SLV Database 4

Thanks again for everyone’s positive comments on my research. This article is installment number 4 in a series and builds on earlier work.  I’ve also written another static snapshot to describe more about the data processing, for anyone interested (a separate write-up, here).

Let’s get straight to the point. The conclusion from my current round of database analysis is that some silver bars currently held in the GoldMoney and BullionVault funds, used to exist in SLV, the percentages are approximately 1 % and 7.8 % (of their LONDON holdings) respectively. As with SLV Database 3, we're calling these signature matches 'Vault Jumpers' because they appear to move between the funds. The hard data is in spread sheet format, with the detailed analysis described in a separate sub-article, here. For a discussion of how the conclusion was reached, and some of the implications, please read on ...

Let's talk definitions. When we say 'Vault Jumpers', it's two primary concepts:

1. A signature match (Refiner, Serial, Fineness, Weight) which appears identical, appearing in two different bar lists (i.e. for different vaults or funds).
2. Time correlations which show the above signature appearing in one and then disappearing, reappearing in another fund (time spent to get there is irrelevant, all we need is a consistent sighting at point A and point B.

There are in fact, different variants. Here's my attempt at a classification:

Vault Jumpers - bars showing up in a vault different to the original one, with the same signature and evidence of transit. Unlike the characters in the film 'Jumpers', bullion bars seem to spend a little bit of time wandering around before they show up again at their new location. This single broad definition can be used to explain the concept.
Vault Jumpers (Adventurous) - these bars not only move vaults but also travel inter-continent. The best example of this is the Perth Mint Vault Jumpers, which get shipped from London to Australia.


Vault Jumpers (Lazy) - bars which show up in a different vault in the same locality. Best example of this is the Goldmoney and Bullionvault bars (today's analysis). For example, a Lazy Vault Jumper might be hanging out in London and can't afford to (doesn't want to) travel abroad to exciting destinations like Zurich or Canada.


Vault Jumpers (Loyal) - bars which move between vaults but stay in the same fund. an example of this is SLV shifting silver around London between it's five different vaults, this is already evidenced over at http://about.ag/SLV and is not something that we've done a study on just yet (but hopefully we will).

Finally there are also 'Fund Switchers' who stay in the same vault but just switch ownership. We don't actually have a study on these just yet, they do exist but they are quite rare — only FIVE of the 1,970 bars in this study have this behavior (that's 0.25%). It is easier to encapsulate all of these in the 'Vault Jumper' concept which is a bit more memorable. Technically most Vault Jumpers are also classified Fund Jumpers with the exception of the loyal ones (note: if someone has a better naming or classification system we could perhaps discuss this in the comments).

There are also two primary types of Vault Jumpers, the ones where we can guarantee their absence, and the ones which have 'jumped' but are still unaccounted for. This is an important premise of Dark Bullion calculations (the topic of my next post) and will need to be captured in my later graphs. Enough theory, onto the practical ...

The existence of these vault jumpers was predicted by Bron Suchecki ahead of time. Based on the proximity of the vaults in London over a long time frame it would be fair to see some bars moving in and out. The only thing which surprised us is the low percentages (we figured the number would be much higher than the 24% Perth Mint Vault Jumpers), and the consensus is this demonstrates market efficiency; generally acting to reduce the cost and risk associated with moving bullion around.

Photo credit: Anglofareast captioned: "Grant & Thornton checking off LBMA Bar numbers, with ViaMat agents at ViaMat Security Vaults in Kloten, Zurich." Original page. Found this page interesting because the Zurich Vault is used by GoldMoney. No idea when the photo was taken, but the point is to show that moving bullion around incurs costs - these guys have to be paid, and the forklifts must be serviced, etc. etc.

So today's study requires some creative explanation from those who are adamant that SLV has no silver — namely it must be explained how the non-existent silver from SLV was taken delivery of, by funds which are generally considered legitimate. So previously the argument has moved already from ‘no silver’ to ‘exactly what percentage of real silver are we talking about?’, but sadly this study is a failure in terms of proving the legitimacy of the entire SLV silver hoard – for we’ve only managed to advance the argument from 0.5 % to 1.0 %. BUT, this still doubles the existing amount of hard evidence and my hope is that any newbie investors buying silver because they have been told that SLV doesn’t have any; will consider the story is not true in its entirety and absolutely requires more narrative.

I have added to the public historical archive of bar list documents, you can find all the updated links on this page; but let’s talk a bit about our source data:

Goldmoney 

I really LOVE GoldMoney’s record keeping – they release a new bar list at the end of every quarter like clockwork, and from their website it is possible to publicly download historical vault reports over the last 11 years! This is the only fund which does this (to my knowledge) and makes things incredibly transparent. *** For consistency I do hold my own copy of these (which are renamed using my own normalized file format).

The Goldmoney records starting in 2001 are the earliest bar lists we have for any fund. For 2001, 2002 and 2003, the records were all graphical and I had to convert these using an OCR service. For 2004, 2005 and 2006 Via Mat used a ‘secured’ version of the PDF which meant that automated extraction was limited, but happily the same OCR service was able to convert these, and finally in 2007 onwards the files are normal (unprotected) PDF’s. Fast forward to mid-2012 and there is already 22 different documents issued for different vaults (gold and silver). Here's a quick look at the different documents currently issued by Goldmoney. The ones for silver used in this study are highlighted in grey.

