88.8% of shoebox owners sold for more than they paid

The articles in this series measured what happened to shoebox prices per square foot. None of them answered the question an owner actually asks: did people who bought one make money? The rest of what we found about shoebox units in Singapore is on the shoebox units guide.

They did. Of 7,523 shoebox units bought and later sold since 2020, 6,680 sold for more than they cost, 88.8%. The median gain was $92,000 on a median hold of seven years.

That is a different measurement from everything else in this series. It is not a median across projects. It is the same unit, bought and sold, with both prices known.

Key takeaways

• 88.8% of shoebox resales since 2020 sold above their purchase price. Median gain $92,000 over a seven-year hold. The distribution
• Annualised, that is 1% to 3.5% a year before any costs, a far more sober number than the headline. What it works out to per year
• Where you bought mattered far more than how long you held. Resales in the core central region (CCR) lost money 44.2% of the time; those in the outside central region (OCR) 6.0%. Where the losses are
• The CCR underperformed in every purchase year, not only the 2011–2013 peak, so this is not simply a timing story. It is not just timing

What actually happened

Every figure below comes from matched pairs: a unit bought, the same unit sold, both transactions in the caveat record. Resales completing from 2020 onward.

	Count	Share
Sold above purchase price	6,680	88.8%
Sold below	822	10.9%
Sold flat	21	0.3%

The spread, measured as percentage gain on purchase price:

Percentile	Gain
10th	−0.8%
50th	+13.3%
90th	+28.8%

Even the worst tenth of these sales roughly broke even in nominal terms. The middle sold for 13.3% more than they paid.

That is the honest headline, and it sits awkwardly beside the rest of this series. Shoebox units underperformed family-sized units on price per square foot. That finding stands. But relative underperformance and absolute loss are different experiences, and almost nine in ten owners had the second one work out fine.

What it works out to per year

A 13.3% gain over seven years is not 13.3% a year. Annualised, the same numbers look far more modest.

Median compound annual growth, by the year the unit was bought:

Bought	Resales	Median hold	Median CAGR
2011	543	10.8 yrs	0.76%
2012	956	9.9 yrs	1.09%
2013	807	9.1 yrs	0.95%
2014	428	8.0 yrs	1.16%
2015	505	6.8 yrs	2.12%
2016	683	6.1 yrs	2.49%
2017	699	5.6 yrs	2.11%
2018	745	5.1 yrs	2.70%
2019	639	4.6 yrs	3.30%
2020	507	4.2 yrs	3.34%
2021	420	3.8 yrs	3.48%

Every year is positive. None reaches 3.5%.

This is before any cost at all. Buyer's stamp duty on purchase. Seller's stamp duty if sold inside the holding period. Agent commission on exit. Legal fees both ends. Maintenance and property tax for every year held. On a shoebox around $700,000, a round trip plausibly runs $40,000 to $60,000, which against a $92,000 median gain is a large share of it.

We are not going to put a net number on it here, because it depends on the loan, the rate, whether the unit was rented, and what the owner paid in duty, none of which is in the transaction record. A worked example with stated assumptions is a separate piece.

The 2021 and 2022 purchase years have short holds, under four years, and their annualised figures rest on a small base. Read them as provisional.

These annual figures cover every sale, including the ones that lost money. A loss annualises perfectly well (a unit bought at $1,000,000 and sold at $700,000 after three years lost 11.2% a year) so no sale is excluded. Pairs held under six months are left out, because annualising a two-month hold produces a number that means nothing.

Where the losses are

The 822 losses are not scattered. They concentrate sharply by region.

Region	Resales	Sold at a loss
CCR (core central)	624	44.2%
RCR (rest of central)	3,317	10.0%
OCR (outside central)	3,582	6.0%

Nearly half of CCR shoebox resales lost money. One in sixteen OCR ones did, a seven-fold difference.

This is the same finding as our regional price article, which reported CCR shoebox psf falling 4.5% since 2017 while the OCR rose 25.9%. What changes here is what it means: that was a median across projects, this is what happened to individual owners.

How much was lost, and how much was made

Frequency is only half of it. A loss that happens often but costs little is a different proposition from one that happens rarely and costs a lot.

Among the sales that lost money, the median loss was 5.2%, about $40,000. Among those that gained, the median gain was 14.8%, about $102,000. Note that this gain figure covers only the profitable sales; the 13.3% quoted earlier is the median across every sale including the losses.

So the typical loss was shallow. One in nine owners came out slightly behind, not ruined.

