Where prices grew fastest

Between 2017 and 2025 the median HDB resale flat rose 52 per cent a square foot. The rise was not evenly spread: the districts that were cheapest in 2017 grew fastest, and the gap between the dearest and the cheapest narrowed.

By district

D25 started the period at $330 psf, among the cheapest in Singapore, and grew 56 per cent. D15 started at $525 psf, among the dearest, and grew 20 per cent. That pattern holds broadly across the list: the further down the price ranking a district began, the more it rose.

Those two figures hold the flat-type mix fixed at 2017, which is what makes them a statement about price rather than about what each district happened to sell. Pooling the mix instead moves some districts by five or six points and leaves the order at the top unchanged, though it moves D25’s own figure from 56.6 per cent to 56.1 per cent. The ranked chart below is on pooled medians, for comparability with the matrix.

Ranked above are the 18 districts where at least two flat types have enough trade at both ends for a mix adjustment to mean anything. D09 would read higher — 56.9 per cent against 56.1 per cent — but its figure rests on a single flat type, so it is not the same quantity as a blend of four or five and is reported here rather than ranked.

By flat type

Growth was broadly similar across sizes, within about ten percentage points. The differences that exist are worth noting all the same: a buyer choosing between a three-room and a four-room was choosing between quite different rates of appreciation.

District and flat type together

The two crossed. Each cell is the change in median price per sq ft for that flat type in that district, districts ordered by overall growth. A dash means the combination traded too little at one end of the period to report — fewer than 20 sales in either 2017 or 2025 — which is itself worth knowing, because it says that pairing barely exists. 44 of 112 combinations fall into that category.

Against private property

Private housing has risen faster. From 2017-Q1 to 2026-Q3, the official HDB resale index rose 51.2 per cent, private non-landed — condominiums and apartments — 59.0 per cent, and landed property 76.3 per cent. The gap between public and landed housing has widened by roughly 25 points in eight years.

It is worth noting what that first figure confirms. The HDB index rose 51.2 per cent over a period in which this site’s own calculation, built from the individual transactions rather than from anyone’s index, found 52 per cent. Two independent routes to the same number is the best check available that the rest of this page is sound.

What this comparison can and cannot do

The comparison above is between indices, which is the only way to put HDB and private growth on one axis back to 2017. The site now also holds individual private caveats — the URA account this page once said was missing exists, and there are per-project medians and a page for every development that has traded.

What the caveat feed cannot do is reach back. It serves a rolling five-year window, so the private history before that cannot be obtained retrospectively and is being collected from the day the archive started. That is why growth over the full period is still read off the indices here rather than recomputed from transactions, and why the private half of the housing ladder is measured over twelve months rather than nine years.

How these figures are measured

Growth is measured between 2017 and 2025, the first and last complete calendar years in the data. The current year is excluded: it covers only the months published so far, and setting a part-year against a whole one would manufacture a fall every January.

Each figure is a median price per sq ft, so a change in the mix of flat sizes sold does not masquerade as a change in price. A cell in the matrix needs at least 20 sales in both the first and last year to appear at all. See the method for the sources and their limits.

Questions people ask

How much have HDB resale prices risen in Singapore?

The median HDB resale flat rose 52 per cent per square foot between 2017 and 2025, the first and last complete calendar years in the data. The current year is excluded, because it covers only the months published so far and setting a part-year against a whole one would manufacture a fall every January. The rise was not evenly spread: by district it ran from 20 per cent in D15 to 56 per cent in D25.

Which district had the fastest HDB price growth?

D25, covering Kranji and Woodgrove, at 56 per cent — from $330 psf to $517 psf. That figure holds the flat-type mix fixed at 2017, which is the figure to quote: a district’s pooled median can move because its flats got dearer or because it sold a different mix of them, and only the first is a price change. Pooling the mix leaves the top of the ranking unchanged.

Did cheap or expensive HDB districts grow faster?

Cheap ones, consistently. The further down the price ranking a district began in 2017, the more it rose, so the spread between the dearest and cheapest districts narrowed over the period. D25 started at $330 psf, among the cheapest, and grew 56 per cent; D15 started at $525 psf, among the dearest, and grew 20 per cent. This is measured on mix-adjusted growth, so it is not an artefact of cheaper districts selling larger flats later in the period.

Have HDB flats or private property grown faster in Singapore?

The chart on this page sets the HDB resale price index against URA’s non-landed and landed private indices, all rebased to 2017. Both official indices are published at 2009-Q1 = 100, which is what makes them directly comparable without any rebasing of our own. The comparison is between indices rather than between transactions, because the private caveat feed serves only a rolling five-year window and cannot reach back to 2017.

How is this growth measured?

Between 2017 and 2025, on medians of price per square foot rather than of price, so a change in the mix of flat sizes sold does not masquerade as a change in price. District figures additionally hold the flat-type mix fixed at 2017. A cell in the district-by-flat-type matrix needs at least 20 sales in both the first and the last year to appear at all; 44 of 112 cells fall below that and are left blank rather than shown as noise.