12 November 2013

More on bullion bank reserves, delta hedging and coat checking

Below is a response to FOFOA's seven part comment to me here on his Gold as a FOREX currency post.
FOFOA: “You can call the slack in the flow their "stock" if you want to, but if there's more flow coming in than going out, then the BBs will accumulate reserves. If there is more going out than coming in, then their reserves will decline.”

This statement makes it sound like BBs are benevolent entities that manage the difference in flow for the market. If investors do not want to buy (sell) unallocated, then the BB will not accumulate (divest) reserves. There is also a separate borrow/lend market, and again, the impact on a BB’s reserves is driven by client action, not the BB.

As you say, the unallocated account holders bear the burden of funding the BB’s reserves, from which it logically follows that BB accumulation/divestment of the “flow” is dependent on unallocated account holder willingness to hold unallocated (or futures etc).

FOFOA: “unallocated credits are not necessarily lent into existence such that the offsetting asset is automatically a promissory note from a borrow, especially ever since the price of gold started rising in 2001 and it no longer made much sense to borrow in an appreciating unit.”

Interesting comment. Certainly the miner hedging has declined over that period, hence the drop in lease rates. I would note that this comment contradicts the goldbug theory that CBs and BBs lend paper gold into existence to supress the price of gold.

FOFOA: “So around 32 LBMA bullion banks which maintain ounce-denominated books (the very definition of a bullion bank) have their reserves stored in the vaults of the 6 bullion banks which are both LBMA-recognized custodians and also the clearing members of the LBMA. Talk about pooling reserves!”

No, this does not follow at all. The non-clearing BBs only need to hold balances with clearing BBs sufficient to settle their net trading with other non-clearing BBs. No BB’s physical or paper (ie on-call liquid) reserves HAS to be held in London with a clearing BB.

The point of the non-clearing BBs is that they have competitive advantages and specialisations with certain client bases in different geographical areas and so deal worldwide and hold physical gold outside of London (as do the clearing BBs).

Loco swaps, for example, are a common transaction and can be settled without needing any clearing in London. For example, Soc Gen could have metal in a Swiss bonded warehouse but have a client in HK who wants to buy. They could call ANZ, whose strategic focus is the Asian region, and do a loco swap of COMEX gold for gold that ANZ holds in HK. EFPs perform a similar function on futures markets.

So non-clearing BBs could hold all their reserves in physical outside of London and only need run some unallocated balances with clearing BBs.

FOFOA: “Are Goldman Sachs' gold reserves, which are deposited at JP Morgan, physically allocated to GS or unallocated bookkeeping credits? The answer is that they are obviously unallocated credits. … Why would the reserves of 30+ non-custodial bullion banks be held in allocated storage in the 6 clearing members' vaults? They wouldn't, of course.”

I don’t see “obviously” and “wouldn’t, of course” as arguments. I don’t see you explaining why it is obvious. Non-clearing BBs could easily hold allocated with a BB if they have exceeded their credit exposure to that bank. While the limits these banks grant each other are large, and they have netting or set-off arrangements, there are still limits and those limits can be used up by other divisions of the bank. In such situations the bullion division of a bank may therefore have to resort to allocated, as I noted here.

FOFOA: “I'm sure you'll dismiss or at least dispute this notion as well, but just think about it. It makes perfect sense, even if it doesn't fit into your limited view of the gold market from your little corner that happens to be mostly physical.”

Very condescending: “limited view” and “little corner”. As they say, if you can dish it out: I’m sure you’ll dismiss the notion that it is not obvious that non-clearing BBs’ reserves are unallocated credits with clearing BBs, but if you think about it, it makes perfect sense, even if it doesn’t fit into your limited view of the gold market with no practical experience of working in the industry and sitting in your little blogging corner making theories up from behind a keyboard.

FOFOA: “They would be unallocated bank-to-bank credits much the same as domestic commercial bank reserves held at the Fed are unallocated dollar credits. … non-custodial bullion banks are essentially depositors or deposit holders at the 6 LBMA clearing banks just like commercial banks have a reserve account at the Fed.”

