Why Central Banks Are Buying Record Amounts of Gold

Gold has spent the better part of three years doing something it rarely does for long stretches: rising steadily while the traditional headwind for a zero-yield asset, rising real interest rates, was blowing directly at it. The usual explanation involving inflation fear or a weak dollar only explains part of the move. The larger and less discussed force sits inside the reserve management departments of roughly two dozen central banks, most of them outside the G7, that have quietly become the single largest category of gold buyer on the planet.

The Reserve Shift Nobody Voted For

For most of the period between the end of Bretton Woods and 2009, central banks were net sellers of gold. European institutions in particular spent decades unwinding bullion they viewed as a costly, non-yielding relic, most visibly under the Central Bank Gold Agreement that coordinated European sales from 1999 onward. That agreement quietly expired in 2019 because there was nothing left to coordinate: the sellers had stopped selling, and a new group of buyers, led by Russia and China in the 2000s and joined later by a wider set of emerging-market and middle-power central banks, had already taken the net flow positive.

What changed in 2022 was the speed and the reasoning. When the United States, the European Union, and their allies froze an estimated $300 billion or so of Russian central bank reserves in response to the invasion of Ukraine, reserve managers everywhere absorbed a lesson that had previously been theoretical: a reserve asset held in a foreign jurisdiction is only as unconditional as the political relationship behind it. Gold held in a country’s own vault, by contrast, cannot be frozen by a foreign government’s sanctions desk. That single event did not create the trend toward official-sector gold buying, but it compressed a decade’s worth of gradual diversification thinking into an urgent, board-level conversation at dozens of institutions within months.

The result shows up cleanly in the data. World Gold Council figures put official-sector net purchases at roughly 1,082 tonnes in 2022, a record going back to when reliable modern data begins, followed by another 1,037 tonnes in 2023 and a broadly similar total in 2024. Compare that to a 2010-2013 average of roughly 470 tonnes a year, or a 2014-2018 average near 500 tonnes, and the step change is obvious: central banks are now buying at roughly double their pre-2022 pace, and they have sustained that pace for three straight years rather than treating it as a one-off reaction.

How a Central Bank Actually Buys Gold

Retail investors buy gold through a handful of familiar channels: a coin dealer, a bullion ETF, a futures contract. Central banks have their own plumbing, and understanding it explains why their buying rarely shows up as a single dramatic transaction.

Direct deals with bullion banks

The most common route is an over-the-counter transaction with one of the major bullion-dealing banks that make up the London market, typically settled as allocated, physical bars held in a recognized vault. These trades are negotiated privately at a price referenced to the London Bullion Market Association’s twice-daily benchmark auction, and because the counterparties are banks used to moving large, discreet blocks for institutional clients, a central bank can accumulate tens of tonnes over a quarter without triggering a visible price spike on any single day. This is the channel a country with no domestic gold mine, and no gold market of its own, almost always uses: Poland, the Czech Republic, and Singapore all buy this way.

Absorbing domestic mine production

Countries that mine meaningful quantities of gold have a second option: a state agency or the central bank itself can simply buy from domestic producers at a formula price tied to the international benchmark, rather than letting that output flow to the open export market. China’s central bank and its state-owned refiners have used variations of this approach for years, as has Russia’s, and it is the primary channel for smaller producers such as Uzbekistan and Kazakhstan. This method has a useful property for the buyer: it never has to compete for supply on the open market, so it does not move the spot price the way an equivalent London purchase might, and it keeps hard currency inside the country rather than paying it out to foreign bullion banks.

Gold swaps and the BIS window

A less visible mechanism, and the one most likely to confuse anyone trying to reconcile the published numbers, is the gold swap. Central banks can swap gold for cash, typically dollars, with the Bank for International Settlements or with another central bank, receiving liquidity while technically retaining the underlying claim on the metal. Because accounting treatment for these swaps varies and disclosure is inconsistent across jurisdictions, a given tonne of gold can appear, at different points, to sit on two different balance sheets, or to have quietly changed hands without a corresponding line item in either country’s published reserve statement. The BIS itself discloses aggregate gold swap and deposit activity in its annual report without naming counterparties, which is precisely why analysts treat swap-driven flows as the murkiest layer of the whole market.

