Feb 012016

The release of Federal Reserve Bank of New York’s December gold stocks report provides and opportunity to analyse the progress of this current phase of withdrawals from its custodial stocks. I say “phase” because in recent times there have been periods of concentrated withdrawal activity in between periods of little or no activity, as the chart below from Nick Laird at Sharelynx shows.


It is interesting that these phase seem to correspond with economic turmoil – dot.com crash 2000/1, global financial crisis 2007/8, and today?

Note that during 2000 and 2001 the FRBNY was able to consistently ship out 40 tonnes a month. That works out at 2 tonnes a day over 20 business days a month. Commercial vaults designed for high throughput can do more than that but if you look at this National Geographic documentary on the Federal Reserve you can see it is not suited to high volumes. As I explained in this post, “those who think Germany could put 300 tonnes in a big plane or warship and move it in one or a few days have been watching too many Die Hard movies”. In any case, Germany’s 300 tonnes could therefore have been realistically  repatriated in one year.

During 2014 and 2015 we know that Germany repatriated just under 190 tonnes and the Netherlands around 123 tonnes. Given the reported net withdrawals from the FRBNY (back calculated as they only report balance in millions of dollars @ $42.22, I calculate the following delivery schedule.


All figures represent withdrawals, except the one highlighted in yellow, which is a deposit. Note that every figure in this table is a multiple of either a 4.420 tonne or 5.157 tonne “lot”, eg 41.991 = (4.420 x 6 + 5.157 x 3). I have tried a number of possibilities but the above is the only realistic one I can find that fits the reported facts in the lot multiples. Out of this comes two observations:

  1. Another central bank(s) have been withdrawing metal from the FRBNY but not disclosing it, close to 40 tonnes to-date.
  2. Someone deposited 41.991 tonnes just as the Netherlands was about to withdraw 123 tonnes.

As the FRBNY is reporting physical custodial stocks, the only explanation for the deposit is either another central bank deposited physical, or the FRBNY moved some of its (ie America’s) gold reserves into the account of another central bank, which could be the result of:

  1. A new FRBNY lease/swap TO a central bank
  2. FRBNY repaying gold leased/swapped FROM a central bank in the past

Given how tight-lipped central bankers generally are, we are unlikely to know who the mystery (and coincidental) gold depositor was.

Jan 062016

Focusing on registered stocks versus open interest is a favourite of many bloggers because it produces dramatic “Comex is about to fail” figures. I have written many times that one also needs to consider eligible stocks as eligible inventory can be converted to registered relatively quickly. Blogger Kid Dynamite noted in passing in an email that December was a textbook example of eligible being used by issuers to make deliveries to stoppers. Not one to take the words of a cartel apologist at face value, I contacted data wrangler Nick Laird for detailed Comex warehouse movements and issuer/stopper figures, to check the facts for myself (and you).

To set the scene, at the beginning of December total registered stocks in all Comex warehouses was 120,967.246 ounces of gold. In retrospect, we know that the total number of contracts that stood for delivery during December was 2033, or approximately 203,300 ounces. Since that number is larger than the total registered stocks, you may be surprised why you didn’t hear about the default of Comex. The reason is that gold was either deposited directly into registered stocks in a Comex warehouse or gold was transferred from eligible to registered.

The table below shows all the changes in registered and eligible stocks per day per warehouse. Note that every single registered change during December was positive. I have coloured them by whether they were directly deposited (yellow), a transfer from eligible to registered (gold), or unknown (blue). In the case of the unknown, these are most likely transfers in my opinion but it is hard to be clear about that as the eligible change is including other movements.


Note that the total increase in registered during December was 154,947.693 ounces – 76% of the contracts standing for delivery – and that 105,086.452 ounce were clearly eligible to registered transfers, given the exact ounces involved. However, what is interesting is how these deposits into registered stocks match up with contracts “issued”, which are listed in the table below.


In this table I have multiplied out the number of contracts by 100 to show them as ounces (and also combined customer and house so the table is easier to read). You will note that the numbers coloured match the amounts being delivered into registered on the same day. For example, on the 15th HSBC issued 223 contracts (22,300 ounces worth) and in the warehouse movements figures we can see a transfer from eligible to registered on the 15th of 22,301.706 ounces.

