Gyroscopic Investing

There is a discussion over on Bogleheads about the expense ratio decay of the asset value of GLD.

I thought this was pretty funny:

let me summarize by saying that a close reading of the prospectus (which I have done) indicates that it would be possible for the vaults to be full of toasted marshmallows without necessarily violating the prospectus.

It got me to thinking about precious metal ETFs versus CEFs.  With GTU, and assuming a favorable premium, does the sale of additional shares rejuvenate/offset the expense ratio to some degree?

There is.  Try "Bogleheads".

Could you please provide a live link to the thread on the Bogleheads forum that you are referring to?

Try http://www.bogleheads.org/forum/viewtop ... 0&t=105526

MangoMan wrote:I meant: Could you please provide a live link to the thread on the Bogleheads forum that you are referring to?

The word "Bogleheads" in the OP is a live link.

MangoMan wrote:sorry, my bad. it was early in the morning, and the contrast of the background made it hard to see.

Hey, no problem.  The link shows up as very lightly underlined on my display and I can see how you could easily miss it.

My initial thought is that GLD is really no different than any ETF or mutual fund in that there is a cost associated with the management of said fund.

WildAboutHarry wrote: It got me to thinking about precious metal ETFs versus CEFs.  With GTU, and assuming a favorable premium, does the sale of additional shares rejuvenate/offset the expense ratio to some degree?

GTU (and CEF) have historically issued new shares when the premium has gotten higher than usual (over 8% or so).  The new shares are issued at what is still a premium, but a lower premium.  The market makers who are involved are thus paying more than the cost of the gold for the new shares (in reality they probably short the new shares before they're created, so their profit is the difference between the current market premium and the "new share" premium).  The difference between the cost of the gold and cost of the new shares is used by these funds to pay for the costs involved in actually buying the gold (and shipping it to their vaults) and I believe typically exceeds these costs - resulting in a profit (held as cash) which the funds use to pay their ongoing expenses. I don't know this for sure, but I would be surprised if these funds have ever sold any metal to pay their expenses - i.e. I believe it's quite likely that the sale of additional shares completely offsets the expense ratio.

The contrast with GLD/IAU/SGOL is interesting.  With the ETFs the authorized participants receive the entire benefit of the difference between the cost of the gold and any difference between this and the cost of as many shares as they create/destroy.  OTOH, they bear the entire cost of buying/selling the physical gold as well.  Meanwhile, the ETFs pay their expenses by selling some of the gold they're holding.

Creating/destroying shares in the closed end funds is a win-win (market makers and shareholders both benefit).  With the ETFs it's a win only for the authorized participants - but they compete against each other which tends to keep the premium/discount quite small.

rickb wrote:I don't know this for sure, but I would be surprised if these funds have ever sold any metal to pay their expenses - i.e. I believe it's quite likely that the sale of additional shares completely offsets the expense ratio.

Thanks for the detailed discussion. I did't really know what the exact mechanism was, other than the transactions are supposed to be neutral as far as existing shareholders are concerned.

WildAboutHarry wrote: There is a discussion over on Bogleheads about the expense ratio decay of the asset value of GLD.

I thought this was pretty funny:

let me summarize by saying that a close reading of the prospectus (which I have done) indicates that it would be possible for the vaults to be full of toasted marshmallows without necessarily violating the prospectus.

It got me to thinking about precious metal ETFs versus CEFs.  With GTU, and assuming a favorable premium, does the sale of additional shares rejuvenate/offset the expense ratio to some degree?

Thanks for the compliment; I'm the original poster of that comment.

I think it's interesting that no one has produced any evidence that my analysis is wrong.

Re marshmallows, the SPDR Gold Trust Prospectus dated 26 April 2012 does not have the previous clause about the Custodian not being responsible for purity and quality.

Anyway, I think if the Custodian took unallocated gold deposited with it by Authorized Participants and instead of allocating physical gold bars it put marshmeallows in the vault that would clearly breach this from page 9 of the prospectus: “Under the Custody Agreements, the Custodian is only liable for losses that are the direct result of its own negligence, fraud or willful default in the performance of its duties."

Slotine wrote: GTU is being paid 0.30% for the first AUM of $100M, 0.225% between $100-$200M, and 0.15% above $200M.  They get more than enough money to cover costs as you can store unencumbered gold at about 0.12% flat and cheaper by bulk weight. 

This is still money coming out of the NAV, although since their fund only requires that at minimum 90% is actually held in gold, they don't really need to ever sell gold to pay themselves.  The actual trustee policy is to target 95% though.

As for premiums affecting their bottom line.  They can't as it's illegal.  Money from new shares minus part of which goes to underwriting fees, has to be split as declared in the 'Use of Proceeds' section of their prospectus.  95% as bullion, up to 5% as certificates, and the remainder as cash.  All three increase the NAV correspondingly.  It doesn't mean they can't withdraw their 0.225% the next day, but there's no pocketing of extra profit involved.

I'm not saying the fund doesn't have expenses, or that the expenses aren't ultimately paid by the shareholders.  Rather, when creating shares there's a profit involved (which indeed becomes an asset of the fund, so reflected in the NAV) and the fund uses this profit to pay its expenses.  In the case of GLD, creating/destroying shares also creates a profit, but all of the profit goes to the APs and none of it to the shareholders (the goal is to induce the APs to keep GLD's market price close to its NAV) - so there's no way GLD can pay any expenses except by selling some of the gold held by the fund (creating an ever dwindling oz/share ratio).

In 50 years, I'd expect GTU's oz/share ratio to still be approximately what it is now.  It's older sister fund, CEF, has essentially done this over the past 40+ years (CEF is a little more complicated to evaluate since it holds a gold/silver blend with a 50/50 value target).

In 50 years, GLD's oz/share will be a little less than 82% of what it is now (assuming the ER stays at 0.4%).

In both cases, the funds will have extracted about the same percentage of expenses from the fund's NAV.  But GTU effectively has an income, while GLD does not.

Oh one shouldn't feel safe at all about the legals of the ETFs. But they do have the gold - try and argue against this sort of massive data analysis http://screwtapefiles.blogspot.com.au/p ... abase.html