The MRT Effect on HDB Prices: I Tested It Four Times, and I Was Partly Wrong

By Gary Lim · ERA Senior Division Director · CEA R009877B · Published 18 September 2026

The claim every agent repeats

The MRT effect on HDB prices is the single most repeated claim in Singapore property marketing, and almost nobody has actually measured it on HDB flats. You have heard the numbers. A new station is worth ten to fifteen percent. Five hundred metres is the gold standard. Buy before the line opens and ride the re-rating. I have said versions of this myself, across seventeen years and more than five hundred transactions, because everyone says it and it sounds obviously true.

So I tested it. Four separate times, on four different rail projects, using HDB’s own resale registration records rather than asking prices or listing counts. Three of those tests I ran earlier this year, and in two of them I published a flat null — I said the new line did not re-rate the town, and I meant it.

This week I ran the fourth and cleanest test, and then I went and read what the academics have actually published on this question. Both of those things made me revise. The strong version of my earlier position — that a new MRT station does not meaningfully move HDB prices — is wrong, and I am correcting it here rather than quietly leaving it up.

What replaces it is narrower, better evidenced, and considerably more useful to you if you are buying or selling a flat near a new line. The premium is real. It is small. It reaches a much shorter distance than anyone tells you. And by the time the first train runs, it is already in the price.

The MRT effect on HDB prices, at a glance

How I tested it, and why it is harder than it looks

The reason this question is usually answered badly is that the obvious method is worthless. If you take flats near a new station and measure what they did before and after, you will find they went up — because everything went up. Singapore HDB resale prices have risen for most of the last decade. Attributing that to the station is like attributing your salary increase to your new haircut.

The method that works is a difference-in-differences comparison. You need a treatment group (flats that got a station) and a control group (comparable flats in the same town that did not), and you measure the difference between their differences. Whatever the market did, it did to both groups, so it cancels. What is left is the station.

That only works if the control is genuinely comparable, which in HDB terms means holding three things constant that most analyses ignore entirely:

• Remaining lease. An HDB flat is a depreciating asset on a clock. Two flats five years apart in vintage are not the same product, and if your treatment group is systematically younger than your control you will measure lease, not rail.
• Flat type and floor area. A town’s median price moves when its transaction mix moves, with no individual flat changing price at all. I demonstrated this on real transactions in my Marine Parade work, where swapping four large-flat sales for four small ones moved the town average nearly nine percent by itself.
• Town. Comparing Ang Mo Kio to Sengkang tells you about Ang Mo Kio and Sengkang. The control has to live in the same town as the treatment.

Every figure below comes from HDB’s own resale registration records, published through Singapore’s open data portal, which is the same registry that produces the official statistics. Not asking prices. Not listing counts. Registered transactions, at flat level, with lease and floor area attached to each one.

Test 1: Mayflower, 2021 — the cleanest experiment I have run

Thomson-East Coast Line Stage 2 opened for passengers on 28 August 2021, adding six stations. One of them, Mayflower, sits at 91 Ang Mo Kio Avenue 4 and serves a cluster of Ang Mo Kio flats that previously had no station within walking distance — the nearest was Ang Mo Kio on the North-South Line, over in the town centre.

That is a genuine natural experiment, and Ang Mo Kio is the right town to run it in: a large, deep, overwhelmingly 1977–1985 resale market where you can find hundreds of near-identical four-room flats on both sides of the line.

That produced three rings:

Restricted to four-room flats, with no month sampling — every transaction in both windows. Pre-period January 2018 to December 2019; post-period September 2024 to August 2026. Twenty-four months each.

Lease drift is not driving this. Ring A lost 5.6 years of lease over the span, the control 6.3 years — near-identical, and the small difference favours A by roughly 0.4 percentage points, about a fourteenth of the measured effect. The regression controls for it directly anyway.