By region, though, the depth of loss follows the same pattern as its frequency:

• Median gain, among sales that gained
• Median loss, among sales that lost

Region	Median gain	Median loss	Worst tenth of losses
CCR	+8.9% / $85,000	−7.7% / −$84,176	−19.6%
RCR	+14.2% / $102,000	−4.4% / −$34,661	−11.4%
OCR	+15.7% / $103,000	−3.7% / −$25,000	−10.6%

The CCR is worse on every measure at once. A CCR buyer was more likely to lose, lost more deeply when they did, and gained least when they did not. The median CCR loss of $84,176 is more than three times the OCR figure of $25,000, and the median CCR gain, $85,000, is barely larger than the median CCR loss.

It is not just timing

The obvious objection is that CCR buyers were simply unlucky with timing: that they bought at the 2011–2013 peak and everyone who did fared badly. The data does not support it.

• CCR
• RCR
• OCR

Bought	Region	Resales	Sold at a loss	Median CAGR
2010–2014	CCR	296	59.8%	−0.23%
2010–2014	RCR	1,278	18.2%	0.85%
2010–2014	OCR	1,400	10.6%	1.38%
2015–2018	CCR	163	39.9%	0.43%
2015–2018	RCR	1,206	5.0%	2.29%
2015–2018	OCR	1,263	4.3%	2.59%
2019–2022	CCR	87	26.4%	1.38%
2019–2022	RCR	739	4.5%	3.19%
2019–2022	OCR	904	1.2%	3.54%

The CCR is the worst performer in all three windows. It improves (59.8% to 39.9% to 26.4%) but never approaches the other two regions in any window. A CCR buyer in 2019–2022, the best window in the table, still lost money more often than an OCR buyer at the 2010–2014 peak.

Timing mattered. Every region improved as the purchase year moved later. But region mattered more, and the two are independent.

Note also that the CCR sample is small (87 resales in the most recent cohort) so read the direction rather than the precise figure.

What this means for judging a specific unit

We are not going to publish a list of projects to buy. A list is stale in six months, and the transaction record cannot see what makes a particular unit good to live in: layout, light, noise, the walk to the station.

What the data does support is a way of checking the unit in front of you.

Region is the single strongest signal in this dataset. Not a rule that CCR shoeboxes always lose (55.8% of them made money) but the base rate is far worse, and it has been in every window measured. If you are looking at a CCR shoebox, the burden of proof on that specific unit is higher.

Entry price relative to its own building matters more than it used to. The psf premium small units carried has gone to zero and inverted. A shoebox priced as though the old premium still applies is priced against a rule that no longer holds.

Check the building's own repeat sales, not the market's. Every figure here is a market-wide base rate. The unit you are considering sits in a specific development with its own history of what units have actually resold for.

And treat the annual number, not the headline. A 13% gain sounds like a result. Spread over seven years, before costs, it is a modest one. Whether that clears your alternative is a question about your circumstances, not about shoeboxes.

What this does not tell you

• Net returns. Every figure is gross capital gain. Duties, commission, legal fees, maintenance, property tax and financing all come out first, and none is in the transaction record.
• Rental income. These are capital outcomes only. Shoeboxes had a gross rental yield of 4.30% in 2025, which over a seven-year hold is a large part of total return and is not counted here.
• Who these owners were. Whether they lived in the unit, rented it out, held it through choice or sold under pressure. Not recorded.
• What happens next. These are completed round trips. Someone buying in 2026 faces a different market.
• Units that have not resold. Owners still holding are not in this data. If holders differ systematically from sellers, that shapes what we can see.

Method and data notes. Data to 17 July 2026.

Private non-landed resale caveats, condominium and apartment, executive condominiums and collective sales excluded. Shoebox is 400–538 sqft, matching URA's 50 sqm threshold.

Matched pairs are built on unit identity: consecutive transactions of the same unit, the earlier taken as purchase and the later as sale. Only pairs whose sale completed from 2020 onward are included, so the exit side reflects the recent market. 7,523 pairs qualify, across 9,137 units with two or more recorded transactions. Matching a unit against itself removes the question of what sold, and charges three prices for it: the ageing of the unit between its two sales, any renovation in between, and the fact that only homes which sold twice are ever seen.

Compound annual growth is computed per pair as the ratio of sale to purchase price raised to the reciprocal of the holding period in years, then the median taken across pairs. Every pair is included, whether it gained or lost: a loss annualises normally, and excluding losses would inflate the result by dropping the owners who did worst. Pairs held under six months are excluded, because annualising a very short hold produces magnitudes that mean nothing.

Where this article reports a median gain or a median loss separately, those figures do cover only the sales that gained or only those that lost, and each is labelled as such.

Regions are URA's market segments, taken from the purchase transaction.

Data to 17 July 2026.

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.