A reserve manager who allocates to gold makes two decisions. The first is whether to hold gold at all – a portfolio question, answered by reference to diversification, inflation and counterparty risk. The second is where and how to hold it – a custody question, answered by reference to security, liquidity, cost and the ability to access or relocate the metal under stress.
These two decisions are frequently made together, but they are governed by different considerations. A manager can be entirely correct about the first and give little independent thought to the second.
The three custody models
Official gold is held under three broad arrangements. Each carries a distinct profile across the four variables that matter – security, liquidity, cost and mobility.
1. Domestic storage, in the country’s own vault
The metal is held at home, under direct national control. This maximises sovereign control and political optionality, and it is the arrangement least exposed to a foreign jurisdiction. It also places the metal behind the full protective apparatus of the state, including, in extremis, its military and security forces – a form of physical protection no commercial arrangement can match.
The costs of this are real but often unstated: building, staffing, securing, insuring and auditing a vault to institutional standard is a permanent capital and operating commitment. Its more consequential limitation is liquidity. A national vault is not necessarily a delivery point that the global dealer market transacts against, so mobilising the metal – selling, lending or moving it – can require physically transporting it to a recognised market centre first. Gold held at home is secure in peacetime and slow to deploy in a crisis.
2. Custody at a major foreign monetary authority
These are predominantly the Bank of England and the Federal Reserve Bank of New York. This is the long-standing default for gold held abroad, and its merits should not be understated. Liquidity is excellent – London in particular sits at the centre of the loco-London market, the deepest pool of gold trading in the world, so metal at the Bank of England can be transacted without leaving the vault. Legal and operational track records are long. And the metal sits behind the security apparatus of a major state, a level of physical protection a commercial operator cannot replicate.
Reported fee arrangements vary: the Bank of England operates a custodial service for which it charges, while the New York Fed has historically not charged storage fees to official account-holders, recovering handling costs only.
The limitation of this model was highlighted by Russia’s invasion of Ukraine in 2022. Metal held at a foreign monetary authority sits within that state’s jurisdiction. For most account-holders, most of the time, that presents no issue. But it reintroduces at the custody layer precisely the single-jurisdiction exposure that motivated part of the gold allocation in the first place. Whether that matters depends on the holder’s relationship with the host state and its assessment of tail risk.
3. Independent commercial custody
Metal held with a private, non-sovereign custodian is the arrangement most institutions use for nearly every other asset on their balance sheet. Its distinguishing features are jurisdictional choice and mobility: metal can be held across a network of commercial vaults in multiple bullion centres, with title held by the owner, and moved or sold at a market centre in a short timeframe.
Its costs are explicit – a storage fee, typically priced in basis points on value. It also carries two limitations the other models do not. It relies on private security and the ordinary rule of law in each host jurisdiction rather than the protection of a sovereign; and it introduces a commercial counterparty. It is also, for official reserves specifically, the least established of the three: there is limited public precedent of a central bank custodying a material share of its reserves this way, and this is a cost to an institution that values precedent.
The tension
Set the portfolio rationale beside the dominant custody choice and a tension appears. Part of the reason for holding gold is to reduce dependence on any single state. Yet most gold held abroad is held with one of two states, and most of the remainder is held domestically. On the specific axis of single-jurisdiction exposure, the custody arrangement does not fully carry through the logic of the allocation.
Reserve managers are aware of this. A rising share have increased domestic storage or diversified their overseas locations in the past year, and more plan to. Since 1972, official holders have repatriated roughly 6,900 tonnes; recent programmes by several major holders moved substantial tonnage out of New York and London.
But repatriation resolves only one version of the concern. It does not remove single-jurisdiction concentration and, for a reserve manager whose independence from the domestic fiscal authority is a live question, it may not clearly reduce this concern. It also forfeits the liquidity and mobility that made the foreign-vault arrangement useful. Domestic storage answers the question of ‘whose soil is it on?’, but not ‘how fast can it move if that soil becomes contested?’
Why the default persists
The Bank of England and the New York Fed are the default because they have been the default for the better part of a century. Reserve managers operate under mandates that prize continuity, precedent and the avoidance of unforced error; the institutions themselves are, by design and temperament, cautious. A custody arrangement with a long, unblemished record carries weight that a newer alternative – however well-structured – cannot yet match, precisely because it has not yet accumulated a comparable record. The incumbents also offer real advantages, above all the liquidity depth of London, that any alternative must equal, not merely approach.
This means the barrier to a different arrangement is not primarily cost or law. A diversified commercial custody network across established bullion centres — New York, Toronto, London, Zurich, Frankfurt, Singapore, Hong Kong and Dubai – is available today, and the fees involved are small relative to a reserve position. The barrier is that being first carries institutional risk. A reserve manager who follows the default and is later proved wrong has plenty of company; one who departs from it and is proved wrong stands alone.
A framework, not a recommendation
This article does not conclude that any particular arrangement is correct; the right answer depends on a holder’s specific circumstances, mandate and assessment of risk. It concludes only that the custody question deserves to be decided on its own merits rather than inherited as a default, and it suggests the variables a deliberate decision would weigh.
An optimal custody policy is unlikely to be a single choice; it is more plausibly a mix, calibrated to the holder’s tolerances: a domestic allocation for sovereign control and political optionality; an allocation at established foreign centres for liquidity depth; and, potentially, a diversified allocation across independent commercial vaults in multiple jurisdictions for the ability to relocate or realise the metal quickly, across borders, under stress. The weights among these are a matter of judgment. The point is that they are weights to be set deliberately, not a default to be accepted.
The tension is narrow and specific: the reasons commonly given for holding gold emphasise independence from any single jurisdiction, while the prevailing custody arrangements concentrate it in one. That tension is not necessarily an error – a holder may reasonably judge that liquidity, precedent and operational certainty outweigh jurisdictional diversification at the custody layer. But it is a trade-off that ought to be made consciously. At present, for many holders, it appears to be made by default.
Steven Feldman is Co-founder and Chief Executive Officer of GBI, which owns the GBI Platform, the institutional infrastructure for physical precious metals that serves global wealth managers, banks and fintechs.
Join OMFIF in Bangkok on 14 October for Redefining resilience in reserve management.
Interested in this topic? Subscribe to OMFIF’s newsletter for more.