Submitted by Andrew Bogan of Casey Research
Are ETFs Really Safe?
Dr.Andrew Bogan is a managing member of Bogan
Associates, LLC in Boston, Massachusetts. He has spoken at many
international investor conferences – his specialty being global equity
investing – and has been interviewed on live television for
CNBC's Strategy Session.
In an attempt to understand the
relatively new but wildly popular Exchange Traded Funds (ETFs), Dr.
Bogan did extensive research into the structures used by ETF operators,
with a special focus on the potential risks that might arise should they
be faced with large and sudden liquidations. Given that there are about
2,000 ETFs in existence, with assets totaling over $1 trillion, we
thought it appropriate to find out what Dr. Bogan has learned in his
David Galland: Our primary goal
today is to give readers a better understanding of exchange-traded funds
(ETFs) and the risks that come with them. Speaking personally, I've
been in this business for a long time, and I find anything that grows as
quickly as ETFs have a bit worrisome.
To begin, maybe you could just talk a little about the difference between an ETF and a traditional stock or bond mutual fund.
Yes. Shares in a traditional mutual fund, whether it's an index fund or
has a managed portfolio, don't trade in the open market. If you want to
own shares, you buy them from the fund. If you want to get rid of your
shares, you sell them to the fund.
A traditional mutual fund takes
its shareholders' capital and invests it directly on a one-to-one basis
in stocks or bonds and holds those securities in custody. Thus it's
always 100% reserved, meaning that the securities it owns correspond
exactly to the shares its investors own. If you want your capital back,
the fund can deliver it to you either in kind or in cash, depending on
That's not the case with an ETF. Shares in an
ETF trade in the open market, which is where retail investors buy and
sell them. An ETF also issues and redeems shares every day, like a
mutual fund. But, unlike a mutual fund, it does so only through
"authorized participants," which are brokers, market-makers and other
DG: Jumping right to the point, has there ever been a problem with an ETF?
have operated pretty well historically, but the mechanics of share
issuance and redemption also creates some unique differences that we
believe may lead to unintended consequences.
There already have
been a few problems with ETFs, some more significant than others. The
Flash Crash on May 6 of last year showed some structural issues with
ETFs and perhaps with our whole market system for equities as well. It's
hard to decide where to draw the line, but a lot of securities departed
from their perceived value during the Flash Crash by very large
amounts. The reasons are still not completely understood, although the
SEC has made a reasonable effort to understand what happened.
incident occurred in September 2008, when the Lehman and AIG mess was
upon us. The commodity ETFs run by ETF Securities, Ltd., in London
halted trading when AIG's solvency came into question. The funds were
investing in derivative contracts, including swap agreements, some of
which were with AIG. It was only the Federal Reserve pumping in tens of
billions of dollars that prevented those products from going. Bailing
out AIG averted a disaster for the funds, and they continued to trade
the next day.
DG: So, the issue with the ETF securities fund was more around the derivatives the fund held, not the structure of the fund itself?
that particular case, it was around the derivative contracts that
underlay the fund, although that kind of arrangement is very common with
European ETFs. Even equity index ETFs in Europe tend to be structured
that way, and that's also not uncommon with a lot of the foreign stock
ETFs as well – including some of those traded here in the United States.
think it's a clear example where you have a counterparty risk wrapped
inside the fund that could be very significant in bad circumstances.
In the case of the Flash Crash, your research paper pointed out that
even though ETFs represent only 11% of the listedsecurities in the U.S.,
70% of the canceled trades during the Flash Crash involved ETFs. Is
there an explanation for that?
AB: Some clarity
is starting to emerge from work done by the SEC and others. But from our
perspective, those statistics are quite alarming. There's no good
reason 70% of canceled trades would be in ETFs while only 11% of listed
securities are ETFs. And even though ETFs trade more actively, they
don't represent 70% of all trading volume. So any way you look at it,
they were badly overrepresented among the canceled trades, i.e.,
overrepresented among the most extremely off-priced trades.
the perspective of financial theory, that makes absolutely no sense.
