Why central banks have been net buyers every year for more than a decade.

When a central bank publishes its “international reserves,” the list usually contains four things: foreign currency holdings (mostly US dollars, mostly held as Treasury securities), Special Drawing Rights at the IMF, the bank’s reserve position at the IMF, and gold. Three of the four are claims on a counterparty — another government, another central bank, the IMF itself. One is not.

This structural distinction is not a matter of taste. It is the operational definition reserve managers actually use. A reserve is what the balance sheet holds when the rest of the balance sheet is being tested. If the test in question is the failure or freezing of a counterparty, the asset whose value depends on that same counterparty is not the asset that does the reserve work.

This is why reserve managers track gold separately from FX reserves, report it under its own line, and explain its presence in the reserve in a way they do not need to explain US Treasury holdings. Gold is the part of the reserve whose role is to be the asset that does not depend on the counterparty being honored.

The line was already running net-positive for twelve years before 2022. What 2022 did was change the slope. To understand why, the relevant fact is the one most US-based readers see only in passing: in late February and early March 2022, the United States, the European Union, the United Kingdom, Japan, and Canada acted together to freeze approximately $300 billion of the Russian central bank’s foreign-exchange reserves — the dollar, euro, sterling, and yen holdings sitting in custodial accounts inside those jurisdictions.

That had not been done before to a G20 central bank. The freezing was not announced as a new reserve-asset doctrine, but in operational terms it was one. Every other reserve manager in the world watched it happen and read it the same way: a reserve held in a foreign currency, in a foreign custodial account, is held at the discretion of the foreign government whose courts and clearing systems can act on it. Whether or not the underlying policy was correct is a separate question. The structural fact is that the discretion is now visibly there.

Gold held in a central bank’s own vault, or in a vault under a bilateral custodial agreement that does not run through the same jurisdictional chain, sits outside that discretion. That is the single sentence that explains most of the slope change.

A household is not a central bank. It does not manage a national reserve, does not face the same liquidity obligations, does not need to settle balance-of-payments flows. The institutional reasoning does not transfer one-for-one. It does transfer in one specific way: the framework for what a reserve is, and what a reserve is for.

Reserve managers concluded that a balance sheet which is entirely composed of claims on counterparties is exposed to the integrity of those counterparties as a single risk. The structural answer they reached is to hold a small portion of the balance sheet in an asset whose value does not depend on any particular counterparty being honored. The same reasoning applies, scaled down, to a household balance sheet that is entirely composed of brokerage positions, custodial-account claims, and fiat deposits.

This is not a forecast about any particular counterparty failing. It is the same posture an institutional treasurer takes when they hold an emergency line separate from their operating cash. The reserve is the part of the balance sheet whose existence is not contingent on the rest of the balance sheet performing as expected. The argument for it does not require predicting which particular thing will go wrong — or whether anything goes wrong at all.

This note does not propose an allocation percentage; the portfolio-efficiency research bracketing that question is read in Office Note 02.

Most public commentary about gold reads central-bank purchases as a tactical signal — central banks are buying, therefore the gold price is going up, therefore one should buy. The structural reading is different and more useful.

Central banks are not the trading desk. They do not buy gold to be early to the next quarter’s price move; they buy gold to put a portion of the national balance sheet into an asset that does not require any particular counterparty to remain solvent or cooperative. The signal is not about the price. The signal is about the posture.

The posture, when it stays in place for fifteen years across nearly every major reserve in the world, is the structural argument. The price does what the price does. The posture is the part with institutional weight.

This is also why HAR does not write price predictions and never will. The structural argument does not have a forecast to fail. The narrative argument attached to the structural argument always fails eventually, and watching it fail teaches the wrong lesson about what the reserve is doing.

At reserve-manager scale, the structural posture is a tonnage figure on a national balance sheet, held in a domestic vault or under a named bilateral custodial agreement. At household scale, the same posture looks like a stated weight in fine ounces, in named refiner-grade form, held under a named depository storage agreement at a named institutional facility, titled to the household or the household’s trust under named governing law.

None of those nouns are decorative. The institutional posture only exists at the household level if every one of those nouns has been written down. The Strategy Brief is the document that does the writing-down. Without it, the household has the idea of a reserve. With it, the household has a reserve.