How property is measured in Singapore

Ask how much Singapore property prices have grown and you will get a number. Ask two people and you may get two numbers, both computed correctly from the same transactions.

Here are four, from our own record of private condominium and apartment sales between 2017 and 2025.

Method	2017 to 2025
Median price per square foot, resales	+46%
The same, holding the 2017 mix of homes constant	+47%
The same, holding the 2025 mix of homes constant	+51%
Repeat sales, the same homes sold twice	+33%

The highest and the lowest are 18 points apart, on one market over the same years. Each method answers a slightly different question, and the answers separate because the questions do.

What each method asks

A median price asks what the middle transaction cost. It tells a buyer what to bring to the market, and it is the figure most often quoted as though it measured price growth. It cannot separate a change in prices from a change in what happened to sell.

A median price per square foot asks the same question after allowing for size. That removes one difference between the homes that sold in each year and leaves the others.

A repeat-sales index asks what happened to the same home, by matching each sale with the same unit's earlier sale. A change in the mix of what sold cannot move it, because both sales are the same property.

A hedonic index, which is what URA and HDB publish, asks what a home with fixed characteristics would have cost in each period, estimating the effect of age, size, floor and location and holding them constant. It controls only for the characteristics in its model.

Why they disagree

Start with the mix of what sold. Hold it fixed, by region, tenure, bedroom count and building age, and median price per square foot growth moves from 46% to 47%, a difference of 1 point. The mix did change, towards the outside central region, towards leasehold and towards older buildings, all cheaper per square foot. But holding it fixed moves the median further from repeat sales, not closer.

Holding the mix fixed is also two answers, not one. Fix the 2017 mix and growth is 47%; fix the 2025 mix and it is 51%, 4 points higher. Both are standard, and they differ because the homes that sold at the start of a period and the homes that sold at the end are not the same basket.

About half of the gap comes from which sales a repeat-sales index starts from. It pairs each sale with the same unit's earlier sale, and in most pairs the earlier purchase was made when the home was new, from the developer or as a sub-sale. Split the pairs by that earlier sale and the index gives two different answers.

Earlier sale	Pairs	Growth
Resale	40,570	+40%
New home	60,350	+29%
All pairs	100,920	+33%

Pairs that started with a new home grew 11 points less than pairs that started with a resale. In those new-first pairs, the earlier price was paid to the developer or, in a sub-sale, to an earlier buyer. A new home's first price is usually set at launch, before the building is finished; its next sale is the resale of a completed, older home. Count only resale-to-resale pairs, the same kind of sale the median counts, and the gap between median price per square foot and repeat sales narrows from 13 points (46% against 33%) to 6 points (46% against 40%), about half.

The 6 points that remain sit inside the method itself. A repeat-sales index carries the ageing of every home between its two sales, cannot see a renovation done in between, and only sees homes that sold twice. We cannot yet say how much each contributes.

The series

Each note goes up as it is written, and this page links them as they land.

• How growth is usually measured, and what more data adds. Price per square foot from one date to another, what it can depend on, including which homes happened to sell, and what a full year of sales and repeat sales add.
• Price indices, Singapore's and Realila's. How URA and HDB build theirs, in Property Price Indices, Explained. What a same home price index is and when to use one, in What is a same home price index?. The sales and settings behind Realila's private index, in How we measure prices and sales.
• Assessing a project. Not yet. We do not have a written method for assessing a development, and this note will not go up before we do.
• The estimated value of a unit. Not yet. We are first rerunning the test of our estimates against later sales, to report how many land within 5% and within 10% of the price achieved.
• Other measures. Bank valuations, annual value, asking prices, rental yield and affordability.

On two well-traded developments, the choice of start and end quarters alone moved the measured growth by 11 and 12 points; at one where the mix of homes that sold changed from quarter to quarter, it ranged from 38% to 125%.

Realila's own indices

Realila computes two price indices. For HDB resale flats it is a hedonic index, the same family of method HDB uses. For private condominiums and apartments it is a repeat-sales index, because a hedonic index built the same way overstated private price growth. Each is compared, quarter by quarter, with the official index for its market: HDB's resale price index and URA's private residential price index. The private index follows URA's across the market as a whole. In the core central region the two usually agree on whether prices rose or fell over a year, but not on by how much, and we are still working on why. Landed homes have neither.

The private index is public, in the Market Snapshot. The HDB index will also be public when it is ready.

If you measure this for a living, or are learning to

This series is written by a practitioner working from transaction records, not by a research department. The clearest open question in it is the 6 points above: we can name ageing, renovation and which homes sell twice as the causes, and cannot yet say how much each contributes. If you work on housing price measurement and have seen this separated well, or you are learning and a question here needs explaining better, write to hello@realila.com. We will publish what we learn and say where it came from.

What this series is not

It does not argue that one method is correct. Each is the right instrument for a question and the wrong one for others, and the failure we see most often is a number computed by one method being used to answer another method's question.

Method and data notes

Private condominiums and apartments, from URA caveat data; executive condominiums are excluded throughout, since their subsidised first sale is not comparable to an open-market one. Median price per square foot is the median of each resale's own price per square foot in 2017 and in 2025. The two mix-adjusted figures hold region, tenure, bedroom count and building age fixed, one at the 2017 mix and one at the 2025 mix, on the 95 combinations present in both years: 10,938 resales in 2017 and 12,059 in 2025.

The repeat-sales figures come from a repeat-sales index (the Bailey, Muth and Nourse method) fitted by year over 100,920 pairs of sales of the same unit, with 2009 = 100. The pairs are as at 28 September 2026. They are rebuilt every week as new sales are recorded, so these figures can move by a point as the pair set grows. The split uses the type of the earlier sale in each pair.

Growth figures are rounded to whole percentages. Every gap between two figures quoted in the text is computed from the figures as printed.

30 October 2026. This note placed homes in market segments by postal district. URA places them by planning area, which moves about one in seven private sales between regions. The figures by region here use the earlier segments and differ from what the same records give now. How regions are set: How we measure prices and sales.