Singapore · Cooling measure
Singapore raises ABSD by 5 to 10 points and tightens LTV limits
MOF, MND and MAS jointly announced higher ABSD rates and tighter LTV limits on the evening of 5 July 2018, effective 6 July.
Announced 5 July 2018 · Effective 6 July 2018
This article has been corrected
3 material corrections have been made. The full record is at the end of this article.
The 30-second brief
On the evening of 5 July 2018, hours before it took effect, Singapore raised Additional Buyer's Stamp Duty by 5 points for most individuals and 10 for entities, introduced a 5% non-remittable levy on housing developers, and cut financial-institution loan-to-value limits by 5 points across the tiers. Developers who could accelerate a launch sold roughly 1,000 units in under five hours before midnight. The rapid price recovery stopped within a quarter: +3.4% in Q2 2018 became +0.47% in Q3. Beneath the aggregate, 2018 produced a clean prime-to-mass gradient the headline index hides entirely, with the Core Central Region weakest and the only segment to decline in 2019. But the collapse in primary-market absorption that followed was a supply story: developer sales of uncompleted units ROSE 19.8% while launches doubled, driven by the en-bloc replacement pipeline and developers' own ABSD deadlines.
Key numbers
What happened
MOF, MND and MAS jointly announced higher ABSD rates and tighter LTV limits on the evening of 5 July 2018, effective 6 July. ABSD rose 5 percentage points for individuals other than citizens and permanent residents buying a first property, and 10 points for entities. Housing developers faced the 25% entity rate, potentially remittable on conditions, plus a new 5% non-remittable component. FI housing-loan LTV limits fell 5 points across the tiers, and mortgage equity withdrawal limits fell 5 or 15 points depending on existing exposure.
What changed
ABSD
| Condition | Before | After |
|---|---|---|
| SC1st | 0% | 0% |
| SC2nd | 7% | 12% |
| SC3rd+ | 10% | 15% |
| SPR1st | 5% | 5% |
| SPR2nd | 10% | 15% |
| SPR3rd+ | 10% | 15% |
| FOREIGNER | 15% | 20% |
| ENTITY | 15% | 25% |
| remittableHOUSING DEVELOPER | n/a | 25% |
| non remittableHOUSING DEVELOPER | n/a | 5% |
0% → 0%
7% → 12%
10% → 15%
5% → 5%
10% → 15%
10% → 15%
15% → 20%
15% → 25%
n/a → 25%
n/a → 5%
FI LTV
| Condition | Before | After |
|---|---|---|
| standard0 | 80% | 75% |
| tenure or age linked0 | 60% | 55% |
| standard1 | 50% | 45% |
| tenure or age linked1 | 30% | 25% |
| standard2+ | 40% | 35% |
| tenure or age linked2+ | 20% | 15% |
| non individual | 20% | 15% |
80% → 75%
60% → 55%
50% → 45%
30% → 25%
40% → 35%
20% → 15%
20% → 15%
MWL
| Condition | Before | After |
|---|---|---|
| no | 80% | 75% |
| yes | 60% | 45% |
80% → 75%
60% → 45%
MIN CASH DOWNPAYMENT
| Condition | Before | After |
|---|---|---|
| 75 no outstanding loan | n/a | 5% |
| 55 no outstanding loan | n/a | 10% |
| one or more outstanding loans | n/a | 25% |
n/a → 5%
n/a → 10%
n/a → 25%
Multi-instrument package. The developer ABSD rows carry null from because the 25% remittable plus 5% non-remittable structure was INTRODUCED here; it later became 35% plus 5% on 16 December 2021 and was UNCHANGED on 27 April 2023 even though the ordinary entity rate rose to 65%. MWL limits were NOT tightened uniformly by 5 points: 5 points where the borrower had no outstanding housing loan on another residential property, 15 points where they did. The lower tenure/age-linked LTV tiers applied where the loan tenure exceeded 30 years for private residential property, exceeded 25 years where the property was an HDB flat, or the loan period extended beyond the borrower age of 65.
