MARKET & POLICIES · INSIGHT 011

Is freehold worth it in Singapore? We checked 224,023 exits. It did not make more money.

Freehold is the one feature Singapore buyers pay for on faith. We tested the faith against the record: every condo unit bought and later sold again since 1995, both ends of the same trade. Over a normal holding period, tenure barely registered.

Farhan Adenan · 29 Sep 2026 · 9 min read

POV matched-pair model: matched pair = same unit bought then sold, held at least one year. Huttons export of URA REALIS caveats, 1995 to 1 Sep 2026. Tenure figures exclude Executive Condominiums. Gross returns, before stamp duty, interest and fees.

Is freehold worth it in Singapore? On the record of what owners actually got back, not as a return. Across 90,267 freehold and 113,923 leasehold condo pairs, freehold returned a median 3.26% a year and 99-year leasehold 3.21%. Both groups held for almost exactly the same time — a median 6.59 years against 6.51 — so this is not a longer hold flattering one side.

The share of owners who sold above what they paid runs the other way: 84.0% of freehold sellers exited ahead, against 86.8% of leasehold sellers. If freehold were the safer asset, this is the number where it would show. It does not.

The one column freehold wins is the one people quote: the dollar gain. The median freehold seller made $252,000; the median leasehold seller made $212,000. That gap is real, and it is not evidence of better growth. Freehold units cost more at entry, so a similar percentage return produces a larger cheque. A bigger gain bought with a bigger cheque is not outperformance. It is scale.

A raw comparison can hide a trick. If freehold stock sits in different districts, or was bought in different years, the averages are comparing different markets rather than different tenures. So we built the fair test. Every pair was placed in a cell of same district, same purchase year, same holding band, and we kept only cells with at least five pairs on each side — 1,455 cells covering 165,408 pairs.

Inside those cells, freehold trailed leasehold by 0.47 percentage points a year, weighted by pairs. The median cell had freehold behind by 0.08 points. And freehold came out ahead in only 46.9% of cells. That is a coin flip that leans leasehold — the opposite of what a freehold premium assumes you are buying.

The two measures differ in size, not in direction. The pair-weighted gap is larger than the median because the cells that favoured leasehold did so by wider margins, or in busier cells, than the ones that favoured freehold. Neither version gives freehold an edge.

Freehold won clearly in D04 Harbourfront / Sentosa (+2.86 points a year), then D01, D23, D28 and D03. It lost hardest in D08 Little India / Farrer Park (−1.82), then D18, D09, D02 and D15. D16 Bedok was flat at −0.01.

Read the names, not just the signs. D09 Orchard and D02 Tanjong Pagar — the addresses where freehold is sold hardest as a mark of quality — sit on the leasehold side. D23 Bukit Batok / Hillview and D28 Seletar, suburban districts nobody calls prime, sit on the freehold side. There is no “prime freehold always wins” rule in this record. There is a district-by-district result, and it has to be read for the district you are actually buying in.

Highlighted rows are the two holding bands where freehold’s median return came out ahead.

Freehold edged ahead in only two of five holding bands: 3–5 years (4.86% against 4.57%) and 15 years or more (3.08% against 3.03%). In the 5–10 year band, the largest on both sides and where the most owners sell, leasehold led 2.45% to 2.16% and sold at a profit more often, 84.0% to 78.3%. The longest band is the closest thing the data has to a long-horizon verdict, and it reads as a draw.

Entry year moves the result far more than tenure does. Buyers in 2005–09 made 5.69% a year on leasehold and 5.10% on freehold. Buyers in 2010–14 made 1.81% and 1.25%. The gap between those two cohorts is close to four points a year; the gap between tenures inside either one is under six-tenths of a point.

Freehold led only for buyers before 2000 (0.99% against 0.76%) and in 2000–04 (5.99% against 5.23%). The 2015–19 cohort was level, 3.855% against 3.86%. Buyers since 2020 favoured leasehold, 4.99% to 4.31%.

This is not an argument for timing. Nobody picks their cohort with hindsight. It is an argument that the part you control — buying a sound unit at a sound price and being able to hold it through a weak stretch — outweighs the tenure label by a wide margin.

The last part of the belief is that freehold costs more, so it must be worth more. We checked the current price. Comparing 2024–26 resale psf for freehold and leasehold units in the same district with the same bedroom count — 72 cells, each with at least 15 sales on both sides — the pair-weighted freehold premium is 0.3%. The median cell has freehold 1.7% cheaper.

That does not mean tenure is worth nothing. It means the comparison is dominated by everything else that differs. The likely reason is mix: freehold resale stock tends to be older, from smaller developers, with fewer facilities, while the leasehold units beside it are often a decade or two newer. Inside the same postcode you are frequently comparing a 1990s freehold with a 2010s leasehold. The price gap you see is mix, not tenure — and the same mix helps explain why freehold’s growth record is ordinary.

It also squares the dollar result in section 01. Freehold owners most likely paid more on average because of what and where they bought, not because the tenure itself commanded a price. The larger cheque on exit simply returned a larger cheque on entry.

There is a real case for freehold. It sits outside the evidence above, and it is worth stating plainly.

Horizons beyond 15 years. Our pairs rarely run that long. The 15-years-plus band holds 8,387 freehold and 8,395 leasehold pairs, against a median hold of about six and a half years. The record is strong on the five-to-fifteen-year owner and thin on the thirty-year one.

Lease decay past about 60 years remaining. Leasehold pricing falls away as the remaining lease shortens, and faster once it drops past roughly 60 years — we measured that curve in The decay curve. A five-to-fifteen-year hold on a newer leasehold ends long before that point. A 30-year hold on an older one does not.

En bloc optionality. Freehold land carries redevelopment value that does not depend on a lease top-up. A collective sale only shows up in our data where one happened, so the option value is invisible in the median.

Financing on old leases. Banks tighten the loan-to-value they offer as the remaining lease shortens against the borrower’s age, which shrinks the pool of buyers who can finance an old leasehold. Freehold never meets that wall.

These are the limits of the evidence, not a hedge on the headline. Within the horizon the record covers, freehold did not make more money.

Within a normal five-to-fifteen-year holding period, tenure is not what decides your return. District, entry year and holding power are. Pay for freehold when you are buying for the 30-year case — a home you intend to keep past the point where a lease starts to bite — or for the land itself. Never pay a premium for it expecting faster growth. The exit record gives that expectation no support.

Before paying up, ask three questions. How long do I really expect to hold? What does this district’s own like-for-like record say? And what is the leasehold alternative’s remaining lease on the day I sell, not the day I buy? If the answers are “under fifteen years”, “leasehold ahead” and “well above 60 years”, the freehold premium is buying you a label.

MARKET & POLICIES · INSIGHT 003

The decay curve: what your 99-year lease actually loses, decade by decade

The exit record by lease decade — from triple digits to 49% — and the freehold twist nobody expects.

MARKET & POLICIES · INSIGHT 002

We checked 67,000 property exits. Here's who actually loses money.

67,354 five-year exits, decomposed by district. D01 lost money most of the time; D28 almost never did.

Dataset — Huttons “Export Transactions” (URA REALIS-derived), 1995 to 1 Sep 2026: 623,578 private non-landed sale rows, 224,023 matched pairs with a hold of at least one year. Tenure analysis on non-EC pairs only. Gross returns, before stamp duty, interest and fees.