The Gold Structural demand
Some buyers do not check the price every morning. Central banks buy gold as an official reserve asset and rarely sell it. Gold ETFs let ordinary investors add or remove exposure at speed. This pillar tracks both together: the slow, structural bid from central banks, and the faster, coincident flow of ETF holdings.
- Weight
- 15%
- Effective weight
- 16.7%
- Source
- IMF IRFCL (CB) + SPDR GLD tonnes (ETF, internal)
- Frequency
- mixed
- Latest date
- 2026-08-04
- Latest value
- n/a
- Staleness
- fresh (1d / limit 45d)
- Evidence grade
- strong for central banks, weak for ETF flows
Some buyers do not check the price every morning. Central banks buy gold as an official reserve asset and rarely sell it. Gold ETFs let ordinary investors add or remove exposure at speed. This pillar tracks both together: the slow, structural bid from central banks, and the faster, coincident flow of ETF holdings.
What it measures
Central bank gold reserves, aggregated from IMF International Reserves and Foreign Currency Liquidity (IRFCL) reports, converted to a 12-month change of holdings. Plus the SPDR Gold Shares (GLD) tonnes-in-trust series, also as a 12-month change.
Why it moves gold
Central banks bought a peer-reviewed record of official gold since 2022. Arslanalp, Eichengreen, and Simpson-Bell (IMF Working Paper 23/14, later published in the Journal of International Economics) tie the acceleration to the use of financial sanctions after 2014 and 2022. ETF flows correlate strongly with contemporaneous price and are published as a coincident indicator with low weight.
Today's reading
Today the structural-demand pillar reads 44/100, near the historical median.
Sub-signal breakdown
| Sub-signal | Sub-score | Details |
|---|---|---|
| cb purchases | 36/100 | percentile: 36.17, change_12m_pct: 0.48, latest_date: 2026-06-30, prior_date: 2025-06-30, staleness_days: 36 |
| etf flows | 52/100 | percentile: 52, change_12m_pct: 5.71, latest_date: 2026-08-04, prior_date: 2025-08-04, staleness_days: 1 |
How it is scored
Each sub-signal (CB net purchases, ETF flows) is percentile-ranked in its own history and combined. Both are oriented so that positive 12-month changes push the sub-score higher.
Orientation. Rising central bank reserves and rising ETF holdings raise the sub-score. Draining reserves and outflows lower it.
Evidence
- World Gold Council, Gold Demand Trends: 2022 central bank purchases 1,081.9 tonnes, a record; 2023 revised 1,050.8 tonnes; 2024 revised to 1,092.4 tonnes; versus a 473-tonne annual average across 2010 to 2021. Fifteen consecutive years of net-buying.
- Central banks bought 863 tonnes of gold in 2025, a 21% drop from 2024's revised 1,092 tonnes and the first year below the 1,000-tonne mark after three consecutive years above it (World Gold Council, Gold Demand Trends: Q4 and Full Year 2025).
- WGC Central Bank Gold Reserves Survey 2025: 95% of respondents expect global CB gold reserves to increase in the next 12 months; 43% expect their own reserves to increase.
- Arslanalp, Eichengreen, and Simpson-Bell, IMF WP 23/14 / JIE 2023: financial sanctions predict central bank gold accumulation.
- ETF flows: the World Gold Council publishes monthly ETF flows and holdings. Flows tend to follow the gold price rather than lead it (coincident indicator).
Caveat
Evidence grade: strong for central banks (peer-reviewed sanctions channel, three years above 1,000 tonnes); weak for ETF flows (coincident with price). The pillar discloses both, weights the CB signal accordingly, and does not treat ETF flows as leading.
Seller-side note
Physical retail sellers face the local dealer bid directly. In periods of strong central-bank accumulation, dealer bids tend to sit firm and physical sell execution is smooth. When CB demand recedes, dealer bids widen. The effect on the seller is on execution friction, not on direction.
Source: International Monetary Fund, IRFCL (central bank reserves). Limited extracts from World Gold Council statistics quoted here for review and commentary; full-series republication is not published on this site. See the methodology page for the composite formula and the provenance page for the per-source status board.