From this you can see that Goldmoney has multiple geographical vault locations, they also have a different document for each vault (unlike SLV which has one single document and each bar has a vault location specified across 5 vaults). In 2011 and 2012 they are also expanding operations to new vaults (note: the vault or document format makes no difference to us since all source data ends up amalgamated in the same table structure – ‘WAREHOUSE_SILVER’).

For most of its life, Goldmoney has been in ‘accumulation’ phase – according to our bar list observations it wasn’t until GGL_30Mar2007 when their bars started disappearing back into the system. 

BullionVault

The bar lists from BullionVault take considerably more effort to collate. In early 2012 they secured access to the documents so that only folk registered with BullionVault could download them (this led to my ‘gram of gold’ story). The BV system has some kind of software-based timestamp on the document which always reflects the current date, even though the base document itself may be older. You can see this in the example image below (original document here).

Reconciled '30-June-2011', but the bar list itself was issued 10 days prior.
Suggests that in that 10 day period, there wasn't any change in the BV bar inventory.

What this means is that at first glance their documents are always current but in practice, the date of issue for the bar list doesn't necessarily get issued every day. In 2012 the number of newly issued unique documents seems to slow down – to be honest I think this was a bug in my software where it was downloading a cached version of the document, and it took me about five months to realize !! Either that or BV didn't actually issue updated documents during that time. Either way I won’t be losing any sleep over the records gap as their records are still better than Sprott's bar lists !! These guys only have one silver repository, which is in London.

This 'bar list quality/frequency' question is best shown visually. I've updated the 'source documents chart' from my previous article to show the records we have for 2011 and 2012 combined (this makes the graph rather large, you'll need to click on the image to see all the detail). Each document is still tracked individually and is represented by a point on the graph. In this, you can clearly see gap in BullionVault's records, the consistency of GoldMoney, the continued rareness of a new bar list document from Sprott's PSLV, and the highly consistent updates from SLV and Perth Mint Allocated Pool Silver.

Updated 'source documents' chart for silver, showing all records from 2011-2012. NOTE: to improve the clarity of this graph, the following modifications have been made: (a) Gold records have been removed as this is a study on silver, (b) Goldmoney's 3 x silver sources have been rolled into one - primarily as they get issued at the same time. The graph is intended to show document issue frequencies only.

This series of articles is intended to discuss the mundane and basic details about the silver industry as well - today’s highlight is the nature of the vaulting operations from GoldMoney and BullionVault. Both of these funds choose to use the vaulting services of Via Mat, and operate it in the same way that a virtual phone network runs a brand off an existing phone provider. i.e. GM and BV  don’t have any of their own vaults, they figure that paying custodian storage fees is cheaper and more efficient than constructing their own vault (fair enough). So as far as we know, Goldmoney Silver London and BullionVault silver London are located in the same vault run by Via Mat – and this metal will be segregated in some fashion (we assume physically segregated but we have no idea how they run their vault – it’s all very secretive, BullionVault talk about it here). If you're wondering what the heck I'm talking about, this is best shown graphically. I took all the recent Via Mat documents (GM & BV) and made a composite of the first page of each - i.e. all 13 documents overlaid on top of each other with white made transparent in Photoshop. The goal of this image is to show they all have the same layout.

Included Source Document links used to create this composite image (you can check them to prove I'm not making this up): BSL, BGZ, BGN, BGL, GSZ, GSL, GSH, GPZ, GPH, GGZ, GGL, GGH, GDH.
Or Photoshop file here.

But I didn't spend 40 minutes stitching that together just to prove they are visually similar - my ultimate point is that these are all documents (different vaults, different metals, different geographies) produced as individual reports from a single database source (a big hello to Neil J. Geddes, whoever you are). I sure wish I had access to their database because it would sure as hell have a lot more interesting details than mine, but basically the same principle (of a central database) is at work. The same composite image could be shown of the JP Morgan output - they have lots of different funds with the same layout.

Anyway, another mildly interesting fact (to me anyway), that there were ZERO fund jumpers between GM and BV by themselves, but based on the comparatively low market share this makes sense. I expect one day to find a Triple Vault Jumper - a signature that appears in more than two different funds (maybe once I combine the ETF Securities data).

Finally, I’d like to talk a bit about the ramifications of this research. In basic form, it demonstrates those who would claim that SLV has no silver, simply hasn’t dug deep enough. That’s all there is to it, case closed.

Just like it’s easier to switch ownership of a bar than going to the expense and risk of shipping it, it’s easier to repeat an opinion rather than going to the effort of researching it. I once remarked to Bron that it would be interesting to construct a catalogue of ‘silver memes’. His feedback was that it would be almost impossible, since new ones pop up every day and new variants are created. In a good sense the stories are alive, with origins as far back as the 1979 silver spike and like good bacterium they have survived in little pockets before flourishing once again in the warm sludge of today’s financial environment using the easy-spread vectors of the internet and social media.

That’s all for today. My next analysis will turn to the less-volatile (and better quality) Gold data. We will also approach the topic of Dark Inventory, since we 'see' it in every analysis.  And of course it also goes without saying that we’ll be thinking of ways to interrogate the data to answer other questions about SLV’s inventory.

*** Update: 6th October 2012,  Goldmoney have 'bullionvaulted' their vault reports - now you can only download the bar list if you're a registered member - I'm in the process of signing up now so that I can access their September 30 documents. So much for transparency and my glowing review of their processes! While I fully doubt the change had anything to do with my work here, I do wonder what their rationale was. I guess it's none of anyone's business unless you have holdings, which to a degree makes sense, but ... whatever.