ETFs are meant to be index-fund trackers. They’re meant to represent a
whole basket of shares, and yet these very securities that are meant to
be diversified actually fell more than their underlying stocks during
the Flash Crash, more often and more deeply.
worrisome; it tells you that in a crisis environment ETFs don't behave
the way financial logic suggests they ought to, which suggests to me
that the theory is incomplete. People haven’t really looked closely
enough at what the unintended consequences of ETF issuance and
redemption mechanics are, and what the realities are in stressful market
DG: At this point, more than half
the American Stock Exchange's daily volume is ETFs, which is quite a
number. These things have only been around for, what, less than 20
years. Yet from everything I've read, it seems they’re not very well
understood, even by you guys. Which is saying something because you’ve
spent a lot of time looking at them, and there are still blank spots in
your knowledge about how they actually operate.
and I think that's an important point. We understand the mechanics of
how an equity trades and from where it derives its value and how it's
priced in the market. The mechanics for mutual funds are well understood
also. The challenge with ETFs is that the process of issuing and
redeeming shares that also are trading is much more complicated than a
lot of people want to talk about. It allows for some unintended
consequences, particularly in connection with short-selling, which
became an important factor only in the last decade.
talk about the process of creating new shares. If I'm running an ETF
that is designed to mimic the S&P 500 index and I have a lot of
people who want to own my fund, I can simply issue new shares based upon
the flow of stocks into my fund, right?
Shares can be created at the end of any day if someone delivers a basket
of underlying stocks to the ETF through an authorized participant. And
shares that are not wanted in the marketplace can be redeemed in kind
for the underlying stocks – or in some cases cash. That's all been
carefully structured and works smoothly. The issue is what happens when
short-selling dominates the trading.
People have been
short-selling ETFs up to shocking levels, like 100% short, 500% short,
sometimes over 1,000% short. That's in a world where stocks like Apple
are 1% short, or IBM is 1.4% short, or General Electric is 0.5% short.
You really don’t see traditional stocks with short positions anything
like this, so clearly something is fundamentally different. The
difference is that ETF short-sellers – including hedge funds, dealers
and arbitragers – are confident they can always create the shares needed
to cover, so they see less risk of being squeezed.
DG: But in a traditional short-selling situation, you typically have to borrow the shares before you can short them.
Yes, and that's true here too. But if you look at the Securities
Settlement Failure data, ETFs are very oddly overrepresented, so it does
look like there is some short-selling that happens before the shares
are borrowed. But that's a small matter. The problem is that there is no
limit to the amount of short-selling you can theoretically do while
still having borrowed the shares. It simply requires the same share to
have been borrowed, short-sold, borrowed from the new owner and
short-sold again down a daisy chain. That's how you get these
arbitrarily large short interest figures.
involves new buyers coming in without the shares being created at all,
and that's the fundamental asymmetry in the short-selling that we're
most concerned about.
DG: Let's get to that,
because you have retail investors, for lack of a better word, and you’ve
got the hedge funds. I suppose they could both own the same fund, but
for completely different reasons; a hedger to hedge another bet, and a
retail investor to pursue a certain goal, but the net result is that the
short interest is still way out of whack from what you'd expect to see
in a traditional stock. I suspect this is something that most of the
retail investors are unaware of. So, where is the potential for the ETFs
to get into trouble?
AB: The trouble could come from a number of different angles.
concern is that the huge short interest building up essentially leaves
the ETF as a fractionally reserved stock ownership system. If you have a
fund, for example, that is 500% net short, then for every one holder of
an actual share there are five other investors who own IOUs for the
shares. Their real shares have been lent out and short-sold to someone
else – usually without the original owner's knowledge, unless they read
and still remember the margin agreement they signed when they opened the
account 10 years ago.