So I assume that your proof that non-clearing BBs’ reserves are unallocated credits with clearing BBs is that this is how fiat money works. The problem with the analogy is that Fed reserves system works because the banks know that there can never be a shortage of reserves – the Fed can just digitally print them. That does not hold in bullion banking, as you can’t print physical gold. I would suggest that this one significant difference means that bullion banking dynamics and institutional structures will be different to those of fiat banking.

My best guess is that bullion banking operates along free banking lines, which I have discussed with costata and Michael H on your blog here. I found it interesting that you never engaged in that discussion, or remember you ever considering the idea elsewhere. Maybe because free banking means the system self regulates the amount of paper gold and thus won’t blow up?
FOFOA: “the system as a whole could be much more thinly reserved since 84% of the banks' reserves are fractionally reserved on a different level, the LBMA interbank clearing tier, for which we have only very limited insight.”

So we have a limited insight into the LBMA interbank clearing tier, but “obvious” insight that the non-clearing BB’s must all hold their reserves as unallocated? To the extent that a BB holds unallocated with other BB’s instead of physical as reserves, then yes it is tiered and the overall fractionalisation is higher. However we have no idea of how much physical reserves clearing or non-clearing BBs hold.

FOFOA: “I don't think I've ever seen you give your explanation for the astonishing volume. I have, in the post above, but I've never heard what you think explains an average of 170 trades per bank, per day, each averaging almost a tonne per trade.”

I’m surprised you are surprised. Your post just said that paper gold traded in FOREX markets and had carry (and funny in a post about carry trades no mention of GOFO was made, which is the carry rate). What is the big deal, any online FX service lists XAU as a currency to trade, the fact that it has an ISO currency code, these two should tell you gold is part of the FX market. As to volume, I covered that in this 2009 post:

“the very fact that gold is no one’s liability and cannot be printed means it attracts a disproportionate amount of trading and speculation. Why is it assumed that 12.7% is excessive and unreasonable? Could not the 12.7% figure be proof of the special monetary nature of gold, proof that it is the King of Currencies?”

That post and the 12.7% turnover figure was pre-survey, which reported daily turnover of 5,400t a day. Latest GFMS figures have CB stocks at 29,597t and investors holding 34,820t. 5,400t by 64,417 stock gives 8.4%. If you said that half of that stock was long-term holders and thus not traded and should be excluded, it gives a turnover of say 17%. When compared to similar figures for fiat money it is not “astonishing” at all and is a great indicator of gold’s liquidity.

As to the number of trades per bank, well I do consider that astonishing - astonishingly low. Knowing how many trades we do with clearing BBs each day, there is no way that is actual number of trades given the number of counterparties the BB deal with at that top tier. There is some sort of aggregation going on in the LBMA survey to give such a low number. Just another reason I am no so quick to draw conclusions or extrapolations from that survey until I understand the methodology.

FOFOA: “such tremendous volume, especially if the BBs are, as Christian says, only leveraged 10:1?”

I am not sure I understand the link between turnover and fractional reserve ratio (which is what the 10:1 is referring to). You can have a 100% reserve system and still have many multiples of turnover. Leverage, yes, will increase turnover, but leverage and fractional reserve ratio are different things.

FOFOA: “what could possibly be offsetting their net exposure given that the volume is ten times higher than the gold futures markets (and 100 times higher than GLD's daily volume)? Physical?”

Again, I don’t get the link between turnover and net exposure. Anyone could buy and sell back the same ounce multiple times during a day and be left with no net exposure at the end of the day. The BBs could intermediate (make markets) between FX desk traders all day and have no exposure.

I would argue that the greater the leverage employed (and these FX platforms and contracts for difference market themselves on the leverage you can use) the less likely these speculators are to hold a position overnight. They add a lot of turnover and liquidity but at the end of the day do not add the sort of massive net positions you think they leave BBs with which require BBs to “delta hedge”.

FOFOA: “Do you even understand how the banks use these? Here's an email from FOREX Trader which I posted back in June of 2012 … The only answer I can think of is that they hedge it by going long correlated (but not identical) assets. What's correlated with paper gold? Silver, copper, euros, crude oil, interest rates, yield curve spreads, whatever.”