Why it matters
The event is the archive's clearest case of a volume ratio moving for supply reasons rather than demand reasons. It also fixes the announcement-window taxonomy: hours of notice produced a compressed same-evening rush plus grandfathering of existing Options to Purchase, which is a different pattern from Thailand's six-month notice and from an overnight measure with no behavioural opportunity at all.
We've seen this before
28 June 2013 · 29 June 2013
Why relevant
Ancestor. The July 2018 package responded to the recovery that ended the TDSR-era downturn: the "close to four years" of gradual decline the Government cited is the 15-quarter fall recorded in evt_sg_2013_tdsr_jun, Q4 2013 to Q2 2017.
Where the comparison breaks
- mechanism: transaction tax plus collateral leverage vs credit-service capacity
- 2013 introduced a framework; 2018 moved parameters within existing ABSD and LTV regimes
- 2013 initiated a 15-quarter decline; 2018 stopped a recovery within one quarter without producing a deep nominal decline
- 2013 applied broadly across debt-funded purchasers; 2018 differentiated sharply by buyer class
- 2018's largest single parameter move, the 15-point MWL tightening, has no 2013 analogue
What happened after
The 2013 outcome profile shows what a framework introduction does to the price cycle. July 2018 shows what parameter escalation does when a recovery is already under way: it truncates the recovery rather than reversing the market.
29 September 2022 · 30 September 2022
Why relevant
Forward direction. When the July 2018 event is read, September 2022 is the later package that tests its central segment finding: CCR was again the weakest price segment and the weakest rental segment under a package whose four instruments were overwhelmingly HDB-directed and which barely touched the private market. That is why this event's falsifier now carries structural CCR underperformance as a standing competing explanation.
Where the comparison breaks
- July 2018 was private-market-directed; September 2022 was overwhelmingly HDB-directed
- July 2018 used a transaction tax; September 2022 used an eligibility bar and interest-rate assumptions
- July 2018 gave hours of notice with OTP grandfathering; September 2022 gave one day
- the September 2022 CCR result arose under a package that barely touched the private market, which is precisely why it functions as a control on the 2018 reading
What happened after
Two structurally different measures four years apart produced CCR underperformance. Taken with the 2019 and 2020 CCR results, that makes structural segment weakness a competing explanation any future CCR attribution must argue against, rather than an effect of either instrument.
Not a precedent for: Introduction of a new macroprudential framework; Prohibitive foreign-buyer tax rates; HDB-specific financing interventions; Measures with a long announcement-to-effective gap; Any clean treated-versus-untreated HDB comparison
What happened next
calendar 2018 and calendar 2019, with quarterly detail Q3 2018 to Q4 2019
Private residential price trajectory
CALENDAR · calendar 2018 and calendar 2019, with quarterly detail Q3 2018 to Q4 2019Private residential, nationalCalendar-year Price Property Index change: 2018 up 7.86%, 2019 up 2.67%. Quarterly quarter-on-quarter, Q3 2018 through Q4 2019: +0.47, -0.07, -0.67, +1.48, +1.33, +0.52. Reconciled against audit figures of +7.9 and +2.7 and +0.5, -0.1, -0.7, +1.5, +1.3, +0.5, all deltas within 0.04.
Interpretation
Rapid pre-measure appreciation stopped immediately. Prices softened in Q4 2018 and Q1 2019, then resumed increasing from Q2 2019. No prolonged or deep nominal decline. An arithmetic note worth stating: 2018's 7.86% was almost entirely earned in H1, BEFORE the measure, at +3.9% in Q1 and +3.4% in Q2, followed by +0.47% and -0.07%. The measure did not reduce 2018's figure; on this reading it constrained what the figure would otherwise have become. The counterfactual is unobservable.
Why this grade
MEDIUM for the initial slowdown in price growth. Timing is consistent, but prices are sticky and the index is affected by transaction and project mix. The 2019 trajectory is MEDIUM at most: supply, project mix, interest rates and broader economic conditions materially confound it.
