May 26, 2000. Gold: Can’t Bank with It; Can’t Bank without It!

The huge jump in the gold derivatives of Deutsche Bank and Dresdner Bank in the last half of 1999 invites all sorts of speculation, particularly when coupled with similar increases at Morgan Guaranty Trust Co. and Citibank. But as intriguing as such speculation is, it should not be allowed to obscure the more important as well as more factual story that resides in the larger BIS numbers of which the figures for these four banks are only a part.

Much has been written recently about the short position in gold. In its narrowest sense, this short position is the accumulated physical gold transferred by deposit (loan) from central banks and others to bullion banks. This gold creates deposit liabilities on the balance sheets of the bullion banks. It must be repaid in gold. Virtually all of it has been sold by the bullion banks into the market, creating various paper assets on their books. Besides central bank deposits, gold deposit liabilities of bullion banks include unallocated gold, often in certificate form, of private parties. But the key point is that this physical gold has left the vaults of the banks and the control of the bankers. It can only be replaced by new production or market purchases, and thus constitutes a net short physical position of the bullion banks.

The total net short physical position of the bullion banks is reliably estimated at between 7000 and 10,000 tonnes, although it could be higher. Ultimately its size is limited by the willingness of gold owners to lend and of gold users (e.g., producers, fabricators, bullion banks, speculators) to borrow. Prudence dictates that the net short physical position be quite reliably hedged, e.g., delta hedging in physical bullion, contracts for forward delivery from gold mining companies, call options on central banks with large gold reserves, or other instruments where the risk of counterparty default on the obligation to deliver physical gold appears minimal.

In a perfectly prudent world, the net short physical position would roughly correspond with the net short gold derivatives position. However, in the absence of such a world, the net short gold derivatives position tends to be larger than the net short physical position. This phenomenon results because while part of the gold derivatives position may be hedged in the physical market or reliable substitutes, other parts may be hedged in less reliable forms of paper gold or even unhedged, such as naked calls.

As discussed in an earlier commentary, writing naked call options can be a very effective means of adding gold to the derivatives market, thereby putting downward pressure on the gold price. Of course, writing naked calls is also a very risky activity. But it demonstrates a key point: the net short gold derivatives position is ultimately limited only by the prudence of the least cautious players and, if applicable, the willingness of governments or other official agencies to back them.

As summarized in tabular form in the prior commentary, the recent figures from the BIS on the total size of the gold derivatives market are important because they suggest: (1) that the central banks may have loaned much more gold into the market than previously thought; and/or (2) that the net short gold derivative position is far larger than suspected or than anyone would deem prudent.

Before addressing these two alternatives, three points about the BIS figures should be emphasized. First, under the Basle Capital Accord, all off-balance-sheet exposures, specifically including gold derivatives, are subject to the capital adequacy standards. Second, the Basle Committee on Banking Supervision has adopted a number of recommendations “to encourage banks and securities firms to provide market participants with sufficient information to understand the risks inherent in their trading and derivatives activities.” And third, the statistics released semi-annually by the BIS on the global OTC derivatives market are an integral part of this risk assessment and disclosure process. A wealth of further information on all these subjects is available at the BIS’s website (

Turning specifically to the derivatives market statistics, the consolidation of notional value at the BIS level is intended to eliminate double-counting between reporting banks and dealers so that the total notional figure is in effect a measure of market size at a point in time. It is not turnover such as reported by the LBMA. For commodities and gold, the closest analogue would appear to be open interest on a commodities exchange. But OTC contracts being non-standard, counting the number of contracts obviously does not work. Accordingly, the best way to measure them is by underlying contract amounts or face values, halving those between reporting parties and taking other steps, as the BIS does, to avoid double-counting.

At the end of 1999, the BIS put the total notional amount of gold derivatives at US$243 billion, up from $189 billion at the end of June. Converting the year-end notional amount to tonnes at the year-end gold price ($290/oz.) gives just over 26,000 tonnes. Using a $300 gold price gives around 25,200 tonnes. However, these 1999 figures are for major banks and dealers with their head offices in the G-10 countries only.

