As an appendix to yesterday’s article, ‘Misunderstanding gold demand‘, I have produced a detailed analysis of The CPM Group’s gold research. Of all of the major research firms, they have the most information available about how they think. While I chose to focus on the CPM Group, as far as I have been able to determine from published reports, GFMS agrees on the basic framework of the quantity model, if not on all of the particulars. The other major research firm, the WGC, uses the numbers compiled by GFMS. From those aspects of the CPM Group’s thinking that are available to the public, I have tried to reverse engineer how they see the gold price from what is available.
In the next section I will explain my interpretation of their thinking. And then, in the following section I will provide my critique of the CPM Group. I have found several problems in their model: the first, that they consider market data on a year-by-year segregated basis; the second, the belief that gold holdings are not part of the market; and third, the premise that net flows drive the gold price. I will discuss each of these points in more detail below.
The CPM Group’s Model
My sources consist of the CPM Gold Yearbook 2012 (cited below as Yearbook) and a 1996 presentation to the Australian Gold Conference by the CPM Group’s founder Jeffrey Christian, Gold: Supply, Demand, Price and Research (below, AGC Presentation) [1]. After reading the sources, I have synthesized what I believe to be the CPM Group’s model of supply, demand, and price into the following propositions:
Gold Supply: The gold supply consists of mine production plus “secondary sales”. The latter consists of melted gold sold in the form of scrap by the jewelry sector and the fabrication sector. See the Yearbook, p. 105: Gold supply, which includes gold output from mines in market economies, gold exports from transitional economies into market economies, and old gold scrap that has been refined, is estimated to have totaled 119.9 million ounces in 2011….
Gold Demand: Gold demand is the sum of the following: industrial use (electronics, dental, medical, other), gold use to make new jewelry, official sector net flow (IMF, US, Canada, other) and investor sector net flow (coins, bullion, bars, Indian). See the Yearbook, pages 6-7. Investment demand is defined by CPM in some places as the net flow of the investor sector and in other places as the net flow of the investor and official sectors combined. My citations in support of this are:
- Yearbook, p. 4, labels on the chart titled Gold Supply and Demand shows investor demand as the difference between supply (defined as mine + secondary) and fabrication demand. From the quantity balance equation (3, above), total supply less fabrication demand must equal net investor and official sector flows.
- Yearbook, p 11: Net purchases of gold by central banks have complemented healthy gold investment demand…. If net central bank purchases have complemented investor demand, they cannot be the same thing.
- See the Yearbook, p. 29, Investment demand (subtitle), followed by: Net additions to private investor gold holdings declined to 34.3 million ounces in 2011
- Yearbook, p. 8: (chart) Net Investment Demand is plotted on the left scale of the chart with the gold price on the right scale. The chart’s title is Investment Demand’s Effect [sic] on Gold Prices.
- Yearbook, p. 9: Gold investment demand is one of the strongest influences on gold prices.
- Yearbook, p. 10: …it is projected that net additions to private investor gold holdings will remain at extremely high levels, which is expected to help keep prices at elevated levels during 2012.
- Yearbook, p. 69: “Central banks were net buyers of gold for the fourth consecutive year in 2011”. (i.e. net flow into the central bank sector was a positive number).
- Yearbook, p. 233: Strong investment demand for gold, particularly during the first three quarters of the year, pushed prices higher.
- AGC Presentation, p. 4: my next slide [not shown in PDF – rb] illustrates the weightings of each supply and demand sector in CPM Group’s amin gold price model. Central bank activity and investment demand trends … exert much more powerful influences in determining the gold price.
- AGC Presentation p. 8: SLIDE SEVEN: Investment Demand’s Effect on Prices. My next chart compares levels of investment demand for gold to changes in gold prices. We call it our most important chart. Herein lies the key to accurate gold price forecasts: Investment demand is the most important influence on gold prices. Markets are made at the margin, and in the gold market investors are the marginal market participants.
- AGC Presentation, p. 8: Over the past few years, investors have not been buying a great deal of gold. As a consequence, gold prices have languished. Investment demand reached a low of 184 mt (5.9 million ounces) in 1994. Investor demand increased 11.7% in 1995, but the level, at 205 mt (6.6 million ounces), remained low. A further increase, to around 239 mt (7.7 million ounces), is projected for this year. The steady increase is being reflected in the slow upward march in gold prices; the low levels overall are being reflected in the fact that prices are not rising sharply.
Quantity Balance: Demand equals supply. See the Yearbook, pages 6-7. The total supply consists of 31.0Moz mined, 4.5Moz secondary, and 13.2Moz from transitional economies for a total of 50.0Mz. Total demand consists of 43.9Moz fabrication, -8.7Moz official sector reserve sales, and 14.8 net investor portfolio additions, for a total of 50.0Moz.
