All posts
Shoebox units: a series

Published 6 August 2026

Shoebox units lost their price premium and kept their rent premium

By Realila

Shoebox units lost their price premium and kept their rent premium. The result is that in 2025 a shoebox has a gross rental yield of 4.30% against 2.87% for a family-sized unit, a gap that has widened every few years since 2017, at exactly the time we and everyone else assumed small units were becoming a worse investment. The rest of what we found about shoebox units in Singapore is on the shoebox units guide.

This also corrects something we published. In our first article in this series we wrote that rising interest rates had crushed the rental yield spread that made small units work as an income play. That was wrong, and the rental data says so plainly. The spread did not narrow. It roughly doubled.

Key takeaways

  • Shoebox rent per square foot has held a premium of roughly 52% over family-sized units for nine straight years, while the sale premium collapsed to zero. The rent premium
  • Gross rental yield on a shoebox rose from 3.56% to 4.30%. Family-sized rental yield fell slightly, from 2.99% to 2.87%. Rental yields
  • This contradicts a claim in our first article, which said the rental yield spread had been crushed. It had not. What we got wrong
  • It does not rescue the explanations we ruled out elsewhere, and it leaves an open question: an asset with a rental yield of 4.3% against 2.9% has lost its price premium anyway. The open question

The rent premium never moved

Shoebox is 400–538 sqft, matching URA's 50 sqm threshold. On the rental side that is URA's 40–50 sqm band, which ends at the same 538 sqft line. Family-sized is 1,000–1,500 sqft on the sale side and the 90–130 sqm bands on the rental side.

Median monthly rent per square foot, across the same projects at both ends of the period.

YearShoeboxFamilyRent premiumProjects (S / F)
2017$4.05$2.65+52.6%318 / 1,053
2018$4.01$2.67+50.5%332 / 1,089
2019$4.03$2.65+51.7%355 / 1,145
2020$4.11$2.64+55.6%347 / 1,112
2021$4.23$2.74+54.4%347 / 1,140
2022$5.16$3.23+59.6%347 / 1,102
2023$6.19$3.98+55.6%345 / 1,118
2024$5.99$3.91+53.1%342 / 1,079
2025$6.09$4.01+51.9%337 / 1,076

Both lines rose steeply in 2022 and 2023, then flattened. What did not happen is any convergence between them. The premium starts at 52.6%, peaks at 59.6% in 2022, and ends at 51.9%. Nine years, no trend.

That is a striking contrast with the sale side, where the same comparison went from a premium of around 21% to below zero. Tenants kept paying more per square foot for small units. Buyers stopped.

What that does to rental yield

Rent that holds while price falls produces rental yield. Gross rental yield here is twelve months of median rent per square foot divided by median sale price per square foot, both measured on projects present at both ends of the period.

YearShoebox rental yieldFamily rental yieldGap
20173.56%2.99%+0.57 pt
20183.37%2.77%+0.60 pt
20193.37%2.72%+0.65 pt
20203.61%2.76%+0.85 pt
20213.53%2.63%+0.90 pt
20224.13%2.85%+1.28 pt
20234.51%3.25%+1.26 pt
20244.16%2.99%+1.17 pt
20254.30%2.87%+1.43 pt

The shoebox line rises from 3.56% to 4.30%. The family line goes nowhere, ending slightly below where it started. The gap between them widens from 0.57 of a percentage point to 1.43: it has roughly doubled.

Two separate things drove that. Rents rose sharply in 2022 and 2023 for both sizes. And shoebox prices, unlike family prices, stopped rising much. The first lifted both rental yields; the second lifted only one.

What we got wrong

Our first article listed investor withdrawal as the explanation left standing after composition, floor effects, space preference and supply were each tested and failed. That elimination still holds: none of those explanations survived, and later articles ruled out tenure and building age as well.

But we gave a reason, and the reason was wrong. We wrote that rising rates had crushed the rental yield spread that made small units work as an income play. The rental data shows the opposite: across the rate cycle, shoebox rental yields rose and their advantage over family-sized units widened.

That article has been corrected. The correction matters beyond a single sentence, because it changes what kind of withdrawal this was. If investors left, it was not because the asset stopped producing income. Something else made them stop buying.

The obvious candidate is transaction cost rather than running economics. Additional buyer's stamp duty for a Singapore citizen's second property rose from 12% to 17% in December 2021 and to 20% in April 2023; the foreign-buyer rate went from 20% to 60% over the same period. That is paid at purchase, on the whole price, regardless of what the unit returns in rent afterwards. A duty that high does not care that the rent is good.

We have not tested that, and this data cannot. It is a candidate, stated as one.

The open question

Here is what we cannot explain.

An asset with a rental yield of 4.30% sitting beside a comparable asset at 2.87% should attract capital. In most markets that gap closes, either because buyers bid the higher-yielding asset up or because its rents fall back. Neither has happened here for four years running. The gap has instead widened every year since 2021.

Either the withdrawal is entirely policy-driven and will persist as long as the duties do, or the market is pricing something we are not measuring: a risk, a cost, or an expectation about the future of small-format stock that does not show up in rent or price today.

We do not know which. We would rather say that than pick one.

What this does not tell you

  • Net rental yield. These are gross figures. Maintenance, property tax, agent commission, vacancy and income tax all come out before an owner sees anything, and they do not scale identically across unit sizes.
  • Total return. Rental yield is one half of what an owner earns. Capital growth is the other, and on that measure the two sizes have behaved very differently over the same period. That is a separate article.
  • Your unit. These are medians across hundreds of projects. Individual buildings, stacks and lease terms vary widely.
  • What happens next. Nine years of history. Not a forecast.
Method and data notes. Data to 17 July 2026 for sales, February 2026 for rentals.

Rental data is URA's non-landed rental contracts. Sale data is private non-landed resale caveats, condominium and apartment, executive condominiums and collective sales excluded.

URA reports rental floor area in 10 sqm bands rather than exact areas, so rent per square foot here uses the midpoint of each band. Shoebox uses the 40–50 sqm band, which converts to 431–538 sqft and shares its upper boundary with URA's 50 sqm shoebox threshold. Family-sized uses the 90–130 sqm bands, 969–1,399 sqft, the closest available fit to the 1,000–1,500 sqft sale band. Because both sides use midpoints, the comparison between them is more reliable than either level on its own.

Both the rental and sale series run on fixed panels: projects with at least three transactions in that band in both the 2017–2019 and 2023–2025 windows, so each column describes the same buildings at both ends. Medians are taken across projects rather than pooled across transactions.

Rental coverage was checked before use: all twelve months are present in every year from 2017 to 2025, and the share of rental records linked to a project profile is stable throughout at 99% for the shoebox band and 96% for the family band. 2026 is excluded: only two months of rental data exist.

Rental yield is gross: twelve months of median rent per square foot divided by median sale price per square foot, both from the same year. It is not a return an individual owner realises.

Data to 17 July 2026 for sales, February 2026 for rentals.

More from Realila

Every number in this post comes from Realila Research, dated to when it was true. For now we publish research notes from the platform to answer the community's questions; the platform itself opens to the public later.

Join the waitlist

Get each new note by email, and be first in when the research platform opens.

Ask for a research note

A question about the market, or data you would like to see.

We read every request. We cannot write a note on every one, but the ones we do write will appear on the Feed, and we will email you if yours does.