Exchange traded funds (ETFs) are hot. Everybody is buying. The mantra of passive investing has created a billion-dollar industry for these “trackers.” And gold is not staying behind. One of the world´s largest ETFs is the SPDR Gold Shares Fund (GLD), a gold-backed ETF, which is supposed to track the gold price. And the GLD ETF is perhaps among the world´s “safest” ETFs, but let us not get started on junk bond index ETFs. ETFs suffer from deep structural flaws whenever there arises a shortage of liquidity. Today we will walk through the case of the largest physical gold backed ETF in the world.
The GLD gold ETF is perhaps the best example of what is wrong with these trackers. In theory, GLD might be the simplest example that exists. GLD only invests in one physical asset (not even futures), that is, physical gold. ETFs are, by definition, open-ended funds, which means that the number of outstanding shares is not strictly limited. New shares are issued whenever new buyers emerge, and the proceeds are exactly invested according to the price or index that the ETF is supposed to track. For many gold investors, this is a reason to view an investment in gold through an ETF as equal to an investment in physical gold for instance through GoldRepublic.
We will show that the supply and demand for ETF shares and the underlying (net) asset value in times of crisis does not follow the predictable path that we would expect.
The GLD is an ETF that knows periods in which both new shares should have been issued, but were not created and issued, as periods in which many shares were liquidated (during which total GLD assets under management dropped twice as fast as the gold price). All this implies that an investor in a gold ETF not only exposes himself to risk regarding the gold price, but also to risk regarding the structure of the ETF itself.
In 2013, the gold price suffered a heavy decline.
||Net Asset Value (NAV) per GLD-aandeel
||Number of outstanding GLD-shares
||GLD Assets Under Management (AUM)
||Gold price per troy ounce (Londen PM Fix)
||Gold price to Net Asset Value (NAV)
||-60 basis points (expense ratio = 0.4%)
From the table above, we can conclude a couple of things. First of all, the size of the GLD (in terms of AUM) fell faster than the Net Asset Value (NAV) per GLD share. But that is not the only thing one might observe: we can also see that the real expenses of the ETF are, in fact, not 0.4% (as is stated in various places), but 0.6%
But the greatest risk of all types of ETFs is the liquidity risk in times of a liquidity crisis.
How does the price formation of an ETF work in practice?
In case of an ETF, two types of different parties are involved: the issuer of the ETF and the designated market makers (sponsors). An ETF sponsor (or “Designated Agent”) has the right to create blocks of new ETF shares and remove blocks of ETF shares from the market.
Is there increasing net demand for the ETF?
In that case, the price of the ETF rises to above the underlying value of the ETF assets (the Net Asset Value, or NAV). The ETF is “overvalued” relative to its NAV; the ETF is sold against a “premium” relative to NAV.
This produces an incentive for the sponsor to go short the ETF and long the underlying assets. This process continues until the sponsor has “enough” underlying assets to do a swap with the ETF issuer (assets for ETF shares).
Is there declining net demand for the ETF?
In that case, the price of the ETF drops to below the underlying value of the ETF assets (the NAV). The ETF is “undervalued” relative to its NAV; the ETF is sold against a “discount” relative to NAV.
This produces an incentive for the ETF sponsor to buy up ETF shares and convert the shares into the underlying ETF assets. Then, the ETF sponsor can sell the underlying assets on the market and obtain an arbitrage profit.
At least, that is how things work in theory. Because this imperfect market of ETF sponsors (or “Designated Agents”) there is one very big “but”, especially in times of crisis. Whenever only one or two parties are responsible for the price formation of an ETF through arbitrage opportunities between the price of the ETF and the value of the underlying assets, then the odds are very high that the following happens:
With declining demand, or in other words a sudden wave of ETF sales, the ETF sponsor is basically forced to take balance sheet risk by parking the underlying assets on its own balance sheet (until the sponsor is able to resell them in the market). During a liquidity crisis, this implies a gigantic risk. ETF sponsor are, moreover, in many cases banks which are heavily regulated and are forced, even during a crisis and times of falling asset prices, to maintain a minimum level of liquid assets. And both gold and illiquid bonds are NOT included in the regulator´s definition of liquid assets.
All this means that these ETF sponsors will not engage in arbitrage operations until the potential profits become so large that they enter the market again.
And this implies that the price of almost every ETF will be discounted severely, and therefore sold far below its Net Asset Value (NAV); the discount to NAV will increase considerably, and the supposedly perfect arbitrage model behind the price formation of ETFs will go up in smoke whenever liquidity is hard to get. And that mean that, in the case of GLD, that you are buying a pig in a poke. Good luck with that.