Paper Gold: What You Own When You Do Not Own Bars

There is a durable argument in gold circles that the market trades far more metal than exists, and that the difference is fraud waiting to be discovered. It is the kind of claim that is easy to shout and hard to examine, because it treats a dozen unlike instruments as one thing. A futures contract, an unallocated account at a clearing bank, a share in a physically backed fund and a gold certificate from a mint are not four versions of the same promise. They differ in who holds the metal, whose balance sheet it sits on, what the holder may demand and when, and what happens if the institution in the middle fails. This structural complexity is the inevitable result of trying to trade a heavy, physical element with the speed and liquidity of a digital currency.

Those differences are invisible while everything works. Every one of these instruments tracks the same price, more or less, on an ordinary Tuesday. They separate in stress — in March 2020, when grounded flights broke the link between London bars and New York contracts; in 2008, when unsecured claims on banks stopped being a theoretical category. The point of understanding the paper market is not to decide whether it is a conspiracy. It is to know which of your risks are price risks and which are structural. In a world of increasing geopolitical tension, the ability to distinguish between a price target and a delivery right is the difference between a hedge that works and a contract that merely reports a price.

Unallocated: the plumbing nobody sees

The overwhelming majority of wholesale gold trading is unallocated, and it exists for a mundane reason: metal is heavy. Moving 400-ounce bars between counterparties requires armoured transport, insurance, weighing, and a chain of custody that survives audit. Transferring an unallocated balance requires a ledger entry. A market clearing tens of billions of dollars a day cannot run on forklifts, so it runs on credit. This system is the 'dark matter' of the gold market — invisible to the casual observer but responsible for the vast majority of the price discovery that takes place every second.

An unallocated balance is a bank's promise to deliver a stated quantity of gold of stated fineness. It is not a bailment. The metal backing that promise is not segregated, not numbered, and not necessarily sitting still — the bank may lease it, use it to settle other obligations, or hold rather less of it than the sum of its obligations, exactly as a deposit-taking bank holds less cash than the sum of its deposits. Nothing about this is hidden; it is the stated architecture of the London market. The system relies on the assumption that only a small fraction of account holders will ever demand physical delivery at once, allowing the banks to maintain deep liquidity with a relatively small physical float.

Allocated accounts invert every one of those properties. The client holds title to identified bars, listed by serial number, refiner and assay. The bank is a custodian, not a debtor. Because the metal cannot be lent or netted, the client pays storage and insurance instead of enjoying a free ledger entry — typically a few tenths of a per cent a year. That fee is the honest price of removing counterparty risk. The fact that most of the market declines to pay it is informative about what most of the market actually wants: exposure rather than metal. For the institutional trader, the cost of allocation is a drag on performance; for the long-term wealth preserver, it is the insurance premium for the only asset that carries no one else's promise.

LPMCL: The Five Banks That Clear the World

To understand why unallocated gold is so dominant, one must look at the London Precious Metals Clearing Limited (LPMCL). This is the inner circle of the bullion market, currently comprising five clearing banks: HSBC, ICBC Standard Bank, JP Morgan, UBS, and The Bank of Nova Scotia. These institutions operate the AURUM clearing system, which settles the vast majority of global gold transactions through a process of 'netting'. Instead of moving bars for every trade, the clearing banks calculate the net difference between what they owe each other at the end of each day and settle the balance through ledger entries.

This concentration of clearing power is what makes the London market the deepest in the world, but it also creates a significant concentration of counterparty risk. If you have an unallocated gold account with a secondary bank, that bank likely holds its own unallocated balance with one of the LPMCL clearers. Your claim is therefore stacked: you are a creditor of your bank, which is a creditor of a clearing bank. The efficiency of the system is remarkable — clearing hundreds of tonnes of metal daily with minimal physical movement — but it is built entirely on the creditworthiness of these five pillars. When one shows signs of stress, the entire unallocated market feels the tremor.

The AURUM system handles two types of transfers: 'allocated' and 'unallocated'. Even within the clearing system, an allocated transfer involves the identification of specific bars, which is why it is slower and more expensive. Unallocated transfers are instantaneous. In the wholesale market, 'Spot' gold settlement (T+2) almost always refers to an unallocated transfer between these clearing banks. The sheer scale of this system is difficult to overstate: in a typical month, the LPMCL clears more gold than the entire annual global mine production. This 'velocity' is only possible because the metal remains stationary while the claims move at the speed of light.

