Launch or resale? On return to capital, resale won four years out of five

Buy at launch or buy a resale? Almost every comparison you will see answers it with a capital gain, and almost every one is misleading: not because the gains are wrong, but because the two buyers are doing entirely different things with their money in the years between.

A launch buyer pays in instalments across three years of construction and collects no rent until the building is finished. A resale buyer pays a quarter of the price on day one, carries a full mortgage from the start, and can rent the unit out that month.

Comparing what each earned on the money actually committed, matched so both bought and sold in the same months and held for the same length of time, resale won in four of the five purchase years we can test. The exception is 2018, and even that flips depending on the mortgage rate.

Key takeaways

• On return to capital, resale led in 2017, 2019, 2020 and 2021 at every mortgage rate tested. Launch led only in 2018. Return on capital
• The reason is rent, not price. A resale buyer collected roughly $105,000 to $123,000 over the hold; a launch buyer collected $35,000 to $51,000. What the construction years cost
• The gap widens sharply after 2019, reaching 12 points in 2021, as launch entry prices pulled away from resale. The entry premium
• Capital gain alone tells the opposite story in the early years, which is why we do not lead with it. Why not just compare gains

How the comparison is built

Two things had to match before any of this means anything.

Same entry and exit. Median purchase dates differ by at most 0.20 of a year between the two groups, and median sale dates by at most 0.17. Both groups bought in the same months and sold in the same months, so both faced the same market on both sides.

Same holding period. Every pair is held between four and six years. Median holds run 4.55 to 5.28 years for launch and 4.29 to 5.01 for resale, with overlapping interquartile ranges throughout.

Bought	Launch pairs	Launch hold	Resale pairs	Resale hold
2017	131	5.28 yrs	64	5.01 yrs
2018	222	4.95	101	4.90
2019	289	4.99	50	4.70
2020	215	4.78	72	4.71
2021	79	4.55	77	4.29

Without that matching, the comparison measures duration and market timing rather than purchase type.

What the construction years cost

This is the single largest difference between the two, and it has nothing to do with the price paid.

A launch purchase completes about three years after booking. On a five-year hold, that leaves roughly two years of possible tenancy after allowing a couple of months to find a tenant. A resale buyer has the full five.

Bought	Launch net rent	Resale net rent	Difference
2017	$51,128	$122,708	$71,580
2018	$42,948	$120,663	$77,715
2019	$44,993	$114,527	$69,534
2020	$38,858	$116,573	$77,715
2021	$34,767	$104,302	$69,535

Net of maintenance, upkeep, property tax, agent fees and vacancy.

Roughly $70,000 to $78,000 of rental income, forgone in exchange for buying earlier. That is the number a gross-gain comparison leaves out entirely.

Return on what was committed

Modelling both purchases properly (the staged payment schedule for the launch, the full deposit and drawn loan for the resale, interest on what is actually outstanding, rent from when the unit can be rented out, and the sale at the end) gives a return on the cash each buyer put in.

At a 2.5% mortgage rate:

Bought	New launch	Resale	Difference
2017	5.85%	7.97%	resale +2.12
2018	8.89%	8.45%	launch +0.44
2019	8.30%	11.42%	resale +3.12
2020	5.45%	14.97%	resale +9.52
2021	1.09%	13.14%	resale +12.05

Internal rate of return on monthly cash flows, which accounts for money going in gradually rather than all at once.

And across the rate range:

Bought	1.5%	2.5%	3.5%	4.5%
2017	resale +3.13	resale +2.12	resale +1.07	level
2018	resale +0.68	launch +0.44	launch +1.60	launch +2.79
2019	resale +4.14	resale +3.12	resale +2.06	resale +0.99
2020	resale +10.43	resale +9.52	resale +8.57	resale +7.59
2021	resale +13.01	resale +12.05	resale +11.06	resale +10.05

Two patterns worth naming. The launch advantage improves as borrowing gets more expensive, because a staged purchase is not carrying a fully drawn loan through the construction years. In 2017 and 2018 that is enough to close or reverse the gap at higher rates. From 2019 it is not close enough to matter, because by then the entry premium had grown too large for financing efficiency to offset.

