We have come across many articles recently (here is an example) discussing the decline in COMEX inventories and/or the decline in inventories held by physical gold ETFs like GLD. Invariably the writers conclude that these inventory declines must be a bullish sign. We don't believe this conclusion to be correct.
We want to get one item immediately out of the way, because there seems to be a lot of misunderstanding on this particular issue. Many observers argue that if the so-called "registered" inventory (metal designated and available for deliveries) were to decline below a certain level, then there could be a "delivery default." This is highly unlikely.
First, historically only between 1.5% and 3% of open COMEX gold contracts ever get to the delivery stage. Second, "eligible gold" can move into the "registered" category and vice versa. Third, gold that is to be delivered may be stored at any warehouse licensed by the exchange and it is up to the seller which one he wants to use. Fourth, movements from one vault to another are actually entirely independent of deliveries, which consist simply of paperwork -- proof that the seller has paid storage up to delivery day, and issuance of a receipt (called a "warrant") that confers ownership. Other than that, the exchange specifies that deliverable gold must be of a minimum fineness of 0.995. The rules can be viewed in detail here (PDF).
The new owner of the warrant can then decide whether to move the gold out of the vault, leave it in the "registered" category, or move it into the "eligible" category ("eligible" gold is gold fit for delivery, and as noted above may at any time be moved into the registered category for delivery purposes). Note that the categories "house" and "customer" accounts that are identified on COMEX delivery reports are not congruent with "eligible" and "registered." Both categories of traders may hold receipts for either types of gold.
So if you take delivery, there isn't going to be a DHL truck showing up one day to dump the gold on your front lawn. Moreover, it is irrelevant (gasp!) if JPMorgan has "enough" gold in its own licensed warehouse to deliver into the contracts it has left open past notice day, since it can deliver gold held at another licensed warehouse. Finally, should a member ever default on a delivery, there are specific rules dealing with such situations. There may be a force majeure situation for instance. In case of a genuine default, the clearing house will refund the cash value of the underlying commodity to the buyer, but it has no obligation to actually deliver the underlying commodity. The specific rules concerning these cases can be viewed here (PDF).
So much for the technicalities. Below is a chart of COMEX warehouse stocks over the past decade. Examine it closely -- it does actually have some informative value as we will explain.
COMEX gold warehouse stocks, via Sharelynx.
Gold Inventories and Market Direction
Let us examine the claim that declines in warehouse inventories are bullish for gold. It seems to us that this idea is based on confusing gold with industrial commodities such as copper. Since copper inventories typically cover only a few months of demand, declines in copper warehouse stocks can be a warning sign of an impending shortage. Copper is largely "used up" -- i.e., it is a fairly non-specific raw or intermediate (in the case of copper cathodes) capital good that is employed in the production of other capital or consumer goods. Since only a few months worth of supplies are held above ground, a decline in warehouse stocks of copper can be a bullish sign for the commodity.
Gold, on the other hand, is primarily a monetary metal and very little of it is used up in industrial applications. Essentially, all the gold that has ever been mined still exists. Therefore, when gold inventories at COMEX warehouses decline, it means only that the gold has moved from vault A to vault B situated somewhere else. This is neither bullish nor bearish, it is essentially meaningless. The gold price does not depend on where specifically gold is held.
Gold comes in various forms. There is jewelry, gold coins, small bars, and finally the so-called "good delivery bars" of approximately 400 ounces each. This is the form in which central banks as well as large-scale investors tend to hold their gold. The standardized bar size and shape makes it easy to stack the gold, and at approximately 12.5 kilograms, the bars are not too heavy to handle. It is, of course, possible that gold held in the form of such bars will be melted down by a refinery to produce, e.g., gold coins, depending on specific short-term demand and supply conditions with regard to gold in certain forms.