Futures: a price instrument that can deliver

A COMEX gold future is a standardised agreement to exchange 100 troy ounces of at least 995-fine gold at a stated month and price, backed by an exchange clearing house that stands between buyer and seller. That interposition is the product's real innovation: neither side needs to assess the other's creditworthiness, because both face the clearer and both post margin daily against price moves. This 'central counterparty' model removes the bilateral risk of the London unallocated market, replacing it with the institutional strength of the CME Group.

Because margin is a fraction of contract value, futures are the cheapest way to carry a large directional gold position. Miners hedging output, refiners hedging inventory and funds expressing a macro view all live here. It is also why the notional volume dwarfs annual mine supply, and why that comparison is close to meaningless. A hedged refinery may turn its position over daily; the same tonne of metal generates volume repeatedly without anyone claiming to own it twice. The futures market is not a warehouse; it is a giant insurance and price-discovery machine that uses a small amount of physical metal to keep its gears greased.

Delivery is nonetheless real, and its mechanics discipline the price. A holder who stands for delivery receives a warrant against metal in an approved depository, having paid the full contract value rather than margin. The narrow, mostly stable spread between futures and London spot — the 'exchange for physical' (EFP) — is the cost of financing and moving metal between the two venues. When that spread dislocates, the message is logistical. A futures contract is a promise of delivery in New York; a spot trade is a promise of delivery in London. Usually, they are fungible. Occasionally, they are worlds apart.

The EFP Crisis of 2020: A Case Study in Basis Risk

In late March 2020, the gold market experienced one of its most dramatic dislocations. The EFP spread — the difference between the price of a gold future in New York and the spot price in London — typically hovers around a few dollars per ounce. As the COVID-19 pandemic shut down global travel, this spread exploded to over $70. To the casual observer, the gold price was rising, but to the professional trader, the market was breaking. The reason was purely physical: London bars are 400 ounces, while COMEX contracts require 100-ounce or kilobars. With passenger flights grounded, the normal mechanism for flying London bars to refineries in Switzerland and then to New York was severed.

Traders who were short futures and long spot — a common 'basis trade' — found themselves in a terrifying squeeze. They had the gold in London, but they couldn't get it to New York to satisfy delivery requirements. The 'paper' price in New York began to trade as if there were a massive shortage of metal, even though London vaults were full. This event proved that the 'link' between different forms of paper gold is not a mathematical certainty, but a logistical one. If you cannot move the metal, the two contracts are no longer the same asset.

The crisis was eventually resolved by emergency measures, including a new COMEX contract that allowed for the delivery of 400-ounce bars. But the lesson remained: the paper market is a skyscraper built on a physical foundation. If the logistics chain stops working, the people on the top floors are trapped, regardless of how much metal is sitting in the basement. This 'basis risk' is the hidden tax on every arbitrage trade, and it only becomes due when the world stops moving.

Exchange-traded funds: allocated metal, one step removed

The physically backed gold ETF was, structurally, the most consequential product of the last quarter century for the metal. It converted gold into something a pension fund could hold without amending its custody policy: a security with a ticker, daily liquidity, an audited holding and no need to insure a vault. Assets followed, and with them a new and price-sensitive source of demand that shows up in flows within hours of a rate decision. However, the simplicity for the end-user masks a highly complex engine that manages the physical metal behind the scenes.

The Creation Loop: How ETFs Manage Physical Supply

Unlike a closed-end fund, a gold ETF can expand or contract its supply of shares to meet demand. This is done through 'Authorised Participants' (APs) — typically large bullion banks or market makers. When demand for ETF shares exceeds the available supply, the shares begin to trade at a premium to the value of the underlying gold. An AP will then buy physical gold in the wholesale market (usually 400-ounce bars), deliver that gold to the ETF's custodian, and receive in exchange a 'Creation Unit' — a large block of new ETF shares (typically 50,000 shares).

The AP then sells these new shares on the stock exchange, capturing the premium as profit and bringing the share price back in line with the metal's value. When investors sell shares and a discount appears, the process works in reverse: the AP buys the cheap shares, redeems them with the fund for physical bars, and sells the gold in the wholesale market. This 'Creation-Redemption' mechanism is the heart of the ETF; it ensures that the paper share remains tethered to the physical bar. But it also means the ETF is a massive 'physical-to-paper' converter, constantly sucking metal into or spewing it out of institutional vaults.

