Lease decay: what a shorter HDB lease actually costs

A 99-year lease sounds like a lifetime, and for the first few decades the market treats it as one. Then it stops. The interesting thing about lease decay in Singapore is not that it happens, it is when it starts showing up in prices, and the answer is much earlier than the arithmetic alone would suggest.

Why does the curve bend so early?

If a lease were simply a countdown, price would fall smoothly and slowly. It does not, because two rules bite well before the lease expires.

CPF. A buyer can only use their CPF Ordinary Account in full where the remaining lease covers the youngest buyer to age 95. Fall short and the usable amount is pro-rated. For a couple in their thirties that condition starts to bind around the 60-year mark, which is why the curve turns there rather than at year 20.

Financing. Loan-to-value falls as the tail shortens, and the maximum tenure shortens with it. A buyer who needs a larger cash deposit and a shorter term is a buyer who can bid less, regardless of what they think the flat is worth.

The result is a price curve shaped by policy as much as by time. Look at the lease band chart on Queenstown or Toa Payoh, two mature districts with a wide spread of lease ages, and the step down between bands is visible without any statistics.

What does the resale record actually show?

Pooled across the 30 districts that publish the cut, a 4-room flat with 90 to 99 years left trades at S$726 per square foot and one with under 50 years at S$575, a ratio of 1.26. That is the figure most accounts of lease decay would print, and it is the wrong one.

Look down the columns and the curve is not even monotone. For 4-room and 5-room flats the under-50 band is dearer than the 50–59 and 60–69 bands above it, which no theory of lease decay predicts. Nothing has gone wrong with the data: the flats with the shortest leases are in the oldest estates, and the oldest estates are the most central. Location is paying back some of what the lease takes away.

That is a composition effect, the same one that makes a national median move without any price moving. To see the lease on its own it has to be held out.

How large is the lease effect on its own?

Much larger. Comparing a 90–99 year lease with a 50–59 year lease inside the same district and the same flat type, the longer lease trades at a median 1.54 times the rate, in 47 of 47 qualifying pairs across 17 districts. Not one pair runs the other way.

So the pooled figure of 1.26 understates the lease effect by roughly half. The spread between 1.13 in Yishun and 2.04 in Queenstown is the second finding: the lease is not priced the same everywhere. Where the land is scarce and the estate central, the market pays far more for the tail; in an outer town it pays comparatively little.

These are frozen at the 2026-07 vintage and are not recomputed with every release, because a sentence written about a specific number should not change under the argument it supports. The live figures on the district pages always carry the current window.

How should the bands be read?

A caution the charts on this site carry and that is worth repeating: lease bands are not a controlled experiment. Older flats are not only shorter-leased, they are often larger, lower, and in different parts of a district from newer blocks. Some of the gap between a 90-year band and a 50-year band is lease and some of it is everything else that correlates with age.

That is why the bands are shown per district rather than nationally. Comparing lease bands within one district holds a great deal of the “everything else” roughly constant. Comparing them across the country does not.

What does it mean if you are buying?

A shorter lease is not automatically a bad purchase. It is a cheaper purchase with a shorter horizon and tighter financing, and whether that trade is good depends on how long you intend to hold it and what you can borrow. What you should not do is treat a low PSF on a short-leased flat as a bargain without checking the lease band it sits in: on the district pages here, that is one chart down from the headline.

Figures on this page track the published dataset and are current to August 2026.