The Annual Gold Fields Silver Surveys : Bullish ? Bearish? or Puzzling?

EXECUTIVE SUMMARY

Gold Fields Mineral Services ("GFMS") publishes widely recognized annual "Silver Surveys". These reports contain a wealth of supply/demand information, and GFMS estimated silver's annual "deficit" at 155.7mm oz. for 1999 – a number that seemingly is quite bullish. However, this is one silver investor who found their recent data about "above ground bullion inventories" both quite troubling and perplexing. Not only because their inventory data undermines the longer term bullish case for silver; but also because GFMS' treatment of this crucial topic was quite vague and opaque.

GFMS divides silver inventories into two categories: "identifiable", and "unidentifiable" (or sometimes "non-identifiable"). "Identifiable" includes exchange warehouse stocks (mainly Comex), Government holdings, silver held by European bullion dealers, and the officially compiled Japanese trade stocks. "Unidentifiable" includes never reported privately held bullion, coins and small bars.

GFMS is realistic in using an analytical framework for "above ground stocks" that allows for unidentifiable stocks. GFMS though, dramatically revised upward,without explanation, estimates given for "identifiable" stocks in prior reports, and that is quite disturbing. For example, in the last two years, 1997 stocks have been revised upwards by about 193,000,000 ounces – from 537mm oz. to 730mm oz., a 35% increase. Yet we could not find any specific mention of this in their reports.

The logical implication of the GFMS data and exhibits, coupled with these revisions is that arguably about 418,000,000 ounces of previously "unidentified" silver inventories apparently came onto the market during 1998 – 1999 (and possibly 1997). GFMS though, only explicitly mentions 128.5mm oz. of this sum. We would like to see GFMS provide some more clarity and detail regarding the rest of that 418mm oz – which at about 290,000,000 ounces is clearly a significant amount of silver.

If our analysis is correct, the last two GFMS surveys have had much more of abearish tone than is commonly recognized. GFMS is basically saying that "unidentifiable" silver bullion inventories are not only filling silver's supply/demand deficit, but they are also adding substantially to the existing level of identifiablestocks at the same time. That is net bearish – even if the 1999 deficit was 155mm oz.

Another key issue is WHY GFMS presents crucial inventory data so vaguely, while presenting less important data so explicitly? IF the GFMS data is accurate they are helping producers avoid lower prices by downplaying the bearish aspects of their reports to potential investors. If though, their data is not accurate, GFMS may be aiding those who want low silver prices at the expense of both silver producers and investors? Or perhaps, GMFS may have simply done a mediocre job of reporting here.

This report is divided into three sections:

1. Identifying where the GFMS reports indicate revised inventory numbers, and underestimates of the amount of silver that came onto the market from "unidentifiable" stocks.
    
2. Addressing why GFMS' handling of the crucial inventory topic was vague and opaque.
    
3. Discussing why this is important for silver producers and investors – and possibly bullish.
    

(Note: GFMS, and the Silver Institute were asked for clarification of their reports. The "SI" did respond professionally, but didn't alter our view. GFMS has not responded. All critical comments are welcome.)

Section I - Evidence of Revisions of GFMS Inventory Data

This section simply presents the pertinent GFMS inventory data.

First: On page 28 of their 1998 Survey, GFMS states that identifiable silver bullion stocks totaled 537mm oz. as of year-end 1997. On page 29 of the year 2000 survey is a chart which indicates that 1997 year-end identifiable inventories were about 730mm oz. – or about 193,000,000 ounces more than the estimate given in the earlier 1998 Silver Survey

Second: On page 5 of the year 2000 report, GFMS states the 1999 silver deficit was 155.7mm oz., and on page 6 of their 1999 survey, GFMS states the supply/demand "deficit" for 1998 amounted to 104.7mm oz. – though this was then revised down to 94.5mm oz in the year 2000 report.

Thus, GFMS estimated the total "deficit" for 1998 and 1999 to be 250,200,000 oz. This equals the amount of industrial demand that had to be met from above ground stocks – be they "identifiable" or "unidentifiable" (with an adjustment for net changes in producer hedging, but this is a minor factor).

