Central bank gold buying and de-dollarisation:the floor underneath a long gold bull market

Three numbers first. Spot gold is quoted at USD 4,121.70/oz today; central banks worldwide hold roughly 36,000 tonnes of gold between them; and in each of 2022, 2023 and 2024 official-sector net purchases exceeded 1,000 tonnes.

That third number is the counter-intuitive one. Between 2010 and 2021 central banks bought only 400–500 tonnes a year on average, and in several of those years they were net sellers. A sudden doubling into four figures, sustained for three years, is not one country acting on a whim — it is a change in how reserves are managed. This article answers three questions: what this buying wave did change, what it did not, and what it has to do with the few ounces of gold in your account.

1. The numbers: three years of buying absorbed a mid-sized country's reserves

On World Gold Council (WGC) data, global central banks bought 1,082 tonnes net in 2022, 1,037 tonnes in 2023 and about 1,045 tonnes in 2024 — 3,164 tonnes over three years. For comparison, average annual net purchases from 2010 to 2021 were about 473 tonnes.

Basis: WGC Gold Demand Trends, central bank net purchases; mine supply estimated at about 3,600 tonnes a year. Percentages are rounded and indicative only.

How much is 3,164 tonnes? At today's price of USD 4,121.70, one tonne of gold is 32,150.7 troy ounces, worth about USD 132.5 million. So 3,164 tonnes is roughly USD 419 billion — more than the annual GDP of many countries. And 1,000 tonnes a year means taking 28% of global mine supply straight out of the market.

Now the stock. Central banks hold about 36,000 tonnes in total, worth roughly USD 4.8 trillion at current prices. The largest holders are the United States at 8,133.5 t, Germany 3,352 t, the IMF 2,814 t, Italy 2,452 t, France 2,437 t, Russia about 2,330 t and China officially about 2,280 t. Those seven account for more than half of all official reserves.

One structural difference is easy to miss: America's 8,133.5 tonnes is more than 70% of its official reserve assets, while China's 2,280 tonnes is only 5%–6% of its official reserves. The global average is around 20%. That gap is the arithmetic behind the market's long-standing expectation that China will keep buying — not a hunch, just a ratio that has room to close.

2. Why central banks buy is nothing like why you buy

Retail investors buy gold to make money; central banks buy gold so that nothing breaks. Those two motives respond to price in opposite directions: retail buys more as it rises, central banks economise as it gets expensive; retail watches quarterly returns, central banks watch solvency a decade out.

1. Gold is the only reserve asset with no counterparty

A Treasury is a liability of the US Treasury; a deposit is a liability of a bank; both depend on an institution performing. Gold is nobody's liability — it sits in a vault and needs no one to redeem it. For a reserve manager this is not a return question, it is a question of whether the asset can still be used in an extreme scenario.

2. It cannot be frozen and cannot be sanctioned

When roughly USD 300 billion of Russia's reserves were frozen in 2022, every emerging-market reserve manager had to rewrite their risk model. Assets held in someone else's custody account may legally be yours, but you do not control them; bars in your own vault are beyond the reach of a sanctions order. That is the most direct explanation for the sudden doubling of purchases in 2022.

3. It is deep enough to absorb size

The global gold market turns over in the hundreds of billions of dollars a day; a central bank buying a few dozen tonnes will not blow the price up. Some small-currency or alternative assets simply cannot absorb sovereign-sized orders. The first requirement of a reserve asset is that you can get in and get out; yield comes second.

4. No credit risk — and no interest

That is the price of gold. A Treasury pays a coupon; gold costs you storage and insurance (typically 0.05%–0.1% a year). So central banks will not convert their reserves entirely into gold — that would mean voluntarily giving up interest income and paying storage on top. The rational course is "a portion", which is why the reserve share settles near 20% rather than racing to 50%.

3. How far has de-dollarisation actually gone: three checkable indicators

The phrase "de-dollarisation" is used far too loosely. Some say the dollar collapses tomorrow; others say the whole idea is a fantasy. Both are wrong. On three verifiable indicators the conclusion is: the direction is clear, the speed is slow, and there is no substitute.

Figures are approximate ranges from public statistics, intended to show order of magnitude and direction; for exact numbers use the latest IMF COFER and SWIFT monthly releases.