By 2021, launch returns 1.09% at 2.5% and turns negative at 3.5%. Resale returns 13.14% and 10.99%.

The entry premium

Here is what changed between the two halves of that table.

Bought	Launch entry	Resale entry	Premium
2017	$750,000	$677,000	+10.8%
2018	$716,000	$690,000	+3.8%
2019	$747,000	$690,000	+8.3%
2020	$846,000	$635,000	+33.2%
2021	$953,000	$680,000	+40.1%

Median purchase price, same size band, same year.

Resale entry barely moved across five years, sitting between $635,000 and $690,000. Launch entry climbed from $750,000 to $953,000.

A buyer paying 40% more for the same size of unit has to earn that back before the purchase breaks even against the alternative. Within four to five years, and giving up three years of rent to do it, they have not.

Why not just compare gains

Because on capital gain alone the answer is different, and flattering to launch:

Bought	Launch gain	Resale gain
2017	12.99%	9.02%
2018	17.83%	10.26%
2019	17.18%	15.44%
2020	12.64%	21.13%
2021	6.80%	16.89%

Launch leads in three of five years here, against one of five on return to capital. The difference is that the gain column ignores $70,000 of rent and ignores that the two buyers committed different amounts of money at different times.

A capital gain is a fact about a property. A return is a fact about an investment. They are not the same, and for anyone choosing between two purchases the second is the one that matters.

Two worked examples follow a single purchase of each type from entry to sale with every cost itemised: a new launch bought in 2018 and a resale bought in 2021.

The assumptions, stated

Everything below is ours, not measured, and the result moves with them. Only the entry price, gain, holding period and pair counts come from the transaction record.

Financing. 75% loan, 30-year amortising, at a flat rate. Four rates shown because a borrower over 2017 to 2025 saw a wide range, and because the answer depends on it.

Launch payment schedule. The standard progressive scheme, completion three years after purchase, the buyer's 25% applied to the earliest stages, interest serviced on the drawn balance during construction. A project taking four years shifts the answer against launch; two years shifts it toward.

Rent. Suburban shoebox medians applied to 474 sqft, from two months after completion for a launch and from purchase for a resale, net of maintenance at $280 a month, upkeep at $1,200 a year, property tax at $2,400 a year, agent fees at half a month's rent a year with GST, and a 3% vacancy allowance. This assumption carries the most weight in the result. A materially weaker rental market narrows resale's advantage.

Costs at both ends. Buyer's stamp duty at the published schedule. Agent commission at 2% plus GST at 9% on sale, legal fees at $3,000 each side. No seller's stamp duty, as every hold exceeds three years.

Not modelled. Additional buyer's stamp duty, which applies to most investment purchases and would reduce every figure, the launch column more in percentage terms since it commits less capital. Income tax on rent, which reduces the resale column more since it collects more rent. Renovation and furnishing.

What this does not tell you

• Your project. Medians across hundreds of developments. A launch that completes early, or rents out well, moves against everything above.
• What a launch buys you that this cannot see. A new building, a full lease, warranties, and the ability to choose a unit before anyone else. Those are real and some buyers rightly pay for them.
• Anything forward-looking. Every purchase here completed its round trip. A 2024 or 2025 buyer faces a different entry premium in a different rate environment.

Method

Pairs are matched from URA caveat data on unit identity, so the purchase and sale are the same unit. Shoebox is 400 to 538 sqft, the upper bound being URA's 50 square metre threshold. Purchases are classified by the transaction type of the buying leg; a third group, sub-sale purchases, numbers 3 to 11 pairs a year and is excluded as too thin to compare.

Holding periods are restricted to four to six years and every sale completed from 2022 onward. Median entry and exit dates were checked directly: the two groups differ by at most 0.20 of a year on entry and 0.17 on exit in every purchase year.

Entry price, gain and holding period are medians across pairs. Return is an internal rate of return computed on modelled monthly cash flows under the assumptions stated above.

Sale data to 4 August 2026, rental data to February 2026.

This article replaces an earlier version that compared the two on capital gain alone, using groups whose holding periods differed by as much as four years. That comparison measured duration as much as purchase type. The correction is dated 13 August 2026.