Consider the following now: There is still the same amount of gold "out there" as before, it has merely moved from one vault to another, presumably changing ownership in the process. This change of ownership has coincided with the following: much lower gold prices and a sharp decline in the commercial net short position and its obverse, the speculative net long position in COMEX gold futures. Moreover, a notable drawdown in other widely watched inventories, such as GLD's gold holdings has occurred at the same time.
Gold held by GLD, in tons, via Trader Dan. This chart is slightly dated -- in the meantime, the trust's holdings have declined further, to 915 tons as of Aug. 6.
While gold was in a bull market, both COMEX and GLD inventories kept growing incessantly, as did the commercial net short position and the speculative net long position in gold futures. All of this changed once gold began its recent price decline.
Note that since GLD is an open ended mutual fund, its holdings mainly change due to arbitrage: whenever the trust's shares trade at a discount to the underlying metal, authorized participants will sell the metal short and buy the trust's shares. Thereafter they will obtain the metal by redeeming the shares. Conversely, if the shares trade at a premium, they will buy the metal and sell short the shares concurrently. Then they will create new "baskets" of shares by delivering the metal they have bought to the trust. Obviously GLD shares will only trade at a premium when demand for them is very strong -- i.e, when investors who want to obtain exposure to gold via GLD show more urgency than those who want to reduce their exposure or are taking a bearish position.
From this context we can conclude a few things. One, the commercial short position in gold futures at least partly hedges inventories held in COMEX warehouses. Thus the decline in the commercial short position occurred concurrently with the decline in inventories. Two, COMEX and GLD inventories both can be regarded as representing a microcosm of investment demand for gold (specifically demand in the West). Therefore, rising inventories tend to coincide with bullish phases and declining inventories with bearish phases. They are therefore a sentiment indicator. Given that the growth and decline of inventories as depicted above coincided with a bull and a bear market respectively, there is empirical evidence for this contention in addition to what we can logically infer from GLD arbitrage.
When looking at a very long-term chart of COMEX gold inventories we can see that exactly the same thing happened in previous bull and bear markets, beginning with the time when it once again became legal for U.S. citizens to hold gold in 1975:
COMEX inventories, long term. Rising inventories roughly coincide with bull markets, falling inventories with bear markets.
The proper conclusion therefore should be that the decline in inventories simply mirrors the bear market and is a coincident indicator of it. It can only be considered bullish if an argument can be fashioned that the decline in gold's price and the associated decline in GLD holdings and COMEX inventories have been excessive and will therefore change direction again. (Note that the change in inventories as such should not matter to gold's price at all -- apart from the fact that it doesn't matter where exactly the gold is stored, the amounts involved are too small relative to the market's total size. As noted above, one can at best regard them as reflections of investor sentiment.) In other words, one must expect that the opinion of market participants will change again in the future, and that gold demand in toto, including its most important element, namely reservation demand, increases again.
Such an assessment cannot be made by watching changes in inventories at specific warehouses. Instead it requires that one form an opinion on the following factors:
- the damage done by central bank policy to date (i.e., its likely future effects),
- the likely thrust of central bank policy in the future (including the growth of the money supply) and the market's perception thereof,
- the likely direction of real interest rates,
- future movements in credit spreads and the state of economic confidence more generally,
- the steepness of the yield curve,
- the desire of people to increase or decrease savings,
- the likely future direction of the U.S. dollar's exchange rate,
- the rate of growth of the federal deficit, and
- the likelihood of other events that may upset market confidence (such as a potential revival of the euro area debt crisis or a general decline in confidence in the monetary and/or fiscal authorities).
These factors can be assessed, estimated, and weighed. If a majority of them is thought to tilt toward a gold bullish configuration in the future, the price of gold is likely to rise and vice versa if that should not be the case. The list above also helps to explain why the gold price has declined since its 2011 high -- too many of these factors have remained in a gold-bearish configuration for longer than many (including us) expected.
This, in turn, has altered market psychology so that a majority of market participants now expects more of the same. This is probably erroneous, especially with regard to the longer-term outlook, but it is what has happened.