“I can think of” – sounds like he is guessing at what it is? If you read through FT Alphaville’s articles and associated links it is not clear that it is totally about using correlated assets. A lot of it is using derivatives of the underlying asset to synthesise the underlying asset’s performance.

It is easy for your FX trader to say “I made a quick delta-0.7 using some regression and eyeballing” and another thing to actual implement it for the sort of volumes you are talking about and sustain that without slippage/loss.

Indeed, as this Reuters article says: “traders say that much of what happens at "Delta One" desks -- the focus of the UBS loss -- and in other parts of the business is driven by banks betting their capital to make a profit in what is known as proprietary, or "prop," trading.” That is, Delta One desk are no more than speculative trading, gambling. No surprise that UBS’ and Soc Gen’s blew up.

FOFOA: “This is the age of complex derivatives, and to dismiss them being used in the "gold" market, especially given the incredible volume reported in the LBMA survey, simply because you don't see them being used in your little physical corner of the market, is simply silly.”

There you go again with the condescension. The Perth Mint may not deal in this sort of risky stuff, but we have deep contacts into the BB market. I’ve asked a BB dealing desk managers about this delta hedging in another context and it was laughed at. He said he had enough trouble getting his middle office (risk) to approve an offset of gold in different locos, let alone something as unstable as complex mix of unrelated but partly correlated assets. You believe your source (by the way, has he/she ever worked on a bullion desk) and I’ll believe mine.

FOFOA: “this coat-check room view since January of 2011, but I didn't come up with it. … But it is obviously the correct view if you give it proper consideration.”

Who cares the source, you’ve endorsed it and appropriated it as part of your view of the market. And again with the “obvious”, I mean I think what I’m saying is obvious but that line of argument doesn’t seem to persuade you for some reason.

As I said, I can see ETFs (any of them) being drawn upon via stock borrowing, but I dispute that a BB can park reserves in it.

FOFOA: “The coat-check room view only applies to GLD, and not to any other ETFs, gold or otherwise. It's not an ETF thing, it's only a GLD thing, because GLD is the LBMA's gold ETF. … because the LBMA is head and shoulders above all other gold selling organizations. And that's very simply why GLD is 10 times larger than other gold ETFs. Not because of marketing (PHYS arguably has better marketing), but because it's the LBMA's gold ETF, and the LBMA had that much gold to put into its ETF!”

What do you mean by “GLD is the LBMA’s gold ETF”. The WGC is the one who sponsored (funded) GLD and created it. The reason it is the biggest is because it was the first and the WGC had been marketing to institutions well in advance of its launch (indeed, some US and UK money was funnelled into the Australian gold ETF because GLD was delayed). It is crazy to say PHYS has better marketing, you have no idea of the machine that the WGC deployed at the institutional level to sell GLD.

And can we please not use “LBMA”. The LBMA does not have any gold, or run vaults or trade. Please be specific. Are you talking about all the BBs, or just the clearing BBs, or only a few of those?

FOFOA: “If someone wants to capture a GLD discount-to-NAV arbitrage opportunity, it suffices to buy GLD and sell spot unallocated (XAUUSD). Then you wait until the selling pressure in GLD has abated, and you square your position. No redemption necessary. GLD is so large that, even if institutional investors dump it, someone else, including the bullion banks, will take it instead of allocated if it is cheap enough. But why redeem?”

The BBs redeem because they have to pay 0.4% management fee on GLD when holding unallocated with a clearing BB cost them a lot less than that and if you have your own vault, the cost of storing the physical yourself is zero. In that case, why wouldn’t they redeem?

FOFOA: “It's one way a bullion bank could turn a dead asset (surplus reserves) into a live one. …”

You didn’t need to go to all that trouble with the T accounts, as steps 1-2 are what DP said and you have answered my question “so you're saying collectively the BBs went naked short 2000t of ETF gold” with a yes, but they delta hedge themselves, which is what I guessed in my response to DP. So the whole coatchecking IN thesis rests on delta hedging. Given I don’t think delta hedging is reasonable for that amount of gold and for the time period involved, we’ll just have to disagree.