Third: On page 29 of the year 2000 report is a chart (figure #21), showing the GFMS estimate for identifiable stocks at year-end 1999 to be approximately 705mm ounces.

(Note: Regarding this 705mm oz. estimate. We couldn't find a number in the text for 1999 identifiable stocks. So we used the chart to estimate the number {by first calculating that each millimeter on the "y" axis equaled roughly 23.5mm oz, and then multiplying by "30" which was the height of the "1999 bar" in millimeters}. This is also how the above 730mm oz. estimate was derived.)

Fourth: Thus, according to GFMS, there was a 250mm oz. deficit for 1998 and 1999, but the level of identifiable bullion stocks rose from 537mm oz. to 705mm oz. – or 168,000,000 ounces during this two year period. How is this possible? The implied "answer" is that about 418,000,000 ounces of previously "unidentified" silver inventories apparently came onto the market during this time. GFMS doesn't actually say this, but it is the logical implication of their data and some exhibits contained in their last three silver reports. This 418mm oz. not only satisfied the cumulative 250mm oz. "deficit" of 1998-1999, but they also (per GFMS data) enabled identifiable inventories to rise 168mm oz.

Fifth: GFMS does acknowledge 128.5mm oz. of "implied disinvestment" for the 1998-99 period in their year 2000 report (p. 5). However, even if all 128.5mm oz. came from "unidentified" stocks, there is still that nagging little question: what about the other 289,700,000 ounces?

Summary: Thus the text and exhibits of the 1998-2000 GFMS reports arguably show that a) 1997 identifiable stocks were revised upwards by 193mm oz. in the next two annual reports, b) about 418mm oz. of "unidentifiable" silver stocks came onto the market during 1997-99, c) 290mm oz. of those 418mm oz. were not explicitly recognized by GFMS, and d) identifiable inventories rose 168,000,000 ounces during 1998-1999 despite a 250,200,000 deficit during that time period.

How might all this be bullish for silver? Well, unfortunately it may not be. However, for two reasons a more cheerful outlook for silver bulls may be warranted. First as detailed in Section II, GFMS has arguably presented their inventory data in a vague and opaque fashion. This increases the odds that their data is wrong; that their year-end 1999 estimate for stocks is too high. Second, to the extent GFMS loyalties lie more with the bullion banks, than with silver investors and producers, their vague treatment of "inventories" could be a bullish indication – because then, if identifiable stocks were really rising, GFMS would probably provide much more precise and explicit detail about that reality.

Section II - Is GFMS being Vague and Opaque regarding Inventories?

There were a number of reasons the GFMS reports gave us pause in this regard:

1. On page 29 of the 1999 report, GFMS states:
   
   "Figure 22 shows the level of identifiable bullion stocks at year-end for the period 1990-98. At the end of 1998 these amounted to 619 Moz (19,250 t). The fall of only 16 Moz (500 t) in 1998 compares to a 156 Moz (4,850 t) decline in identifiable stocks in 1997." (our emphasis).
   
   However, in order "to show" that 1998 identifiable stocks "fell" by 16mm oz. to 619mm oz., GFMS had to FIRST revise upwards the 1997 inventory estimate by 98,000,000 ounces – from 537mm ounces to 635mm ounces. Only THEN, could GFMS say identifiable stocks "fell" 16mm ounces during 1998 to 619mm ounces. GFMS though, (as best we can tell) neglects to explicitly tell the reader about their 98mm oz. upward revision of 1997 identifiable stocks; and if not for that revision, GFMS would have needed to state that identifiable stocks increased during 1998.
    
