The PropKaki Gap™

Definition

The PropKaki Gap is the percentage by which the median transacted price per square foot (PSF) of freehold private condominiums exceeds that of 99-year leasehold private condominiums in the same quarter, island-wide:

A reading of +20% means freehold buyers paid 20% more per square foot than leasehold buyers in the same quarter, on a whole-island median basis.

Scope and data source

Property types included: URA caveat records with propertyType equal to "Apartment" or "Condominium" only. Executive Condominiums are excluded (they are a quasi-public housing type subject to HDB resale restrictions and their freehold status does not apply). All landed property (Terrace, Semi-detached, Detached) is excluded: landed stock is overwhelmingly freehold and its inclusion would conflate tenure with property-type effects.

Tenure classification: A transaction is classified as freehold if its URA tenure field starts with "Freehold" or contains "999" (covering both "Freehold" and "999-year leasehold from..." titles). All remaining transactions are classified as 99-year leasehold. This mirrors URA's own published category system.

Transaction window: URA's rolling ~5-year private residential transaction data, accessed via the in-memory cache maintained by PropKaki's geo intelligence module. The same source powers PropKaki's District Price Tracker and the PIE property valuation tool. No direct URA API calls are made during the weekly computation — only the cached data is used.

PSF basis: Strata floor area in square feet (URA area field converted from square metres at 10.7639 sqft/sqm).

Computation steps

Suppression rules

A quarter is suppressed from all three series (freehold PSF, leasehold PSF, and Gap) if either tenure bucket contains fewer than 10 transactions in that quarter. This threshold matches PropKaki's district PSF suppression rule. Suppressing the entire quarter when either bucket is thin prevents a sparse freehold market in a given quarter from producing a gap reading that could be driven by a handful of high-value outliers rather than market conditions.

History accretes across weekly sync runs: once a quarter's values are written to market_series, they remain even after they fall outside URA's rolling 5-year window.

Interpretation notes

The Gap is an island-wide median comparison. It does not control for location, floor level, unit size, age, or project quality. A freehold condo in Bishan is compared to a leasehold condo in Tanjong Pagar in the same quarter. In periods where freehold and leasehold condo supply is geographically uneven — for example, if most new freehold completions are in the CCR while leasehold completions are in the OCR — the Gap may reflect the CCR/OCR price differential as much as the tenure premium.

PropKaki publishes the Gap as an observed market reading, not as investment advice. The tenure premium in a specific project or district can differ materially from the island-wide figure. For project-level tenure comparisons, use the PIE valuation tool with the tenure_split=true parameter.

The like-for-like basis (v2.0)

The interpretation note above — that an island-wide median may report the regional price differential rather than the tenure premium — was written as a caution before it had been measured. In August 2026 it was measured, and it was the dominant effect.

Two compositional differences separate the tenure arms. 89.3% of new sales are 99-year leasehold, so new-launch pricing inflates the leasehold arm with stock that has never been resold. And freehold projects have a median completion year of 2001 against 2011 for 99-year projects — a ten-year vintage gap, meaning the island-wide figure was substantially reporting building age. Removing both reverses the sign of the result.

The like-for-like basis therefore compares only resale transactions, within the same URA market segment, and within the same completion decade. It is published alongside the island-wide Gap rather than replacing it: the island-wide figure remains the correct answer to “what did freehold and leasehold transact at”, and the like-for-like figure answers “what is tenure worth, holding other things equal”. They are different questions and they have different answers.

Comparable-quality standard

A transaction count is not a comparable count. Repeat sales within one development are a single comparable observed several times, so a cell resting on 300 sales from two projects is not a 300-comparable sample — and a transaction floor cannot tell the difference. Every published like-for-like cell must additionally satisfy:

Thresholds were set from the measured distribution, not chosen in advance. On first application all 15 region cells passed every test; roughly half of district cells did not, and are withheld. The suppressed count is published rather than hidden — a cell that fails is omitted entirely rather than shown with a caveat, because a caveat travels separately from the number and the number is what gets quoted.

Completion-year source, and its gate

URA’s caveat feed carries no completion date. The only age-like field it does carry — lease commencement, parsed from the tenure string — exists for leasehold only; using it would adjust one arm and not the other, which is worse than no adjustment because it appears rigorous. Completion years are therefore compiled separately and maintained by PropKaki, with the compilation date published as the dataset vintage.

Because that source is maintained rather than fed, what degrades is coverage rather than age. If the completion-year join falls below 93% on either tenure arm, or the two arms diverge by more than 5 percentage points, no like-for-like figure is published at all for that run. An uneven join does not control for age; it introduces a fresh selection effect while looking more rigorous than before.

What is not adjusted for

The basis is median transacted PSF on strata floor area. No adjustment is made for floor level, facilities, project scale, condition, orientation, or movement in the market within the eight-quarter window. Two consequences follow, and both are readings of the data rather than defects in it.

In the Rest of Central Region, 2010s stock shows freehold transacting below leasehold. That cell passes every quality test — approximately 200 distinct freehold developments against 42 leasehold ones — and the ratio is itself the likely explanation: freehold sites in the city fringe are small, producing boutique blocks with few facilities, while leasehold sites of that vintage are large integrated developments. Where scale and facilities differ that sharply, they can outweigh tenure. PropKaki reports the observation and names the probable cause; it does not adjust for it.

In the Core Central Region the leasehold arm is structurally thin, because CCR stock is predominantly freehold. Those cells clear the comparable-quality thresholds, but rest on materially fewer leasehold developments than freehold ones. That is a feature of the market rather than a fault in the measurement, and is stated rather than smoothed over.

Series keys and API access

All three series are available via PropKaki's free JSON API (no key required):

gap_premium_qtr returns percentage values (e.g. 18.5 = 18.5%).median_psf_freehold_qtr and median_psf_leasehold_qtrreturn SGD per square foot.

Citation

Preferred citation for the PropKaki Gap:

When quoting a specific quarter's reading, include the quarter, e.g.: "The PropKaki Gap stood at 18.5% in 2026Q1 (PropKaki Intelligence, methodology v1.0)."

Full citation formats, BibTeX, and licence terms: /market/cite.

Version history