By Bradley Kay
It's another autumn since the financial crisis of 2008. The air is getting brisk, children are returning to school, and financial pundits are fresh from their summer holiday and ready to identify the next innovation to destroy the global markets and economies. That can only mean one thing: a new crop of scare stories about exchange-traded funds. The most recent offender came out a little over a week ago, under the title "Can an ETF Collapse?" by investment firm Bogan Associates LLC. It quickly spread to the typically excellent FT Alphaville and even front page commentary on CNBC.com.
So, can an ETF collapse? Well, no. At least, not via the process detailed in Bogan Associates' paper. The white paper focused on ETFs with substantial short interest relative to the assets in their trust. These are almost always smaller funds focused on a corner of the market with a very negative outlook, which causes most long-only shareholders to bail out of the fund and entices short-sellers looking to benefit from the underlying shares declining in value.
In some instances, the value of the ETF shares being shorted can even exceed the value of shares that long-only investors actually hold. If this happens, the few long-only shareholders may not like the future returns on their investment (professional short-sellers tend to be a savvy lot in their choices of investment), but the standard safeguards in the institutional share lending market prevent this short-selling from creating "phantom shares" in the market that lack backing assets. Financial blogger and former trader Kid Dynamite wrote an excellent and detailed post on this exact subject earlier this week, with a great discussion in comments, but I will outline the basics of why short interest does not pose a problem for standard ETF shareholders in the rest of this article.
Any Shares to Borrow? I'm a Little Short
We can start this example with a hypothetical new ETF. There is no short interest and 2 million shares outstanding. If shares are worth $50 a piece, then there are $100 million in assets in the ETF trust and $100 million in shares held by the investing public. Every fund starts out this way, with little to no short interest and a share count precisely matching assets.
Now let's assume that trader Hedgie Shorterson wants to bet on this ETF going down in the future, because it focuses on an industry, region, or stock type with particularly poor prospects. In fact, Hedgie wants to place a particularly massive bet of shorting 1 million shares. Hedgie cannot simply sell the ETF shares in the market without holding them, as then he could not deliver the shares three days later (the required settlement date for equities and ETFs on U.S. and European exchanges). Instead, Hedgie needs to find an investor already holding those million shares, arrange to borrow them, and then sell those borrowed shares on the market.
If some institution holding a million shares of the ETF does not plan to sell anytime in the future, it can lend those shares to Hedgie for a fixed payment rate. In return, Hedgie pledges collateral, typically composed of stocks, bonds, or cash, that is worth the entire value of the lent shares. That way, the institution does not need to worry about whether Hedgie makes money on the short position or not. If the position fares poorly and Hedgie defaults, the institution only needs to sell the collateral and use the proceeds to buy back (or create) its lent ETF position. In the meantime, the institution knows that those million shares are lent out, and it can neither sell nor redeem those shares without demanding that Hedgie return the lent shares. In fact, redemptions of lent shares are explicitly banned in ETF prospectuses.
Hedgie borrows and sells those million shares on the exchange, and the new owners of those shares could take them in to redeem. However, that is perfectly offset by the million shares held by the institution that will not sell or redeem. So, here's the final outcome after the short sale:
$100 million (2 million shares) in assets in the ETF
$50 million in unsalable and unredeemable shares held by the institution (offset by $50 million-plus in collateral from Hedgie)
-$50 million short position for Hedgie
$50 million held by previous shareholders aside from the institution who lent shares (able to be redeemed)
$50 million held by new shareholders who bought the borrowed shares from Hedgie (also able to be redeemed)
So, every dollar in redeemable shares has a dollar in assets backing it. And that's with a short interest of 50% in our hypothetical ETF. The situation can get a bit more complicated if no institution has the million shares to lend to Hedgie. Then some institution will actually have to create the million shares before lending them out (putting more money in the ETF trust while allowing an equal value to be sold short by Hedgie), but it will still result in the same end point--where the institution holding the ETF shares knows it cannot sell and the new shareholders have full assets backing their redeemable shares.
In both of these situations, the ETF ended up with more shares in existence than the number of shares sold short. So, how do we get these situations where the number of shares sold short is 5 times as high as the number of shares outstanding in the market? To explain this, we need to keep following those short-sale shares after they enter the market.
The Chain Gang and the Incredible Shrinking Fund
Let's revisit our case of an ETF with 2 million shares outstanding after Hedgie short-sold 1 million shares. There's a chance that Hedgie sold those shares to another institutional investor willing to lend shares and another short-seller (perhaps Hedgie's sister, Peggy Shorterson) borrowed them. Now, Peggy also has to pony up $50 million in collateral to protect the new institutional investor against her short position, and there are now $100 million in short sales offset by the $100 million of long positions owned by institutions and protected with $100 million-plus in collateral that would be used to create new shares in case of a problem. There are still only $100 million in the ETF trust, which perfectly matches the $100 million of unlent shares on the market, but now we have a short interest of 100%. These chains of sales to institutions and more lending can lead to very high short interest numbers but not to more danger because every link in the chain requires full collateral to protect the position.
