Classically an exchange-traded fund is a derivative: It’s a contract whose value is based on the value of a basket of underlying stocks. (Bonds, whatever.) The contract trades on an exchange — thus the name — and so it has a trading price, but it also has a value that you can calculate.1 You can replicate it by buying the underlying stocks, or hedge it by selling them. So the value of the ETF is very well defined and transparent, and its price should closely reflect that value. That’s nice!
In addition to being a derivative, an ETF is also the cool way for cool people to invest in stocks. Mutual funds are lame: They have lots of expenses, you can only buy or sell them once a day, you have to pay capital-gains taxes, many of them are even actively managed. ETFs are sleek and awesome and efficient-markets-y. So they’re growing, as a class, faster than mutual funds,2 and they have a marketing appeal that old-timey actively managed mutual funds lack.
Most of the time there’s no reason to distinguish between these two ways of thinking of an ETF. They work together. The derivative mechanics allow the cool-stock-investing tool to work: Because the ETF can be replicated and hedged, market-makers can constantly provide liquidity in ETFs by offering to buy your ETF shares from you (or sell them to you) and then hedge by selling (buying) the underlying components. The trading keeps the price accurate, so the price tracks the value. It’s a paradise of liquidity and zero-arbitrage pricing. It’s a derivative that makes the world better for regular investors.
But there are disconnects. On the one hand, there are ETFs that are less “cool way to invest in stocks” and more gee-whiz derivatives structuring, giving regular investors exciting ways to lose money trading levered products that they don’t understand.3 And on the other hand, there are ETFs that are basically mutual funds and don’t really work as derivatives, but are nonetheless marketed as ETFs because ETFs are cool; that is how branding works.
The Pimco Total Return ETF, for instance, combines the cool branding of ETFs with the formerly cool branding of the Pimco Total Return Fund, but trades in odd lots of illiquid bonds and can’t really be replicated with much confidence by market makers. But actively managed mutual funds are so 10 years ago, and the temptation to make them into ETFs is strong.4
Once you’ve gone that far you might as well go further, or so BlackRock and Precidian thought, incorrectly.5 The Securities and Exchange Commission this week rejected their applications ”to start the first non-transparent ETFs.” A non-transparent ETF is a contract that trades on an exchange and whose value is based on the value of a basket of underlying stuff, but you don’t know what’s in the basket.
The basket is actively managed by BlackRock or Precidian, and the ETF provider tells you every 15 seconds how much it’s worth, but the contents are a secret.6 Basically it’s a regular actively managed mutual fund that you can buy and sell all day.
There’s nothing obviously wrong with that as a concept. I don’t own any ETFs, but I own shares in some mutual funds, even some active ones. If I want to buy or sell I do that at the end of the day, buying from or selling to the fund itself at the net asset value that the fund reports. But if at noon I wanted to sell you some of my shares, and you wanted to buy them, and I was happy to sell at $20, and you were happy to buy at $20, what would be wrong with that? I mean, mechanically, probably lots of things — that’s not really how mutual funds operate — but it doesn’t, like, affront my sense of how markets work.
I have a thing, you want the thing, we each make a reasonable guess about how much it’s worth, our guesses overlap, and we come to an agreement on price. We don’t know the value of the shares, exactly, but that’s true of almost everything in life. (Certainly it’s true of buying and selling shares of regular companies, which is why people keep doing it.)
And the non-transparent ETF is actually a bit better than that: Like other ETFs, it would distribute an updated estimate of net asset value every 15 seconds so you don’t have to guess too much at the value.7
So a non-transparent ETF seems perfectly reasonable as a cool way to invest in stocks. (Bonds, whatever.) But it’s not a derivative. Or not a good one anyway; it’s a contract whose value is based on the value of a basket of stuff (this is secret) and that BlackRock/Precidian can change at any time. It’s hard to replicate, or hedge, the mind of a money manager, or even her portfolio if she won’t tell you what’s in it.8
So the derivative structure of the classic ETF, which allows for instant liquidity and a price that always closely matches the value, is broken in the non-transparent ETF.
And the SEC said no on more or less those grounds. Here are the SEC’s letters rejecting BlackRock’sand Precidian’s applications. They basically say, this derivative is totally dumb, it doesn’t work, don’t do it. From the BlackRock one:
The Commission preliminarily believes that, even under normal market conditions, market makers could be unable to deconstruct the portfolio holdings of a proposed ETF with sufficient accuracy in order to construct a hedge portfolio that is closely aligned to the NAV per share of the ETF. The proposed disclosures by the Applicants would likely be useful in narrowing down the pool of securities and other assets that may be held by the ETF, but only to a limited extent.
So the price of the thing in the market couldn’t be arbitraged to its underlying value. That’s bad for price discovery. It’s also bad for trading costs. Here’s another thing the market makers said:
They indicated that they would likely use the pieces of information provided by the Applicants (IIV, quarterly portfolio holdings disclosure and prospectus disclosure) to construct hedge portfolios using sophisticated algorithms. Their ability to construct hedge portfolios that are generally predictive of the portfolio holdings of the ETF is critical to their management of their exposure to the ETF. If there is a break in the alignment between the market makers’ hedge portfolios and the NAV per share of the ETF, the market makers’ risk of loss increases. The greater the risk of loss, the more the market makers will seek to cover that risk by quoting wider price spreads of the proposed ETFs. This would result in market prices, at which investors would buy and sell the ETF shares, not being at or close to the NAV per share of the ETF.