One critical detail often overlooked is that this process happens 'in-kind'. The AP delivers bars, not cash. This is why the ETF is considered 'physically backed'. However, the retail investor is not part of this loop. You cannot take your 100 shares to the bank and ask for a gold coin. Only the APs, dealing in multi-million dollar 'baskets', have the right to touch the metal. For the retail holder, the gold is 'allocated' to the trust, but the investor only owns a pro-rata share of the trust's assets, which is a very different legal standing from owning a specific serial-numbered bar.

Forensics: How to Audit a Custodian's Bar List

For the sceptical investor, the most powerful tool is the 'Bar List'. Major physically backed ETFs publish this document daily. It is a massive spreadsheet listing every single gold bar held by the custodian on behalf of the fund. To verify this, one doesn't just look for a total weight; you look for the 'Good Delivery' identifiers: the refiner's stamp, the year of manufacture, the serial number, and the exact weight in troy ounces (calculated to three decimal places).

A substantive audit involves more than just checking the numbers add up. A forensic approach cross-references the refiners on the bar list with the LBMA 'Good Delivery List' — the official record of refiners whose bars are accepted in the London wholesale market. If a fund holds bars from a refiner that has been 'former listed', it may indicate a lack of quality control. Furthermore, independent auditors like Inspectorate are hired twice a year to physically enter the vaults, count the bars, and randomly weigh a sample to ensure they match the list. The results of these audits are public documents, and a serious investor should read the 'Physical Inspection Report' with care.

The ultimate test of a bar list is its uniqueness. Because each bar has a serial number and a specific weight, the same bar should never appear on two different funds' lists. In theory, a researcher could aggregate every major ETF bar list and check for duplicates — a process that would expose 'double-counting'. While no such fraud has been discovered in major ETFs, the very fact that this data is public provides a level of transparency that unallocated bank accounts simply cannot match. If you can't see the bar list, you don't own the gold; you own a secret.

The Geometry of Frictional Costs: VAT, Premiums and Spreads

Arguments about paper gold are usually conducted in the language of trust, when the decisive variable over a decade is cost. A physically backed fund charges an annual fee (the expense ratio) that compounds quietly against you. A futures position pays or receives the roll each time a contract is rotated forward, which in a normal upward-sloping curve (contango) is a persistent drag. But physical gold has its own geometry of cost that is often even more punitive.

In the European Union, 'investment gold' is exempt from VAT, but this exemption does not apply to silver or to gold jewellery. In other jurisdictions, a Goods and Services Tax (GST) may apply unless the gold meets specific purity and storage requirements. Furthermore, the 'bid-offer spread' on physical coins can be 3% to 5% for a one-ounce coin, and as high as 10% for smaller denominations. This means the gold price must rise by 10% just for the investor to break even on a round-trip trade. In contrast, an ETF share can be traded for a spread of a few cents.

There are also hidden 'geographical costs'. Gold is worth more in some places than others. The 'Shanghai Gold Premium' or the 'India Gold Premium' reflects the cost of importing metal into those restricted markets. If you buy gold in a high-premium market, you are betting not just on the gold price, but on the continuation of that local supply-demand imbalance. If you ever need to sell that metal outside that jurisdiction, the premium vanishes, resulting in an immediate loss. Paper gold products generally avoid these local dislocations, trading instead at the 'global' price, which is both a benefit (liquidity) and a risk (you lose the local hedge).

It is tempting to treat physical gold as the real thing and everything else as a shadow of it. The market does not work that way. The paper layer is what allows a Swiss refiner to lock in a price for metal that is still in a shipping container, a jeweller in Mumbai to fix costs before the wedding season, and a miner in Ghana to finance a shaft against production three years out. Remove it and the price of physical gold would not become purer; it would become far more volatile and far more expensive to access, because every participant would carry their own unhedged inventory risk. The paper market provides the 'liquidity buffer' that makes physical ownership affordable for the rest of us.

What the paper layer cannot do is settle a crisis of confidence in the institutions that write it. That is the one job physical metal has never delegated, and the reason serious holders of gold, from central banks to family offices, keep a portion of the position in bars they can point at. The instruments are not rivals. They answer different questions, and the discipline is in knowing which question you are asking. In a stable world, paper is superior for its efficiency; in a collapsing one, physical is superior for its independence. Most of history happens somewhere in the middle.

