Policy Impact & Market Dynamics

Analysis Date: 2026-02-06
Data Period: 2017-2026
Coverage: HDB, condominium, and EC market dynamics

Key Takeaways

The clearest finding

Cooling measures do not transmit evenly across Singapore housing segments. The condo market showed measurable suppression in prime districts, while HDB prices remained resilient and in this sample even moved higher after a major policy event.

What this means in practice

• Investors should avoid treating policy news as a market-wide signal.
• First-time HDB buyers should be cautious about delaying purchases purely on expectations of policy-driven price declines.
• Cash buyers may find selective lease-related mispricing where financing constraints distort demand.

Core Findings

1. Prime condos reacted differently from HDB

Impact

• Private and public housing should not be modeled as one policy-response market.
• Buyers need segment-specific expectations when interpreting regulation.

2. Timing matters, not just the existence of a policy

The causal work suggests that immediate level changes and post-announcement slope changes can tell different stories from simple before-and-after comparisons.

Impact

• Policy effects can look muted in a broad average while still creating short-lived buying or selling windows.
• Analysts should avoid relying only on simple period splits.

3. Lease mispricing is concentrated, not universal

Impact

• Not all short leases are bargains.
• Near-fresh leases may offer value mainly to buyers not constrained by typical financing behavior.

Decision Guide

For investors

• Separate policy-sensitive condo exposure from resilient HDB exposure.
• Use policy events to improve entry discipline, not to justify broad market calls.

For first-time buyers

• In HDB, affordability and unit fit still matter more than trying to “game” policy announcements.

For upgraders

• If selling HDB and buying condo, remember you are moving between segments with different policy sensitivity.

Technical Appendix

Data Used

• Primary input: data/parquets/L3/housing_unified.parquet
• Coverage: 2017-2026, split by property type (Condominium, HDB)
• Condo DiD sample: Condominium transactions post-2021, CCR vs OCR
• HDB RDiT sample: HDB transactions, Jun 2022 - Jun 2025 (18 months before/after policy)

Methodology

• Regional classification: CCR (10 planning areas — Bukit Timah, Downtown Core, Orchard, River Valley, etc.), RCR (12 planning areas — Bishan, Queenstown, Toa Payoh, Novena, etc.), OCR (remainder)
• Difference-in-differences (DiD) via scripts/analytics/analysis/causal/analyze_causal_did_enhanced.py:
  • Treatment: CCR condos; Control: OCR condos
  • Policy event: Sep 2022 ABSD changes for foreigners
  • Regression: price ~ treatment + post + treatment × post
  • Minimum 100 transactions per group
• Regression discontinuity in time (RDiT) via scripts/analytics/analysis/causal/analyze_rd_policy_timing.py:
  • Running variable: months since policy (negative = before, positive = after)
  • Policy event: Dec 2023 cooling measures
  • Bandwidth: ±6 months (default), robustness tested at 3, 6, 9, 12 months
  • Jump test: price ~ post; Kink test: price ~ months_since + post + post × time
  • Minimum 100 transactions for analysis
• Lease mispricing: empirical market pricing vs theoretical Bala depreciation schedule, ≥5% deviation flagged

Technical Findings

• Condo DiD: ~-$137,743 relative effect (CCR vs OCR, Sep 2022 ABSD) — prime condos meaningfully suppressed
• HDB RDiT: ~+$13,118 immediate jump after Dec 2023 measures — HDB did not show intended cooling
• Lease mispricing patterns:
  • 30-50 yr range: market appears richly priced relative to Bala theory
  • 80-84 yr range: mild undervaluation in some cases
  • 97-98 yr range: large discount relative to theoretical schedule
• Robustness: RDiT bandwidth sensitivity (3-12 months) confirms direction is stable, magnitude varies

Conclusion

The causal evidence supports segment-specific policy transmission: prime condos absorbed a measurable price suppression (~$138K) from ABSD changes, while HDB showed resilience or even upward movement (+$13K) after cooling measures. This asymmetry is robust across multiple bandwidths and timing windows. The lease mispricing analysis suggests that the market does not price strictly to theoretical Bala depreciation — shorter leases (30-50 yr) may be richly valued while near-fresh leases (97-98 yr) trade at unexpected discounts, possibly due to financing constraints distorting the buyer pool. Key limitations: DiD relies on comparability of CCR and OCR condo segments; RDiT assumes no confounding events at the cutoff; causal estimates are regime-specific and should not be treated as timeless constants.

Scripts

• scripts/analytics/analysis/causal/analyze_causal_did_enhanced.py — DiD with CCR/OCR regions
• scripts/analytics/analysis/causal/analyze_rd_policy_timing.py — RDiT around policy events