In a recent article, The Case Against Gold David Berman interviews a Canadian fund manager who is bearish for all of the wrong reasons. As do the vast majority of analysts and writers, he totally misunderstands how the gold price is formed.
While Berman is correct that “gold must obey the law of supply and demand”, his explanation of how the law works is fatally flawed. The supply and demand numbers presented in the article are meaningless and tell us nothing about the future direction of the gold price.
The key to understanding the gold price is that gold is in demand as an asset, not as a commodity. A commodity is a good that is purchased in order to be permanently removed from the market (usually destroyed), while an asset is a good that is purchased because the buyer values it more as a holding than for its direct use. The most important consequence being held rather than destroyed is that the accumulated stockpiles of exceed annual production by a large margin. In the case of gold the ratio is between 50:1 and 100:1 while for most (consumed) commodities it is less than 1:1, (i.e. less than one year’s total production is held in above-ground stockpiles).
For a commodity, everything that is produced is sold; everything that is sold is purchased; and everything that is purchased is consumed. The act of consumption removes the supply from the market permanently. For the moment, let’s idealize the situation a bit by assuming that there are no above-ground stockpiles of a commodity at all. Subject to this simplifying assumption the following is true: the amount traded is the same as the amount produced which is the same as the amount consumed.
This mathematical identity is not true for an asset. Because so much of the asset already exists and so little is produced, most of the trading takes place among buyers and sellers who are only shifting the existing stock piles among themselves. The equality of the quantities produced, supplied, demanded, and consumed does not hold. Small amounts of gold are consumed for industrial uses but the quantity permanently consumed can be considered effectively zero because most gold is either held in investable form (bars and coins) or as jewelry. The quantities bought and sold are by definition equal, but the quantities produced and consumed are not equal, nor does the quantity produced necessarily equal the quantity bought and sold. The quantity bought and sold can exceed quantity produced by an enormous ratio because there is no limit on how much existing stockpiles of the asset may be traded during a one-year period.
The different relationship between the quantities produced, consumed, and traded between a commodity and an asset is the source of much of the confusion about gold. For a commodity, increases or decreases in the quantity produced can have a dramatic effect on the price because the quantity consumed must move either up or down along with the quantity produced. Because there are no stockpiles to buffer the difference between production and consumption, the only way for them to come into balance is through price. But for an asset, things work differently. For an asset, most trading consists of the change of ownership of existing stocks. The gold that gets produced will certainly be sold, but trading volumes during a year will far exceed new mine production because of the large volume of trade of existing holdings.
To quantify the impact of mine supply on the market I have (elsewhere) estimated the trading volume on the London Bullion Market Association (LBMA). The LBMA can absorb an entire year’s mine output in about twelve trading days. And consider that the LBMA is only one of several investor bar markets worldwide, and that the investment market as a whole includes the coin market as well. While I do not have a precise number, I estimate that an entire year’s worth of mine production turns over in the physical market in several days.
Because mine supply is small compared to existing holdings and because the volume traded exceeds mine supply by a large ratio, mine supply is quickly absorbed into the market and has little impact on the gold price. Far more important is the price at which existing sellers are willing to part with their ounces.
With this in mind, let’s return to the article. Berman states ”the underlying fundamentals are now looking distinctly negative for the metal’s long-term prospects.” As we will see reaches this conclusion by looking at the market only on an annual production basis:
“According to GFMS, the London-based precious metals consultancy, global jewellery purchases plunged to 1,759 tonnes last year from more than 3,000 tonnes at the start of the decade, as the rising price of gold turned off consumers and jewellery makers cut back on gold content.”
As explained above, looking only at annual production ignores most of the gold market. The entire gold market is a single integrated market with one price. There is not one market with one price for gold produced this year and another market selling at a different price where existing stockpiles of gold are traded. Looking at gold on an annual basis gives the appearance that the supply – and therefore the demand – are highly variable. If the supply goes from 1759 to 3000 over a few years – almost a 100% increase — then demand must also nearly double, otherwise the price would plunge, right? Well, no. The supply of gold is not 1759 or 3000 tonnes, it is about 165,000 tonnes – the total all mined gold in history that has not been lost or destroyed. And what about demand? The demand is also 165,000 tonnes, exactly equal to the supply. And the supply has not doubled over the last few years, it has grown modestly by about 1-2% each year for a cumulative increase of less than 20%.
What does it mean to say that the demand for an asset is equal to the supply? For a commodity, the demand consists of off-take from the market for destruction. An asset, on the other hand, is not destroyed; it is acquired in order to be held. The demand for an asset consists of what economists call reservation demand, meaning that investors demand an asset by holding it off the market while the price is below their selling price. Demand for gold has not and does not need to double to keep pace with mine supply. As the supply grows by about 1% each year, reservation demand must only grow by the same amount in order to keep the market in balance.
Here is where we get into the meat of Berman’s case for the bearish gold supply fundamentals:
“At the same time as demand is falling, gold supply is rising. Most central banks, for instance, are jettisoning their gold holdings. According to the World Gold Council, total gold holdings were about 30,600 tonnes in December, down nearly 2,900 tonnes since the start of the decade – and these overall declines take into account an increase in gold holdings by China’s central bank
Adding to supply is ramped-up production from mines. This rose to a four-year high of 2,572 tonnes in 2009. In addition, consumers are now happily mailing in their “scrap” gold jewellery to the host of gold-dealing companies that have sprouted up in recent years. Add in this recycled gold and total world supply hit 4,034 tonnes last year, the highest level in at least a decade.”