Who Is Buying, and Why Their Reasons Differ

Lumping every official buyer into a single “de-dollarization” story flattens some real differences in motive and behavior.

Reserve diversifiers. Poland’s central bank has run one of the most disciplined, publicly explained accumulation programs anywhere, buying methodically toward a stated target of roughly 20% of reserves in gold, up from around 4% a decade ago. The Czech National Bank, India’s Reserve Bank, and Singapore’s Monetary Authority fall into a similar category: institutions with historically low gold shares relative to their reserve size, working toward an internally set target on a multi-year clock that has almost nothing to do with this week’s gold price.

Sanction-proofers. China’s central bank resumed publicly reporting gold additions in late 2022 after an 18-month reporting gap, adding modest, regular increments most months since. Independent estimates from trade data and refinery flows have long suggested the real pace of accumulation, spread across the central bank and other state-linked vehicles, runs well ahead of the official monthly figure. Russia, effectively locked out of Western custodial infrastructure since 2022, has leaned even more heavily on domestic gold as one of the few reserve assets it can still freely use.

Domestic-market absorbers. Turkey is the case that trips up anyone reading the monthly tables too literally. Turkey’s central bank has, in different years, both bought heavily and sold heavily, because part of its gold activity is really about managing domestic bullion market liquidity and supporting local banks’ gold-denominated deposit products rather than pursuing a steady reserve target. Kazakhstan and Uzbekistan show similar volatility, buying up domestic mine output in some periods and releasing it to local jewelry and investment demand in others. Netted over several years, most of these countries still come out as buyers, but the month-to-month swings can look, on a spreadsheet, like the opposite of what is actually happening strategically.

Reported Numbers vs. the Real Number

Every figure quoted above rests on a reporting system that is voluntary and lagged by design. Countries that subscribe to the IMF’s data standards disclose reserve holdings, including gold, through International Financial Statistics on a schedule of their own choosing; there is no obligation to report in real time, and no penalty for a long silence. China’s central bank went quiet on gold reserves for a year and a half between 2019 and 2022, then resumed disclosure with small, round monthly additions that many analysts view as a deliberately smoothed, conservative signal rather than the full picture.

The World Gold Council addresses this gap by publishing both a reported figure, drawn from official disclosures, and a separate estimate of unreported buying, built from trade flows, refinery throughput, and other indirect evidence. In recent years that unreported component has, at times, been large enough to meaningfully change the read on which country is actually the largest buyer in a given quarter. Add in the ambiguity created by gold swaps described above, and the honest conclusion is that nobody outside the buying institutions themselves knows the true, complete number in real time. What analysts track instead is the trend across several data sources pointing the same direction, which is precisely why one strong or weak month should never be read as a reversal of the broader pattern.

What Price-Inelastic Buying Does to the Gold Market

The reason this buyer matters more than its tonnage alone suggests comes down to how it behaves relative to price. A hedge fund or an ETF holder buys and sells gold partly in response to price: a sharp rally invites profit-taking, a sharp drop can invite bargain-hunting, but both responses are sensitive to where the price sits today. A central bank pursuing a multi-year reserve-diversification target is not making a trade. It is executing a strategic allocation decision on a horizon measured in years, and it tends to keep buying through both rallies and pullbacks because the target share of reserves, not the daily price, is the variable that matters to the board that approved the program.

That price-inelastic behavior helps explain one of the stranger features of the 2022-2024 gold market: the metal kept climbing even as real interest rates, the traditional headwind for a non-yielding asset, rose sharply. The historical relationship between gold and real yields, which had held with reasonable consistency for two decades, weakened noticeably once official-sector buying reached record levels, because a buyer that is not discounting future cash flows at the prevailing real rate does not respond to that variable the way a financial investor does. Several sell-side strategists now describe central bank demand as a structural floor under the gold price. A floor is a useful description, but not a guarantee: the pace of official buying can slow, as some monthly readings through 2025 suggested it was doing at the margin, and a floor that erodes gradually is still a floor that is lower than it was.