Can you just clarify one thing – how much of GLD are you claiming came from excess reserves, that is, how much of it was delta hedged (at its peak)?

FOFOA: “Before GLD, there was a storage cost associated with sitting on excess reserves, but the banks had no choice but to eat that cost.”

As I said to DP, “the marginal cost to a bullion bank to hold physical reserves is zero (vaulting is primarily a fixed cost business), there was/is no pressing need to create ETF's to save costs by parking metal in an ETF structure.”

FOFOA: “now there was also an opportunity cost in addition to the storage cost of sitting on "excess" reserves.”

Not sure what you mean by opportunity cost. I’m guessing the fact that you can invest the excess cash from selling GLD shares less the lesser amount needed for that perfect delta hedge. Have you considered as well that under your theory the BBs didn’t need GLD to manage any excess reserves as they could have just delta hedged any client long positions rather than accumulating physical (and it is just as easy to borrow physical gold as it is to borrow GLD). GLD coatcheck in theory requires delta hedging, but once you believe in delta hedging you don’t need GLD coatcheck in theory.

FOFOA: “the LBMA could be using GLD shares in its internal clearing process, only redeeming them when the underlying physical is needed to restock the subterranean stream, that is, for physical allocation or delivery. No need to redeem simply for interbank clearing, just transfer the shares.”

No, the LPMCL Clearing software, AURUM, and the associated SWIFT metal instructions do not allow for equity transfers, it is a metal (ounce) system only.

FOFOA: “I realize that you're trying to be Bron-the-Debunker, superhero destroyer of precious metals charlatans and defender of rational analysis, but with this GLD drain issue, I sometimes wonder if you really can't see how irrational your physical arbitrage view appears, or if you're just sticking to your guns.”

It seems pretty irrational to me for you to stick to your guns about a complex coatcheck/GLD/delta hedging when there exists another Occam’s razor explanation: after 2008 gold developed a narrative among institutional investors that it was an uncertainty hedge, lots of money piled in, pushed gold price and ETF holdings up, the narrative changed now that everyone thinks the economy is getting better and the money (and gold) flew out.

FOFOA: “And yes, I did see your latest post asserting that PHYS redemptions might be due to an arbitrage opportunity rather than a general shift in preference toward physical.”

I said “It is a bit hard to say if it is arbitrage or just a holder(s) from long ago getting out” the latter being a shift in preference toward physical. So it could be either.

FOFOA: “PHYS lost what, a little over one tonne in the last 5 months? GLD has lost an average of 46 tonnes per month for the last 10 months. That's well over a third of its holdings in less than a year.”

What interests me, is if BBs are desperate for physical and failure the fractional reserve system is imminent as some claim, why are we not seeing much physical redemption from PHYS, especially when a BB is paid to do so?

FOFOA: “I must admit, I don't envy you gold sellers right now.”

Thankfully the Perth Mint doesn’t “sell” gold, so I don’t envy them either. Insurance/protect wealth reasons for buying gold are boring and don’t attract clicks but are evergreen. Those who went for clicks and dramatic headlines and tried to sell gold on other reasons now face a credibility problem.

FOFOA: “in general, what is your argument for buying physical gold right now?”

Because even “if you don't expect a financial, economic or currency crisis”, you can’t be 100% sure at this time. Nor can you know if Freegold is right. If either happen then you have some protection. If neither happen, then any loss was just your insurance premium.

FOFOA: “We can resume the survey discrepancy debate if you want.”

I’m just going to have to find the time to write it up in an appropriate way and send questions to the LBMA, that is the only way to resolve it.


  1. In regard to an argument for buying physical gold right now:

    Because even “if you don't expect a financial, economic or currency crisis”, you can’t be 100% sure at this time.

    This seems rather weak in light of there being a financial system reset in 3 of the past 100 years (i.e. 1914, 1939 and 1971). Personally, I think the odds of another financial system reset within the next 10 years are closer to 50:50, justifying a higher insurance premium...