Raw median psf growth: treatment +43.1%, control +37.0%. A regression on log psf, controlling for remaining lease and floor area, puts the difference-in-differences at:

So the station is worth about six percent, and that is a real, statistically significant number. If I had stopped there, this article would simply have said “the conventional wisdom is roughly right, just smaller than advertised.” But the next test is where it gets interesting.

The timing decomposition: the whole gain landed before the trains did

A six percent effect measured from 2018–19 to 2024–26 tells you the flats were re-rated at some point across those eight years. It does not tell you when. So I split the span into three window-pairs and ran the same comparison on each.

Read those three rows together, because individually none of them is the story.

The first row is the quality check. Across a period when nothing was happening, treatment and control moved together — a precise null. That is what a valid control group looks like, and it is the step almost no industry analysis bothers with.

The second row carries the whole effect. Between 2018–19 and the eighteen months ending in August 2021, flats within 360 metres of a future Mayflower exit gained 5.7% relative to comparable Ang Mo Kio stock. The station was under construction, the alignment and station names had been public since 2012–2014, and the market priced it.

The third row is the one to tape to your monitor. From the run-up through to August 2026 — five years spanning the opening and everything after it — flats next to Mayflower gained nothing relative to control. Minus half a percent, statistically indistinguishable from zero.

How far does the premium actually reach?

The industry answer is 500 metres, sometimes a kilometre. My three rings say something tighter.

By somewhere between 530 and 740 metres, the premium is statistically indistinguishable from zero. And those are straight-line distances — real walking distances run roughly 1.2 to 1.4 times longer, so the 360-metre boundary is more like a 450-metre walk.

This replicates what I found at Bukit Panjang, where the premium attached to tight integration rather than to rail access in general: a block at the station entrance carried a large premium, a block 281 metres away carried a small one, and a precinct served by the same line but not adjacent to it carried nothing at all.

The practical translation: “near MRT” in a listing is doing a lot of work. Eight hundred metres is not near. If you are paying a premium for a flat that is a nine-minute walk from the station, the data says you are paying for something the market does not actually price.

A three-room cross-check on the same streets pointed the same way and harder — the inner ring outgrew the 530–740 m ring by 13.4 percentage points, against 3.7 points for four-rooms. That fits: smaller flats house more transit-dependent households. It is directional support rather than a second full test, because I did not pull a matched control street for three-rooms, and I would rather say so than dress it up.

Test 2: Bukit Panjang and the Downtown Line

Downtown Line Stage 2 opened on 27 December 2015, bringing heavy rail to a town that until then had only the Bukit Panjang LRT feeding into Choa Chu Kang. If any new line in Singapore should have re-rated a town, it was this one: Bukit Panjang went from a single LRT loop to a direct ride into Bugis and the CBD.

I ran the same design on matched 1986–89 four-room stock. Over the eleven years spanning the opening, the Cashew precinct grew 28.9% in psf while pooled LRT-only controls grew 34.6%. The difference-in-differences came out at −5.7 percentage points, and at −6.1 and −7.0 points under two wider specifications.

Negative. The precinct the new line was built to serve underperformed the parts of the town it did not serve.

What did pay in Bukit Panjang was integration, on a very short radius. Jelebu Road, where the flats sit essentially on top of the interchange, ran 27.0 points ahead. Lompang Road at 281 metres managed 6.3 points. Cashew, served by the same line but not integrated with anything, managed nothing.

Read alongside Mayflower, the Bukit Panjang result is less strange than it looks. DTL2’s alignment was announced in 2007 and confirmed in 2008 — my measurement window opens well after the market had years to price it. I was measuring the aftermath, not the event. The honest reading of Bukit Panjang is not “the Downtown Line did nothing.” It is “by the time I looked, there was nothing left to find” — which is precisely what the Mayflower timing decomposition predicts.

Test 3: Canberra, 2019

Canberra station opened on the North-South Line on 2 November 2019 — an infill station dropped into an existing line, serving a Sembawang catchment that had been walking or riding a bus to Sembawang station.

