Flash Crash: Attack of the Clones!
While many were relieved by the short duration of the flash crash on May 6 and the fact that it didn’t go nearly as far as the crashes of 1987 or 1929, in important respects, it was far worse than either of those. True, the Dow only dropped five and a half percent. But that drop took just five minutes, a speed of decline that exceeds anything in U.S. stock market history. Moreover, the decline in the averages sugarcoats the real carnage, which includes some stocks that went to zero for a few brief moments. That didn’t happen in ’87. And in ’87 there were no stocks whose values momentarily doubled while others went to zero. And while the ’87 crash turned into a good buying opportunity, the recovery took months, long enough for investors to participate if they wanted to. The recovery from the flash crash – both stocks that melted down and those that melted up – took only 90 seconds.
What went wrong? The National Market System, authorized by Congress in 1975 and built by the SEC, malfunctioned. NMS envisioned the use of modern telecommunications technology to tie together the regional stock exchanges and the NYSE into a unified national system. NMS had two primary goals: to bring down the NYSE’s monopoly and to route orders to the exchange with the best price. In 2007, Rule 611, the order protection rule known simply as “Reg. NMS,” accomplished both goals spectacularly. The Big Board’s percentage market share plunged from the eighties to the twenties. And so efficient is Reg. NMS at routing to the best price that it can instantaneously sort through all the visible venues and hit it, even if it is zero.
This flash crash result was unexpected, of course. Which brings up another dimension of the NMS failure: its vetting process. Every market structure-related rule since 1975 has been required to contain a mantra stating that it will “remove impediments to, and perfect the mechanism of, a national market system.” To avoid unintended consequences, the SEC offers and participates in extensive public comment periods, hearings, roundtables, operational oversight groups, industry testing groups and the like to make sure that nothing has been overlooked when a new rule rolls out. In the case of Reg. NMS, this process took many years, many thousands of pages of rule proposals and comments on them, and many thousands of man-hours of testing by the best stock market minds of Washington and Wall Street. How could they possibly have missed what would happen on May 6?
The few overlooked factors that surprised the market that day were well-known features of the landscape and, in hindsight, obvious menaces to safe operation of Rule 611. They are: stop loss orders, market orders and stub quotes. How did they get through Reg. NMS’s extensive vetting process? The simple answer is that they were old, presumably innocuous order types that were not considered at all. But in the new Reg. NMS environment, they were definitely not innocuous. While they had never been problematic when the NYSE was a manually operated monopoly, their inner demons were released when Rule 611 forced immediate execution at the best price on May 6.
Another piece of the Reg. NMS structure that played a role was a controversial remnant of the Big Board’s manual floor auction: its “liquidity replenishment points.” In volatile times, LRPs allow the exchange to momentarily disengage from the electronic markets while its floor auction restores balance. The disengagement occurs because New York’s prices are not immediately available during LRPs, which makes them ineligible for the trade-through protection that Rule 611 normally provides, thus forcing all orders in the National Market System to bypass New York and go to the electronic markets where execution is immediate. The electronic markets are not required to continue trading during LRPs, but generally choose to do so, partly to show that they can get along without New York’s liquidity and pricing help, and partly to take market share from New York while it is disengaged. New York’s traders don’t like the loss of market share, but no doubt take some comfort from watching their competitors flail without them.
May 6 was nothing more than such flailing writ large. The NYSE did fine while disengaged. The electronic markets flailed hopelessly and nearly died. Their high frequency market makers, sensing trouble, disappeared. With little else in their books, the market orders pushed prices to where the stub quotes were, producing ridiculous trade prices. With no floor governors or other manual processes to spot the difference between real trades and market structure failure, the electronic NMS printed them all.
With hindsight, it is easy to see how this happened and how to repair it. In fact, it may be repaired already. Everyone is talking now about the dangers of unlimited market orders and stop loss orders, so they are undoubtedly being used less now and may be on the way out altogether. And markets that allowed stub quotes are embarrassed and no doubt moving swiftly to clean up their rules and habits in this area. Even the electronic markets may temper their practice of unconstrained market share grabbing when the NYSE is in LRP mode, which caused their high frequency market makers to flee. In any case, they will be more alert to potential problems at such times, which could prevent them from happening in the first place. One of them is even planning to introduce its own version of LRPs. Others may follow.
Critics have long suspected that any benefits from NMS may have come with drawbacks that more than offset their value. Cheaper liquidity, for example, may have come at the cost of unstable price discovery and excessive volatility when high frequency market makers disappear, a fear realized in spades on May 6. But the reality is actually far worse. By eliminating human traders, NMS killed off the culture of honest service that underpinned capital formation, freeing the former investment banks to focus instead on speculation, transforming them from socially useful capital raisers into socially harmful, too-big-to-fail problems for the U.S. taxpayer.
Critics and admirers of Goldman Sachs alike were awed by the unending string of profitable days in its latest earnings report. Few noticed what this means, namely, that our biggest investment bank has become perhaps primarily a high frequency trader, since this is the only trading activity that almost never produces a down day. This shift was not Goldman Sach’s idea; it was the SEC’s. NMS destroyed the environment that once made capital raising an attractive and profitable activity. And NMS created the electronic trading environment and its high frequency trading opportunity. The SEC left the investment banks no choice but to leave the unprofitable activity and enter the profitable one.
