A 0.5% Print Is Telling You the Market Cooled. A District 10 Detached House Says You’re Lying to Yourself.

I’ve spent years pricing collateral for mortgage-backed structures, and the Q2 2026 Singapore print is the kind of number that gets misread in committee. The URA headline says private residential prices rose 0.5 percent quarter on quarter, the smallest gain in seven quarters. HDB resale fell 0.3 percent to 202.7, its second straight quarterly decline, something that index hadn’t done in about seven years. First-half 2026 ran 1.4 percent against 1.8 percent in the same stretch of 2025, the smallest half-year gain since 2020. Read that on a Bloomberg terminal and you’d mark the whole book as cooling. That read is wrong, and it’s wrong in a way that costs money. Because inside the same final dataset, core central region landed prices were up 20.9 percent year on year in the first half, on 869 transactions worth roughly S$5.4 billion. One market, two stories, and only one of them made the headline.

Pull the composite apart and the mechanics are simple. The 0.5 percent is a weighted average of landed at plus 2.5, core central non-landed at plus 1.8, rest of central non-landed at minus 1.8, and outside central non-landed at minus 0.1. The declining segments carry the weight because that’s where the volume lives. Developers moved 2,141 new private units in Q2, up from 2,013 in Q1, with a 77.5 percent take-up rate on launches. Private resale ran 2,634 units, HDB resale 6,268. Healthy digestion, not collapse. But volume concentration means a 0.1 percent suburban dip and a 1.8 percent rest-of-central slide are what the average is actually averaging. The core central jump is real and moves the citywide index by a rounding error. Then look at landed on its own ledger: minus 0.4 percent in Q1, plus 2.5 percent in Q2, a 2.9 point swing in one quarter. A family closing on a District 10 detached house in June 2026 paid a price no part of that 0.5 percent headline describes. Even the flash-to-final revisions ran one direction. Every segment published in both releases was revised down, by 0.1 to 0.2 points, core central non-landed from plus 2.0 to plus 1.8, landed from plus 2.6 to plus 2.5.

Here’s how this maps to risk. The index measures price per square meter of transacted stock, not transaction value, not volume. Small denominators make violent percentages, so a 20.9 percent print on a thin, lumpy landed market is signal and noise at once. But the direction is unambiguous. Prime landed is appreciating through a period when the composite says deceleration, and the composite cannot see it because so few transactions sit in that bucket. Any buyer waiting for prime landed prices to follow the headline down is waiting on a signal the instrument doesn’t carry. If you’re holding or underwriting against Singapore residential collateral, stop marking to the composite. Segment the book, price landed separately, and treat the 0.5 percent print as what it is: an average of suburbs, quoted as if it were the city.