Tuesday, August 23, 2016


Today is the eighth anniversary of this blog, but as you know, last year I moved most of my activity to a subscription blog called the Freegold Speakeasy. Anyone can join, and a subscription is only about 60 cents a day, but for those of you who don't, can't or just plain won't pay for this stuff, I'm going use this anniversary post to give you a little taste of what you've been missing. ;D

One of the many topics I've been covering this year is GLD and its incredibly increasing inventory. If you didn't know, GLD has added 316 tonnes of gold to its inventory since January 1st. That's a third of its total, which currently stands at 958 tonnes. And a third of that addition, or 108.04 tonnes to be precise, was added in February alone. And that's why, on February 25th, I published a post titled "Plausible Explanations for GLD’s Recent Behavior".

On that day, I wrote:
"I’m theorizing that APs have been creating shares for hedge funds to bet on “gold”, and HSBC has been leasing bar numbers from other entities with allocated gold stored in their vault. They could even be leasing some from the BOE vault."

That was just a theory when I wrote it, but it became much more than a theory when, two weeks later, BlackRock filed a Schedule 13G disclosing that it had acquired 13% of GLD (more on this in a moment), and two months later, when GLD disclosed in its 10-Q filing that at least some of the new gold came from the Bank of England vault. Again, this was just a theory I came up with, using my Freegold lens and coat-check-room view of GLD, which was later confirmed by SEC filings.

Now, a 13G (or 13D in some cases) must be filed by anyone who acquires more than 5% of a publicly traded company. It must be filed within 45 days of a 5% acquisition, or within 10 days if a 10%-or-more threshold is passed. According to BlackRock's filing, it passed that 10% threshold on February 29th, and therefore filed the 13G ten days later. On Feb. 29th, GLD had 777.27 tonnes in its inventory, and 13% of 777.27 tonnes is 101 tonnes. GLD had only added 135 tonnes so far this year at that time, and most of it was in February, so the majority of the additions at that point in time had apparently come from one company, BlackRock. This is important because it supports my theory that someone was soliciting, promoting, possibly even incentivizing, these new share basket creations, rather than their creation being any kind of organic, market-driven phenomenon.

So who might be promoting new share creation, over the purchase of existing shares or bullion bank unallocated, as a way for hedge funds to make big bets on gold, and why? Well, I think the World Gold Council (WGC) might be doing just that. The WGC is a trade association and market development organization for gold mining companies who paid membership dues which totaled $60.2 million in 2012, $28.2 million in 2013, dropped to $12.5 million in 2014, and were phased out entirely in 2015 after several miners resigned from the WGC to avoid paying dues. [1]

No longer collecting membership dues, the WGC now relies on Sponsor fees paid from GLD to its fully-owned subsidiary, the World Gold Trust Services LLC (WGTS), for nearly all (about 97%) of its revenue. But that income stream declined too, from $104 million in 2012 to $71.5 million in 2013, down to $47.8 million in 2014. This was due in part to the 47% decline in GLD inventory during those years. The fee was 0.15% per year of the NAV, or Net Asset Value of GLD, which is the price of gold times the number of ounces in GLD. So changes in the price of gold, and the overall size of GLD, both affect the WGC's fee, which is 97% of its income.

The WGC's combined revenue from membership dues and GLD fees plunged 63%, from $164.2 million in 2012 to $60.3 million in 2014, a decline which the WGC arrested through cost-cutting, restructuring and fee-raising in 2015. In 2015, the WGC cut its number of employees by 44%, and also managed to get its GLD Sponsor fee raised to 0.40% of the NAV. What it actually achieved was a net increase from 0.15% to about 0.24%, after paying out the Marketing Agent, Custodian and Trustee fees. In 2015, that net increase amounted to about $15 million in added revenue. [1]

A Puzzling Disclosure

The BOE disclosure, even though it more or less confirmed my hunch, is a bit of a puzzle itself. Firstly, the disclosure said that no gold was held by subcustodians as of the last day of the quarter, but that some had been held at times during the quarter, the greatest amount of which was 29 tonnes, which was held by the BOE. Here's the actual wording from GLD's 10-Q filing at the SEC:

Subcustodians held no gold on behalf of the Trust as of March 31, 2016. During the quarter ended March 31, 2016, the greatest amount of gold held by subcustodians was approximately 29 tonnes or approximately 3.8% of the Trust’s gold at such date. The Bank of England held that gold as subcustodian.