All three point the same way: the dollar's share is eroding slowly, and no other currency can carry the load. The euro is limited by an incomplete fiscal union, the renminbi by the degree of capital-account openness, and gold by supply — 3,600 tonnes of mine output a year is under USD 500 billion at today's price, which cannot hold USD 12 trillion of global reserves.

So what central banks are buying is not a "replacement for the dollar" but "a layer of insurance alongside the dollar". The distinction matters: if it were replacement, the gold price would be pushed to a level nobody can calculate; because it is insurance, the upside is paced, capped and moves with the risk premium.

4. From the official bid to your account: four transmission chains

1. Supply: removing the marginal tonne

This is the hardest link. Annual mine supply is about 3,600 tonnes and recycled gold about 1,300 tonnes, so total supply is roughly 4,900 tonnes. On the demand side, jewellery takes about 2,000 tonnes, technology about 300, investment (bars, coins, ETFs) about 1,200 and central banks about 1,000. The 1,000 tonnes the official sector buys does not come back to the market; it goes into a vault and may not emerge for a decade.

Put differently: the tradeable increment falls from 4,900 tonnes to 3,900 — a 20% cut. The same private demand now has to be filled from a much smaller pool, so the price clears higher. That is why, after 2022, the same level of real yields corresponds to a gold price nearly double that of the 2018 cycle.

2. Stock: locking up the float

Unlike oil, gold is barely consumed. Of the roughly 210,000 tonnes ever mined, most sits with private holders (jewellery, bars, ETFs) and about 36,000 tonnes with central banks. The defining feature of the official holdings is that they only go in — apart from a handful of countries selling in a crisis, almost no central bank actively sells gold.

The smaller the float, the more sensitive the price is to flows. With today's numbers: 210,000 tonnes above ground, 17% locked up by central banks, roughly 45% in jewellery, and less than 30% genuinely tradeable (bars, ETFs, metal behind futures). That is why relatively modest flows move the price so far — the tradeable pool is shallower than you think.

3. Signals: shifting market expectations

Central banks are the market's smartest and slowest money. They do not read charts, they do not day-trade, and their decisions rest on a five-to-ten-year reserve strategy. When the market sees three consecutive years of four-figure official buying, it draws an inference: this group believes the monetary environment of the next decade needs gold.

That inference alone can lift prices, because it moves the private sector's anchor for what a "fair" gold price is. In 2024 gold ran from the 2,000 area to above 2,600 with no big drop in real yields — much of that was the anchor itself moving up.

4. Currency: the price of credit

The dollar price of gold is, at bottom, "how many dollars one ounce buys". When the market doubts the long-run purchasing power of the dollar — wider fiscal deficits, repeated debt-ceiling standoffs, sanctions used as a tool — the same ounce gets quoted in more dollars. Central bank buying is a symptom of that doubt, not the cause. Get the causality backwards and you end up concluding that gold collapses the moment central banks stop buying.

5. The other side: three reasons this is overrated

An article that only makes the bullish case is not worth reading. This section takes apart the claim that "central bank buying determines the gold price".

1. The data lags, and it is incomplete

IMF monthly reserve data usually lags by one to two months, and some countries do not report monthly at all. WGC quarterly figures are aggregated — by the time you read that "country X added 30 tonnes", the trade is long done. Worse, many central banks build positions in tranches through the OTC market, so the market never knows who the buyer was. Trading short-term off a two-month-lagged indicator is driving with the rear-view mirror.

2. Central banks do not chase, and buy less as prices rise

This is the most overlooked counter-mechanism. Reserve managers have cost discipline: when gold spikes they slow down or pause and wait for a pullback. During the acceleration in the second half of 2024, monthly purchase volumes did visibly slow. Official demand is therefore a cushion under the price, not the engine of the price — it supports at lows and steps back at highs.

3. The real short-term driver is still real yields and the dollar

Regress the last two decades of gold against TIPS real yields and the negative relationship is still the strongest single line. Central bank buying explains why the floor is rising; it does not explain why gold fell 3% this week. If your horizon is days to weeks, watching the dollar index and Treasury yields is far more useful than watching reserve data.

There is also a deeper constraint: gold pays nothing. The limit on official accumulation is the share at which storage plus opportunity cost becomes unbearable. The 20% global average is already close to that ceiling. Expecting central banks to put half their reserves into gold does not survive contact with fiscal logic.

