Sep 162015


Bullion banking is integral to the function of the modern gold market. Unallocated and allocated gold accounts and associated clearing mechanisms centred in London facilitate the efficient transformation of gold from mine to end consumer. However, banking involves risk – for an individual bank should borrowers fail to repay their loans and also at a systemic level.

This series of posts on the fractional reserve bullion banking system explain how bullion banking works and where the risks are.


Unallocated bullion bank accounts are fractionally backed, no different to fiat banking. Indeed most unallocated accounts are fractional, as it is impossible to offer a 100% backed account with no storage fees unless you are a physical user of gold like The Perth Mint.

Unlike refiners, manufacturers or distributors, a bullion bank has no real need for physical gold itself. Unless they are storing it on an allocated basis on which they can charge storage fees, having a (free, or very small fee) unallocated account backed by physical gold in a vault is, if not an outright loss, at least a not very productive and profitable use of their client’s gold deposits.

Therefore bullion banks are incentivised to lend gold. This naturally leads to the question of how much do they lend and how much do they keep as physical reserves. The fact is no one really knows. Jeff Christian of CPM Group gave us an insight into the possible fractional reserve ratio here, where he says that most banks are operating on a 10:1 ratio, but notes that AIG was operating at 40:1.

In terms of fractionalisation, it is important to distinguish between on-call deposits and term deposits. For example, if a bank borrows gold for a term of 1 year and lends it out for 2 years, that does not present any immediate risk of a bank run as the lender to the bank has no right to the gold now, only in 1 year.

Bullion banks lend their on-call gold deposits (that is, unallocated account credits) to borrowers for fixed terms into the future. This is called maturity transformation and I tend to agree with this blogger that “that, without any government protection, it is incredibly unstable and will melt down at a drop of the hat. With full government protection, it is stable”.

So in addition to how much physical does a bank hold relative to on-call deposits (fractionalisation), how long a bank has lent out gold for also matters. For example, if a bank only had 10% as physical but had lend the remaining 90% for no longer than, say, 1 week, then you may conclude that they are unlikely to suffer from a bank run as they will quickly get gold back to repay those on-call depositors.

However, that assumes the people they lent it to actually deliver against their promise to repay their gold loan. In other words, to whom did the bank lend and how credit worthy are they? This is called credit risk.

But, the bank may claim, we have collateral against the loan, so if the client doesn’t pay up we can sell their collateral and buy the missing gold. This of course assumes that the collateral does not go down in value or that the gold price does not go up, in a market stress situation. This is called price risk.

So the risk a bullion bank’s unallocated accounts presents to them depends on the bank’s:

  • Fractionalisation – percentage of physical gold reserves they hold
  • Maturity transformation – degree of mismatch between maturity of assets and liabilities
  • Credit risk – the credit worthiness of unsecured counterparties
  • Price risk – amount and quality of collateral and gold price exposure for secured counterparties

All of these factors apply to fiat banking as well, but as our blogger notes, fiat banking is ultimately backed by government. This is possible because a government can print fiat and exchange it for a bank’s long term assets, suddenly increasing the bank’s physical (banknote) fractional reserves to give to depositors, and thus avoid a bank run.

However, while one can’t print gold, a bullion bank experiencing a run where its unallocated holders want physical delivery can approach central banks to borrow physical gold on the basis that its gold assets will mature into gold eventually, with which it can repay the loan.

Gold Assets

Understanding the types of gold “assets” a bullion bank can hold, and the risks associated with each of them, is essential to assessing the stability of the bullion banking system.

In the case of gold lending, there are two types of borrowers as there are only two things you can do with borrowed gold (no one borrows gold just to keep it at home to look at):

  1. Use it as inventory in your gold business (eg jewellery, minting)
  2. Sell it (that is, short the gold price to benefit from it falling), the sellers being either mining companies or investors/speculators (hedge funds, individuals)

If you are someone without creditworthiness, which just means that a bullion bank makes an assessment that you may not repay your debts, then a bullion bank will require some security or collateral which they can access if you don’t pay. An example of this in consumer lending is a bank holding a mortgage on “your” home.

For gold manufacturing businesses the bullion bank can be reasonably sure you have physical gold to repay and can put in place some sort of lien or mortgage type arrangement against the physical inventory and/or other assets of the business. There is still a risk that the business goes bankrupt with the gold being sold and not replaced or maybe the owners just steal the gold. However, lending to and monitoring business is what banks do, and generally do well and while people want to buy jewellery and invest in gold the risk of default is low for these businesses (if the gold market was to go into a protracted bear market that may be a different thing).

The borrower who is short selling gold presents a bigger risk because neither the bullion bank nor the short selling borrower has any physical gold to mortgage as it has been sold. As a result, bullion banks will lend short sellers gold on the condition that they sell it and keep the resulting cash from the sale as collateral. Since the gold price is volatile, the bullion bank will also require the short seller to put up additional margin. So a bullion bank has both cash from the sale and margin to cover themselves.