    So these things [resets] do happen. They have not happened recently, but they do have a tendency to happen. And when it happens, it is not the end of the world. It does not mean we all go live in caves and eat canned goods. What it means is that you have to reboot the system. That means the major economic powers have to get together and sit down at a table and rewrite the rules of the game. So I spend a lot of time thinking about what the future of International Monetary System will look like after this collapse. I think that is the most likely outcome."

    Bron, always enjoy reading your blog, lots of good info and needed debunking of goldbug conspiracy theories.

    But I do wish you would spend 25% of your time exploring which theories may have merit, especially in a financial system reboot. What could go wrong with bullion bank positions and the likely impact on the price of gold. What are the likely precursors in the gold market that would signal a reset was coming?

    And it would be nice if you would talk your book sometimes, since as a Perth Mint employee, most all your retirement plan is invested in gold; and that you have very positive feelings toward gold investment that are not always reflected in this blog.

    So please throw us faithful readers of your blog a bone and inform us which ideas for a rising gold price have merit and what impact a financial system reset may have on gold.

  2. I really enjoy reading your comments because of your deep knowledge of the physical market. Thanks for all your helpful information.

    But I do not understand why you are wasting your time with people who do not understand the difference between Gold as a FX instrument and physical Gold.

    I would be more interested on your thoughts on SGE physical delivery or on Sprotts theory that there is far more demand than supply in the physical gold market.

  3. Anon - I thought FOFOA did a good job on Sprott here http://fofoa.blogspot.com.au/2013/09/what-about-sprotts-4500-tonne-export-gap.html

    Marks - good topics for a post, will put on my to-do list.

  4. "exists another Occam’s razor explanation"

    Pot calling the kettle black Bron. The idea that somehow the quantity of gold compared to its annual production gives gold its value is not just convoluted but preposterous.

    How can you possibly quantify this ratio to come up with a value? You have no idea how much gold has been mined since the days of ancient Greece. And when was the last time you calculated the stock to flow ratio of a loaf of bread to decide whether you would pay the ask at the market?

    The simple & logical explanation is that it is the quality of something that determines its value. You pay for a loaf of bread if you think it has the qualities you desire. As it happens the quality of gold is constant, the perfect standard, the perfect measure of value.

  5. Thank you Bron for the link on the Sprott Analysis. Very interesting!

    Do you have any thoughts on the physical delivery of SGE as reported by Koos Jansen?
    Is there any flaw in his analysis?


  6. Justin,

    I'm at a loss at where your comment comes from - where in the post was I talking about quantity of gold vs annual production?

    I only mentioned the turnover, which is volume traded by stock. And in that case I wasn't using it to make any statement about the "value" of gold.

  7. Koos has done some good work there but I would note in the comments to some of his posts it is not clear that SGE deliveries are necessarily reflective of demand, depending how you define that.

    For example, consider the following figures for a year:

    Gold stock in country at start of year: 1000t
    Mine production 100t
    Imports 50t
    Scrap 50t (came from 1000t stock)
    Gold stock in country at end of year: 1150t

    SGE would report deliveries of, and Koos say demand is, 200t.

    Alternatively you could argue that scrap should be excluded because that was just gold changing hand and demand was 150t, as that was what was added to stock (ie net accumulated).

    Alternatively, you could argue demand is 50t as that was the net accumulation by the country from the rest of the world.

  8. Thank you for your thoughts on Koos Jansen

  9. I note http://www.zerohedge.com/news/2013-11-14/qeeen-yellen-nomination-hearing-post-mortem-dovish-32-18-hawkish Yellen said that "NO ONE HAS A GOOD MODEL ON WHAT INFLUENCES GOLD PRICES"

    Maybe FOFOA's FX friend should contact her

  10. You've done well addressing this long-winded attack without resorting to the same level of petty personal vindictiveness shown by your tormentor.

    I forced myself to read the original. I regret that.

    I note elsewhere in his blog that he has just discovered London trading volume. Remarkable.

  11. Was good to hear your thoughts on Big Al's show last weekend Bron.....keep on blogging !