 yrs | 78% | 2,947 |
| GCB | +7.3% | 5.8 yrs | 85% | 565 |
Read it honestly: annualised return is price appreciation only — it excludes buying/selling costs (BSD, ABSD, agent fees), holding costs and any rebuild spend, and it ignores rental income. It is a like-for-like appreciation benchmark, not a total return.
Investigate a district & type
Pick a district and landed type to see recent pricing, activity and the repeat-sales return for that pocket:
–
median price (3y)
–
median annualised return
–
caveats since 1995
Where to buy landed — the estates that grew, trade & rebuild, ranked by the numbers ›
Estimate a specific landed home’s fair value — district, type, land size & tenure ›
Rebuild economics — does teardown-and-rebuild pay in your district? ›
Latest landed market report — prices, volumes & top estates ›
Latest landed transactions — the most recent deals on record ›
Landed rental yields — the thin income side, by type & district ›
Landed value screener — recent resales below the comp model ›
Investor verdict
Landed is an appreciation and scarcity play, not a yield play — rental yields are thin and it trades infrequently, so entry price and holding power matter more than in the condo market. The repeat-sales record shows steady long-run appreciation, but with wide dispersion: the estate and the entry price decide the outcome, not the type alone.
Favour freehold in an established landed enclave with real transaction depth (so you can exit), buy on land $psf versus recent same-type comps in the same district, and underwrite conservatively — a single outlier sale can make a street look cheaper or dearer than it is.
How to use this before you buy
Compare on land $psf, same type, same district
A terrace and a detached house don’t share a $psf. Benchmark like-for-like against the district league above, then the specific estate.
Check the exit, not just the entry
Thin volume means slow exits. Prefer estates with steady yearly sales over a “rare” pocket that almost never trades.
Separate the land from the house
Old houses on good land can be a rebuild play; new builds price the construction in. Judge the land value first, the building second.
Cost the full round-trip
The returns here are appreciation only. Add stamp duties, agent fees and holding costs — and any rebuild — before you call a deal a win. See the stamp-duty & cost rules.
GCB corner
Good Class Bungalows — the top of the landed market, on plots of at least 1,400 sqm in gazetted GCB Areas. The thinnest, least-transparent segment of all: 1,845 caveats since 1995, but many change hands quietly and never appear.
1,845
GCB caveats since 1995
$26.00M
median price (3y)
$2,033
median land $psf (3y)
D10, D11, D23, D21
most active districts
GCB caveat caution: because GCB deals are frequently done privately and caveats are optional, quoted GCB volumes and medians are a floor, not a complete picture — use them directionally.
How is this worked out? — data, index, returns, sources & limits
Dataset
79,350 landed sale caveats, 1995–2026, compiled from URA caveat data (Terrace, Semi-Detached, Detached and GCB). $psf outliers (outside $50–$10,000) are dropped.
Prices
Median land $psf by year, land-titled sales only so the series is comparable; strata-titled landed is excluded from the price trend. Medians (not means) throughout — landed is skew-prone.
Returns
Repeat-sales: the same address bought then sold later forms a pair; annualised return = (sell/buy)^(1/years)−1. Holds under 9 months or price ratios outside 0.3–8× are excluded as flips or errors. Appreciation only — excludes stamp duties, fees, holding and rebuild costs, and rental.
Explorer
District × type cells show the 3-year median land $psf and price, all-time caveat count, and the repeat-sales annualised return where at least 8 pairs exist.
Limits: caveats are optional, so quiet deals (especially GCBs) are missed; some addresses are masked or unmatched; a thin cell can swing on a single sale. Directional, not a valuation.
For educational purposes only — not financial advice.