Mining companies are sort of like our jeweller or minter, in that they are a business involved in physical gold, the only difference being that the gold they hold is in the ground and not in a factory. This is a bit more risky than a gold manufacturer as they may not be able to get the gold out of the ground at a reasonable cost or have some other operational problems. They are also more risky than a speculator as the mine used the cash to pay expenses or buy equipment, so there is no cash left to use as collateral (the bullion bank could mortgage equipment etc, but resale value of that and an unprofitable mine would be low).

If you are someone with creditworthiness, then the bank will let you do the above things without a need for margin or collateral, at least up to whatever credit limit they set for you. This is obviously a lot more risky than secured lending.

Finally, a bullion bank can also “lend” gold to themselves in the process of creating derivative products, which may be best explained by two examples.

If there are speculators who want to sell futures contracts a bullion bank can take the other side and go long a futures contract. To offset that obligation to buy gold in the future, the bullion bank can borrow gold (that is, from their on-call depositors) and sell it. They then invest the resulting cash and when the futures contract is delivered, the bullion bank uses the cash to pay the short futures contract holders and receive their gold. That gold goes back into the bank’s physical reserves to back their on-call depositors’ accounts. The result is that the on-call depositors’ accounts were being “backed” by the long futures contract the bullion bank was holding.

Another more complex example would be someone wanting to buy a put contract on gold (they have the option to sell gold to a bullion bank). If a bullion bank sells a put contract that means they have a potential obligation to buy gold in the future. As with the long futures example above, the bullion bank hedges that obligation by borrowing gold and selling it (technical note, with options the amount of gold the bullion bank will sell varies depending on the volatility of the gold price, for example, against a put option for 1000oz, a bullion bank may only sell 100oz of gold – this is called delta hedging).

From the above, we can construct what sort of gold “assets” a bullion bank can hold:

  • Unsecured mine short sales
  • Unsecured speculator short sales
  • Unsecured gold business lending
  • Secured mine short sales
  • Secured speculator short sales
  • Secured gold business lending
  • Futures (long)
  • Options (sold puts, purchased calls)
  • Other derivatives
  • Unallocated gold held with other bullion banks or central banks
  • Allocated gold held with other bullion banks or central banks
  • Physical gold in vaults under their control

In addition, each of these (except for the last three, which are on-call) will have different dates at which the contracts mature, that is, when the bullion bank gets the gold back.

Gold Credit

The discussion above about on-call deposits funding gold assets implies a traditional view of banking that banks take in deposits and lend them out. This overlooks the creation of credit money directly by banks, as the Bank of England explains here. In the same way, bullion banks can create credit gold (that is, unallocated). So their gold balance sheet can consist of unallocated liabilities created “out of thin air” backed by promises to repay gold.

Consider a simple world with one Miner, one Refiner, one Bullion Bank, and an Investor.

A Miner delivers dore to the Refiner for refining. Due to competition, these days Refiners pay Miners for their dore once an assay has been completed, which is usually in a couple of days and well before the Refiner has been able to actually refine the dore. The assay reveals that the dore contains 12oz of pure gold and the Refiner quotes an outrageous (but easy for me to calculate) charge of 2oz in refining fees.

As the Refiner does not have any gold to pay the Miner, it asks the Bullion Bank for a 10oz gold loan. The Bullion Bank agrees to do so at an outrageous rate of 10%, and creates unallocated gold credits out of thin air. At this point the Bullion Bank’s balance sheet looks like this:


The Refiner then instructs the Bullion Bank to transfer unallocated gold to the Miner, as payment for the dore (usually done via loco swaps). The Bullion Bank’s balance sheet now looks like this:


Note: while the Bullion Bank does not hold one ounce of physical, the Refiner is holding physical, making their promise to repay the gold loan credible, but not without risk. The Miner needs cash, rather than gold, to pay wages and other expenses, so they enter the marketplace to sell their “gold”. As it happens there is an Investor who is interesting in holding some (unallocated) gold. The Miner instructs the Bullion Bank to transfer gold to the Investor. The Bullion Bank’s balance sheet now looks like this:


Meanwhile, the Refiner diligently works to turn the dore into 99.99% pure 1 oz gold bars. After one year (very inefficient but easy to calculate interest), the Refiner delivers 11 x 1 oz gold bars to the Bullion Bank as repayment of the loan and interest of 10%. The Bullion Bank’s balance sheet now looks like this:


The process above in effect is no different to a gold real bill (see here for a discussion of real bills by Professor Fekete), where the Refiner issues the Miner with a real bill for 11 oz of gold and the Miner discounts that bill with the Bullion Bank for 10 oz of gold. It is why Keith Weiner says that the gold lease rate is really a discount rate. People who understand the real bills doctrine may find it interesting that in the professional market, bullion banks charge a small fee on unallocated balances – discounting in another form perhaps?

The gold credit creation process above is in my opinion a legitimate function of bullion banking that facilitates the business of gold manufacturers and distributors getting gold into investors’ hands, a good thing we would all agree. The Investor in our example is saving in gold and financing the industry by the act of holding unallocated and deferring a desire for physical gold.

