On May 6, 2010, the stock market crashed, but only for a few minutes, an event called the Flash Crash. Between 2:30 and 3:00, the market fell by 5% and recovered that loss. The results for some stocks were much worse. Accenture (symbol ACN) fell from about $40 to $.01, and recovered. Proctor & Gamble (symbol PG) fell from just under $55 to under $40, and recovered to $55, all within 2 minutes. The commodity markets also crashed and recovered. The Securities Exchange Commission and the Commodity Futures Trading Commission are conducting a joint investigation. Their preliminary report, dated May 18, 2010, is here (.pdf).
The SEC/CFTC report thinks the problem is “liquidity.” They say that the number of buyers and sellers dropped precipitately beginning at about 2:35, and point to other indications of loss of liquidity. For their purposes, liquidity means that there are enough buyers and sellers to make sure that individual trades don’t affect the price inordinately. As John Maynard Keynes tells us, there is no liquidity for the market as a whole; that is, if too many people want to sell, they won’t be able to find a buyer. And you have to ask what liquidity means if it can be withdrawn so rapidly.
It now appears that professional traders exited the market during the stormy period. That certainly would reduce liquidity. The result is that retail traders, you and me trying to manage our 401(k)s and IRAs, took the losses. Some of the trades were canceled, those in which the price was 60% above or below the price of the instrument at 2:40. That will help some of the people who traded ACN, but not those who traded PG and about 200 other stocks.
The chart above is taken from the SEC/CFTC report at page 51. The x-axis is the time, in increments of one second. The left y-axis is the bid and ask prices, shown on the chart by the red and green lines. The right y-axis is volume in thousands of shares. The blue lines at the bottom show the volume by second. The black lines show that trades are executing both at high and low prices. Just eyeballing the chart, it looks to me like trades in PG averaged about 6,000 shares per second during the decline. It looks like a fairly even drop of about $16 per share, so we can estimate that the average loss was $8 per share. That puts a total estimated loss of $2,880,000 on mostly retail customers, like you and me.
The market data firm Nanex says that the problem happened in part because the New York Stock Exchange computerized quote system was delayed, and High Frequency Trading programs took advantage of the situation. The Nanex report also discusses “quote-stuffing”, the practice of dumping seconds long bursts of thousands of quotes in specific patterns of bid and ask prices into a stock exchange. They say that this slows down the quote system and can lead to collapses like the Flash Crash. Others agree.
At least one member of the CFTC, Bart Chilton, is wondering whether Nanex is on to something.
We had our Technology Advisory Committee in this room last month and I asked the experts if this type of thing [High Frequency Trading programs taking advantage of price disparities due to delays in quote reporting] was possible or if it was just a conspiracy theory. Four of the panelists assured me that this could take place. In fact, they even acknowledged that some algorithmic programs were geared to not only take advantage of market circumstances, but could be used to instigate certain market conditions in order to then initiate their own program of buying or selling.
Therefore, I urge the staffs of the CFTC and the SEC to not leave this and any other stones unturned as they continue to investigate the flash crash. Did algo price pirates seek false profits on May 6th? Inquiring minds want to know.
I bet the people who sold PG in that rotten minute want to know if they got cheated by Wall Streeters.