Structured notes and gold-linked products

A step further from allocated metal sit structured notes: bank-issued debt instruments whose return is linked to the gold price through a formula, often with capital protection, a cap, or leveraged participation attached. These products hold no gold at all. The issuer manages its own exposure through derivatives and its own book, and the investor's return depends on both the gold price path and the issuing bank's ability to pay — a second layer of counterparty risk stacked directly on top of the market risk. These are 'gold-themed' credit risks rather than gold investments.

Their appeal is genuine for a specific need: an investor who wants defined downside protection with some upside participation cannot construct that payoff cheaply with a plain ETF holding. The cost of that structure is usually opacity — the embedded option pricing, the issuer's funding spread and the distributor's commission are rarely broken out clearly, which makes these among the hardest paper-gold products to compare on a like-for-like basis. You are paying the bank to take a risk you could not manage yourself, and the bank is charging a premium for the convenience of that wrapper.

Sharia-compliant gold products

Islamic finance treats gold as a ribawi item under classical jurisprudence, meaning exchange must be immediate and in equal measure to avoid riba, which historically excluded gold from conventional deferred-settlement trading for observant investors. The AAOIFI Shariah Standard on Gold, published in 2016, resolved this for the modern market by defining conditions under which gold-backed ETFs, allocated accounts and even certain futures-adjacent structures can qualify, provided possession — actual or constructive, through a recognised custodial mechanism — occurs promptly and the underlying asset is genuinely allocated. This standard opened the door for trillions of dollars in Islamic capital to enter the formal gold market.

• The product must reference physically allocated gold, not a synthetic or purely cash-settled exposure.
• Constructive possession must be established promptly, typically same-day, rather than deferred to a future settlement date.
• The gold must meet a minimum purity threshold, generally 995 fine or better, consistent with wholesale market standards.
• Ownership records must be auditable to a specific, allocated pool rather than a general claim on the issuer's balance sheet.
• The transaction must involve an immediate transfer of ownership, avoiding the 'deferred payment' (nasi'ah) that characterises conventional futures.

What past dislocations actually taught the market

March 2020 is the case study every serious paper-gold analysis returns to, because it is the cleanest recent example of structural risk separating from price risk. Passenger flight groundings during the pandemic's onset disrupted the physical logistics chain that normally keeps London loco spot and COMEX futures prices arbitraged into a tight band — refiners could not fly large bars to New York vaults to be recast into deliverable weights, and for several tense sessions the futures price traded at a premium to spot wide enough to draw public attention and a rare joint statement from the exchange and major bullion banks. This was not a failure of the gold price; it was a failure of the 'paper' promise to be interchangeable with the 'physical' metal at zero cost.

The resolution came through unglamorous logistics fixes — new refining capacity brought on to recast 400-ounce bars into 100-ounce and kilobar deliverable sizes, and eventually a new COMEX contract specification accepting London-standard bars directly. The lesson generalises well beyond gold: paper markets fail at their physical seams, not at their pricing formulas, and the fix is almost always operational rather than regulatory. If you hold a contract that relies on a specific logistics chain, you are long that logistics chain, whether you realize it or not.

A short framework for choosing

Every product discussed here answers a different question, and the honest starting point is naming the question before naming the product. An investor wanting cheap, liquid, short-term price exposure with no interest in ever touching metal is well served by futures or a large physically backed ETF. One wanting a decade-scale store of value uncorrelated with any institution's solvency should be paying storage fees on allocated bars, or holding coins directly, and should stop comparing that cost against a fund's expense ratio as though the two products did the same job. The 'hidden' cost of allocated metal is the price of total independence.

The mistake this piece has tried to dismantle is treating 'paper gold' as a single suspect category to be either trusted wholesale or dismissed wholesale. It is a shelf of differently engineered promises, priced off the same number on a screen, and the discipline that protects a buyer is not scepticism about gold — it is precision about which promise is actually being purchased. If you don't know the difference between a clearing bank ledger and a serial-numbered bar, you don't yet know what you own.

Digital gold apps and the rise of micro-ownership

The newest layer of paper gold did not come from a bank trading floor but from a phone. Digital gold apps let a buyer purchase a fraction of a gram of gold for the price of a coffee, with the underlying metal held in vault by a partner refiner or bullion custodian on the platform's behalf. The pitch is savings discipline dressed as a commodity purchase: round up a card transaction, divert the change into gold, watch a balance accumulate in grams rather than currency units. This has democratised access to gold, but it has also added a new layer of intermediation.