The warehouse movement ounces of course generally will not match exactly with number of contracts as 100oz bars are usually odd weight where the bar’s weight varies between each bar within a specified tolerance, or where three kilobars are being delivered (which while being exact weight, are 32.15oz each so will rare exactly equal multiples of 100oz, see here for more info). Nevertheless, the repeated movement from eligible to registered matching issued amounts on the same day is proof that eligible stocks can be drawn upon by shorts to meet their obligations. Accordingly, solely looking at registered stocks to open interest is not a reliable indicator of the ability of shorts to make delivery.

For Trainspotters Only

In the table below I’ve done two things:

  1. Split Scotia Mocatta’s eligible warehouse movement figure from the first table above into an assumed eligible to registered transfer and the balance into an eligible other.
  2. Divided all ounce warehouse figures by 32.15. If the resulting amount is an integer (highlighted in purple) then it indicates the movement of kilobars.

I think the fact that the “eligible other” figures are kilobars strongly indicates that my blue “unknowns” were all eligible to registered transfers. It is also interesting that Scotia seems to deal a lot in kilobars in smallish quantities, at least during this snapshot on December. JP Morgan dealing in tonnes of kilobars is nothing new for my long-term readers.


Dec 152015

Last week I wrote about the gold warehouses associated with the CME’s kilo futures contract. Today I’ll have look at the Comex New York warehouses but rather than focusing on the eligible/registered debate, which has been done to death, I want to look at the fight between the warehouses for storage customers and the entrance of JP Morgan into this $30 million per annum business in early 2011 (hence my tacky topical title).

The stacked area chart below shows Comex gold stocks by warehouse. It shows the build up from 3 million ounces to nearly 12 million ounces during gold bull market and then a fall, similar to what we have seen with ETF stocks. The point marked (1) is the entrance of JP Morgan into the vaulting business in March 2011 and their impact on the other two major warehousers, HSBC and Scotia.


A better sense of the impact of JP Morgan’s entrance into this business can be seen in a percentage stacked area chart, see below.


Before JP Morgan’s arrival, HSBC and Scotia had over 90% of the market but this chart shows clearly that JP Morgan quickly cut HSBC’s market share from 50% to 35% and also into Scotia’s as well. During 2013 HSBC fought back and regained market share, initially from JP Morgan but then in 2015 eating into Scotia’s business.

The storage market dynamics are a bit difference in silver, with total silver stocks being somewhat consistent around 100 to 120 million ounces during the bull market and in contrast to gold, showing a ramp up to 180 million ounces as silver has fallen.


The storage business is also more competitive, with Brink’s and Delaware having larger roles and market share being spread out. Again we see the impact of JP Morgan, although it took until mid 2012 before they gained business. CNT also entered the market in late 2012.


From a market share point of view, it would appear that JP Morgan and CNT together have mostly taken business away from HSBC and Scotia and been able to maintain it.

While both the gold and silver warehousing market seems quite dynamic and competitive, it is interesting that the players don’t compete on price, with all of them charging according to CME $15 per 100oz gold contract per month and $8.50 per 1000oz silver bar per month (excepting CNT, who charges $6.75). At current metal prices that $15 fee equates to 0.17% per annum and for silver the rates are around 0.75%.

In addition, those fees also have not changed at least since June 2014, which is as far back as the Wayback Machine recorded the fee file. No one charges a Delivery In fee (which makes sense, you don’t want to dissuade people from putting metal in), but Delivery Out fees do vary, indicating some competition for those clients who do use Comex to source physical.

The lack of price competition is unusual in what is not an insignificant market. The table below shows the total estimated storage fees earned from 2011 to today.


Given that vaulting has large fixed costs, every additional ounces stored generally represents clear profit – the marginal cost of additional ounces stored is close to zero. With that set up one would expect more jockeying on storage fees.

The big loser in the warehouse wars has been Scotia, who had 32% of the $30 million per year storage revenue on offer during 2011, falling to 14% in 2015 ($4 million worth). The winners were CNT which moved from zero to 8% and JP Morgan who currently sit at 24%, equivalent to $7 million.

Dec 112015

Back in July I pondered what happened to 110 tonnes of gold in one of the Hong Kong gold warehouses registered by CME for its kilo futures contract. The chart below updates the figures and shows that wherever the gold went, its is gone for good.


Note that the straight line marked (1) is just an extrapolation between the last figure reported in the CME’s submission and the first figure reported on CME’s warehouse stocks report (see previous post for an explanation) and is not reflective of the actual movements during this “blackout” period.