Retail investors thinking through their own reasons to hold gold are working with a different set of trade-offs than a reserve manager, even when the asset is identical. Anyone weighing a real-assets allocation for portfolio protection rather than sovereign reserve management purposes will recognize the framework used in our guide to hedging a portfolio with real assets, though the time horizon and the size of the check being written are rarely comparable.

A Worked Example: What It Costs a Central Bank to Swap Treasuries for Gold

Numbers make the trade-off concrete. Picture a mid-sized reserve manager, call it the Central Bank of Meridia, holding $50 billion in total reserves split 92% U.S. Treasuries and 8% gold. Its board votes to lift the gold share from 8% to roughly 12% by selling $2 billion of short-dated Treasuries and buying bullion.

• Capital deployed: $2,000,000,000
• Spot gold price assumed: $3,650 per troy ounce
• Ounces purchased: $2,000,000,000 ÷ $3,650 = 547,945 troy ounces
• Converted to tonnes: 547,945 oz ÷ 32,150.7 oz per tonne ≈ 17.0 tonnes

Now the opportunity cost. The Treasuries being sold were yielding roughly 4.1% a year. Over a three-year holding period, the forgone interest income, compounded annually, comes to approximately $2,000,000,000 × (1.041³ − 1) ≈ $256.7 million. Gold pays no coupon, so that $256.7 million is a real, quantifiable cost of making the switch, not a hypothetical one.

For the trade to simply break even against that forgone income, with reserve principal otherwise held constant in both scenarios, gold would need to appreciate by roughly 12.8% cumulatively over the three years, which is $256.7 million divided by the $2 billion deployed. If gold instead rose 25% over that window, a plausible outcome given the realized gains between 2022 and 2024, the position would show a price gain near $500 million against $256.7 million of forgone yield, a net advantage of roughly $243 million before the modest cost of vaulting, insurance, and assay, which for large allocated holdings typically runs a handful of basis points a year, far below the yield gap it is being weighed against. If gold instead moved sideways or fell, the forgone Treasury income becomes a straightforward, uncompensated cost with no offsetting gain. That asymmetry, upside if gold performs, a clear and calculable drag if it does not, is the actual trade-off every reserve manager making this switch is accepting, whether or not their public statements frame it that plainly.

Sovereign Gold Buyers Ranked: Who Added the Most

The table below groups the largest publicly identifiable official-sector buyers by their approximate net tonnage added across 2022 through 2024, alongside the primary motive attributed to each program and roughly how large gold has become as a share of that country’s total reserves. Figures are rounded and drawn from official disclosures and World Gold Council compilations; several entries, marked accordingly, are believed to understate the true pace of accumulation.

Figures are rounded approximations compiled from official central bank disclosures and World Gold Council reporting; several are subject to later revision and, in China’s case, widely believed to be conservative relative to actual accumulation.

Three Years of Record Buying, in One Chart

The step change in annual buying volume is easier to see side by side than in a table of totals. The chart below plots approximate annual net official-sector purchases, in tonnes, against the pre-2022 baseline, marked with the dashed line.

Common Mistakes Investors Make Reading the Central Bank Gold Story

The topic gets misused in financial commentary almost as often as it gets used well. A few recurring errors are worth naming directly.

Treating one month’s data as the whole picture. IMF disclosures arrive with a lag, get revised, and are voluntary in timing. A single strong or weak month from one country tells you almost nothing about the trailing trend, which is the only version of this data that actually matters.

Assuming the buying guarantees near-term price gains. Central bank demand operates on a multi-year strategic clock, not a trading horizon. It is a structural support, not a signal that gold will be higher next month, next quarter, or even necessarily next year.

Confusing sovereign buying with ETF or retail investment flows. These are different buyer types with different sensitivities. ETF holdings can swing sharply with price and sentiment inside weeks; official reserves essentially never do, because the decision-making process behind them runs through a board or a governor’s office, not a trading desk reacting to a chart.

Reading gross purchases where net is the honest number. Turkey, Kazakhstan, and Uzbekistan can each show large “buying” headlines in a given month that are really domestic market operations offset by selling elsewhere in the same year. Netting matters.

Taking China’s official figure at face value. The People’s Bank of China’s small, round monthly additions since resuming disclosure in late 2022 are widely viewed by analysts as a conservative floor on the true number, not a complete accounting of state-linked accumulation.