For the ETF itself, it means that the fund
holds only 15% of the underlying securities implied by the gross number
of fund shares that investors think they own. The other 85% isn't
totally missing, it just isn't held by the fund.
commented that the money is all there, it's just in hidden plumbing in
the financial system, and we agree with that exactly. The question is,
how many investors understood they were storing their money in the
DG: So walk us through what
might happen if there were large-scale redemptions. Let's just say that
for whatever reason, people decided this was the time to get out of a
particular fund. How do things get unwound?
Redemptions have to flow through an authorized participant, which is
usually a broker or market-maker, and it's only that institutional layer
that can actually redeem. If for some reason a significant portion,
say, half or 80% or so, of the total fund ownership wanted to redeem and
get the underlying stocks from the ETF through the authorized
participant layer, you would fundamentally have a crisis in a
The ETF could not deliver the
underlying stocks to all the would-be redeemers. The investors who
really owned just an IOU on shares that had been lent to short-sellers
wouldn't have a direct claim on the fund, so their demand to redeem
would force an unwinding of the short-sales.
it seems that it's not so much the fund that might have a problem. The
fund is only liable for the shares it has issued. The risk seems to lie
in the counterparties – the brokers or the investors that brokers lent
AB: Right. Essentially you have just
that. You have quite a bit of counterparty risk here, because if you
think your shares can be redeemed and then the fund halts redemptions
because they’re running out of the underlying stocks, you're stuck.
Normally ETF shares are redeemable through the authorized-participant
channel, but an ETF or any other institution that issues something that
is redeemable but fractionally reserved could be hit with a run, like a
Now the big question is, in practice, would this happen?
It's up to everyone to form their own conclusion, but interestingly the
first argument we heard when we began looking into ETFs was that this
was just a theoretical topic and that there would never be a really big
redemption in a large ETF. But we have since learned that's actually not
the case, because a giant redemption in IWM, one of the largest ETFs,
occurred in 2007.
Now we think that 2007, being one of the best
markets for equities since maybe the late ‘90s, was a pretty forgiving
time to test the crashworthiness of an ETF that runs into a massive,
unexpected redemption. But IWM was redeemed from millions of shares
outstanding down to something on the order of 150,000 shares, and in one
day, and that's because somebody tried to crash the fund.
DG: Was that a really lousy fund, and somebody just said, "Enough, I'm going to punish you guys and get out of it,” or –
Oh, no, no, IWM is one of the largest and most liquid ETFs in the
entire market. It's the Russell 2000 iShares ETF. It is the poster child
of why ETFs are great. But even so, what's interesting is that the
first argument we got from industry insiders was that our misgivings are
nonsense, growing out of some theoretical conversation about what might
happen but is never going to happen, and now we're being told it
already has happened and nothing broke too badly, so what are we worried
DG: Let’s stick with this potential
problem of a huge bunch of redemptions. People say, "Oh my god, I've got
to get out of my ETFs," and there is a wholesale run on the funds.
Because of the way ETFs are structured, it would seem that if they post
net redemptions for a day, that the broker that had lent fund shares to
short-sellers would just force the borrowers to buy back and cover their
AB: That's exactly right, but
remember, for an ETF to create units requires someone to deliver the
underlying stocks, so there's somebody who's on the hook to buy those
stocks en masse all at the same time.
DG: No matter what has happened to the price in the interim.
Yes, which gives rise to the question of who's on the hook and what's
their creditworthiness when they get put on the hook. Have their prime
brokers really been keeping appropriate track, as they’re required to do
and on most days have done, of the creditworthiness of those, say,
hedge funds or other kinds of short-sellers?
DG: Because you're not talking about small amounts of money.
No. In fact, in one ETF, IWM again, short positions recently amounted
to 14 billion dollars. That's not an enormous amount for the capital
markets, but it's a pretty significant amount with respect to 2,000
small stocks. If there were a run, actually doing that unwind and
getting those 14 billion dollars' worth of extra ETF shares would
require buying 14 billion dollars’ worth of Russell 2000 stocks. If you
didn’t want to be more than, say, 10% of volume, it would take 40
trading days to buy all you needed.