Q3 2018 - Q2 2019 against Q3 2017 - Q2 2018
Primary market absorption, like-for-like on uncompleted units
CALENDAR · Q3 2018 - Q2 2019 against Q3 2017 - Q2 2018Private residential primary market, all segments, uncompleted unitsUnits sold, uncompleted only: 7,349 to 8,801, up 19.8%. Units launched: 5,418 to 10,902, up 101.2%. Absorption ratio 135.6% to 80.7%. Outcome-window quarterly detail for sold uncompleted: Q3 2018 2,910, Q4 2018 1,793, Q1 2019 1,797, Q2 2019 2,301. For launched: Q3 2018 3,754, Q4 2018 1,657, Q1 2019 2,989, Q2 2019 2,502. Developer sales of uncompleted units rose 19.8% in the four calendar quarters containing the post-event period, while units launched rose 101.2%, reducing the absorption ratio from 135.6% to 80.7%. The ratio decline reflects a doubling of released supply rather than a contraction in sales.
Interpretation
This REVERSES the intuitive reading. In the baseline year sales exceeded launches, meaning the market was clearing standing stock faster than developers were releasing it. DO NOT STORE: "the measures caused developer sales to fall." AUDIT DISAGREEMENT ON RECORD, retained deliberately: the audit reported approximately 8,760 sold and 9,769 launched for Q3 2018 to Q2 2019, a ratio of 89.7%. The Prevo metric store gives 9,036 and 10,902 on the headline pairing and 8,801 and 10,902 on the like-for-like pairing. The audit's own calendar-2018 and calendar-2019 figures reconcile exactly with the metric store, which pins the constituent quarters, so its four-quarter window figures are internally inconsistent with its annual figures. The computed values are preferred. HEADLINE-SERIES NOTE: URA's headline developer sales series, uncompleted plus completed, gives 8,795 for calendar 2018 and 9,912 for 2019, both reconciled against audit with delta 0. Those figures are correct for citation but are not used here, because launches are uncompleted-only and a like-for-like ratio requires like-for-like inputs.
Why this grade
LOW-MEDIUM for the 2019 developer-sales trajectory. Launch volume was a major determinant, and the launch surge itself was driven by the 2017-18 en-bloc replacement pipeline and developers' five-year ABSD deadlines, both independent of this measure.
calendar 2018, 2019 and 2020, Q4 over Q4
Segment divergence, calendar-year (the evidentiary framing)
CALENDAR · calendar 2018, 2019 and 2020, Q4 over Q4Private residential non-landed, by market segmentCalendar-year change, Q4 over Q4. 2018: CCR +6.73%, RCR +7.42%, OCR +9.35%, all residential +7.86%. 2019: CCR -1.69%, RCR +2.82%, OCR +4.22%, all residential +2.67%. 2020: CCR -0.37%, RCR +4.69%, OCR +3.20%, all residential +2.21%. Index levels for CCR: 136.0 at Q2 2018, trough 132.3 at Q1 2019, partial recovery to 134.1 by Q4 2019, still 133.6 at Q4 2020, below its pre-measure level two and a half years later. Over the same span RCR rose 148.5 to 160.6 and OCR 169.7 to 183.6.
Interpretation
In 2018, the year the measure landed, segment performance formed a clean monotonic gradient with prime weakest and mass market strongest: OCR +9.35% above RCR +7.42% above CCR +6.73%. The aggregate index at +7.86% conceals this ordering entirely. CCR was the only segment to decline in 2019, and it declined again in 2020. DATA HANDLING RULE: CCR is the thinnest segment and its index is correspondingly noisy, with single-quarter moves of plus or minus 3% appearing repeatedly. Annual and point-to-point figures are the defensible ones; do not quote any single CCR quarter as evidence.
