One-Bedders Yield More. In the Same Building, They Appreciate Half as Fast.

The case for a one-bedroom condo is an income case, and the income case is correct. Pair every private rental contract filed in the year to August 2026 with the resale prices in the same project, and the one-bedders yield 4.21% gross against 2.98% for the three-bedders in the same market. That is not a rounding difference. It is 41% more income on every dollar deployed.

The part that rarely gets run alongside it is what those units did on the way out. Take the 12,933 resale transactions in the year to October 2026 where the caveat records what the seller originally paid, and compute the annual compound return on each one. The one-bedders returned 1.85% a year. The three-bedders returned 4.50%. The income premium is 1.23 points. The capital gap running the other way is 2.65.

The gradient runs both ways

Every size band earns more rent per dollar than the one above it, and every size band compounds more slowly than the one above it.

The last column is the one that gets no airtime. Of 1,812 one-bedders resold in the year to October, 240 went out below what the seller paid for them, after a median hold of 7.9 years. Among three-bedders the figure is 106 of 5,525. A one-bedroom owner is seven times more likely to take a nominal loss on exit than a three-bedroom owner in the same market.

Add the two columns and the picture is not a trade-off at all. Gross yield plus capital return comes to 6.06% for a one-bedder, 6.49% for a two, 7.48% for a three and 7.43% for a four. The small unit does not win on income what it gives up on price. It loses on both, once both are counted.

Hold the building still

Pooled numbers across bedroom counts are confounded, and obviously so. One-bedders are concentrated in central districts and in newer launches, where entry prices were highest. Three-bedders are concentrated in the suburbs, where the last five years were kindest. That comparison asks two questions at once.

So compare units inside the same building, where the district, the tenure, the age, the managing agent and the MRT station are all identical by construction.

In 102 projects with enough rental and resale volume to compute both, the one-bedder out-yields the three-bedder by a median of 1.09 points, and it does so in 101 of the 102.

In 100 projects with at least three matched repeat sales of each, the three-bedder out-compounds the one-bedder by a median of 2.22 points a year, and it does so in 93 of the 100. Median one-bedder: 2.19% a year. Median three-bedder: 4.19%.

Same building, same street, same lease. The capital gap is almost exactly twice the income gap, and the income gap is the only one most buyers have seen written down.

What survives the cuts

The ordering does not depend on how the sample is drawn.

• By region. One-bedders trail three-bedders by 2.75 points in the CCR (0.18% against 2.93%), 2.41 points in the RCR, and 2.67 points in the OCR. Monotonic in all three.
• By when the seller bought. Among units bought in 2006-2012 the gap is 2.26 points; in 2013-2016 it is 2.84; in 2017-2020 it is 3.54; in 2021-2025 it is 2.91. The 2013-2016 cohort is the one that bought at the old peak, and the gradient is no steeper there than elsewhere, so this is not a story about one bad vintage of shoeboxes.
• By time window. Sales in the last 24 months instead of 12: 2.08% against 4.64%. All five years of caveats: 1.99% against 4.37%. Holds of at least three years only: unchanged at 1.85% against 4.50%.
• By floor area instead of bedroom label. Under 600 sq ft returns 1.85% a year; 600-899 returns 3.04%; 900-1,199 returns 4.53%. The bedroom count is not doing any work the square footage was not already doing.

Twelve of twelve region and cohort comparisons run the same way. The direction is not a specification choice.

Where the gradient stops

It is not "bigger is better." Sorted purely on floor area, the return peaks in the 900 to 1,599 sq ft range at roughly 4.5% a year and then falls back to 3.52% above 1,600 sq ft, on 1,318 transactions. The very large unit has the same problem as the very small one at the other end of the market: a thin buyer pool and a quantum that only a few households can finance.

The sweet spot in this data is the ordinary family unit. That is also, not coincidentally, the unit most households are buying to live in rather than to let.

The arithmetic in dollars

Take $2.10 million, the median three-bedroom price, and run it two ways at the medians above.

As one three-bedroom unit, it earns $60,000 a year in rent and compounds at 4.50%, which is $94,593 in the first year.

As 2.36 one-bedroom units at $890,000 each, it earns $87,828 a year in rent — $27,828 more — and compounds at 1.85%, which is $38,842. That is $55,750 less.

Net of both, the shoebox basket is about $27,900 a year behind, before anyone has paid for the second set of maintenance fees, the second property tax bill, the second agent and the second round of vacancy.

And before the tax that actually decides it. A Singapore Citizen buying a second residential property pays 20% ABSD, and 30% on the third, rates unchanged since 27 April 2023 (IRAS). On an $890,000 unit that is $178,000 on the second and $267,000 on the third. The multi-shoebox portfolio only ever made sense as arithmetic on a page where that line was missing.

What this does not show

It does not show that one-bedders are a bad asset. It shows that the thing they are good at is smaller than the thing they are bad at, measured the same way.

The yields here are gross. Net of maintenance, property tax, insurance, agent commission and vacancy, every number in the yield column comes down, and the arithmetic above gets worse for the basket, not better — see gross yield versus net yield for where the leakage sits.

They are also medians of project medians, not the yield any individual landlord booked. A project's median rent and its median resale price are not the same unit on the same floor.

The repeat-sale returns are measured on units that sold. Owners sitting on a loss often do not sell, so the realised returns here flatter every band, and probably flatter the one-bedders most, since that is where the losses are. The 13.2% loss rate is a floor, not a ceiling.

There is a real case the data cannot answer against. A one-bedder costs $890,000 and a three-bedder $2.10 million. For a buyer who cannot finance the larger quantum, the comparison is not between two investments, it is between one investment and none. Nothing here says a cheaper entry is the wrong entry. It says the yield number is not the reason to make it.

And nothing here changes the published scorecard weights. The Resale Project Scorecard scores rental yield as one factor of seven, at 10% of the total, and this is a reminder of why it sits at 10 rather than higher: a yield score is a measure of income, not of total return, and in this data the two point in opposite directions.

How to read it

If the plan is to hold a unit and live off the rent, the yield gradient is real and the small unit is the efficient way to buy income.

If the plan is to hold for eight to ten years and sell — which is what the median owner in this dataset actually did — the capital line is the bigger number, and it has run against the small unit consistently, across every region, every purchase cohort and every time window tested, and inside 93 of the 100 buildings where both can be compared directly.

Run both lines before deciding. The one most people have run is the smaller one.

---

Gross yields pair 87,760 private residential rental contracts filed September 2025 to August 2026 with 22,158 resale and sub-sale caveats from October 2025 to October 2026, matched at project and bedroom count, minimum three of each. Capital returns use the 12,933 caveats in that window that record the seller's own purchase price and date, held at least one year, across 1,611 projects. Figures are medians. Stamp duty rates are as published by IRAS and current as of 27 April 2023. This is general information, not investment advice; confirm any duty position with IRAS or your conveyancing lawyer.