Quantities Drive the Gold Price: Each of the terms in the quantity balance equation is a contributing cause of the gold price, with net investment demand being the most important cause. If the net investor sector flow is a large positive number, this is deemed a cause of a higher gold price, while a small positive or a negative number is deemed a cause of a lower gold price. My reason for thinking this is the following citations:
Forecasting the Gold Price: Knowing the cause of an event does not necessarily help to forecast the event. A cause can only be used to forecast an effect if the cause occurs far enough in time before the effect that the cause can be identified and acted on.
Mr. Christian states that net investor flows can be used to forecast the gold price. I am not clear from the following whether Christian is saying that the net investor flows in one year can be used to forecast the following year’s gold price, or whether he means that a correct forecast of next year’s net investor flows is the key to forecasting the next year’s price. Below is a relevant passage:
AGC Presentation p. 8: SLIDE SEVEN: Investment Demand’s Effect on Prices. My next chart compares levels of investment demand for gold to changes in gold prices. We call it our most important chart. Herein lies the key to accurate gold price forecasts: Investment demand is the most important influence on gold prices. Markets are made at the margin, and in the gold market investors are the marginal market participants.
Summary: Based on the above, it appears The CPM Group posits the following logic regarding the gold market, though I cannot say for certain because their quantitative model is proprietary.
- Supply is defined as mine plus secondary
- Demand is the sum of the various components.
- The price on any market is set by supply and demand. If the number that CPM has defined as “demand” increases, the price will be higher.
Critique of CPM Group’s Model
Consumption versus Asset
My first criticism of the CPM Group’s model is that its conceives the price formation process only in the context of annual data. In fact, annual production and consumption quantities are quite unimportant in the overall picture of gold price formation. The error of looking at gold as an annual market is widespread and follows from a failure to understand the difference between a consumption commodity and an asset.
Most commodities are produced primarily for consumption. Consumption permanently destroys the economic value of the commodity (or in some cases, makes the commodity costly to recover back to a form where it has economic value). The market demonstrates that a commodity is produced mainly for consumption by the lack of large above-ground stockpiles. A small stockpile measured in terms of production output would be several days, weeks or a small number of months at most. All commodities other than gold have stockpiles that are small by this definition.
In consumption commodity markets, production and consumption must remain very nearly in balance because on the one hand, reserves will be depleted quickly if consumption exceeds production and, on the other hand, production can only exceed consumption if stockpiles increase. Stockpiles generally will not increase much beyond a few weeks or months of production flow because users and speculators have historically demonstrated that they are not willing to hoard larger supplies. For a consumption type commodity, the supply that is produced during a given year, plus small stockpiles, is the only supply available for consumption during that year.
An asset is a good that people buy in order to hold, rather than consume. Some examples of assets are land, property, money, stocks, bonds, and gold. Nearly all of the gold ever mined still exists either as bars, coins, or jewelry. The total quantity of gold stockpiles grows by about 1-2% per year, implying a ratio of stockpiles to one year production in the 50 to 100 x range. Only a small amount of gold is truly consumed. Even jewelry fabrication is not consumption because its bullion value is retained and can be reclaimed at the relatively low cost of melting. Gold is held in a continuum of products which can be transformed from one to the other, such as bars, coins, and jewelry.
Price formation in an asset market works differently than in a consumption market. Because consumption goods are bought in order to be permanently destroyed, buyers must bid for newly mined product. The reservation demand to and from stockpiles does not play much of a role in the pricing process. All possible supply (regardless of price) is from recent production, and any possible demand (regardless of price) is demand for current consumption. In a consumption good market, the lack of stockpiles ensures that:
quantity supplied = quantity produced
quantity demanded = quantity consumed
In an asset market production and consumption (destruction) do not significantly impact the supply or the demand because they both are small compared to the above-ground supply. In a consumption market, the trade is mostly from producers to consumers, while the majority of gold trading consists of movement of gold from one stockpile to another stockpile. The supply schedules are dominated by the offers of existing stockpiles at a range of prices. The demand schedules are dominated by reservation demand to hold existing stockpiles. Selling by producers and buying by (destructive) consumers is small in comparison.
As I have explained here, the gold market is a single integrated market where all sellers compete against all buyers. It is not an annual market for the current year’s supply. Buyers compete to buy any gold, not only for gold mined in the last year. All sellers compete to find buyers. The reservation demand for existing stocks, because it is so large, is the major player in the price formation process.