Why this grade
MEDIUM-HIGH as a CEILING, not a floor. It rests on exactly two things, stated because a third was withdrawn: (a) the 2018 monotonic gradient, OCR above RCR above CCR, which is the ordering a buyer-class measure predicts, falls inside the valid attribution window, and is invisible in the aggregate index; and (b) the 2019 divergence, CCR the only decliner at -1.69%, by a wider margin but further from the event and correspondingly more confounded. Neither is decisive alone; together they support MEDIUM-HIGH. It does NOT rest on the four-quarter point-to-point comparison, which was computed, found to flip sign on baseline choice, and discarded as evidence; see the rejected_inferences row on this event. CAPPED at MEDIUM-HIGH because CCR underperformance persisted into 2020, when attribution to this measure is void, indicating causes independent of it.
The 2020 column is recorded for trajectory only. Attribution to this measure is void from approximately March 2020, when COVID-19 becomes the dominant confounder. See the outcome_windows ladder on this event.
not applicable: the comparison is invalid rather than unmeasured
HDB resale market as a comparison group
CALENDAR · not applicable: the comparison is invalid rather than unmeasuredHDB resale, nationalInterpretation
HDB resale is a PARTIAL AND CONTAMINATED comparison, NOT a clean natural control. The July 2018 LTV tightening did not affect loans granted directly by HDB, but it DID affect covered financial-institution loans used to purchase HDB flats. HDB resale was also exposed to ABSD depending on buyer profile, and to different eligibility, ownership, financing and supply conditions. Any clean treated-versus-untreated HDB design is invalid for this event.
Why this grade
Not assessable for this event on available data.
Q3 2018 to Q3 2019, point to point
Segment divergence, anchor-derived window (RECORDED, NOT EVIDENTIARY)
CALENDAR · Q3 2018 to Q3 2019, point to pointPrivate residential non-landed, by market segmentCCR non-landed 137.8 to 138.0, up 0.15%. RCR non-landed 146.6 to 155.4, up 6.00%. OCR non-landed 169.5 to 173.1, up 2.12%.
Interpretation
This is the machine-generated view of segment performance and it is a DIFFERENT window from the calendar-year table in outcome 3, which is the evidentiary framing. It is recorded so that anyone who computes it finds it already here, with the reason it is not used. Its sign is sensitive to baseline choice: on a Q2 2018 baseline the same CCR comparison gives 136.0 to 135.3, down 0.51%. See the rejected_inferences row on this event.
Why this grade
No causal weight is carried by this framing. Its sign turns on which quarter is chosen as the baseline, because CCR moved +1.32% in Q3 2018, lifting the baseline, and +2.00% in Q3 2019, lifting the endpoint. A figure whose sign turns on a single quarter of a thin index is an artifact of the window. The MEDIUM-HIGH grade on CCR relative underperformance rests on outcome 3 and not on this row.
+0d, the evening of 5 July 2018
Same-evening transaction rush
ELAPSED · +0d, the evening of 5 July 2018 · FROM ANNOUNCEMENTPrivate residential primary market, three named projectsApproximately 1,000 units were reportedly sold across Park Colonial, Riverfront Residences and Stirling Residences in under five hours on 5 July 2018, before the midnight commencement. Contemporaneous reports give Riverfront Residences more than 500, Stirling Residences about 200 and Park Colonial about 300. The figures were preliminary and based partly on developer, agency and industry-source reports.
Interpretation
CLASSIFICATION: ULTRA-SHORT ANNOUNCEMENT WINDOW with SAME-DAY PULL-FORWARD and PRE-EXISTING-OTP GRANDFATHERING. Not a general three-week anticipation period. This is a third distinct announcement-window pattern in the archive, alongside Thailand April 2019 (six-month notice, large multi-month pull-forward, measurable payback) and Singapore December 2021 and April 2023 (overnight, minimal behavioural opportunity).
Why this grade
Directly triggered by the announced midnight commencement: developers capable of accelerating launches could issue Options to Purchase before the deadline, and did. The mechanism is not in doubt. The magnitude is TIER 2 sourcing, well reported but not project-level verified, so the figure carries approximation and the grade attaches to the occurrence rather than to the exact count. Separately, the immediate post-measure transaction disruption is graded MEDIUM-HIGH in the canonical file: timing is strong, but comparisons are distorted by transactions pulled into 5 July, and no figures are stated for it, so it has no row of its own.