The Malca-Amit warehouse dropped down to one tonne when the contract started trading and has only increased to 1.148 tonnes since. The Loomis warehouse I haven’t shown as there was no history reported by CME and it is only held around half a tonne since it came online mid this year.

All the action in Hong Kong is in the Brink’s warehouse, which appears to average 1 million ounces (or 31,103 kg). However, this has no relation to the volume that is being put through the CME kilobar contract.


As the chart above shows, the average daily volume is about 300 contracts and open interest at say 30 contracts (30 kg). In terms of an Owners per Ounce metric the contract is running at 0.001, or to put it another way, there is 1 tonne of gold “backing” each contract. Those that get worked up about Comex “leverage” ratios should be interested in the fact that the Hong Kong warehouse report only shows eligible stocks and has never shown any registered, which probably has something to do with the fact that the delivery notice report shows no issues/stops for the kilo contract so far this year. Forget your 300:1 Comex “leverage” – that would put the OI/Registered Stocks ratio as divide by zero error, or in other words, the CME’s Hong Kong contract has infinite leverage!

While the CME Hong Kong kilo contract is basically dead (even though they have 12 firms on their Market Maker Program), the Brink’s warehouse is far from morbid. This chart from Nick at Sharelynx shows that there has been over 840 tonnes of gold withdrawn (and pretty much that much received in) since the futures contract started trading.


Obviously this activity has nothing to do with the kilo contract and must be related to other over-the-counter (OTC) trading. For example, company A has gold with Brink’s and does a private sale to company B, the trade is settled between the two banks by their settlement departments, and then company B then instructs Brink’s to ship it out, at which point Brink’s reports that to CME as a withdrawal. These movements are only now visible to the market because the gold in the Brink’s warehouse is in the form of kilobars, which are eligible for the contract, and therefore have to be reported even if they have nothing to do with futures trading.

So even though the CME kilo contract doesn’t seem to be getting any traction, we can at least thank them for doing it because it now gives us visibility into the Hong Kong gold trade. One part of that trade that can be shown from the warehouse stocks data is the inventory build up prior to the Chinese New Year. The chart below shows the warehouse stocks in Brink’s leading up to Chinese New Year, indexed to 100 so they are easily comparable between each year.


The data is a bit chunky as CME only reported monthly average historical stocks in its submission, but it is clear that there is generally an inventory build two to three months before the new year. So far for this year, leading to the 2016 new year, we see the stock build up and now it is being worked down as metal gets delivered to jewellery firms for production into finished pieces.

Just one final observation. As per CME storage fees note, Brink’s charges $6.50 per contract per month, which works out at around 0.22% per annum storage rate. Malca-Amit only charges 46% less at $3.50 a month, but that hasn’t got them any business. By my estimation, Brink’s has earned over $13 million in storage fees since January 2011, and $2.3 million so far for 2015 – a lot of money Malca-Amit and Loomis are missing out on.

Dec 092015

In July I did a post on How much gold does China really have? which looked at the rate China was accumulating gold reserves based on their occasional announcements of how much gold they had. Extrapolating out their average monthly rate of 8 tonnes gave the chart below, which projected to 2,186 tonnes by the end of 2020.


Since then China has begun to report its gold reserves every month and has accumulated 86 tonnes in five months – an average of 17.2 tonnes. That is a lot more than their previous rate and you can get a better idea of the acceleration if I update the chart above with the new reserves figures and then project that rate into the future.


If China keeps at this rate then it will have 2,792 tonnes by the end of 2020, which is 606 more tonnes than I initially estimated. In my previous post I noted that official reserve additions plus commercial bank additions seemed to average 45% of monthly flow of gold into the Chinese market (that is, imports and newly mined domestic gold). The chart below shows these two estimates in green and blue.


What is interesting is that the average of the green and blue over the May 2009 to June 2015 (the two dates where reserves were announced) comes out at 17.1 tonnes a month (the red dotted line) which is very close to the average Chinese accumulation rate of 17.2 tonnes (the purple dotted line) since July 2015. The average of the June 2011 to June 2015, a period where inflows into China increased significantly, is 20.5 tonnes, which is close to the two largest monthly additions to China’s reserves.