I tested whether it re-rated the town and concluded it had not. That conclusion needs the same qualification as Bukit Panjang: the station was announced in 2015 and the opening date was public from May 2019, so any capitalisation had years to happen before the window I measured.

There is a second reason Canberra is a weak test, and it is worth stating because it cuts against me. Sembawang’s flats are young — it is one of only three towns in Singapore with no block yet under sixty years of remaining lease — and the town was absorbing a very large volume of new BTO supply across the same period. Both of those things move prices independently of rail, and neither is fully controlled by a town-level comparison. I would not now cite Canberra as strong evidence in either direction. It is the weakest of my four tests and I am demoting it.

Test 4: does a station at least make your flat easier to sell?

This is the fallback argument, and it is a fair one. Maybe the station does not show up in price. Maybe it shows up in liquidity — your flat sells faster, to more people, with less hassle.

I tested that last week, across all twenty-six HDB towns, using every resale registration from January 2017 to August 2026 — 238,162 transactions — divided by each town’s sold dwelling stock to get a true turnover rate rather than a raw count.

Two things came out of it. First, the entire country collapses into a band 1.30 percentage points wide: Queenstown at the bottom on 1.99% a year, Punggol at the top on 3.29%, with twenty-three of twenty-six towns between 2.0% and 2.7%. There is no illiquid HDB town in Singapore.

Second, and relevant here: what moves turnover is town age, not rail. Turnover against the share of blocks under sixty years of lease gives a correlation of −0.659, explaining about 43% of the variation between towns. The three towns with no blocks under sixty years — Punggol, Sengkang and Sembawang — rank first, second and fifth. Rail access explains none of what is left over.

So the liquidity fallback does not rescue the story either. A new station does not measurably make a town’s flats change hands more often. That work is not yet published on this site; when it is, it will sit alongside this piece as its companion.

Where I was wrong, and what the researchers found

Here is the part most agent blogs would skip.

After running these tests I went looking for serious published research on the same question, partly to check myself. There is a real academic literature on Singapore rail and housing, most of it out of NUS, and the strongest paper in it contradicts the strong version of my earlier position directly.

Three other pieces of published work fill in the picture:

• Diao, Leonard and Sing (2017), in Regional Science and Urban Economics, studied the Circle Line’s effect on private non-landed housing using a spatial difference-in-differences design. They found +8.6% with spatial controls against 10.6% without them — meaning naive estimates overstate the effect by about 23%. They also identified the treatment zone empirically at 600 metres, and found that anticipation effects appear as early as a year before opening and diminish as the opening date approaches. Professor Sing Tien Foo directs the NUS Institute of Real Estate and Urban Studies.
• Fesselmeyer and Liu (2018), same journal, asked a cleverer question: what happens to flats that were already connected when the network expands? They found the North East Line expansion lifted already-connected flats by 1.5–2.0%, worth at least $455 million — roughly 9% of the line’s $5 billion cost. Benefits spread across the network, not just to the blocks beside the new stations.
• Tu and Wang examined Circle Line neighbourhoods with a multistage panel model and found the effect on risk premiums peaks around 24 months after opening and then normalises, with public housing less volatile than private.

Taken together, that literature tells me something specific: the effect is real, it is in the low single digits to at most around 7%, it is concentrated within roughly 600 metres, it is substantially anticipated, and it decays. My Mayflower estimate of about 6% within 360 metres sits comfortably inside that range. Where I diverge is on timing — Chia and Lim attribute the largest slice to the opening phase, while my decomposition puts essentially all of it before opening.

I do not think that is a contradiction so much as a boundary-definition problem. Phase windows are drawn differently by different researchers, and a study spanning 2000–2016 across many stations has to draw them coarsely. My window boundaries are specific to one station with a known opening date. What both of us agree on is the magnitude and the tight radius. If you want the conservative reading: somewhere between “all of it” and “most of it” is priced in before the trains run.

What the MRT effect on HDB prices actually is

Putting my four tests and the published research together, here is what I now believe, stated as precisely as I can make it.