So we think that if you
actually had a very sudden redemption run on IWM, there is a real
likelihood of a short squeeze occurring in the Russell 2000. We don’t
expect that at any particular time, it's just something that could
happen if enough things went wrong.
The short position in an ETF
like IWM being over 100% means that a large amount of the money
investors think they have placed in Russell 2000 stocks has in fact been
lent to hedge funds and other short-sellers. You take that across the
entire ETF industry and you're looking at about 100 billion dollars in
short interest – money that did not go into the underlying shares or
gold or whatever the ETF represents. It was instead lent to hedge funds.
It has been deposited in a shadow banking system where ETFs allow
short-sellers to borrow money from institutional and retail investors.
DG: And what are they doing with that money?
Well, no one knows. Presumably they invest it in what they think is
going to make a better return than what they shorted, because you can't
score the 10% or 20% those guys are all trying to make every year by
buying the index. So it's anybody's guess.
One question that Terry Coxon asked as I prepared for this interview was
whether there is any way for the marketplace to let the fund's share
price deviate for long from NAV?
AB: The tracking
of an ETF's price with the fund's NAV, which historically has been
extremely close, is totally dependent on an arbitrage mechanism. The
arbitrager can make money by continuously pushing the price of the ETF
toward its NAV. The question is... what NAV? What they mean by NAV is a
value per share outstanding of the fund's underlying stocks. But of
course you have this huge implied ownership through short-selling, and
the short-sellers' shares are not being counted in the shares
DG: A lot of our readers have
money in GLD, which is the ETF that invests in physical gold. You've
looked at GLD, and it's based upon the premise that as investors pour
money in, the operators of GLD turn around and buy physical gold and
store it. And likewise with redemptions, they just sell the gold. My
understanding is that there isn't anywhere near the same level of short
interest on GLD.
AB: The short position in GLD
isn't nearly as large as it is for some equity funds – but we have
looked at GLD, and it has the same structural issues, just to a lesser
extent, at least for now. The short interest in GLD has fluctuated
around 20 million shares. Now, GLD is a pretty big fund. With 20 million
shares short, it is roughly 95% fractionally reserved. So for all the
investors who think they own the underlying physical gold, the fund
actually has 95% of it in the vaults.
But GLD does not have to
stay at 95% fractionally reserved. If there were a massive wave of
short-selling in GLD, you could end up with a very significant
fractional-reserve situation. If that were followed by heavy
redemptions, you'd have the same kind of problem I described earlier –
not enough gold to redeem all the shares.
DG: Could they just say, "From here on, we're not issuing any more shares"? Would that stop the short-selling?
Not necessarily, because, you know, the short-sellers are selling – in
fact, it would probably exacerbate the short-selling. So as long as a
fund is issuing shares, aggregate buying demand can be satisfied by
expanding the fund. If they stop issuing shares, aggregate demand would
get satisfied by short-sales of existing shares. So, if anything,
closing the issue window should make the problem worse, not better.
Working through the mechanics of this, let's say gold drops by a few
hundred bucks. Say, for instance, that there is some major change in the
market along the lines of when Volcker raised interest rates back in
'79-'80. And at that point a lot of short-sellers say, "Okay, this is it
for gold," they pile on, they start shorting the hell out of GLD, and
now all of a sudden you’ve got a real problem because the fractional
aspect of it balloons, if you will.
AB: Well, you
don’t necessarily have an immediate problem. It depends on the market
conditions and the level of panic. You certainly would have a ballooning
fractional-reserve situation, meaning that the reserves held in actual
gold versus the implied ownership by people who think they own GLD (even
though the shares have been hypothecated by the broker) will shrink.
Those investors may believe they are still entitled to the metal, but
the reserve of gold held on their behalf starts to shrink very quickly
under those conditions.
The bigger challenge might be if there
were an actual redemption wave. If that happened when GLD was already
substantially fractionally reserved, then you're back to an 1800s gold
bank problem. Fractionally reserved banks can be hit with a run.