We can calculate the date on which the BOE held 29 tonnes for GLD using simple math and the daily spreadsheet (you can download the Excel spreadsheet by clicking here). 29 tonnes is 3.8% of 763 tonnes, and there was only one day during the quarter when GLD held close to that amount. That was Friday, Feb. 26th (coincidentally the day after my post), when GLD held 762.4 tonnes. The day prior, it held 760.32, and the day after, it held 777.27 tonnes. So it was on Feb. 26th that the greatest amount of gold was "held by subcustodians" (the BOE being the subcustodian).

If we subtract 29 from 762.4 tonnes, we get 733.4 tonnes, and once again, there was only one day during the quarter when GLD held approximately that much. On Feb. 19th, GLD held 732.96 tonnes. The day prior, it held 713.63 tonnes, and the day after, it held 752.29 tonnes. The point here is that we can pinpoint precisely the specific dates and additions that constituted the 29 tonnes disclosed in the filing.

From Feb. 19th to Feb. 26th, 29.44 tonnes were added to the GLD inventory from the Bank of England, most likely because of new share creations on behalf of BlackRock hedge funds. It gets even more interesting when we look at the specific chugs. They were:

That's right! You might remember those two days, Friday then Monday, with the same exact amount added: 19.33 tonnes added two days in a row! Now, perhaps, we finally have a plausible explanation emerging. Now, perhaps, we know why they split a mega-chug of almost 39 tonnes into two equal parts on subsequent days. Wherever the first half came from, we now know that the second half came from the BOE! And, to me, that implies the epic tightness and scarcity that fits my current view of the LBMA.

The second thing about this BOE/subcustodian disclosure which I found puzzling was the request from the SEC to the WGTS, which ostensibly motivated the disclosure. It was in a letter dated 3/29 and filed on 3/30. Here's the official pdf of the letter. Simply stated, it asked GLD to "please disclose the amount of the Trust’s assets that are held by subcustodians." One day later (even though they were given ten business days to respond), dated 3/30 and filed on 3/31, the final day of the quarter in which the disclosure appeared, GLD's response letter replied, "We will, to the extent material, disclose in future periodic reports the amount of the Trust’s assets that are held by subcustodians. Please be advised that during fiscal 2015, no gold was held by subcustodians on behalf of Trust."

No big deal, right? Except that GLD has been disclosing subcustodian holdings in every 10-Q and 10-K filing since 2009! Here, here, here and here are just a few random examples which show that the SEC's letter was as unnecessary as GLD's response (really the WGC's response if you look at the letterhead) was strange. Just do a CTRL-F search for the word subcustodian in those filings, and you'll see that each one says, "Subcustodians held nil ounces of gold in their vaults on behalf of the Trust."

Of course now, in hindsight, it appears that those statements referred only to the last day of the quarter, that specific day of reporting, which wasn't obvious or even implied prior to the 4/29/16 10-Q filing, which for the first time differentiated "during the quarter" from the end date with regard to subcustodian holdings. Think about this, because I think there is a simple and plausible explanation.

The SEC's letter doesn't make sense the way it is written, at least not to the casual reader, because it asks for something that GLD had already been doing for seven years. Neither does GLD's response make sense, because it should have brought attention to the fact that GLD had been reporting nil ounces held by subcustodians since 2009. Both letters do make sense, though, if both letter-writers knew they were referring to "during the quarter" as opposed to on the end date of the quarter. But because this difference wasn't explicit in either letter, they only make sense if they were essentially staged, filed in order to explain and justify a change in reporting whose real motivation lay elsewhere. Indeed, both letters barely squeaked in before the end of the quarter.

If this explanation is correct, then it raises the question of why the change in reporting needed to happen for that specific quarter. There are basically three players involved—the SEC, GLD and the BOE. Of the three, the BOE is the new one, so perhaps it was the BOE that, for whatever reason, wanted to get this subcustodial disclosure on the record. Perhaps a request by the BOE was the real motivation. Think about the events we know about in terms of a timeline.

First, we have BlackRock creating a massive amount of new GLD shares in February, which requires HSBC (the GLD Custodian) to allocate actual bar numbers to the Trust. HSBC adds 90 tonnes-worth of bar numbers from Jan. 1 to Feb. 19, and then something changes. Wherever those new bar numbers were coming from suddenly runs out, so HSBC turns to the BOE.

The straw that apparently broke the camel's back was a massive 38.66 tonne share creation request on Friday, Feb. 19, which had to be split in two over a weekend. 19.33 tonnes-worth of bar numbers were allocated to GLD on Friday, probably from somewhere within HSBC's own vault, and then another 19.33 tonnes-worth of bar numbers were provided by the BOE on Monday… and another 8.03 tonnes-worth on Wednesday… and 2.08 tonnes on Friday. The total for that week from the BOE… 29.44 tonnes.