The discount nobody puts in the brochure

There is a finding in the literature that cuts directly against the sales pitch, and I have never once seen it in an agent’s marketing.

Diao, Li, Sing and Zhan, publishing in 2023, examined 100,445 HDB resale transactions from 2011 to 2020 and measured the effect of living close to aboveground MRT tracks. Using noise-barrier installation as an exogenous shock, they found that flat prices rise by roughly 3% to 4% for every kilometre of distance away from aboveground tracks. Installing noise barriers recovered 2–3% of it.

Think about what that means in combination with everything above. The station premium is about 6% and reaches roughly 360 metres. The track penalty is 3–4% per kilometre and applies to anything near the line itself. A flat that is close to the tracks but not close to a station entrance can plausibly be carrying the penalty without the premium.

This is not hypothetical. It is exactly the profile of a great many flats along above-ground stretches of the North-South, East-West and Bukit Panjang LRT alignments. If you own one, the noise you live with is in your price. If you are buying one, “near MRT” in the listing is not automatically a plus — ask whether you are near the station or merely near the train.

Why the industry’s numbers are so much bigger than mine

When I went through the pages that currently rank for this question, the pattern was consistent enough to be worth naming.

I want to be fair about two things. One analysis I found — on a listings portal, of all places — actually ran transaction data around seven stations, five years either side, indexed to the national price index, and concluded that condos within a kilometre underperformed those one to two kilometres out. It is undated, unattributed and covers condos rather than flats, but it is honest work and it reached a conclusion close to mine. Credit where it is due; I am not the first person in Singapore to notice this.

And the PropNex piece deserves credit for at least separating announcement from opening, which is more than most manage — even as its own caveat dismantles its headline.

The deeper problem is structural. A 29% market-wide rise and a 65% headline are not two facts. They are one fact and a misattribution. Until an analysis puts a control group next to its treatment group, it cannot tell you which is which, and almost none of them do.

Who should care, and who should not

Crossing the line: CRL and JRL towns

If the mechanism I have described is right, it makes falsifiable predictions about the two lines now under construction. I am writing them down in advance so they can be checked against me later.

Jurong Region Line Stage 1 has been pushed back. LTA’s March 2026 rail factsheet confirms it will now open around mid-2028 rather than the previous end-2027 target, after construction and testing delays — viaduct works over the Pan-Island Expressway restricted to overnight closures, and heavier-than-planned ground reinforcement for a viaduct over a canal. Stages 2 and 3 have no confirmed dates at all. Anything you read that still says end-2027 is out of date.

Cross Island Line Phase 1 targets 2030, with twelve stations running from Aviation Park through Loyang, Pasir Ris, Tampines, Defu, Hougang, Serangoon North and Ang Mo Kio to a Bright Hill interchange. Phase 2 and the Punggol Extension target 2032.

My predictions, on the record:

1. Most of the CRL premium in Hougang, Pasir Ris, Tampines and Ang Mo Kio is already in the price today, four years out. The alignment has been public since 2019 and construction is visible.
2. Buyers entering those towns in 2029–2030 “before the line opens” will capture very little. They will be late, not early.
3. The gains will concentrate within about 400 metres of the confirmed station sites, and be statistically invisible beyond roughly 700 metres.
4. JRL’s delay to mid-2028 should have produced a small, temporary softening in blocks nearest the affected stations — a deferred benefit is worth less today.

I will re-run all four when the data allows. If I am wrong, I will say so on this page, the same way I have corrected myself here.

Frequently asked questions about the MRT effect on HDB prices

How much does an MRT station actually add to an HDB flat’s price?

About 6% in lease- and size-adjusted price per square foot, based on my Mayflower test, for flats within roughly 360 metres of a station exit. Published NUS research on Downtown Line 2 found up to 7% within one kilometre at the opening phase, and 1.8–3.0% at the announcement and construction phases. Anything in the 10–15% range is almost certainly a rising market being mistaken for a station effect.