While I have argued above that the quantities traded do not drive the price, my point here is a different one. Consider a gold market without producers or (destructive) consumers. The market would clear at a price and quantity where the supply and demand schedules come into balance. Adding a relatively small quantity to either side of the market would not move the price much, even if the buyer (seller) were totally price-insensitive because of the large depth of the supply and demand schedule on either side of it.
Recent Activity Sets the Price
In the AGC Presentation (p. 100), Mr. Christian explains that he does not consider gold holdings to be part of the market if those holdings have not changed hands recently. This is consistent with the view discussed in the previous section, that the market price clears only the current year’s supply against the current year’s demand. If I understand the reasoning behind this, Mr. Christian believes that only transactions participate in price formation. After a particular gold ounce has not been traded for a long enough time, he no longer considers that ounce to be part of the market. At that point, in Mr. Christian’s view, that ounce has no impact on the market price. This view is consistent with the CPM’s mistaken focus on quantities traded as the keys to the gold market:
Gold also has a multiplier. For the Australian dollar, as with the U.S. dollar, I believe the multiplier is around 3 or 4. For gold, our estimate is that the multiplier is around 9 at present. That is, an ounce of gold entering the bullion market, from mine output, scrap recovery, central bank sales, or wherever, will be involved in 9 transactions before it exits the bullion market, either being used in a fabricated product or being dumped into an investor’s inventories somewhere.
Mr. Christian’s view is entirely mistaken. As I have explained, the supply side of the market is formed by all of the owners of existing gold, who offer it at a range of reservation prices. The price is an emergent property of the decisions of all of the owners of gold stockpiles to not to sell below their reservation prices, and the decisions of gold bidders not to offer above their reservation prices. The reservation demand of the sellers of existing gold, no matter how long ago it last traded, is the primary reason for the gold price being where it is.
Net Sector Flows
My second criticism of the CPM Group’s model is over-emphasis on net sector flows. In the preceding section, I discussed Mr. Christian’s incorrect view that gold holdings are not part of the market. Removing holdings (the primary locus of price formation) from consideration leaves only quantities recently traded as a possible object of investigation. The CPM Group goes to the far regions of the earth to measure quantities. The effort expended to get the numbers is impressive. While these numbers contribute to our understanding of what is going on in the market, they are largely irrelevant to an understanding of the price.
As explained above, net sector flows are caused by preference changes, and are not a cause of the gold price. Quantity balance rules require that “total supply” as the CPM Group defines it be equal to the sum of gross fabrication demand and net investment flows (private and official). This identity is logically valid and true at all times, but it contains no information about the cause of the price, for the reasons given above: a net flow into, or out of, one sector of buyers is not the cause of the price being higher, or lower. For every inflow into one market sector, there are equal and opposite outflows from other market sectors.
As discussed above, I believe that the CPM Group’s model attributes causality to net flows based on the mistaken belief that the market’s price equilibration process only balances net flows during a year. As explained, flows are not a cause or driver of the price; they are a reflection of changes in preferences that have occurred since the last trading activity. These changes in preferences determine where the gold flows.
CPM Group: Conclusion
The CPM Group’s approach to the gold market is consistent with common practices in the industry. Their approach is not any more mistaken than that of the other analysts who do things the same way. I have chosen CPM as the subject of this appendix because the clarity and detail of their reports has made it easier for me to follow their thinking. I do not take issue with the entire contents of the yearbook, only with their misuse of quantities.
I cannot rule out the possibility that the CPM Group has found statistical correlations between net flows into or out of one of the sectors, and the gold price and that these correlations are employed in their proprietary model. This could occur if it were generally true over many years that actors in one sector (or country) tended to be more price sensitive, or those in other sectors less price sensitive. If that were the case, then buying by the traditionally price-insensitive buyers would be correlated with a higher gold price and selling by price-sensitive buyers would be coincident with a lower gold price.
I should also give CPM Group credit for their study of macro-economic factors that do impact the gold price. These factors influence the price through their effect on investor preferences. I agree with the 2012 Gold Yearbook Press Release that “a host of economic, financial and political problems” are driving investor interest in gold. The Yearbook covers in detail macro-economic factors such as monetary policy, the business cycle, currency exchange rates, debt, and political uncertainty.
[1] I had hoped to use only free sources but I found that the yearbook, which is priced at $150, contained invaluable information in my understanding of the CPM Group’s model. The CPM Group originally provided me with a free copy of the yearbook, but prior to writing this article, I purchased a copy at the normal retail price.
Robert wishes to thank Mr. James Hickling of GoldMoney.com for assistance in copy editing the final draft.
