Worthy to add to above figures is this: GFMS revises recent X years ALOT.
This means that drawing conclusions from those recent X years, can seriously deviate from reality. The more recent, the less the figures are reliable.
Especially 'investment' related data seems to be revised. Apparently, GFMS lacks alot data. Or, knows it, but for some reason decided to not take it into account.
A red line I noticed over the years is this one: they assume(d?) every inequality between supply and demand as investment (read: stockpiling/destockpiling), causing the 'implied investment' figures (based on the unknown/inequality) to be consistently higher than what they later get revised to.
This has had serious consequences, which can be identified by comparing the World Silver Survey in that year released figures with subsequent years released figures:
First column = year
Second column = original figure Moz
Third column = (so far) revised figure Moz
2000 87.1 > 94.7 >>> error -7.6
2001 -11.4 > 1.9 >>> error -13.3
2002 12.6 > 24.9 >>> error -12.3
2003 -6.0 > 7.8 >>> error -13.8
2004 -37.4 > 67.9 >>> error -105.3
2005 -67.6 > 0.7 >>> error -68.3
2006 -64.0 > 24.9 >>> error -88.9
2007 -22.0 > 53.5 >>> error -75.5
2008 -48.2 > 48.1 >>> error -96.3
2009 -136.9 > -61.9 >>> error -75.0
GFMS told us investors in those years, that there was X Moz investment (negative = stockpile adding/investing, positive = stockpile selling/disinvestment).
But apparently, as indicated by the later revisions, those figures had an error.
Now look at the evolution of that error.
I should collect all the individual SilverSurvey's for every year as to have a complete picture but this isn't evident.
So I'll use what I found.
During 2000-2003, when the price hung flat at $5, the average error was 11.75 Moz.
During 2004-2009, when the price tripled to $15, the average error rose to 84.88 Moz.
So the average error in the GFMS data over 7-folded since 2004.
Anyone want to make an attempt to explain this?

And this isn't peanuts, because 1) the yearly stockpiling / destockpiling is an accumulating element with equally great impact on the future and 2) an error of 85 Moz is 10% of the annual supply / demand.
And I noticed over the years quite some of those "glitches" in the released silver data. Ex a comparison between the 2012 and 2013 data leaves an unexplained gap of about 200 Moz. Then a funny coincidence of nearly exact same over the years totals of coins&medals and implied investment, as like for every ounce sold as a coin, there is another ounce sold as bar outside fabrication (so exchange/ETF depositories).
What explanations are out there?
- Somebody is messing with data deliberately?
- Some registered/official entitities are able to trade without reporting duties? Institutionals along black pools?
- Some claimed silver simply doesn't exist?
One may also wonder why GMFS, after all those years, finally revised its "Implied Net Investment", only this year (figures for 2013), while they apparently DID have the specialized older data (as illustrated by Coins & Medals having been replaced by Coins & Bars), and ETF / exchange inventory added.
Especially this last is remarkable: they DID have the figures, yet inserted them in what was basically nothing but an error category, as alike they didn't want "investors" to know.
Then, post $50 $40 $30 $20, post mortem, they finally publish the original data.
If GMFS / Silver Institute considers it their job to inform silver market people, well lol, apparently that goes only as far as it fits their vested interest.
