Who owns the gold in London’s vaults?

What the September count measures

LBMA’s 7 October release put end-September gold holdings at 9,821 tonnes across the London vaults in its survey, including the Bank of England. August’s published figure was 9,632 tonnes. The difference between the rounded observations is 189 tonnes, consistent with the reported monthly rise of 1.96%. September is the latest month I could verify for this edition.

That change answers a narrow question about location. If the same reporting perimeter is used and a bar arrives from outside it, the citywide stock rises. A bar transferred between two vaults already inside it leaves the combined weight unchanged. A sale can also leave the weight unchanged. This is why I would retain the word ‘held’ in a headline about the series. ‘Bought’ introduces an owner and a transaction that the count does not identify.

Source notes: [1] LBMA — September London vault release · [2] LBMA — August London vault release

Follow a bar through a change of title

The Bank of England stores bars on behalf of HM Treasury, other central banks and some commercial firms. Under its allocated custody arrangements, the customer holds title to specific bars. A researcher who adds the Bank’s separately reported custody weight to LBMA’s London total counts that part of London twice.

Imagine an owner sells a bar to another customer and the relevant accounts are updated while the bar stays in place. In this simplified example, the seller has disposed of gold and the buyer has acquired it. A physical count at either end of the day produces exactly the same answer. Reverse the example: the owner moves its own bar from overseas into a reporting London vault. Now the count changes while the owner’s total gold weight is constant.

Source notes: [1] LBMA — September London vault release · [3] Bank of England — Gold custody and ownership · [4] Bank of England — Gold statistics

Can a dealer actually buy the metal?

Suppose a dealer needs a binding offer for delivery in London on Tuesday. The monthly release helps establish the scale of the local custody network. It gives no seller’s name, asking price, offer size or commitment for Tuesday. Even a complete list of bars and their owners would leave the owners’ willingness to sell at that price unknown.

A useful cross-check would be contemporaneous firm bids and offers for the required size, acceptable bar specification and settlement date. A price for a small transaction could conceal a very different cost for a large one. If only monthly inventory is available, I would leave executable supply unquantified. Turning an assumed percentage of total stock into a precise estimate of available metal gives the assumption more authority than the evidence.

A reproducible entry for the physical-market notebook

I record the reporting boundary, month-end weight and previous comparable observation on one line. On a separate line I record the actual question: for example, whether a specific buyer encountered delivery delays. Resolving that second line would require dated evidence about delivery terms or completed settlements. The September count can support a statement about more metal under reported London custody. It cannot date a delay, identify a new official buyer or tell us what a dealer could have sourced on Friday afternoon.

Source notes: [1] LBMA — September London vault release · [2] LBMA — August London vault release · [4] Bank of England — Gold statistics