Behind the Opaque Silver Statistics

The annual Silver Surveys, sponsored by the Silver Institute, have shown a sizable supply-demand deficit for almost a decade. Yet silver has basically traded sideways for the last seven years, resulting in much confusion. The key to this riddle involves silver bullion inventories.

Unfortunately, the last two Silver Surveys, prepared by the consulting firm "Gold Fields Mineral Services," were quite vague and opaque on this crucial topic.

This undermined attempts to analyze the data's market implications. GFMS' actual inventory data, which is quite difficult to ferret out, clearly seems bearish. However, there is also a solid bullish case due to doubts about the thoroughness of the GFMS inventory discussion. Finally, various related issues are unclear—some bearish, some bullish, creating additional ambiguity.

Underlying the above is the question: Why was GFMS so vague about inventories? Because, depending on what inventories truly are, the GFMS reports could benefit one market group, such as bullion banks, at the expense of another—silver producers.

GFMS categorizes silver inventories as identifiable, or unidentifiable. Identifiable includes silver held by European bullion dealers and exchange warehouses, government holdings and Japanese trade stocks. Unidentifiable includes never reported privately held bullion, coins and small bars. This framework is passable but simplistic, and GFMS magnifies its shortcomings.

To start, GFMS dramatically revises upward, without explanation, data for identifiable stocks in prior reports. 1997 inventories were revised from 537mm ounces to 730mm the following two years.

Their surveys also imply that roughly 418mm ounces of unidentifiable inventories flowed onto the market during 1997-1999, but they only specifically account for 128mm.

Thus, despite chronicling a 250mm-ounce deficit during 1998-99, GFMS also reports identifiable inventories increasing 168mm ounces. Moreover, this counterintuitive result is omitted in the survey's text.

The surveys also indicate governments sold 127mm oz. in 1998-1999, but their inventories increased by about 95mm oz. Curious about that? Annoyingly, data for individual government stocks are unavailable—but one can approximate aggregated government holdings from a chart.

In the year 2000 survey, 68 charts and 46 tables provide a wealth of data (Romania used 193,000 oz. in photography). Why not a table for clarity on the many millions of ounces of upward revisions, and "unidentified" bullion flows? To uncover the above numbers, one has to link the GFMS reports for 1998, 1999 and 2000 (the year 2000 report alone won't suffice).

Additionally, some inventory estimates must be derived from small charts since specific numbers are omitted. For instance, no figure is given for 1999 identifiable stocks.

So we calculated from the bar chart on page 29 of the year 2000 survey that each millimeter on the y-axis equaled 23.5mm ounces. We then multiplied by 30, the bar's height for 1999. The result, 705mm ounces, is 168mm greater than the 1997 inventories shown in the 1998 report. The 168mm + 250mm equals the above-mentioned 418mm ounces of unidentifiable inventory flow.

Availability of bullion inventories is key for judging the long-term direction of silver prices. If reports of supply-demand deficits are accurate, then inventory attrition should eventually trigger a major rally.

GFMS though, basically says that "unidentifiable" inventories not only filled silver's 1998-99 deficit, they also concurrently increased identifiable stocks. That is bearish.

However, GMFS groups under identifiable very different types of inventories in terms of "transparency" and "closeness to market". This may grossly overstate the bearish tone of their inventory data. For example, on a 0-10 scale, Comex inventories warrant a transparency ranking of near 10 whereas a ranking for European dealers, and the Chinese would be about a 2. Thus, there is far less certainty that these latter silver stocks actually exist.

GFMS may profess to know, and may truly believe they know, but common sense, and history, argue otherwise. We estimate China and European dealers account for both 65 percent of the identifiable stocks GFMS contends existed at yearend 1999, plus most of the 1998-99 increase in identifiable stocks.

This casts some doubt on the core bearishness of the GFMS inventory data. GFMS' data may be right, but to implicitly equate somewhat hypothetical inventories with Comex stocks could be quite misleading.

Additionally, some identifiable stocks are probably not readily available to the market—most U.S. and Indian government holdings and Japanese industry stocks. China's stocks are probably undeliverable, which would lessen their significance in tight market conditions.

Then there are related issues. For the bulls, the question is, have some European dealer inventories been leased and-or include any of Buffett's silver? Considering leasing, Buffett, this year's deficit, the possibility of error in the China-European dealer components, the likelihood some Comex stocks are investor held and the arguably unavailable and undeliverable identifiable stocks, an investor request for delivery on 75-150mm ounces of Comex and-or LMBA contracts, could plausibly trigger an explosive rally.

On the bearish side, it is unclear if GFMS is counting inventories in some private depositories, like Brinks, or bullion held under certificate programs by banks such as Scotia of Canada.

Additionally, China's inventories may be underestimated. Whether bullish or bearish, the silver industry would likely benefit if GFMS chose to upgrade the inventory section of next year's survey.

Michael Dorfman
Managing Director of Kenrock Investments LLC