The case for
The price recovery stopped within one quarter and the 2018 segment ordering is exactly what a buyer-class measure predicts, monotonic from mass market to prime, and invisible in the aggregate index. The same-evening rush is directly attributable to the announced midnight commencement, graded HIGH.
The case against
The absorption collapse supports a supply-timing explanation over a demand one. Developer sales of uncompleted units ROSE 19.8%; the ratio fell because launches more than doubled, driven by the en-bloc replacement pipeline and five-year ABSD deadlines, both independent of this measure. CCR underperformance persisted into 2020 and recurred under the September 2022 package, which barely touched the private market, so a structural explanation now competes on three instances.
What this view assumes
- the LTV exemption follows the LENDER, not the property type; an HDB flat bought with a bank loan was inside the measure
- quarterly aggregations are labelled as calendar quarters containing the post-event period, never as elapsed 12-month windows
- the ABSD grandfathering protected existing OTPs and did not give the market a three-week anticipation period
- CCR is the thinnest segment; single quarters are never evidence
- the summary phrase "at the peak of the 2017-18 collective sale cycle" is a MARKET-CONDITIONS CHARACTERISATION, not a sourced figure, and belongs to the analyst layer this event keeps separate from its factual layer. It is retained after the 2017 collective-sale aggregate was dropped as DISPUTED on definition, and it is now BETTER supported than when written: the URA figure behind the event, approximately 11,100 potential units from awarded collective-sale sites without planning approval within approximately 19,500 including GLS sites, is primary evidence of an intense cycle, where the dropped 28-sites aggregate was TIER 2 with an unstable definition. Do not read the phrase as resting on the dropped figure, and do not remove it as unsourced: it rests on the replacement.
What we don't know
- the counterfactual price path without the measure is unobservable
- the decomposition of the 2019 sales figure between demand response and launch supply is not recoverable
- whether CCR's 2018 underperformance is the measure or the tightening of China's outbound capital controls from 2017
Prevo view
InterpretationSingapore's July 2018 package stopped a rapid price recovery within one quarter and produced a clean prime-to-mass gradient invisible in the aggregate index, with CCR the weakest segment in 2018 and the only one to decline in 2019; but the collapse in primary-market absorption that followed was driven by a doubling of released supply from the en-bloc replacement pipeline rather than by falling demand, since developer sales of uncompleted units rose 19.8% over the same period.
Confidence: MEDIUM-HIGH
What would change this view: If developer sales and price movements during 2019 are better explained by launch timing, project mix and the en-bloc replacement pipeline than by distance from the 6 July 2018 policy event, the event's contribution to the medium-term trajectory should be downgraded. The immediate same-night and short-term transaction effects may remain strongly attributable even if medium-term demand-side attribution is weakened. The computed outcomes trigger this falsifier in one direction and resist it in another: the absorption outcome supports a supply-timing explanation over a demand one, downgrading the medium-term demand-side reading to LOW-MEDIUM, while the segment outcome pushes the other way, because the 2018 prime-to-mass gradient is what a buyer-class measure should produce and is invisible in the aggregate. A further test would separate the two: if CCR's 2018 underperformance is matched by comparable prime-market weakness in jurisdictions with no equivalent measure over the same period, the segment evidence weakens and structural explanations, including capital-control tightening, should be preferred. AMENDMENT, from evt_sg_2022_cooling_sep: that competing structural explanation is now three instances strong, not one. CCR was ALSO the only declining price segment in the September 2022 window, down 1.32% against RCR up 7.20% and OCR up 5.98%, and the weakest rental segment over the same period, up 14.47% against RCR up 18.43% and OCR up 19.61%, under a package whose four instruments were overwhelmingly HDB-directed and which barely touched the private market. A segment that underperforms under a private-market transaction-tax measure in 2018 and again under an HDB eligibility package in 2022 is displaying a structural feature, not the effect of either instrument. The MEDIUM-HIGH grade is RETAINED, because the 2018 evidence is a monotonic gradient across all three segments rather than CCR alone, and that gradient remains what a buyer-class measure predicts. But structural CCR underperformance is now a standing competing explanation that any future CCR attribution must carry and argue against.