I don’t know if “sophisticated” is the market makers’ word or the SEC’s,9 but there you go: This structure would create uncertainty for market makers, which would lead to wide bid-ask spreads.
Also everything is just too slow:
Market makers operate at speeds calculated in fractions of a second. In today’s markets, 15 seconds is too long for purposes of efficient market making and could result in poor execution.
You can see the SEC’s priorities here: instant liquidity, tight bid-ask spreads, and prices that rapidly incorporate information and reflect market consensus about value.10 Those are totally reasonable priorities. That’s sort of a classic set of efficient-market priorities. I occasionally use the phrase“efficient markets fundamentalism” about the SEC, and this decision fits right into that world view.
You could have other priorities, other world views. A striking number of market commentators actually think, no, a few seconds is a pretty reasonable time to wait to do a trade. Relax, they say, that trade isn’t going anywhere, no real fundamental investor will be harmed by that delay. Have some batch auctions, even! Sure, non-continuous trading will make market-making riskier, requiring electronic market makers to widen their spreads. But not by that much. And sure, it will take longer for prices to fully incorporate information. But not that much longer. “There is such a thing as too much liquidity.”11
That forms the backdrop of a lot of the debate about high frequency trading, of course. HFT advocates say that they’re providing liquidity, improving price discovery, and reducing bid/ask spreads. Those claims encounter a lot of skepticism: Would slowing down by a few milliseconds really make markets that much less efficient? Shouldn’t we ban high-frequency traders, or at least slow them down?
What’s fun here is that it’s that debate in reverse. BlackRock and Precidian proposed a product that was too slow, that wouldn’t allow for incorporation of information into prices on a microsecond-by-microsecond basis, that wouldn’t let market makers make near-risk-free profits by trading quickly in multiple markets. And the SEC said, no, that’s not allowed. Modern equity markets are about speed and efficiency. If your product doesn’t align with those values, you can’t trade it.
1 And it’s a pretty simple derivative to model; there are not a lot of debatable assumptions in “add up the price of a bunch of stocks and take a weighted average.” It’s a delta-one derivative, as they say.
3 No I’m sure your levered inverse VIX futures ETF is a great product for retail investors, tell me more.
4 Or maybe call your not-especially-index-y mutual fund an index fund, same basic idea.
5 Also other guys — BlackRock and Precidian got their rejections this week, but other applications (from Capital Group, T. Rowe Price, Eaton Vance, etc.) are pending.
6 Like any mutual fund, the non-transparent ETF would disclose the contents of the basket at the end of each quarter, on a 60-day lag. So that’s something, though it would presumably trade frequently enough that this picture would not be that useful. There would also be general prospectus disclosures about investment strategy, limits on what can go in the basket, etc.
7 Technically it’s called an “intraday indicative value” and is not exactly the NAV. From the BlackRock letter:
Investors and others acquiring the proposed ETFs’ shares would primarily have to rely on the intraday indicative value (the “IIV”), which would be disseminated by an exchange every 15 seconds during the trading day, to assess the value of a proposed ETF due to the lack of portfolio transparency. The IIV would be calculated by a calculation agent who would receive the daily list of securities constituting the proposed ETF’s portfolio from the ETF sponsor. As acknowledged by the Applicants, the IIV is based on the value of the proposed ETF’s portfolio and is calculated by the calculation agent using the last available market quotation or sale price of the proposed ETF’s portfolio holdings. As further acknowledged by the Applicants, the IIV is not the NAV; rather, it is a reference produced by a third party seeking to approximate the proposed ETF’s underlying per share net asset value. Applicants also concede that the IIV is not intended as a “real-time NAV” and (unlike the NAV) would not include extraordinary expenses or liabilities booked during the day. As discussed below, an ETF’s portfolio could contain securities and other assets all (or most) of which need to be fair valued in order for the IIV to be accurate.
For illiquid securities, etc., it could be very wrong.
8 Tangentially related, and delightful, are Renaissance Technologies’ options on whatever it’s thinking of.
9 I have a bee in my bonnet about “sophisticated algorithms,” especially as applied to market making, which is so simple an algorithm that human floor traders did it in their heads for generations. But I am willing to believe that replicating a secret portfolio based on periodic value snapshots is challenging and would require sophisticated algorithms.
10 I am perhaps simplifying and stylizing a bit for my argument. The SEC is also worried about redemption fees, and inequitable treatment among investors, and the risk that some investors will be redeemed at prices higher or lower than the actual net asset value.
11 Volcker said that in a different context, but plenty of people have said it about electronic market-making in equities.Felix Salmon summarizing Joseph Stiglitz is a good overview. E.g.:
But is faster price discovery better than slower price discovery? Let’s say good news comes out about a company, and its share price moves as a result — does it matter how fast it moves? Is any particular purpose served to seeing the price move within a fraction of a millisecond, rather than over the course of, say, half a minute? It’s hard to think of a societal benefit to faster price discovery which is remotely commensurate with the costs involved in delivering those faster price moves.