Goldman was once a prominent member of the capital raising community built around the Nasdaq dealer market. That community was pressured out of existence by NMS reforms that began with the Order Handling rules in 1997, which had a similar effect on that market to that which Reg. NMS had on the Big Board in 2007. The net effect of both NMS reforms is that, where once there was a monopoly with a number of highly differentiated but coordinated functions, now we have a twitching mass of linked clones with no differentiation. All of the exchanges and ECNs today are built on the same business model, have the same structure and, with few exceptions, trade every stock. It is difficult to overstate how different this is from the way it was before 1997.
Once the NYSE was responsible for all of the trading in its listed companies, and a listing there was much desired, as the Big Board was where a seasoned and successful company would list its shares if it could. Nasdaq was responsible for all of the trading in its list of newer companies, which was where the IPOs came out. New York was a floor auction with narrower spreads. Nasdaq was a dealer market with the wider spreads that seemed appropriate for its less seasoned companies and in any case played a role in providing incentives for investment banks to underwrite new companies. The aggregate monopoly, where there was only one primary market for each stock and almost all the trading in that stock was done there, made the whole concept of best price routing moot – best price was what those markets did. We were the envy of the world because we had the most prestigious market for the biggest and best companies, The New York Stock Exchange, and because we had a phenomenally successful means of keeping the pipeline full by starting new companies on Nasdaq. Both of these advantages have been eliminated by the National Market System.
Granted, liquidity was more expensive then, but the pre-NMS market wasn’t prone to flash crashes. Apparently, coordinating the clones is a tougher task than was first thought. And the clones are still multiplying. Investors might be surprised to learn that the top four markets alone operate ten exchanges, each of which is separately licensed and labors under its own Rule 611 routing requirements, even if it is housed in the same building with one or more of the other clones. The lesser markets will no doubt catch on soon that they will get more total market share, too, if they do a little cloning themselves. So the number of clones will only increase, which can’t make coordinating them any easier. Most important, raising capital is not part of the clones’ business model. The old monopoly market could and did launch innovators like Microsoft and Intel, Amazon, Ebay and Starbucks. Those days appear to be over.
NMS’s philosophical foundation touts the electronic markets as a leveling force that will spread the profits and advantages of the exchanges and their members amongst average investors. Whether any net benefit has come of it is highly questionable. But it is certain that NMS has led to a dramatic increase in the SEC’s ranks and in its control of the market structure. And it is indisputable that those gains have come at the expense of the human traders and capital raisers that NMS’s machines have replaced.
According to a variety of anecdotal reports, investors are most troubled today by two things: the flash crash and the fact that, one month later, we still don’t know what caused it. Both of these problems could be cleared up by removing the SEC from its role as the chief investigator of such problems. The Commission simply has no incentive to uncover its own errors.
Although the party line is that we still don’t know what caused the crash, the solution has been decided upon: coordinated single stock circuit breakers. We should be careful what we wish for. The proposed circuit breakers, which are being rushed for a June 14 rollout, will stop trading for five or ten minutes, depending on a variety of circumstances, if a stock moves ten percent in a rolling five-minute period. These will be in addition to another set of circuit breakers coming November 10 in the form of a new and operationally complex short sale restriction that kicks in when stocks have declined ten percent from the opening price. Such remedies will be difficult to understand and implement and the conglomeration will certainly be confusing, especially if added to multiple versions of LRPs at the various exchanges. How they will coordinate with each other, too, is a potential problem. We should remember that NMS is itself primarily a top-down order flow coordination scheme and it failed spectacularly on May 6. Circuit breakers are the nuclear option among order flow coordination schemes. May 6 proved how difficult it is to see around corners in the vetting process. But even that event could have been so much worse, if the almost forgotten market wide circuit breakers put in place after the 1987 crash had kicked in, as they very nearly did.
The very rapid moves and their diverse directions prove beyond a doubt that what we were witnessing on May 6 was a market structure failure, not some mood swing of investors or efficient pricing of stocks based on new information. But it could have been much worse. What if prices had been frozen at the peak of the failure? The theory on circuit breakers is that investors will get more rational – read courageous – if markets take a time out so that they can receive and evaluate new information. But the theory rests on the assumption that the market is falling because of panicking investors. On May 6 there was no sudden pessimism or panic in the falling stocks, much less sudden optimism in the rising ones. Nothing, in short, that could have possibly caused such extreme moves so quickly. While a time out to soothe frayed nerves in a panic may have some value if the market is functioning properly, a time out that merely gives a clearer picture of how dysfunctional the market structure is will actually cause investors to panic. Most investors were unaware of the flash crash until it was over. That would not have been the case if the market wide circuit breakers had kicked in. They would have given investors in this country and around the world plenty to panic about and plenty of time to do it.
About Steve Wunsch
Steve Wunsch is an inventor of stock exchanges (Arizona Stock Exchange and International Securities Exchange - ISE) and market structure consultant living in New York City.