2. GFMS further states on page 29 of the 1999 report:
   
   "The more modest drop in identifiable stocks in 1998 fits well with our lower overall disinvestments number of just over 47 Moz (1,470 t) last year. It also implies that the balance, i.e. around 30 Moz, of net disinvestments would have come from non-identifiable bullion stocks." (our emphasis)
   
   This "30mm" though, does not include the 98mm oz. upward revision in year-end 1997 stocks. Unless GFMS wants to acknowledge that they badly miscounted "IDENTIFIABLE" stocks at year-end 1997, then another 98mm ounces had to have flowed out of non-identifiable into identifiable during 1998.
    
3. GFMS states on page 29 of the year 2000 survey:
   
   "Net supply to the market from sales of above-ground bullion stocks came to 155.7 Moz (4,844 t) last year. Of this, 87.0 Moz (27,06 t) can be identified as government sales (…). This leaves the balance of 68.7 Moz (2,138 t) as having come from a net fall in private sector holdings of bullion (…). However, as Figure 21 implies, private disinvestment was not at the expense of identifiable bullion inventories: their two largest components, Comex and European Dealers' stocks were, in aggregate, up year-on-year. The mathematical conclusion has to be that all the private disinvestment last year came from stocks that were unidentified (…).
   
   That is pretty much their entire comment under the heading "Identifiable Bullion Stocks". They do not mention their estimate for year-end 1999 identifiable stocks in the text. The chart though, (p.29) yields an estimate of 705mm oz. which is 86mm oz. More than the 619mm oz. of the prior year. Furthermore, the same chart shows they revised upwards their yearend 1998 estimate for identifiable stocks by about 106mm oz. (again with no explicit comment). As a result the chart on p.29 of the year 2000 survey shows that identifiable inventories declined a little bit during 1999.
   
   Therefore this chart obscures the fact that GFMS "really said" identifiable stocks ROSE by about 86mm oz. during 1999. Since GFMS also said the 1999 deficit was 155.7mm oz., it would have been "nice" if the chart gave a hint of that 86mm oz. increase in identifiable stocks.
    
4. GFMS, however, in their Year 2000 silver survey, did not just upwardly revise their estimate for year-end 1998 inventories. They also, once again, revised upward their estimate for year-end 1997 inventories – from 635mm oz. to about 730mm. Thus the 1997 figure was revised up another 95mm oz. on top of the 98mm oz. revision from the prior year. Once again, we could not find a peep about this in the text, nor exact data for the chart (is it common knowledge that one should simply always read the GFMS reports with the help of a micrometer?).
   
   Please stop and think about this for a moment. First, GFMS is implicitly saying that at year-end 1997 there were actually 730mm oz. of identifiable silver bullion inventories in the world, BUT they missed accounting for 193mm ounces of them (over 25%) in their 1998 report. Over the next two years, GFMS then retroactively adds these 193mm ounces back into their yearend 1997 estimate, BUT try and find a single word about this "revision" in the text of their reports.
   
   Second, some components of "identifiable" stocks are well known e.g. Comex stocks, U.S. and Indian government stocks, and the Japanese trade inventories, probably totaled about 280mm ounces at year-end 1997. If so, that means the above 193mm oz. "upward revision" was made on a base of about 257mm oz. (537 – 280 = 257). In other words, GFMS upwardly revised the other components of year-end 1997 "identifiable inventories" by roughly 75% (193/257).
   
   Third, these "quiet revisions" then help obscure the implication in the GFMS data that they are alleging that an additional 289,700,000 of previously unidentified silver inventories came onto the market during 1997-1999 – resulting in above ground identifiable inventories being almost 290mm ounces higher at year end 1999 than they would otherwise be.
   
   But don't tell anybody. May we suggest that some silver producers and investors just might like to be informed about those 289,700,000 ounces?
    
5. What further highlights GFMS's "reticence" about these "upward revisions" and their implications is that there is such a wealth of other data in their reports. We counted 46 "tables" of assorted sizes, plus 68 charts. For example, if you need to know that last year Malaysia recycled about 65 thousand ounces of scrap, or that Romania used about 193thousand ounces in photography – then no problem. The actual numbers are easily found.
   