Structurally these platforms sit closer to an unallocated or pooled-allocated claim than to a numbered bar in the buyer's name. The operator typically holds a bulk allocation of physical metal against the sum of all customer balances, rather than carving out an identified sliver of a bar for every micro-purchase, because doing the latter at gram-and-below scale would be operationally absurd. That pooling is not dishonest, but it reintroduces the counterparty question: whose balance sheet is the metal sitting on, and what happens to the customer's claim if the platform, rather than the vault operator, fails first? In many jurisdictions, these apps operate outside the strict regulations that govern banks or mutual funds.

• Confirm whether the app's partner custodian is a recognised, audited vault operator, or an unnamed 'partner refiner' with no public assurance report.
• Check the buy-sell spread on small transactions; some platforms charge several per cent round-trip on sub-gram purchases.
• Look for an explicit, exercisable right to convert a balance into physical delivery above a minimum threshold, and read the delivery fee.
• Treat any platform that also lends out customer gold balances for yield as an unallocated structure regardless of its marketing language.
• Understand that 'digital gold' is often a contractual claim against a fintech start-up, not a direct ownership of metal in a vault.

What regulation actually protects, and what it does not

Buyers often assume that a regulated financial product is, by definition, a safe one, and regulators have spent considerable energy trying to correct that assumption. Regulation of paper gold products typically governs disclosure, market conduct, capital adequacy of the issuing institution and the accuracy of marketing claims. It rarely eliminates the underlying structural risk the product is built on; it makes that risk visible and enforceable rather than removing it. You can have a perfectly regulated unallocated account that still leaves you as an unsecured creditor.

A physically backed ETF regulated under a mainstream securities regime must hold assets that are legally separated from the fund manager's own balance sheet, publish an auditable bar list, and use an independent custodian bound by a trust or similar structure that keeps the gold outside the manager's insolvency estate. That is real protection, and it is the reason serious institutional allocators are willing to hold ETF shares as a proxy for metal at meaningful scale. An unallocated account at a bank, by contrast, can be entirely properly regulated as a banking product — capital requirements, deposit-style conduct rules, disclosure obligations — while still being, in law, an unsecured claim in an insolvency. Regulation made the risk clear on paper years before any crisis tested it; it did not make the risk disappear. The regulator's job is to ensure you were told about the risk, not to prevent you from taking it.

This matters most at the edges of the market: platforms operating from jurisdictions with light-touch commodity regulation, gold-linked products marketed to retail investors through channels with weaker suitability requirements than mainstream brokerages, and cross-border digital gold apps whose ultimate custodial arrangements are harder for an ordinary buyer to verify than a domestic bank's terms and conditions. None of this is an argument against using regulated products. It is an argument for reading the custody clause with the same attention usually reserved for the price. In the end, your legal status as a claimant is determined by the law of the jurisdiction where the metal is held, not where you clicked 'buy'.

A worked comparison across three holding periods

Put numbers to the abstractions and the choice of format sorts itself by horizon more cleanly than by ideology. Over a holding period measured in months — a tactical view around a data release or a currency move — the dominant cost is transaction friction, and futures or a large, liquid ETF win comfortably against coins, where the dealer spread alone can exceed a plausible short-term price move. For the active trader, physical gold is a trap of high entry and exit costs.

Over a holding period measured in several years, the annual expense ratio of a fund starts to matter more than any one-off spread, and the comparison against the all-in cost of storing allocated bars — insurance, vault fees, occasional audit — becomes genuinely close, tilting toward metal only once the position is large enough that fixed storage costs are a small percentage of the total. At this stage, the decision often comes down to tax efficiency and the investor's desire for custodial independence.

Over a holding period measured in decades, intended explicitly as a hedge against institutional failure rather than a tactical trade, the calculus changes again. A fund's counterparty structure, however well designed, is still a structure that depends on a custodian, a trustee and a jurisdiction remaining functional across a horizon long enough that none of those things can be taken for granted. That is the scenario in which the extra cost of coins in a safe, or allocated bars with a private vault operator outside the banking system, stops being an inefficiency and starts being the entire point of the exercise. When you are buying for your grandchildren, you don't buy a contract; you buy an element.