On a more irregular basis, the BIS collects similar information for as close to the whole world as it can. Its last larger survey as of the end of June 1998 showed a total global figure for gold derivatives of $228 billion compared to a G-10 figure on the same date of $193 billion, indicating that at that time there was an additional $35 billion (or 3629 tonnes @ $300/oz.) in gold derivatives outside banks and dealers headquartered in G-10 countries. Accordingly, assuming a continuing difference of around the same magnitude, the total global gold derivatives market is on the order of 26,000 to 28,000 tonnes, more than twice the higher estimates of the net short physical position, and almost as large as the stated gold reserves of all the world’s central banks put together.

What does this number mean? How should it be interpreted? Is it really as large as it looks? Analysts are unlikely to agree. I look at it this way. If the net short physical position is 10,000 tonnes, and if that position has been fully hedged (far from certain), the total notional value of all that business should be around 20,000 tonnes. In that event, using 26,000 tonnes as the total notional amount for all gold derivatives, the net short gold derivatives position — over and above the net short physical position — is about 6000 tonnes, and the bullion banks have undertaken to deliver this amount in addition to what they must deliver to cover the net short physical position of 10,000 tonnes.

In other words, as I define it, the net short gold derivatives position is the amount by which the total notional value of all gold derivatives exceeds twice the net short physical position, and it must be added to the net short physical position to get a total net short position, i.e., the amount of gold that the bullion banks are committed to deliver on their outstanding paper. This formula effectively assumes that the net short physical position is fully hedged with zero risk of default but that any derivatives position in excess of twice the net short physical position represents a short position without reliable hedges in physical gold. [For a fuller explanation of the reasoning behind this formula, see the explanatory note at the end of this commentary.]

To return to the two alternatives posed earlier, is it possible that the physical gold borrowings underpinning total notional gold derivatives are larger than thought, thereby reducing, assuming they are credibly hedged, the net short gold derivatives position?

Of course it is. Here are the obvious possibilities. One, the central banks have on a net basis leased more gold than thought, thereby increasing the liquidity base of the gold derivatives market. Two, the central banks have on a net basis sold more gold than thought, and this gold has remained within the gold banking system, i.e., sold to investors or others who have deposited it in unallocated accounts. Three, the central banks have written covered calls in far greater volume than thought, thereby providing an additional sound base for gold lending. All these possibilities require that the central banks on a net basis reduce their combined official gold reserves by much more than they have suggested they are willing to do or have done.

Indeed, looking at what central banks say and allowing for significant private lending as well, it is very hard to come up with even 7000 tonnes as a net short physical position. In that case, the net short gold derivatives position would be 12,000 tonnes (26,000 – 2 x 7000), and the bullion banks are in even more parlous condition under my rough formula. But whatever the exact numbers, and whatever the motives of some of the bullion banks or the officials who have supported them, there is a far more fundamental but apparently unrecognized problem here. By historic standards, the bullion banks are operating with wholly inadequate gold capital and gold reserves for the very extensive gold banking business that they have built.

Everyone, bankers and regulators alike, appears to have assumed that gold derivatives are a more or less ordinary variety of commodity derivative. However valid this assumption may be for markets like the COMEX or the TOCOM having open interest of only a few hundred tonnes, or for a single bullion bank with gold derivatives of similar size, it is dangerously false as applied to a total gold derivatives market where deposits or their equivalent exceed 10,000 tonnes. Exposures that amount to but fractions of annual new production are one thing; a total exposure equal to several years of new production is quite something else.