6. What this means for you: how much, and with what

The direct use of all this is not forecasting the gold price, it is calibrating position size. If the most conservative money in the world keeps 20% of reserves in gold, then an individual living on dollars (or a local currency) allocating 5%–15% is reasonable, not aggressive.

First work out what "one ounce" costs you

At USD 4,121.70/oz and an FX rate of 6.6913, that is CNY 886.70/g, CNY 27,579.59/oz and CNY 886,700/kg. If you have USD 100,000 of investable assets, a 10% allocation is USD 10,000 — about 2.43 ounces, or 75.5 grams, at today's price. Get that order of magnitude straight before you argue about instruments.

Costs are typical ranges; check your broker, the ETF prospectus and the dealer's counter terms. ETF fees and spreads vary considerably by product.

The costs of these three differ by an order of magnitude, and mixing them up is the most common way to lose money: run a three-month swing trade in physical bars and a 3% premium eats half your stop; use leveraged spot for a ten-year allocation and one 20% drawdown can liquidate you.

Three compliance and tax questions you must answer yourself

1. Moving money across borders. The annual individual FX facilitation quota is the equivalent of USD 50,000, the purpose must be declared truthfully, and it may not be used for capital-account items such as overseas property or securities investment. Splitting transfers through friends' and relatives' quotas (known as "ant moving") is detected and can cost you this year's and next year's quota. For the legal routes see remitting money to buy gold: compliant routes and red lines.
2. Local tax. The US classifies physical gold and gold ETFs as collectibles, with a top long-term capital gains rate of 28%; Canada includes 50% of capital gains in taxable income; Australia gives a 50% discount after 12 months of holding; Hong Kong and Singapore levy no capital gains tax, though trading as a business may be taxed as profits. CFDs and futures have yet another tax character, and many jurisdictions treat them as ordinary income.
3. Anti-money laundering. Large deposits will require proof of source of funds. Keep payslips, tax records and property contracts for at least five years. Receiving or paying through a third party's account can trigger a freeze in either jurisdiction.

7. How to track it yourself: four data sources and their release rhythm

A simple rule of use: central bank data tells you direction, COT and ETFs tell you pace. Persistent official accumulation means a rising long-term floor; but if managed-money net longs hit a record while ETFs take in size for several weeks, short-term sentiment is crowded and chasing has poor odds. Conversely, official buying while speculative money exits is often the best window to allocate.

There is also a free real-time indicator: the live gold price and bid/ask. Today's spread is USD 0.66, or 0.016% — that number is itself a thermometer for liquidity. A widening spread usually means market makers are cutting risk, and volatility tends to follow.

8. Five common misreadings

1. "Central banks are buying, so gold must go up." The error is treating a slow variable as a fast one. Official demand sets the floor over three years, not the direction over three days. Central banks bought 1,082 tonnes in 2022 and gold first fell, then flatlined; the real rally came after late 2023.
2. "De-dollarisation means the dollar collapses." The dollar's reserve share fell from 71% to 58% over more than twenty years, and the dollar index spent that time oscillating between 90 and 115. Reserve share and exchange rate are two different things: the first is official allocation, the second is market pricing.
3. "China is secretly buying gold." China publishes official reserves monthly: about 2,280 tonnes, or 5%–6% of official reserves, far below the 20% global average. The ratio is public and needs no guessing. The market's expectation of further buying rests on exactly that gap, not on a conspiracy theory.
4. "Central banks buy bars, so physical will run out." Central banks buy 400-ounce London Good Delivery bars through wholesale and OTC channels — a completely different supply chain from the 100-gram bar in a retail shop. Retail shortages are usually local refining and logistics problems.
5. "Follow the central banks and you cannot lose." Their holding period is measured in decades, they use no leverage and they have no redemptions to meet. Mirror them with leverage and you may be liquidated in year three while they are still under water. Right direction, wrong instrument, still a loss.

One last sense of scale: over the past 12 sessions gold has traded between 4,076.23 and 4,221.53, a range of 3.6% and a net change of −1.62%. Two weeks of ordinary noise can consume a full round-trip cost on physical bars. Choose the instrument and the size first, then the direction — reverse that order and no amount of correct macro calls will save you.

9. Frequently asked questions

Related tools