These figures would seem to imply that now that China is reporting its reserves it can accumulate officially at the rate it was doing so unofficially in the past. Of course that theory implies that Chinese commercial banks would no longer be adding to their gold inventories, which seems unlikely as long as the gold market in China is developing and expanding. On the other hand, as financing trades unwind possibly the commercial banks are reducing inventories and this is being absorbed into official reserves.

I would also note that the above analysis assumes China adds to its reserves by sourcing from the domestic market only. Koos Jansen is of the view that China is acquiring 400oz bars from overseas markets. I do not discount that this may occur but my view is that even if done via the Bank Of China as Koos speculates (and I think that is the most likely candidate) such buying would be obvious to the Western bullion banks. If PBOC via BOC gold buying was consistent and ongoing I think it would make them prey to traders and hence my view would be that any such activity would be sporadic and tactical, taking advantage of times when western gold market demand was weak. As we receive more data over the next year on China’s reserves, commercial bank stock, imports and mining we should be able to confirm which theory is correct.

Jul 272015

On Friday I posted on the messaging China may have been sending with its central bank gold reserves announcement. Today I will update this analysis from 2012 to estimate how much gold the Chinese government unofficially holds and how much the population holds. I estimate that the total amount of gold in China is approximately 10,950 tonnes, with the population holding 6,490t, commercial banks holding 2,060t and the government, officially and unofficially, holding 2,400t.

How much gold is in China?

Koos Jansen estimates the total amount of gold within China at 13,781 tonnes. In large part the difference between Koos’ figure and mine is due to Koos assuming that the Chinese held about 2,500t of jewellery prior to 1994. In my 2012 post I quoted a source that notes that after the revolution all gold held by citizens, and gold mined, went to the government and was used to pay for imports. The analysis that follows does not rely on this total stock figure to work out official and other government gold holdings but it does affect the balance the population holds. If you agree with Koos then you can add the 2,500t to my 6,490t estimate of private stocks.

Where does China buy its gold?

It is my view that Chinese government acquires gold both domestically and from overseas, that all of it is held with China, and that any imports are reported in customs figures. Koos disagrees with this, arguing that as we see no figures in the customs category “monetary gold” from any country reporting gold exports to China, and since all SGE transactions are non-official, the government must be buying its reserves gold from overseas and importing it without having it declared.

I agree with Koos that “the PBOC buys gold in utmost secret or it would influence the market and geo-politics” and that they may make overseas purchases, but I find it hard to believe that China can dictate to the customs department of another country that their gold exports should not be reported at all (which would draw attention to the movement and negate secrecy). I also find it hard to believe that the PBOC would buy in its name from the overseas markets. It would be impossible to hide such activity from Western bullion banks and secure carriers and the information would leak out eventually, even if it could get the movements not reported in customs figures.

In my opinion, if the PBOC did not want to buy from domestic sources, it would request a Chinese commercial bank or a sovereign wealth fund to acquire gold in their name, import it as “for non-monetary purposes” and then get them to hold it until the PBOC wanted to officially acquire/report it. This method blends any PBOC purchases in with general gold importation flow, providing the secrecy it requires. As Huang Guobo, Chief Economist at SAFE, noted: “private demand for gold purchases is actually large but it is fragmented and intangible, and it is conducted through multiple channels and by multiple subjects that have less influence on the market, so this is more efficient in terms of the gold trade”, by which I take “more efficient” to mean “minimise price impact and maximise secrecy”.

Secondly, China does want its gold reserves to have some believability. If it was importing gold without any customs reporting, then Western analysts would have no basis on which to determine whether the reported reserves existed. By acquiring gold via reported imports (albeit in non-monetary form) and domestically, it allows the sort of analysis I will do below, which means that reported Chinese gold reserves can be assessed as to their reasonableness.

Koos also argues that the PBOC mostly buys gold from overseas. In Koos’ quote of the announcement, it says the purchases occurred “through various domestic and international channels … major channels of accumulation include: purifying domestic gold scraps and gold of various grades, direct purchase of production, transaction in domestic and foreign markets”.

I think the mention of “domestic” twice and the specific forms indicates that the purchasing was not solely from overseas sources. If it was mostly from overseas why mention all that detail? I note this Telegraph article where it says that a “division of the People’s Liberation Army mines gold and transfers the metal to the Chinese finance ministry, acting outside normal commercial channels. The government also buys gold directly from Chinese producers.”