   However, if you are a little bit curious about the details regarding how governments apparently sold 127 MILLION ounces in 1998/1999, and yet their official holdingsincreased by about 95 MILLION ounces, the breakdown of GFMS estimates for individual government holdings is unavailable – all we could find was a small chart which required a good ruler to derive a crude estimate for all governments in the aggregate.
   
   Couldn't one more table (#47) be added showing "upward revisions" in above ground inventories, and greater clarity on the quantity of bullion flows from unidentified sources?
    

Section III – Why GFMS' Reporting on Inventories is Important

The quantity and availability of above ground bullion inventories is a crucially important factor for judging the longer term future direction of silver prices – especially to the extent the reports of annual supply/demand deficits by GFMS (and CPM Group) have been accurate. Because then the "stocks to deficit" ratio projects into the future a time when above ground stocks (at a given price) could plausibly reach such low levels that a substantial rally would be extremely likely. For example, the above discussed 290mm ounces is the difference between the GFMS estimate for 1999 year-end identifiable bullion stocks being 705mm oz. as opposed to just 415mm oz.

In addition, investor and industry perception of the level of silver's above ground inventories also affects future prices. If perceived as ample, investors are more likely to shun the metal and/or dispose of what they already have, and industry is more likely to see less of a need to carry much inventory. Thus, the perception that inventories are ample can help create that reality.

For example, if prospective investors believed identifiable stocks were only 415mm oz., they might be far more likely to acquire silver. Especially since there is a reasonable case to be made that about 250-300mm oz. of the identifiable stocks are probably not readily available to the market at near current prices; and this would mean just one more 155mm oz. deficit could bring available identifiable stocks to extremely low levels relative to the basic industrial supply/demand deficit.

Silver investors and producers should be aware that to the extent the inventory reporting in recent GFMS surveys become widely accepted as valid, then the case for higher silver prices will be weakened. The implication of recent GFMS reports is that "deficits" may not be that bullish because they don't necessarily cause identifiable stocks to decline toward a point that would exert compellingly upward pressure on prices. For if there exists such a huge quantity of "unidentifiable" silver stocks, that these stocks can continue to fill the deficits, andalso add to the existing level of identifiable bullion stocks at the same time, then THAT IS NOT BULLISH !

One intriguing question here is: WHY did GFMS discuss identifiable silver stocks in such a vague and opaque manner? That it happened two years in a row suggests it was not "accidental". "Incompetence" is also unlikely as there are some arguments supportive of their data (e.g. both the Buffett rally, and the Asian crisis resulted in much silver being melted down, and sources confirm heavy silver flows out of China). Indeed, GFMS could be correct in their (implicit) estimate of how much unidentified bullion came onto the market in 1997/99. However, 418,000,000ounces is a rather BIG number; so big that it is hard not to think about other possibilities.

Thus, we wonder: Could GFMS be trying to help producers avoid even lower silver prices? Because IF their data IS accurate, it is probably better for silver producers that GFMS highlight the deficits, and downplay the "fact" identifiable stocks arerising - as this is more alluring to investors. On the other hand if far less than 418mm oz. of silver came onto the market from "unidentifiable" sources in 1997-99, and the GFMS inventory estimate is NOT accurate, (i.e. too high) then GFMS would be doing a very big favor for all those who want low silver prices.

GFMS states that European bullion dealers had 330mm oz. at yearend 1999. That 330m oz. "estimate" is a key number. Up 130mm ounces from two years ago, it is a number open to dispute (would European bullion dealers tell GFMS, or anyone else, information not to their own advantage?). A much lower number would be quite bullish. Govt. stocks are also key. It is hard to accept on faith that their stocks really rose 95mm oz. during 1998-99 despite sales of 127mm oz. China, in particular, is not exactly renowned for transparency (and nor are the bullion banks).

Silver investors and producers should insist GFMS provide more detail about their inventory data (or admit a higher degree of ignorance which would be fine). It also seems to us that the manner in which GFMS treated the silver market should perhaps prompt a closer inspection of whether the GFMS GOLD Reports have also been "vague" on some crucial matters relating to that market.