Goldman Sachs, on gold, estimates central bank purchases of 79 tonnes in August, (vs pre-2022 average of 17 tonnes); China was the largest identifiable buyer in August at 25 tonnes

• Continues to see central bank demand as a structural support for gold

Broker estimates of official-sector gold buying are built on inferred flows rather than timely reporting, since the IMF data arrive with a lag and a large share of purchases historically goes unreported, meaning revisions to these tallies are routine and the identifiable buyer list understates the true total. The relevant framing is the step-change in the run-rate of central bank accumulation relative to the pre-2022 norm, a shift that desks have consistently tied to reserve diversification and sanctions-risk hedging following the freezing of reserve assets in past episodes, rather than to price-sensitive demand. That distinction matters for the transmission channel: official-sector buying is price-inelastic and recurring, so it functions as a persistent bid underneath the market that cushions drawdowns when speculative or ETF positioning unwinds, in contrast to the historical pattern where gold's marginal buyer was rate-sensitive investment demand and the metal traded tightly off real yields. China featuring as the largest identifiable buyer fits the established sequence in which reported pauses in its accumulation have tended to coincide with price consolidation before resumption. The follow-ons of interest are the official monthly reserve disclosures, whether purchases hold at elevated levels at higher prices, which would confirm inelasticity, and any widening gap between reported and estimated buying. Confidence in the structural-support thesis rests on the pattern persisting through price rallies, which it largely has.