Mark Stys had warned his client about the biomedical/pharmaceutical ETF called PJP—not because the stocks in it were a bad bet, but because the ETF was small and some of the stocks were thinly traded.
“Our warning was: ‘There’s not enough liquidity in the ETF,’” Stys, a financial advisor in Northern Virginia who manages about $130 million in assets for clients, remembers saying.
The client did some trading on his own, and bought it anyway. PJP, run by Invesco, was based on an index holding 25 stocks, including some big companies (Eli Lilly and Amgen, for instance) and some smaller ones.
Stys was worried that if the market fell and the client wanted to sell, there wouldn’t be buyers; he was also worried that the lack of trading in some of the underlying stocks meant there wouldn’t be a clear price.
On Aug. 24, 2015, he turned out to be right, and agreed to share his records of the day to help illustrate what can happen to individual investors during what are colloquially known as “flash crashes.” One of the big questions about ETFs is the extent to which their rapid growth has led to an increasing number of flash crashes and a greater risk for the mother of all flash crashes.
Here’s the background: ETFs have a unique structure, in which investors, often powered by algorithms, trade groups of stocks instead of individual shares. Professional and institutional investors, like hedge funds and brokers, can buy baskets of securities at any time of the day without buying and selling the underlying individual securities (eventually the trades are settled). Individual investors use ETFs to do things like save for retirement (or to gamble a bit, as the investor in this case was doing).
Institutional investors use them for many sophisticated market moves like futures, hedging risk, or making money on trillions of small prices changes every day. The number of transactions undertaken by professional investors dwarfs the transactions by and for individual investors, likely by a ratio of something like 9 to 1, according to an analysis by Vanguard.
But sometimes the small investors get burned by the bigger market.
One of the reasons Stys was worried about PJP is that he knew what had happened five years earlier, during the flash crash of 2010. During a horrifying 36 minutes in the market on May 6, “over 20,000 trades across more than 300 securities were executed at prices more than 60% away from their values just moments before,” according to a report by regulators issued a few months later.
The report found the flash crash was triggered when a large mutual fund company tried to sell $4.1 billion worth of E-mini contracts. (E-minis are S&P 500 futures contracts.) The algorithm the trader opted to use was set to sell based only on volume—there was no flexibility for time or speed. High-frequency traders initially bought some of the orders but ran into pre-set caps for purchases. Without enough buyers for E-minis, the price started dropping. The price declines in the market of professional investors confused the automated trading systems used by brokers, including those acting on behalf of individual or “fundamental” investors. Those trading systems briefly shut down, causing what regulators called a “second liquidity crisis” in the market for the individual stocks in the ETFs. Without enough buyers for the sell orders in the market, the prices for the stocks started falling, too.
After the market closed, the exchanges and FINRA met and jointly agreed to cancel (or break) all such trades under their respective “clearly erroneous” trade rules.
There have been many small flash crashes, or liquidity events since 2010. Most ETFs, especially small ones and especially a new kind of ETF called synthetic ETFs, are at the mercy of algorithms because their pricing is derived from the underlying securities they represent—placing them one step away from actual buyers and sellers talking to each other. Prices are supposed to be close to the underlying value of the stocks contained within an ETF basket. But if there’s a glitch—not enough buyers and not enough sellers in a given window of time—the theory is that algorithms don’t have the information they need to set prices well.
If a truly big market crash happens—perhaps triggered by a macroeconomic event—no one knows exactly where or how much liquidity would dry up or what the effect on prices could be.
So, back to Stys’s client. The stage was set for the small flash crash in Aug 2015 the night before, as CNBC reported: “On Sunday night, Aug. 23, a large drop in equities in Asia triggered a drop in index futures in Europe and the U.S. U.S. stock futures went down (7%) prior to the U.S. open.”
Stys’s client was watching. He asked Stys to put a sell-at-the-open order on the small ETF.
Stys put the order in with his broker, one of the authorized participants or market makers that are supposed to keep ETFs trading near the value of the securities they hold. But three of the underlying stocks in the ETF hadn’t opened with a market price; the algorithm priced the ETF at 40% lower than it had been the day before, and sold it. It’s still not clear to Stys how the broker’s algorithm arrived at the 40% down price. A few hours later, the market had recovered, and the ETF was selling at down 5%.
Stys sent a note to the broker. “I want a response, both the client and me are livid. I have a lot of problems with an execution down 40% from the open and then ten minutes later, down about 5% the rest of the day. Also, the reporting on the executions took forever. I know Monday morning was a shit show, but I mean that trade adjusted VWAP would do fine.”
VWAP stands for volume weighted average price. The broker responded by noting that, “In the overall case of the ETF’s pricing from Monday, FINRA made no ruling, and trades were not declared erroneous.” (In the US, FINRA regulates brokers.)
The client lost about $10,000, and gave up trading on his own. Stys remains reluctant to take on clients that want to do their own trading, and also investigates ETFs that he uses closely for liquidity. Invesco has since reconfigured PJP, so that it’s now based on an index of 30 stocks.