Evidence check
15
Material claims
14
Primary confirmed
1
Corroborated
12
Qualified
0
Government estimates
View evidence report
| # | Claim | Type | Class | Verification | Evidence location |
|---|---|---|---|---|---|
| 1 | Additional Buyer's Stamp Duty was raised by 5 percentage points for individuals, excluding Singapore Citizens and Singapore Permanent Residents purchasing their first residential property. | Number | Observed fact | VERIFIED PRIMARY | MOF/MND/MAS joint media release, 5 July 2018; IRAS historical ABSD rate schedule |
| 2 | Additional Buyer's Stamp Duty for entities was raised by 10 percentage points, from 15% to 25%. | Number | Observed fact | VERIFIED PRIMARY | MOF/MND/MAS joint media release, 5 July 2018; IRAS historical ABSD rate schedule |
| 3 | An additional 5% non-remittable Additional Buyer's Stamp Duty was introduced for housing developers, on top of the 25% entity rate. | Number | Observed fact | VERIFIED PRIMARY | MOF/MND/MAS joint media release, 5 July 2018; Stamp Duties (Housing Developers) (Remission of ABSD) Rules |
| 4 | Foreigner Additional Buyer's Stamp Duty was raised from 15% to 20%. | Number | Observed fact | VERIFIED PRIMARY | IRAS historical ABSD rate schedule |
| 5 | Singapore Permanent Resident rates for a second residential property and for a third or subsequent residential property were both 10% before the change and both rose to 15%. | Number | Observed fact | VERIFIED PRIMARY | IRAS historical ABSD rate schedule |
| 6 | Loan-to-value limits for financial-institution housing loans were tightened by 5 percentage points across the applicable tiers; loans granted by HDB were unaffected. | Number | Observed fact | VERIFIED PRIMARY | MOF/MND/MAS joint media release, 5 July 2018 |
| 7 | The Mortgage Equity Withdrawal Loan limit for a borrower with an outstanding housing loan on another residential property fell from 60% to 45%. | Number | Observed fact | VERIFIED PRIMARY | MOF/MND/MAS joint media release, 5 July 2018 |
| 8 | The measures took effect on 6 July 2018. | Date | Observed fact | VERIFIED PRIMARY | MOF/MND/MAS joint media release, 5 July 2018 |
| 9 | The Government stated that private residential prices had declined gradually for close to four years, began rising in Q3 2017, and had increased sharply by 9.1% over the preceding year. | Policy characterisation | Policy characterisation | VERIFIED PRIMARY | MOF/MND/MAS joint media release, 5 July 2018 |
| 10 | Under the ABSD transitional provision, the previous rates applied where the OTP was granted on or before 5 July 2018, was not varied on or after 6 July 2018, and was exercised on or before 26 July 2018 or within its original validity period, whichever was earlier. | Date | Policy characterisation | VERIFIED PRIMARY | MOF/MND/MAS joint media release, 5 July 2018 |
| 11 | The Property Price Index for all private residential property rose from 144.1 in Q1 2018 to 149.0 in Q2 2018, an increase of 3.4% quarter on quarter. | Market metric | Observed fact | VERIFIED PRIMARY | URA Q2 2018 private residential statistics, Annex B-1 |
| 12 | Developers sold 7,972 private residential units excluding executive condominiums in 2016 and 10,566 in 2017. | Market metric | Observed fact | VERIFIED PRIMARY | URA Q2 2018 private residential statistics, Annex B-1; URA Q4 2017 final statistics |
| 13 | Approximately 1,000 units were reportedly sold across Park Colonial, Riverfront Residences and Stirling Residences in under five hours on 5 July 2018. The figures were preliminary and based partly on developer, agency and industry-source reports. (Unnumbered in audit; the distinctive-finding figure, stored so it is not prose-only.) | Market metric | Observed fact | VERIFIED MULTIPLE SECONDARY | Contemporaneous market reporting of the 5 July 2018 evening launches |
| 14 | Unsold private residential units with planning approval excluding executive condominiums stood at 23,514 at end-Q1 2018 and 26,943 at end-Q2 2018. (Unnumbered in audit; the corrected pre-event supply figures.) | Market metric | Observed fact | VERIFIED PRIMARY | URA Q2 2018 private residential statistics, Annex B-1; Prevo metric store ura_unsold_with_approvals_all_uncompleted |