Accordingly, I believe that the PBOC or its non-official arms could easily purchase from the domestic market without negatively affecting their policy of private gold accumulation. Indeed, during temporary slumps in the domestic market, non-official purchases could be a way of supporting the local mining industry and avoiding deep discounts within China to the London price.

How much is China buying?

In yesterday’s post I presented the chart below.


It makes a simple extrapolation between the dates of the official reported gold reserves figures (in red). The green area therefore represents an estimate of accumulation of gold by non-official Chinese organisations, which was then moved into official stocks at the date of reporting.

The chart below graphs this red/green accumulation from the chart above as a percentage of the monthly flow of gold into the Chinese market (that is, imports and newly mined domestic gold).


Notice that in the early years China was officially acquiring around 45% of all the flow of gold into the Chinese market. The percentage then declines from Jan 2003 (when China reported 600t of gold reserves) because the level of imports and mining increased greatly while we are assuming that the PBOC “gradually accumulated” during that period. While the accumulation was probably not in a straight consistent line, it nevertheless shows in general how little of China’s gold flow was being accumulated by the PBOC.

However, my green line is only an assumed official accumulation based on reported reserves. No doubt there has been other non-official accumulation but the question is how much of China’s gold flow is it reasonable of the Chinese government to have acquired without restricting its stated policy of encouraging private accumulation?

To answer that I turn to some estimates made by Matthew Turner, Senior Analyst at Macquarie Group (see this tweet) on how much gold Chinese commercial banks hold, based on their annual reports (see chart below).

If I add these figures (with some extrapolation back from 2007 to 2003) to our percentage flow chart, we get the following.


This I think is interesting, as it implies that as the Chinese commercial banks expanded their gold lending (mostly to domestic gold manufacturers to support the increasing jewellery and investor demand), naturally the government had to restrict its own acquisitions (in percentage terms, not in ounces) to ensure that a reasonable amount of gold was reaching the domestic market.

You will note that the purple line seems to run very much at the 45% rate that applies prior to 2003 (the red line). Does this constitute an official policy, namely that the Chinese government aims for the private citizens to accumulate 55% of gold flows? Pure speculation on my part and possibly reading too much into a pattern on a chart, but the best basis I think on which to make an estimate of unreported non-official gold accumulation.

If I then assume a policy of no more than 45% of gold flows into government (in the broadest sense) stocks, any time the purple line falls below 45% could be points at which other Chinese government organisations are accumulating gold. On this basis I produce the chart below.


The dark blue “Unreported Non-Official Accumulation”, which totals 745 tonnes by June 2015, is the calculated plug figure, if you will, to make government accumulation equal 45% of Chinese gold flow. Adding this to the official reserves gives a figure of 2,403t. If one wants to consider government controlled gold stocks in its broadest sense, then adding the commercial banks stocks gives you a figure of 4,460t.

It is possible that the PBOC leases some of its gold reserves to its commercial banks, in which case there is some double counting. However, if the 45% thesis is correct, then all that would do is increase the “Unreported Non-Official Accumulation” figure.

I would also note that this analysis does not include any gold purchased overseas and held overseas. This is a possibility but one that I give a low probability to because it means that China is taking on some exposure to Western banks to hold their gold for them and also because, again, it would be unlikely that such information would not leak out to Western governments, negating any strategic privacy that China may wish to have with respect to its gold activities.

One final chart, which takes the data in the chart above and expresses it as a percentage share of all gold within China.


This demonstrates the “sharing” of China’s gold stocks between the government and its people and that “the policy of ‘gold held by the people’ has been well achieved”.

Jul 062015

A reader asked for some information on Comex warehouse stocks, open interest and deliveries, which is timely as for gold we are currently in a very unusual phase historically. First, warehouse stocks (all charts from Nick at Sharelynx.com).

comex stock

Over the past two years we have seen inventories building but the standout feature is the reduction in registered stocks, both in ounces and as a percentage of total stocks for gold. When we divide open interest by registered stocks the unusual nature becomes more clear.

comex opo

For gold, ignoring the temporary spikes, we are at historically high owners per ounce, or if you want to convert the 85.455 into a percentage, only 1.2% of open interest is covered by gold in the warehouses. For silver we are at elevated levels, but nothing as dramatic.