Right. Is there anything else that would make this whole "house of
cards" collapse? Suppose a highly visible ETF stumbles and is unable to
meet redemptions, or they just have to postpone redemptions. That might
be the sort of trigger that could really send people off.
You know, one of the big risks, by the way, that no one has really
discussed much, is if an ETF were to have a big redemption run in
panicky market conditions and halted redemptions. Halting redemptions is
a complicated decision, because it breaks the symmetry that allows the
arbitragers to go long or short both the basket of stocks and the ETF
shares to move price toward NAV.
So it's quite possible that if
redemptions were halted for any length of time, the arbitragers wouldn't
be keeping the share price in line with NAV. We already know from the
Flash Crash that significant price departures from NAV are quite
possible for ETFs.
DG: Knowing what you do, I
mean, obviously you deal on an institutional level with your
money-management firm, do you own ETFs personally?
AB: We do not. We do not own any ETFs either personally or on behalf of the funds we manage.
DG: Is it because of the research you’ve done or just because it's not what you guys do?
would say it's primarily because it's not part of our strategy, but
obviously we did the research because we were interested in
understanding the product better.
DG: So, any advice for readers? Is there a short interest over which a person should be concerned about his holdings?
Well, I don’t know if I could set a threshold, but I would certainly
encourage people to make sure they know what the short interest is in
any fund they are considering. That's a metric that is starting to
become more accessible. Since we published in September, some of the ETF
sponsors, like BlackRock, have begun reporting on ETF short interest,
which I think is terrific – kudos to those guys. We would like to see
better transparency and disclosure, so that institutional and retail
investors alike are aware of the counterparty risks that are "hidden in
the plumbing," to use Morningstar's term, and are aware of the actual
and somewhat complicated mechanics of the products that they’re buying.
Do the ETFs with a mandate to magnify an index 2 or 3 times (e.g., RSW)
have an elevated level of risk, due to the additional leverage?
The underlying "assets" from which these funds get their NAV are
derivatives to begin with, which introduces another layer of
counterparty risk – one that has already experienced serious
problems. We find it surprising that packaging complex derivatives in an
exchange-listed security (the ETF) seems to remove all of the
sophisticated investor standards usually applied to derivatives trading
by SEC, CFTC, etc.
One ETF recently launched in the U.S. is PEK,
the Market Vectors China A Shares ETF. This is another great example of
where the industry is headed.
It is illegal for most foreign
investors – except a few licensed global institutions – to buy A shares
on Shanghai or Shenzhen, China's two mainland stock markets, and Market
Vectors is not one of the exceptions. So instead of owning A shares, the
ETF owns swaps with brokers that are licensed in China to own A shares.
The fund holds the swaps as its underlying "assets." So PEK is an
NYSE-listed China A shares ETF that does not own a single Chinese A
If PEK were to become significantly short in the secondary
market, it would mean a fractional-reserve ownership of a derivative
representing a basket of stocks that would be illegal for nearly all of
the ETF's investors to own directly. More confusing still is what it
means to be short PEK in the first place, since it has historically been
illegal to be short A shares in China at all.
In essence, ETFs
are being used to package and securitize products that are at best
poorly understood and in some cases are used to circumvent securities
regulations. An example closer to home is when the SEC briefly banned
short-selling of essentially all financial stocks in 2008. The
financial-sector ETFs were not on the list, so many hedge funds kept
right on shorting financials using those ETFs.
DG: Certainly a lot to think about here. Any other questions I forgot to ask about, but that I should have?
AB: No, I think that was a pretty good coverage of a little bit of work we've done.
Is there a good publication that would help people better understand
the mechanics of the ETFs, because it is obviously very complicated,
something that people might want to be able to study?
the best place to look is in the fund's prospectus. The prospectuses
are long and impenetrable, because they’re written by the legal team,
but they really do have a tremendous amount of information. If you can
float through one of them, I think it's definitely to your advantage.
DG: Thank you for your time.