What also happens, as we would somewhat expect, is that the short interest rises because long-only assets start to leave the ETF. Let's start back at the fund with 2 million outstanding shares and 1 million shares sold short by Hedgie. If you were a shareholder with some of those 2 million unlent shares in our hypothetical ETF, and you saw the same warning signs that made Hedgie want to short it so badly, would you really want to stick around to find out how that fares? Redemptions by shareholders can shrink the ETF's asset base while leaving the short positions more-or-less intact (because some institutions specialize in creating and hedging ETF shares that they then lend out to short-sellers). Say that half of the investors don't like the writing on the wall and redeem their million shares. Now, our hypothetical fund still has that $50 million short position offset by a collateralized institutional holding, but it only has $50 million in assets held for the $50 million in unlent shares still in the market. Shrinking assets due to redeeming current shareholders led the ETF's short interest to grow from 50% to 100%.
What if the ETF shrinks away to nothingness? What if the holders of those $50 million in remaining unlent shares want out? Well, thankfully, there are processes to address even this scenario. Nearly every ETF has a level of assets that triggers liquidation. Once that many shareholders have sought redemption of their shares, the fund is no longer viable, so the fund company will simply sell everything left in the ETF trust and divvy it among the remaining shareholders who have not lent out any of their shares.
Where does that leave our short-seller and the institution that lent the shares? The short-seller will have to cover his short because the ETF that underlies the whole investment will soon be gone. Hedgie can either go on the market and buy up 1 million shares from the investors who want to sell out or redeem (unlikely, as he would have to offer a premium to get all the shares), or he can create the million shares on his own by buying the underlying stocks in the ETF or paying an authorized participant to create the shares for him. If poor Hedgie has gone bust, the lending institution simply takes the collateral he supplied and uses it to buy up the underlying securities for the ETF, creating the 1 million in shares needed to keep its long position. No matter what, the money is there for the new shares and that 100% short interest goes back to 0% with no harm done and no individual shareholders left holding the bag.
Ultimately, the danger for these ETFs with bearish prospects appears because all we see are the eye-popping short interest numbers and the small assets in the ETF portfolio. We don't see the hundreds of millions in collateral that back up each of those short positions and that ensure new units could always be created by the lending institutions to cover those short positions. Individual investors and anyone else who does not lend out their shares need not worry about "phantom shares" crowding them out in a rush for redemptions. Their shares can all be redeemed, and every one has full assets backing it in the ETF trust.
For those of a mathematical and mildly masochistic bent, here is more rigorous detail in the form of an induction proof on why there can not be an asset crunch for an ETF so long as naked shorting does not exist. The language in this supplement is more technical and contains hazardous amounts of jargon, because it originated as an e-mail string among our research team. Still, given that this is a weak induction proof, technical language should be the least of any readability issues.
Assume there's a short interest of $a in the fund and a total of $b in other shares out on the market (held by individuals, institutions, etc.--all that matters is that those are shares that are being held and not lent out, making them eligible for sale or redemption).
When an ETF first starts up, $a = $0 in short interest, so there are exactly $b in the fund's trust to match the $b of shares held by long-only investors. Now assume that Hedgie wants to borrow $c in shares and short them. One of the following has to happen:
1) Hedgie sources the shares from one of the existing shareholders.
- Immediately, the number of shares eligible for redemption falls from $b to $b-c, because an institution just lent $c of their holdings and now has collateral to cover their counterparty risk.
- Hedgie takes the $c in shares and sells them on the open market, increasing the short interest from $a to $a+c.
- Market participants buy up those $c in shares and consider them unencumbered, increasing the value of shares eligible for redemption from $b-c to $b.
- Meanwhile, assets in the trust have remained steady at $b this entire time, as there were no creations or redemptions, so there are still assets to match every share eligible for redemption.
2) Hedgie sources the shares from a create-to-lend program.
- The prime broker and authorized participant creates $c in new shares, increasing the assets in the ETF trust to $b+c, then lends out the $c in shares to Hedgie and takes his collateral.
- The hedge fund takes the $c in shares and sells them on the open market, increasing the short interest from $a to $a+c.
- Market participants buy up those $c in shares and consider them unencumbered, increasing the value of shares eligible for redemption from $b to $b+c.
- Meanwhile, assets in the trust have increased to $b+c due to the create-to-lend process, so there are still assets to match every share eligible for redemption.
By this process, the short interest can get as large as possible without ever allowing there to be a big pool of investors holding "phantom shares" that can't be redeemed. The short positions are always fully offset by institutions who know their shares are lent, will not sell without calling the shorts back, and hold collateral to cover the cost of new share creation in case Hedgie goes broke. If all the unencumbered shareholders start selling off, they can make the fund assets fall from $b to zero, but it can't go negative.
A short interest of greater than 100% simply means that $a > $b, but those $a of shorts are still fully offset with collateralized positions of institutional investors.
Disclosure: Morningstar licenses its indexes to certain ETF and ETN providers, including Barclays Global Investors (BGI), First Trust, and ELEMENTS, for use in exchange-traded funds and notes. These ETFs and ETNs are not sponsored, issued, or sold by Morningstar. Morningstar does not make any representation regarding the advisability of investing in ETFs or ETNs that are based on Morningstar indexes.