Ignoring gold’s real opportunity cost just because a sovereign is buying. A central bank managing $50 billion in reserves and a household managing a retirement account are not making the same trade-off. Gold’s zero yield is a genuine, ongoing cost for both, but it is proportionally a much smaller strategic concession for an institution diversifying a small slice of a very large balance sheet than it is for an individual investor allocating a meaningful share of personal savings.

A Practical Checklist for Tracking Central Bank Gold Demand

• Read the World Gold Council’s quarterly Gold Demand Trends report for both the reported and the estimated unreported official-sector figure, not just the headline total.
• Cross-check individual country reserve disclosures through the IMF’s International Financial Statistics database rather than relying on a single secondary source.
• Watch the People’s Bank of China’s monthly gold reserve statement, but treat it as a floor, not a ceiling, on actual Chinese accumulation.
• Net out known domestic-market operators, chiefly Turkey, Kazakhstan, and Uzbekistan, before drawing conclusions from any single month’s headline purchase or sale.
• Follow the Bank for International Settlements’ annual report for aggregate gold swap and deposit activity as a rough proxy for behind-the-scenes central bank gold financing.
• Judge the trend over a trailing twelve-month window at minimum; a single data point is noise, not signal.
• Separate the strategic, price-insensitive official-sector bid from the price-sensitive ETF and futures market before using either as a timing tool.

Key Takeaways

• Central banks bought over 1,000 tonnes of gold a year in 2022, 2023, and 2024, roughly double the 2010-2018 annual average.
• Most official buying happens through private deals with bullion banks or by absorbing domestic mine output, not through visible open-market purchases.
• The 2022 freezing of Russian reserves reframed gold, for many reserve managers, as the one major asset a foreign government cannot unilaterally restrict.
• Reported figures understate true buying; China’s disclosed additions are widely viewed as conservative relative to actual state-linked accumulation.
• This buying is price-insensitive on a monthly basis, which is why analysts describe it as a structural floor rather than a trading catalyst.
• The trade-off for any reserve manager is concrete and calculable: a real, quantifiable forgone-yield cost against uncertain future price appreciation.

Frequently Asked Questions

Why are central banks buying so much gold right now?

The two biggest drivers are reserve diversification away from a small number of currencies and, since 2022, a sharper awareness that dollar and euro reserves held in foreign custodians can be frozen for political reasons, as happened to Russia. Gold held domestically cannot be restricted the same way, which makes it attractive to reserve managers regardless of where gold’s price happens to be trading.

Which central banks have bought the most gold since 2022?

Poland, China, India, Turkey, and Singapore rank among the largest publicly identifiable buyers by net tonnage added since 2022, based on official disclosures and World Gold Council compilations, though China’s true pace is widely believed to exceed its reported figures.

Does central bank buying guarantee gold prices will keep rising?

No. It provides a structural, price-insensitive source of demand that supports gold over a multi-year horizon, but it is not a short-term trading signal, and the pace of official buying can slow, as some data through 2025 suggested it was doing at the margin.

How is China’s actual gold buying different from what it reports?

The People’s Bank of China resumed public reporting in late 2022 after an 18-month gap and has since disclosed small, round monthly additions. Independent estimates based on trade flows and refinery data suggest the real pace of accumulation across the central bank and other state-linked vehicles likely runs ahead of the official monthly figure.

What’s the difference between a central bank buying gold and a gold ETF buying gold?

A central bank is executing a multi-year strategic reserve decision that is largely insensitive to the day’s price and rarely reverses quickly. An ETF’s gold holdings reflect the aggregated, price-sensitive decisions of its shareholders and can swing sharply within weeks as sentiment or price momentum shifts.

References

• World Gold Council, Gold Demand Trends, quarterly official-sector purchase data and unreported-buying estimates.
• International Monetary Fund, International Financial Statistics, national reserve composition disclosures.
• Bank for International Settlements, Annual Report, aggregate gold swap and deposit activity.
• National Bank of Poland, public statements on reserve diversification targets.
• People’s Bank of China, State Administration of Foreign Exchange, monthly reserve asset statements.