Correction, Mr. Berman: a seller selling gold is not an increase in supply. It is only a transfer of existing supply from one owner to another. It does not matter whether this seller is a central bank or someone melting scrap. And moreover, is central bank selling a huge bearish indicator? 3,000 tonnes per year is about annual mine supply. This is, as noted above, a few trading days’ volume. If there are no buyers near the market price, then it could be bearish, but not nearly as bearish as it sounds when you consider that the total supply of gold is 165,000 tonnes – 50 to 100 years annual production. The real influence on the price is not possible to quantify. Once the gold changes hands, the price depends on the reservation price of the new buyers, which could be higher than that of the recent seller.
Berman also misunderstands the relationship between investment products and jewelry:
“Unlike most commodities, gold isn’t consumed in high quantities. Industrial uses account for only about 10 per cent of total demand, which is why jewellery makers have traditionally provided most of the market for the metal.
However, global demand for gold jewellery has been in steady decline, replaced only by demand from fickle investors – a shift that could have a disastrous impact on the price of gold if those buyers turn squeamish.”
“Of course, gold is still in high demand. But the source of this demand is now investors, who have become gold’s biggest buyers for the first time in about 30 years. These buyers have no uses for gold, other than the hope of selling it to someone else at a higher price somewhere down the road.”
In reality, investment and jewelry are not so different. They are two different ways of holding gold. In either form, the gold is not destroyed. The boundary between them is somewhat fluid based on the relative intensity of investment demand versus jewelry demand. As gold becomes more in demand by investors demanding bars and coins, the opportunity cost of using it for jewelry increases. At the margin, more people will sell back unwanted jewelry as scrap and (also at the margin) more buyers will substitute silver or other metals for gold. If the investment demand fell, then at the margin people would hold onto more of their family jewelry and use more gold in new jewelry fabrication. What Berman describes as a fall in demand is only a natural shift between different ways of holding gold as an investor and consumer valuations change.
The intentions of the jewelry buyer and the investment buyer may be different. Investment buyers hold gold as a store of value, while jewelry buyers may use it for ornamental purposes. But the gold is still held and most jewelry can and does eventually return to the market as melt. And furthermore, in some parts of the world a lot of jewelry is purchased as a store of value – a wearable investment. In the West investors prefer to hold gold as bars or coins while in Asia and the Near East, some jewelry purchases are investments rather than ornamental. Bullion jewelry has a low workmanship value-added compared to ornamental jewelry which sells for substantially more than its gold melt value.
Berman also makes the common error of quantifying gold demand by counting trading volume. As explained above, for a commodity the quantity traded is equal to both production and consumption, but not for an asset. For an asset, the supply and the demand are both equal to the total quantity stockpiled. There is no obvious relationship between production and trading. High market volume in the gold market means a high level of demand and an equally high level of supply. This does not provide a clear indicator of the price direction in the market, though this does not stop many analysts from pointing to high trading volume as evidence of either a lot of demand (if they are bullish) or a lot of supply reaching the market (if they are bearish). We cannot use the trading volume as a measure of supply and demand because the price at which the existing supply changes hands does not depend on the trading volume: the price can rise, fall, or go sideways on increasing, or decreasing volume.
An example that I like to use to illustrate this point is if an analyst studying a particular stock (corporate equity) wrote “demand for the shares of corporation XYZ were 30M shares…but we expect demand for the shares to increase next year to 40M shares, which is very bullish”, then anyone reading this would have to scratch their head a bit in trying to understand what the analyst meant. In fact, the statement is meaningless — the analyst would be spouting gibberish. The deep confusion exhibited by writers and analysts on this issue contributes to the lack of understanding by investors and the continuing stream of articles such as Mr. Berman’s piece.
One of the best indicators that the an analyst is in a state of deeply misguided thinking is that he provides two different numbers for supply and demand. The numbers result from adding up some fraction of the total gold trades on each side of the market and calling those numbers “supply” and “demand”. As noted above, trading volumes do not measure supply differently than demand. If the supply and demand number are not equal, then the analyst has counted some trades as supply only and others as demand only. These numbers do not mean what these analysts say they do and they tell you nothing about the direction of the gold price.
When reading interviews and articles by prominent gold analysts – such as Jon Nadler and Jeffrey Christian – providing price forecasts based on quantitative figures for supply and demand, keep in mind that these numbers are meaningless and so then are their interpretations of the numbers. Whenever I listen to or read a Jeffrey Christian interview I feel the urge to grab him by the shoulders, shake him, and shout “Stop talking nonsense, man!”.
In the end, the gold price is not analyzable by quantitative measure of supply and comparing it to a quantitative measure of demand. Supply and demand are identical and both are equal to the total stock of gold. The price depends on the reservation price of the owners of existing stockpiles as compared to the buy price of existing holders of fiat money. This cannot be measured other than by the gold price itself.
How then do we forecast the price of gold? Personally I am sympathetic to methodologies that compare the purchasing power of an ounce of gold over time to other assets such as a man’s suit or the DOW index. Another approach that makes some sense to me is looking at historical trends in gold’s total “capitalization” compared to that of other financial assets. But it’s hard to point to an arbitrage that would bring the price of gold in line with historical values of these metrics.