| 15 | At end-Q2 2018, awarded collective-sale sites without planning approval represented approximately 11,100 potential private residential units. Together with Government Land Sales sites without planning approval, the potential supply was approximately 19,500 units. (Unnumbered in audit; replaced the 2017 collective-sale transaction tally on the founder pass.) | Market metric | Observed fact | VERIFIED PRIMARY | URA Q2 2018 private residential statistics, supply pipeline |
Claim → Evidence → Source
Sources
Primary sources
- MOF/MND/MAS joint media release: Raising Additional Buyer's Stamp Duty Rates and Tightening Loan-to-Value Limits to Promote a Stable and Sustainable Property Market
Monetary Authority of Singapore · 5 July 2018
- URA quarterly private residential property statistics (series)
Urban Redevelopment Authority
- Stamp Duties (Housing Developers) (Remission of ABSD) Rules and Stamp Duties (Non-licensed Housing Developers) (Remission of ABSD) Rules
Ministry of Finance
- IRAS historical ABSD rate schedule
Inland Revenue Authority of Singapore
Corrections
13 August 2026
Original: Developers sold 7,972 private residential units excluding executive condominiums in 2016 and 10,682 in 2017.
Corrected: Developers sold 7,972 private residential units excluding executive condominiums in 2016 and 10,566 in 2017.
THE ARCHIVE MADE THIS ERROR WHILE WARNING ABOUT ITS OWN ERROR CLASS. 10,682 was URA's preliminary excluding-EC estimate from the monthly-sales release of 15 January 2018; the final Q4 statistics establish 10,566. The claim's own qualification rejected 14,707 from that same release as preliminary, and adopted 10,682 in the same sentence. Rejecting one preliminary figure produced enough confidence not to check the other. The 14,707 reference is correct and remains, because it is explicitly labelled preliminary. Found in the founder primary-source pass.
13 August 2026
Original: Approximately 28 residential collective sale sites transacted for approximately S$8.7 billion in 2017, against approximately three sites for about S$1 billion in 2016.
Corrected: At end-Q2 2018, awarded collective-sale sites without planning approval represented approximately 11,100 potential private residential units. Together with Government Land Sales sites without planning approval, the potential supply was approximately 19,500 units.
DISPUTED on definition, so replaced rather than repaired. Published tallies conflict because they count different property types, transaction stages and cut-off dates: approximately 26 deals at S$8.5 billion in some, approximately 28 at S$8.7 billion in others. The claim never stated which universe it used, and "approximately" does not cure a definition problem. The substitute is a URA figure measuring the supply pressure that actually faced policymakers, which is what the claim was doing analytical work for. Found in the founder primary-source pass.
13 August 2026
Original: Under the ABSD transitional provision, qualifying Options to Purchase remained exercisable at the previous rates until 26 July 2018.
Corrected: Under the ABSD transitional provision, the previous rates applied where the OTP was granted on or before 5 July 2018, was not varied on or after 6 July 2018, and was exercised on or before 26 July 2018 or within its original validity period, whichever was earlier.
The claim row dropped the earlier-expiry condition that the canonical file states correctly in its policy layer. As stored, it read as an entitlement to 26 July for every qualifying OTP; in fact an OTP whose own validity expired sooner expired sooner. The database had diverged from the canonical text, which is the divergence co-location checks exist to catch. Found in the founder primary-source pass.
Prevo provides research and informational analysis only. It is not a broker, investment adviser or fiduciary, and nothing on this site constitutes investment, legal, tax or financial advice. Verify independently.