Is it too late to buy near a new MRT line once it has opened?

For the purpose of capturing the premium, yes. At Mayflower, the entire measured gain landed before the station opened; the window from the run-up through to 2026 shows −0.5%, statistically zero. That does not make it a bad flat — it makes it a fairly priced one. You are buying convenience, not upside.

How close to the station do I need to be?

Closer than you have been told. My three distance rings put the significant effect at 80–360 metres straight-line, with 530–740 metres statistically indistinguishable from zero. Academic work on the Circle Line identified the treatment zone at 600 metres. Straight-line distance understates walking distance by roughly 1.2 to 1.4 times, so treat 350 metres on a map as about a 450-metre walk.

Does a flat near MRT sell faster?

Not measurably. Across all 26 HDB towns and 238,162 resale registrations from 2017 to 2026, turnover sits in a band just 1.30 percentage points wide, and what explains the differences is town age, not rail — the correlation with the share of blocks under sixty years of lease is −0.659. Nationally, about 1 in 42 flats changes hands per year, which in a 100-unit block means two or three sales in a whole year.

Is being near the MRT tracks the same as being near the station?

No, and the difference can cost you. Research on 100,445 HDB resale transactions found that prices rise roughly 3–4% per kilometre of distance away from aboveground MRT tracks — a noise penalty. Noise barriers recover 2–3% of it. A flat close to the tracks but not close to an entrance can carry the penalty without the premium.

Why did your earlier articles say the MRT had no effect?

Because I measured two corridors — Bukit Panjang’s Downtown Line and Sembawang’s Canberra station — in windows that opened well after the alignments were announced, so the pricing-in had already happened before my data started. I found nothing because there was nothing left to find, and I over-generalised from that to “no effect.” The Mayflower decomposition, which catches the run-up, corrects it. That is the honest account and it is why this article exists.

What about the Cross Island Line and Jurong Region Line?

CRL Phase 1 targets 2030, serving Loyang, Pasir Ris, Tampines, Defu, Hougang, Serangoon North and Ang Mo Kio. JRL Stage 1 has slipped to around mid-2028 from an end-2027 target, with Stages 2 and 3 having no confirmed dates. My prediction, written down in advance: most of the premium in those towns is already in today’s prices, and buyers entering shortly before opening will capture very little.

Should MRT proximity change my buying decision at all?

As a tiebreaker between two flats you would otherwise rate equally, yes — a few percent is a few percent, and daily convenience has real value that no price series captures. As a primary reason to stretch your budget or accept a worse lease, no. Remaining lease, floor area, storey and town all move price more than rail does. If you want to see how much each of those is worth in your own numbers, run them through the free calculators at listings.sg/tools before you commit to anything.

Related reading on this site

This article is part of a run of pieces where I test a widely repeated claim against transaction data and publish whatever comes out, including when it contradicts me.

• RTS Link opens Jan 2027: threat to your home’s value, or just noise? — the same question asked of a cross-border rail link. The answer there was closer to “noise” than the headlines suggested, for reasons that will now look familiar.
• HDB Resale Prices 2026 — the market-wide baseline you need before attributing anything to a station. This is the number that has to be subtracted first.
• Hougang HDB Resale 2026 — a Cross Island Line Phase 1 town, and the beachhead for million-dollar deals in non-mature estates.
• Bedok HDB Resale 2026 — a Thomson-East Coast Line town where lease decay, not rail, is the variable that moves the price.
• Million Dollar HDB Flats 2026 — where the top of the market actually is, and what drives it.
• The Lucerne Grand Review — a new launch sitting directly beside Lakeside MRT in the Jurong Lake District, which is the private-market version of this same question.

My Ang Mo Kio, Bukit Panjang, Sembawang and Marine Parade town deep-dives, and the two series pillars on lease decay by town and resale volume by town, are written and queued for publication — they carry the underlying town-level data behind several of the figures above.