Our simple example can be expanded to many more participants, like bullion distributors and the like. Indeed, most of The Perth Mint’s large bank distributors pay for coins by unallocated credits and the Mint uses these unallocated credits to pay Miners for dore, which is made into coin and so on in a continuous flow. Another quote from Professor Fekete is relevant here to explain why this type of fractional bullion banking is OK (and 100% reserve banking is flawed):

“The notion that the bank’s promise, if it is to be honest, forces it to have a store of gold on hand equal to the sum total of its note and deposit liabilities stems from a fundamental confusion between stocks and flows. The promise of a bank, as that of every other business, refers to flows, not stocks. The promise is honest as long as they see to it that everything will be done to keep the flows moving. In the case of the bank, the promise is honest as long as the bank carries only self-liquidating bills, other than gold, in the asset portfolio backing its note and deposit liabilities.”

Note the last sentence in that quote. Bullion banking is safe “as long as the bank holds only gold and self-liquidating bills [ie loans to the gold industry] to cover the bank note [ie unallocated] issue, it changes neither the supply of nor the demand for credit”. Maturity transformation risk is the extent to which the gold assets a bullion bank holds do not mature into physical gold within a short time and instead are being used to fund outright speculative short selling and much longer term financing.

Risk of the Gold Assets

The many gold “assets” mentioned earlier have varying levels of certainty that they will mature into physical gold, or risk of non-payment. These assets are often referred to generically as “paper gold”, but this hides great differences in riskiness between them and is so overused that people fail to appreciate the real risks involved. So what are these risks?

For some paper gold instruments it is quite easy to estimate the size of the exposure, for other more complex derivatives, a bullion bank would rely on something like Black Scholes model and it is here that a lot of risk is introduced. Consider these limitations of Black Scholes from that Wikipedia link:

  • the underestimation of extreme moves, yielding tail risk, which can be hedged with out-of-the-money options;
  • the assumption of instant, cost-less trading, yielding liquidity risk, which is difficult to hedge;
  • the assumption of a stationary process, yielding volatility risk, which can be hedged with volatility hedging;
  • the assumption of continuous time and continuous trading, yielding gap risk, which can be hedged with Gamma hedging.

The article I think naively says some of these risks can be hedged, with other derivatives! But then how are these valued, using similar formulas? Ultimately, there is just another counterparty on the other side and we get back to these assets being either an outright promise (unsecured) or a promise covered by collateral or margin. But that collateral itself needs to be valued – by those same formulas in many cases. And how to determine the amount of margin? By those same formulas.

It is the false assumption underlying much of the formulas used by the bullion banks to work out how to “hedge” themselves that can introduce systemic risk, as this article The mathematical equation that caused the banks to crash explains.

In it Professor Ian Stewart notes that even though the Black-Scholes equation was based on false assumptions “the model performed very well, so as time passed and confidence grew, many bankers and traders forgot the model had limitations.” Are the people within bullion banks considering tail, liquidity, volatility and gap risks? And if they are, are they looking at it from the same viewpoint that gold investors do, which is one that looks over a long timeframe and is more adverse to extreme events?

By way of example, some years ago The Perth Mint was looking at Treasury software packages. I remember the salesperson saying that the software had all the complex “formulas” inside it and worked them all out for you. I asked where it got the key inputs from, like volatility. The answer was from one year’s worth of data of the underlying asset! That didn’t seem to me to capture events like the 1980 $850 boom and bust.

In addition, will a bullion bank’s gold assets be robust in the face of extreme events? Consider the new branch of mathematics called complexity science, which Professor Stewart explains “models the market as a collection of individuals interacting according to specified rules” and which reveals that “virtually every financial crisis in the last century has been pushed over the edge by the [traders] herd instinct. It makes everything go belly-up at the same time.”

The liability, or sources of funding, side of a bullion bank’s balance sheet are also relevant here. It is obvious that people buying and leaving gold with a bullion bank, as unallocated, is a big source of funding. But bullion banks can also acquire funding via derivatives, or to be more accurate, net off their assets with opposite ones of a similar type. For example, long futures against short futures, or options against options.

However, these would rarely line up in terms of maturity, so on top of the misestimation of the value of these paper gold instruments, outright counterparty exposures, inadequacy/variability of the collateral/margin calculation, you have maturity transformation – a deliberate mismatching of maturities of these products to their sources of funding, which requires that if needed, new sources of funding can be found or existing ones rolled over with little problem.

Considering all this complexity and room for error one would conclude that we have a highly unstable system, one that Nassim Taleb would call fragile and sensitive to stress, randomness and disorder.

However, the fact is that the bullion banking system has not failed, notwithstanding the many calls that it would default or fail. For example, in 1998 open interest vs stocks exceeded 40:1 yet there was no failure. What about LTCM, or AIG (40:1 gold leverage as per Jeff Christian) – shouldn’t that have been enough to blow up the system? So how do we explain this apparent robustness? In our next post we will delve deeper into the risks and the mechanisms that control it.