⚠️ Correlation, not causation
This piece walks through a period in which two things — central bank purchases and regional gold premiums — both moved. Two variables moving together is not proof that one caused the other. Central bank reserves reporting is also lagged (usually one to three months) while premiums are daily, so any short-window overlap is easy to misread. The framework below is meant to help distinguish what the data can support from what it cannot.
The Number That Changed Everything, Almost
Central bank net gold purchases have been running well above the pre-2022 average for several years, according to World Gold Council and IMF data — averaging roughly a thousand tonnes a year across 2022–2025 by the World Gold Council's estimate.
That is a large number.
It is often cited both to explain gold's spot-price move over the period and to argue that regional premiums should have widened along with it. The first argument — that central bank buying is one of the supports for the spot price — is broadly consistent with the data. The second — that central bank buying visibly widens regional premiums — turns out to be much weaker than it looks, for the reasons this piece works through.
The rest of the piece walks through what central banks actually do when they buy, why the flow shows up in some price signals and not others, and what a premium-focused reader should and should not conclude from the reported flow data.
How Central Banks Actually Acquire Gold
A central bank does not walk into a bullion dealer with a bag of dollars.
There are essentially three channels through which reserve gold is acquired.
Domestic production purchase
The largest source for the largest buyers.
The People's Bank of China, the Reserve Bank of India, and the Central Bank of Russia have all bought significant quantities directly from domestic mine production.
Domestic producers sell into the sovereign at pre-agreed formulas that reference international benchmarks but settle in local currency.
This flow never touches SGE, MCX, or LBMA in the visible way people expect.
It is a bilateral relationship between a state buyer and state-connected miners.
Over-the-counter London market purchase
The traditional channel for European and Middle Eastern central banks.
A reserve manager instructs a bullion bank — typically JPMorgan, HSBC, UBS, or ICBC Standard — to accumulate a target quantity in the London OTC market.
The bank aggregates the purchases across multiple counterparties, often over weeks or months, and delivers a set of Good Delivery bars into the buyer's Bank of England allocated account.
This does touch LBMA supply, but the aggregation window is long enough that it does not usually generate visible premium spikes.
Bilateral swap or purchase from another sovereign
Less common but material at the margin.
Central banks occasionally acquire gold from each other, or from the IMF, in bilateral transactions that never touch a public market at all.
Why the Flow Rarely Widens Visible Premiums
Even in years of elevated reserve manager buying, the aggregate reported purchases are modest compared with global daily paper-and-physical turnover across LBMA, COMEX, and SGE. Reserve manager flow is a small fraction of the daily activity in these markets.
Even the largest single-day central bank transactions are typically well under 10 tonnes, an amount that the London OTC market absorbs without visible premium impact.
The flow is also deliberately smooth.
Reserve managers do not want to move the market against themselves.
A central bank that intends to acquire 100 tonnes over a year will typically split that across the year rather than execute in a single week.
This intentional smoothing means the flow never concentrates enough at any moment to overwhelm normal market-making capacity.
Where the Flow Does Show Up
Three effects of the central bank flow do register in market data, and one of them shows up in premium data specifically.
📊 What central bank flow actually does
- Chronic price support. Sustained reserve manager buying removes physical metal from private-sector float. This tends to appear as a slow-moving background tailwind for the dollar price of gold rather than as sudden spikes.
- ETF-flow decoupling. In the pre-2022 era, gold's price tracked developed-market ETF flows tightly. The 2022-2025 period broke this correlation: gold rallied even as GLD outflows continued for stretches. Central bank buying is the most-cited explanation, and it fits the data.
- Occasional London-versus-Shanghai divergence. When Asian central banks accelerate purchases and take delivery via SGE-adjacent channels rather than through London, LBMA to COMEX to SGE premiums can shift for weeks. This is the one place where central bank flow shows up as a premium signal — but only when the destination bank explicitly wants Asian delivery.
The Chinese Case in Particular
The People's Bank of China reported gold purchases during parts of this period. SGE prices were also relatively firm in several stretches, but the public disclosures and the site's market series are not sufficient to estimate a causal effect.
This is the strongest single piece of evidence in the recent data that a central bank flow can shift a regional premium regime.
But even here, the effect is modest.
The size of any apparent shift depends on the selected dates, benchmark, exchange rate, and missing-data treatment.
It has persisted for years, which is itself informative, but it is not the kind of dramatic premium spike that headline stories about central bank demand might lead a reader to expect.
What This Site Shows and Doesn't Show
The dashboard displays daily premiums against the USD spot benchmark for SGE, MCX, and LBMA.
Central bank flow is not directly visible in any of these numbers.
What is visible is the second-order effect: a persistent shift in the SGE premium regime that started around late 2022 and has continued into 2026.
We do not attempt to attribute the SGE regime shift specifically to PBOC buying.
There are other candidate explanations — Chinese retail demand strength, capital controls tightening, changes in domestic gold refinery output — and the honest answer is that separating these causes from public data is very difficult.
What we can say is that the shift exists in the data, and that it correlates in time with the PBOC's disclosed buying acceleration.
What a Reader Should Not Do With This
The temptation, given the above, is to use central bank purchase announcements as a signal for gold price direction or premium direction.
This does not work well in practice.
Central bank purchase data is reported with a lag of one to three months.
By the time it is public, the market has already had the flow priced in through OTC counterparties who saw the accumulation happen.
The idea that a private investor reading the World Gold Council's quarterly report is going to front-run a trade based on that data is not supported by the historical record.
The more useful frame is: central bank buying is a slow, chronic tailwind for the price and a smaller chronic tailwind for Asian regional premiums.
It is not a trade signal.
Anyone using it as one should be prepared to be wrong for extended periods.
📌 Key Takeaways
- Central bank net purchases were elevated relative to earlier years, but exact annual figures should be verified against the World Gold Council's reports for the years in question.
- Most of this flow is executed via London OTC or domestic production purchases, in smoothed accumulation programs that do not spike visible premiums.
- SGE pricing was relatively firm during parts of the buying period, but the overlap should be treated as a question for further analysis rather than evidence of causation.
- Central bank flow provides chronic support for the dollar gold price but does not work as a short-term trade signal because the data is reported with a lag.
- Attribution of the SGE regime shift specifically to PBOC buying is a correlation, not proof of causation — other explanations remain in play.
Related Posts
Why We Don't Smooth the Premium Series
A short editorial note on why our daily premium chart shows every wobble rather than a nicer-looking moving average.
Central Bank Gold Holdings by Country: Who Owns What, and What the Trends Show
A tour through the top central bank gold holders as of mid-2026, the buyers and sellers of the past decade, and what these numbers do and do not tell you about future prices.
Gold and Inflation: What Fifty Years of Data Actually Shows
The "gold is an inflation hedge" story is repeated so often it feels like established fact. The historical record is more nuanced. Here is what the data actually says, and how to use it.