The Basle Capital Accord provides two methods for determining capital adequacy for off-balance-sheet gold derivatives: the current exposure method and the original exposure method. Without going into all the details of these formulas, which are basically the same as for exchange rate derivatives, suffice it to say that the capital adequacy requirements range from 1% of total notional value for maturities under one year to 7.5% (or higher under the original exposure method) for contracts over five years. These percentages are less than those for equity, other precious metals or other commodities derivatives. Furthermore, there is no requirement that any of the necessary capital be held in gold bullion.

Under this regime, largely on the basis of deposit liabilities by the central banks and others amounting to something like 10,000 tonnes, the bullion banks have built a gold lending business of equal size, to which they have added a net short derivatives position of another 6000 tonnes. All this has been done at gold prices in the $300/oz. neighborhood. And so far as can be told, almost all the gold deposited with the bullion banks has been sold into the market and disappeared from the gold banking system to India and other parts of the Middle East and Asia.

Accordingly, where by traditional standards the bullion banks should be holding in their own vaults on the order of 5000 to 6000 tonnes of gold reserves, under the new gold banking paradigm they are apparently almost completely naked to about this same amount of gold liabilities. On an individual basis, perhaps, the gold derivatives business of some bullion banks may look like commodities trading. But taken as a whole, the gold derivatives business of all these banks has evolved into nothing less than full-scale gold banking, which done prudently has always required immediately available gold reserves equal to 35% to 40% of deposits.

This situation is ironic testimony to the true nature of gold as permanent, natural money. Bankers and governments could not abide the discipline of the gold standard, even in watered down forms such as the gold exchange standard or Bretton Woods. But even after expunging from the banking system any formal role for gold, neither the central banks nor the private banks capable of acting as bullion banks could resist the temptation to engage in gold banking. As money just lying around, the allure of gold proved too strong for the bankers, who now calling it a commodity, proceeded to reestablish an enormous gold banking business while disregarding all the prudential rules that several hundred years of gold banking experience had taught.

In the process, they fostered the illusion that low gold prices demonstrated confidence in their paper money system whereas in fact these low gold prices reflected only the reckless abandon with which they were creating paper gold liabilities in lieu of physical gold. These low gold prices had two further deleterious effects. First, they encouraged the flow of physical gold to parts of the world where gold’s true value is still appreciated, and from which only much higher prices will cause it to return. Second, they decimated the gold mining industry worldwide, all as brilliantly set forth in John Hathaway’s newest essay, The Folly of Hedging (

An old saying, long forgotten, is about to take on new life: “There’s no rush like a gold rush, and no run like a bank run.” In these circumstances, the safest gold is not in bank storage of any variety. It rests in more imaginative places: the snake pit, the closet with the black widow spiders, or buried in the backyard near the Doberman’s bone and well-within range of his leash.

[Explanatory Note: The formula is essentially arbitrary, but reflects my view of market events. If the net short physical position is less than 10,000 tonnes, the net short derivatives position is larger, in which case more of it is likely to be effectively hedged. If the net short physical position is more than 10,000 tonnes, the assumption that it is fully hedged with no effective risk of default is even less tenable. A cursory review of Cambior’s hedge book shows how shaky some so-called hedges on this short position were even prior to the Washington Agreement. Accordingly, it seems quite doubtful that bullion banks writing calls for the likes of Barrick in the wake of that agreement were able to hedge effectively, particularly with many mining companies trying to reduce their forward exposure. That the big increases in gold derivatives during the last part of 1999 were concentrated in just a handful of well-connected banks only adds to the doubts. However, I am assuming that any official support for the net short derivatives position is not included in the BIS figures on notional values.

Both sides of my equation have flaws: the net short physical position is certainly not perfectly hedged; and some of the net short derivatives position may encompass adequate hedges — from official sources or otherwise — for what appear to be mostly short positions. What I am guesstimating is that assuming a net short physical position of around 10,000 tonnes, these two sets of errors more or less offset each other. And what I am further suggesting is that even if the net short physical position is adjusted up or down, the exposure on the net short derivatives position remains at roughly the same order of magnitude.]