Central Bank Gold Buying: What It Actually Does to Regional Premiums

The Number That Changed Everything, Almost

Central bank net gold purchases averaged roughly 1,050 tonnes per year from 2022 through 2025 — more than double the average annual pace of the prior decade. That is a large number. It has been cited to explain gold's move from around $1,830 at the start of 2022 to over $4,000 by mid-2026, and it has been used to argue that regional premiums should have expanded to reflect central bank demand pressure on physical supply.

The first argument — that central bank buying supports the spot price — is largely correct. The second — that central bank buying visibly widens regional premiums — turns out to be mostly wrong, for reasons that are worth working through. This piece walks through what central banks actually do when they buy, why it shows up in some price signals and not others, and what a premium-focused reader should and should not conclude from the 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

Given that reported central bank buying is around 1,000 tonnes per year — a bit under 3 tonnes per day on average — and given that the global daily turnover across LBMA, COMEX, and SGE is well above 5,000 tonnes of paper-and-physical combined, central bank flow is a small fraction of daily turnover. 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.

The Chinese Case in Particular

The People's Bank of China reported net gold purchases in almost every month from November 2022 through mid-2025. During the same window, the SGE premium against COMEX behaved unusually — it stayed positive for much longer stretches than in the 2015-2021 baseline, averaging +1.5% rather than the historical +0.5%. 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. A shift from +0.5% to +1.5% is one percentage point. 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

Our dashboard displays daily premiums for COMEX, 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.