Then, four and a half weeks later, the SEC files a puzzling letter asking GLD to disclose subcustodian holdings, using a questionable pretense, and GLD responds immediately, on the last day of the quarter no less. I should also note that, according to the 10-Q in question, subcustodians held no gold on behalf of GLD at the end of the quarter. So either the 29.44 tonnes were physically moved from the BOE to HSBC's vault during the weekend of Feb. 27/28 (because another 15 tonnes were added on Monday), or else some other arrangement was made between HSBC and the BOE. It may even be as simple as HSBC leasing some space inside the BOE's vault. Who knows?

The point is that this is the simplest, most plausible explanation I could come up with for the puzzling BOE disclosure and SEC/GLD letters. For whatever reason, it seems to me that the BOE must have wanted it on the record that it was now providing GLD with bar numbers. Incidentally, I emailed Warren James back in June to see if he was still collecting bar numbers for his database. I asked him if he noticed any difference in the bar numbers coming in this year versus past years, and here's what he told me:

"regarding the GLD additions, it was just a once-off analysis and I didn't publish anything because it wasn't a comprehensive survey. i.e. it only looked at a period of about 6 weeks. But yes, the normal flow of 'dark bullion returning' markedly stopped at some point and was replaced with a flow of 'random' bars as the 'new' source."

I then asked him to confirm that, by "random bars," he meant old bars coming from existing stock, as opposed to new bars coming into the LBMA from the mints:

"re: Random bars. Yes that's right --> there are three main sources of stock for 'GLD Addition' ...
very broadly speaking, the following pattern applies:

approximately 30% is always 'previously-seen-stock-previously-absent' (dark bullion).
another 30% is consistently the Johnson Matthey newly minted stuff (now branded as 'Asahi', since they bought out JM)
the remaining 30 - 40% is always a mix of randoms, with proportions as expected from Refiner market share.

So yes, this event was notable that for a number of additions the proportions from the three sources changed to roughly 0:0:100, suggesting that temporarily source #1 and #2 were unavailable.
... so the question now is whether that trend continued or not, and whether it is still present.

Just for the record, the BOE vault stores approximately 5,134.37 tonnes of gold. [2]

As you can see from the chart, the BOE only owns 310.3 tonnes of that gold. Most of the rest is held on behalf of other central banks, and 1,355 tonnes is held on behalf of unknown entities, which may or may not be central banks. From BOE day, Feb. 19, to GLD's peak for this year of 982.72 tonnes on July 5th, 250 tonnes were added. Since then, the inventory has receded by about 24 tonnes to 958, but, to put it into perspective, if all of that 250 tonnes had come from the BOE's own stockpile, it would have represented 80% of the UK's gold.

This raises the question of whether the BOE would actually be lending its own gold to HSBC, or someone else's. For this, I turn to ANOTHER:

ANOTHER (Sun Oct 05 1997 21:29 - ID#60253)

People wondered how the physical gold market could be "cornered" when its currency price wasn't rising and no shortages were showing up? The CBs were becoming the primary suppliers by replacing openly held gold with CB certificates.

ANOTHER (Sun Nov 16 1997 10:20 - ID#60253)

Does a CB have collateral to lend it's gold? Understand, they only lend their good name on paper, not the gold itself. The gold that is put on the market in these deals belongs to someone else! The question is not "Are the CBs worried for the return of gold?" but, "Has our paper been lent to the wrong people?".

And FOA:

Friend of ANOTHER (Mon Aug 10 1998)

The process: An oil country (or others
[say, BlackRock]) goes to London and purchases one tonn of gold [or 100 tonnes-worth of GLD shares] from a Bullion Bank. The BB [HSBC] borrowed this gold [the physical representing the bar numbers allocated to GLD] from the CB (leased). The one tonn gold certificate [new GLD shares] is transferred to the new owner. The gold stays in the CB vault and the owner goes home.

You can draw your own conclusions. Obviously ANOTHER and FOA weren't talking about GLD in those quotes, because it didn't even exist yet. But they were talking about one way that central banks supply gold when needed.

Curiously, in its latest 10-Q, filed just three weeks ago today, GLD neither disclosed nor denied that it's still getting bar numbers from the BOE. Instead, they changed the wording to reflect the greatest amount held during the last nine months, rather than the last quarter as they did in April. To me, this obvious ploy implies the obvious, and then some.