As gold’s open interest has been relatively stable over the past few years at around 40 million ounces the decline in “coverage” is entirely due to the drop in registered stocks. For silver, open interest has increased from 500 million ounces at the beginning of 2012 to close to 1000 million today but registered stocks have also almost doubled over that time to 60 million ounces, hence we don’t see much change in silver’s “coverage” ratio.

When gold first moved into these very high owners per ounce figures there was a lot of chatter about the potential or certainty of failed settlement and Comex default. At the time, I noted that such commentary was ignoring eligible inventory and the fact that such inventory can be converted to registered relatively quickly. After two years of low coverage rates and no Comex default, one would have to consider eligible to be relevant and when you look at owners per ounce for total warehouse stocks, you get a very different picture.

comex opo.eligpng

The chart shows that owners per ounce for gold has been stable between 4-6 and is currently 4.9, or a coverage ratio of 20.5%.

Another thing that has to be considered when looking at Comex warehouse stocks is the actual percentage that stand for delivery. The charts below show cumulative deliveries made each month versus open interest.

comex deliveries

For gold we can see that it used to range between 2-4% taking delivery but in mid 2013 around when registered inventory declined, we can see that the delivery rate does not exceed 2% and is historically quite low.

So even if you want to ignore total stocks of 9.1 million ounces versus 44.6 million ounces of open interest and just focus on 0.522 million ounces of registered stock versus the 44.6, that coverage ratio of 1.2% is well within current delivery rates, with plenty of metal in eligible stocks potentially available.

I would note that the decline in registered gold stocks and delivery rates occurred soon after gold’s dramatic crash through $1550 and into the 1300s, and the stocks and rates have stayed low since then during the subsequent weak/sideways gold price phase we are currently in. A reflection of lacklustre western investor interest in gold? However, I note that eligible inventories have increased from 6 million ounces at the time of that price drop to 8.5 million ounces today, and such accumulation has usually been considered an indicator of positive western investor sentiment to gold.

I don’t have an answer at this stage as I hadn’t noticed this discrepancy until delving into the figures for this post but it is another sign of the exceptional state of the current gold and silver markets, which in my opinion hasn’t been this pessimistic since 1999 when gold was pushing $250. If ever there was a contrarian trade …

Dark Gold

 Posted by at 9:09 am  Stocks
May 122015

The gold market is often referred to as being “opaque”, lacking transparency. This is something of an understatement. The chart below is something I worked with Nick to put together. It shows all of the publically reported holdings/stocks of gold.

Transparent Gold

It is an extensive list and much wider in scope than many others, which usually just add up the major ETFs. The first thing that strikes you is the dominance of the ETFs in terms of traditional stores of gold like COMEX or TOCOM. I have no doubt that if the ETFs weren’t created the blue COMEX stock in that graph would have been a lot bigger.

Also striking is the huge drop off in 2013 from over 100 million ounces down to today’s 67.2 million ounces. At first glance this looks bad, like people don’t want gold, but the chart is only showing the preferences of the specific type of people who invest in these products (mostly Westerners). It is not like the gold was just thrown away, and import/export analysis does show that gold ex-ETFs has been imported into Switzerland from the UK, made into Kilo bars and then sold into China and India.

In the end, the amount of gold stays the same, it is just the ownership of the gold that changes. This ownership change does have some information value, as I discussed here, and while I will comment on what is going on with these visible stocks from time to time, it is worth putting the chart above into context.

The latest GFMS Gold Survey estimates the total amount of gold mined since ancient humans were first attracted by the “tears of the sun” at 5,902,877,700 ounces or 183,600 tonnes, which they say is currently held by the following groups:

Jewellery – 2,797,115,250 oz
Investors – 1,183,147,600 oz
Central Banks – 993,458,175 oz
Industry – 803,768,750 oz
Unaccounted For – 125,387,925 oz

So the 67,213,000 ounces of “transparent” gold represents only 1.1% of all gold or 5.7% of all the gold held by investors. The chart below demonstrates the relative insignificance of the transparent gold – the little blue bump at the bottom is Nick’s chart above.

Gold Stocks

When it comes to gold held by investors, Dark Gold dominates. While a lot of this dark gold is held privately, much of it would also be held in vaults operated by the major security firms like Brinks or G4S or run by bullion banks. Knowing what is going on with the gold in these vaults would give the operators some information advantage in the market. We, however, can only guess at what total stocks sit in these vaults (although Warren James is giving it a go with his analysis of ETF bar lists, see here for how he does it).