Why Central Banks Started Buying Gold Again

In the late 1990s, a person could hold a serious conversation in a European finance ministry about disposing of the national gold reserve entirely. The argument was not eccentric. Gold paid no coupon, cost money to store, sat on the balance sheet earning nothing, and had spent two decades falling. Sovereign bonds paid a yield, settled instantly, and were issued by governments that everyone in the room assumed would remain solvent and friendly. Several countries acted on that logic and sold, some of them close to the bottom of the market.

Three decades on, the official sector buys. Not loudly, not all at once, and not in the countries that sold — but consistently enough that central banks are now among the most reliable sources of demand in the market. Understanding why requires setting aside the two explanations that dominate popular commentary, inflation fear and speculative enthusiasm, neither of which describes how a reserve manager actually thinks.

What a reserve portfolio is for

A central bank's foreign reserves are not an investment fund. Their purpose is to let the state meet obligations denominated in currencies it cannot print, to defend or smooth the exchange rate if it chooses to, and to survive an episode in which the country loses access to international funding. That last mandate is the important one, because it is the only one that cannot be met by holding assets that perform well in ordinary conditions.

Reserve managers therefore optimise in an unusual order: safety first, liquidity second, return a distant third. A pension fund that loses two percent has a bad quarter. A central bank that cannot settle an import bill for fuel has a political crisis. The asymmetry explains why reserves are dominated by short-dated sovereign paper of a handful of issuers, and why the residual allocation to a non-yielding metal has always been defended on grounds that sound, to an asset manager, faintly irrational.

The selling era, and why it was orderly

The 1990s disposals are often recalled as a collective loss of faith. The record is more procedural than that. European central banks emerging into monetary union held very large gold allocations, inherited from a settlement system that no longer existed, and several concluded that the allocation exceeded any plausible use. The problem was that uncoordinated selling by institutions of that size would have collapsed the price they were selling into.

The 1999 Washington Agreement on Gold, and its successors, existed to solve that coordination problem: signatories capped their collective annual sales and pledged not to expand gold leasing. It was, in effect, a cartel formed to make an exit orderly. It worked. It also had a second effect nobody advertised at the time — by publishing a ceiling on official supply, it told the market exactly how much selling to expect, which removed the fear of an unbounded overhang.

The reversal, and who drove it

Around 2010 the arithmetic flipped, and it flipped in a different set of countries. The institutions accumulating were largely emerging-market reserve managers whose foreign holdings had grown enormously through trade surpluses and were overwhelmingly concentrated in the sovereign debt of two or three issuers. For them the question was not whether gold yielded anything. It was whether it was prudent for ninety-odd percent of national savings to consist of claims on governments over whose policy — and over whose willingness to honour those claims in a dispute — they had no influence at all.

Diversification into other currencies only partially answers that concern, because every currency reserve is still somebody's liability, held in an account that somebody else administers. Gold in a domestic vault is the only major reserve asset with no issuer, no counterparty, no maturity and no administrator. It is, in the precise sense, the asset of last resort: the one that still works when the relationships underpinning everything else have failed.

Sanctions changed the discount rate on trust

The immobilisation of large sovereign reserve balances in recent years converted an abstract tail risk into an observed event. Every reserve manager on earth watched a peer institution discover that the liquid, safe, highly rated portion of its portfolio could be rendered unusable by administrative decision in another capital. Whatever one thinks of the politics, the professional lesson was unambiguous, and it was drawn in treasuries that had no involvement in the underlying dispute.

The response has not been dramatic reallocation. Reserve portfolios move slowly, and the alternatives to deep sovereign bond markets are thin. What has changed is the marginal allocation: a somewhat larger share of new accumulation directed to gold, a preference for holding it domestically, and a noticeably reduced appetite for lending it out for a few basis points of return.

How the buying is actually done

A central bank does not call a broker and buy a thousand tonnes. Purchases are executed in ways designed to be invisible: standing arrangements to acquire domestic mine production at a benchmark price, discreet over-the-counter accumulation through a small number of bullion banks, occasional purchases from other official institutions. Where domestic production exists, buying it locally has the additional advantage of paying in local currency and never crossing a border.

• Domestic production offtake — the mine sells to the central bank rather than exporting, often at the London benchmark less a small discount.
• Over-the-counter accumulation — small, irregular tranches through bullion banks, deliberately sized not to move the fix.
• Official-sector transfers — direct purchases from another central bank or a multilateral institution, settled by title transfer without the metal moving.
• Refining and upgrading — converting existing holdings of older, lower-fineness bars into current Good Delivery specification so they are tradeable without assay disputes.

What this does and does not mean for the price

It is tempting to read official buying as a bullish signal in the way a fund manager's position would be. That reading misunderstands the participant. A reserve manager buying gold is not forecasting a higher price; in most cases the institution would prefer a lower one, since it intends to keep buying for years. The flow is slow, price-insensitive and largely indifferent to the news cycle, which makes it a poor predictor of any given quarter and an unusually durable component of demand across a decade.

The practical effect is on the shape of drawdowns rather than the height of rallies. When investment demand retreats and exchange-traded funds are net sellers, a bid that does not care about momentum absorbs metal that would otherwise have to clear at lower prices. Several of the shallow corrections of recent years look, in the flow data, less like enthusiasm and more like the absence of a seller of last resort.

None of which makes official buying permanent. Reserve policy is policy: it can reverse when the institutions driving it decide their allocation is adequate, or when a government under fiscal pressure discovers a saleable asset in the basement. That has happened before, in living memory, in countries that now teach it as a cautionary tale. The current direction is well established. It is not a law of nature.