I was emailing about this just the other day with someone who asked me about my thinking in terms of why the BOE might want this on the record. He wanted to know if I thought the BOE might be covering its bases against potential legal problems with getting its gold back in the case of a market dislocation and GLD dissolution. I will share with you my reply:

Yes, that’s very close to my thinking. But it’s not so much that the BOE will run into legal problems (it won’t), but that it wants to avoid looking bad or getting blamed in such a situation. Imagine if it happened fairly soon, and in the aftermath it came out that the BOE had just started supplying GLD this year, and did so secretly, then it might look bad for the BOE.

So you have the WGC promoting this (temporary) increase in GLD shares and inventory because it makes them millions in Sponsor fees (316t added x $1,340/oz. x 0.24% = $32M), and the BOE is willing to go along, wittingly or unwittingly (doesn’t really matter which), but not secretly.

Look, GLD is going to be terminated and liquidated if/when the London gold market dislocates, period, full stop. It's right there in the prospectus under termination events. So all GLD shareholders are going to be cashed out at the frozen paper gold price, and the physical will be “sold” in order to “liquidate” the fund. In other words, in Freegold, the overnight gold revaluation windfall profit won't go to GLD shareholders, even though GLD is fully allocated with bar numbers of real physical gold. It will go to the BOE or whoever owned those bars in the BOE vault that were leased to HSBC and then allocated to GLD.

So, when someone like Warren in the future goes through the bar numbers and finds that a bunch of them came from the BOE in 2016, at least the BOE can point to the SEC filing and say that it wasn’t done in secret!

One question I see repeatedly is, Why do I even care about GLD? Why do I spend so much time discussing it? Well, there are at least a couple of reasons why. First, GLD is a significant component of the LBMA and the global gold market in general, but it wasn’t around when A/FOA were writing. They stopped writing in 2001, and GLD was created in 2004. So it has always been kind of a pet project of mine to integrate it into the big picture.

Second, the LBMA is the very center of the global (paper) gold market arena, not COMEX, not the SGE, only the LBMA, so, as goes the slack in the flow of physical through the LBMA, so goes paper gold and its dominance in pricing. However, the LBMA is extremely opaque, so we cannot see directly whether there is an abundance or scarcity of unallocated physical reserves standing behind the massive paper gold market, but sometimes we can see the shadows it casts. Shadows like, for example, HSBC having to go to the BOE for dusty old bar numbers for the first time ever.

You see, I view GLD as a unique window into the otherwise-opaque world of the LBMA. It's a window that didn't exist when A/FOA were writing, so in a way, they were the window in their time. For certain things, they told us then what GLD can tell us today, like whether or not the physical side of the most overwhelming segment of the global (paper) gold market is currently tight and under stress.

Looking for signs that it's tight is as good as it gets. We have solid theory and logic that says it should be, so we look for evidence that it is, and from that we deduce that time is short. The precipitous drain in 2013 was a good sign that it was tight then, and now this BOE disclosure is a new sign that it’s tight right now, even though the previous sign reversed.

In "Plausible Explanations for GLD’s Recent Behavior", I explained how I like to think of the life of GLD in terms of three phases—growth, plateau and drain—and in 2016 we have apparently entered a new phase, with a spurt of (inorganic) "growth" driven by something entirely different from the growth in phase 1.

Here is a list of my GLD-related posts this year, for anyone who's interested in exploring this topic further:

Plausible Explanations (for GLD’s Recent Behavior) February 25, 2016 – 178 comments
Coat Check 101 March 4, 2016 – 291 comments
Bullion Banking 201 March 19, 2016 – 366 comments
GLD Recap June 12, 2016 – 290 comments
GLD and the BOE June 30, 2016 – 366 comments
GLD and the WGC July 29, 2016 – 280 comments

In addition to those, and just as many non-GLD-related posts this year, the Speakeasy also has new video "Debriefs" of me interviewing, via Skype, Michael dV, tEON, Sam, Indenture, Delusional Investing, byiamBYoung, and an all-new Debrief with Aquilus. So if you would like to join the fun, or just lurk silently like a fly on the wall, you can click here and join for the low, low price of just $110 for six months. ;D

Thank you for stopping by, and if I don't see you at the Speakeasy, I'll probably see you back here again in a few months. :D


[1] Credit for the WGC statistics goes to Ronan Manly of BullionStar.com and his excellent sleuthing and reporting on this topic, especially in The funding model of the World Gold Council: GLD Fees and Gold Miner Dues and Execs flee GLD – The revolving door at the SPDR Gold Trust Sponsor.
[2] Once again, Ronan Manly has done some very impressive digging, calculations and reporting on the topic of gold at the BOE, especially in How many Good Delivery gold bars are in all the London Vaults?….including the Bank of England vaults and Central bank gold at the Bank of England.