- Managing Me
- Big Ideas
- Managing People
The financial markets "flash crash" on May 6, 2010 – when the Dow Jones Industrial Average index plunged almost 1000 points before bouncing back – was one of the most terrifying events in recent Wall Street history. It might have only lasted 20 minutes, but it wiped US$1 trillion from shareholder equity. Although share prices recovered, the incident left investors and regulators reeling. How robust and sound could the financial system be when the stock market can fall 25% in a few minutes during the day?
"The decline and rebound of prices in major market indexes and individual securities on May 6 was unprecedented in its speed and scope," said the US Securities and Exchange Commission (SEC) in a report. But almost two years on, it still does not know what caused this breathtaking intraday panic.
Months of investigations yielded only one clear fact: the US securities regulators have no clue how to stop another crash. The SEC was forced to acknowledge that it did not have the information to explain what caused the May mess and announced it would build a $US4 billion centralised data system to track all orders, messages and trades in real time. Yet not knowing what caused the flash crash did not stop the SEC from taking swift and dramatic action.
Turbulent times lie ahead as exchanges around the world, including in Australia, are following the SEC's lead as it instinctively brings out new regulations aimed at preventing another acute crisis.
There has been no shortage of suggestions as to the cause of the flash crash and plenty of people have been accused. High-speed algorithmic trading has taken the brunt of the blame, as have hedge funds, derivatives traders and proprietary-trading (when a firm trades for direct gain instead of commission dollars).
There was a number of contributing factors to the flash crash but, according to Michael Aitken, chief scientist of the Sydney-based Capital Markets Cooperative Research Centre and chair of Capital Market Technologies at the University of New South Wales, the New York Stock Exchange (NYSE) did not help by closing down for 30 seconds during the crisis. "In doing so they sucked 25% of the liquidity out of the market," he says. "Little wonder that trading algorithms, which can trade at the speed of light, were sent into a frenzy."
Trades on the NYSE and NASDAQ were shut down through what is known as a "liquidity replenishment break". This curb on trades kicks in when stocks are in freefall. But on May 6 when these exchanges halted trading, the market went haywire and sell orders flowed to smaller exchanges which remained open and quickly became swamped by the panicked selling. Prices plummeted. In short, a system geared to stop the market going into freefall added to the damage.
The SEC, under enormous political and public pressure to act, proposed brand new circuit breakers. Under the new rules, stock trades are now halted if they swing more than 10% in a span of five minutes. The SEC admitted that part of the problem was that some markets stayed open during the crisis and the securities' regulator appears to believe the only way to avoid crashes in the future is to force all markets to close.
Heading the wrong way?
Aitken believes something is seriously amiss with this conclusion. He insists there's a fundamental problem with making decisions about market design changes based on authority rather than evidence. The SEC had no evidence that closing the market could be good, Aitken points out. Available evidence suggests that markets that stay open ultimately are less volatile than those that are forced to close. Empirical evidence also suggested the decision to ban short sales in 2008 – when the global financial crisis hit – was not a smart idea.
Now, Australia looks set to introduce a new set of market design changes, including circuit breakers to limit big price swings, in part quoting the authority of the SEC. However, there's plenty of evidence that circuit breakers or kill switches don't work – rather they postpone the inevitable. "I feel regulators haven't grasped the implications of halting trading if there is a crash, and have opted for a kneejerk reaction that achieves little," says Aitken. He argues that Australian securities regulators need to break out of the cycle of authority-based decision-making and should instead follow the lead of Canada's Ontario Securities Commission and prioritise evidence-based decision-making.
Market structure decisions are often made by following overseas trends or through a consensus view after calling on market participants. "Few [regulators] provide evidence in support of their proposition," says Aitken, who points to the 20 submissions to CP 145, the Australian Securities and Investments Commission's document on market structure change, released in 2010.
Maureen Jensen, executive director and chief administrative officer of the Ontario Securities Commission (OSC), supports this view. Regulators continually benchmark against what other regulators are doing, which is not the way to make effective policy, she notes. Typically, they call on all market participants for their opinions and reach a consensus view. "That is not research," says Jensen.
"The right answers require a significant amount of data and time for analysis. Even then, you're not always going to get clear answers for every problem, but you would certainly have better answers than just selecting from the choices from vested interests. We all need to look at what works best for our market – we can't make policy based on what everyone else is doing. We saw this in response to the flash crash when everyone around the world was implementing short-selling bans but not really understanding what prompted the market fall."
Jensen's Canadian colleagues looked at the impact of short sale bans after the 2008 market crash and found that the bans did not help to reduce market volatility, instead they stalled legitimate trading decisions. When the bans were removed, the legitimate sells came pouring into the market. "Short sales are a legitimate third of our market. They are not predatory, but are hedges to risk," Jensen says.
Although Duncan Fairweather, executive director of the Australian Financial Markets Association (AFMA), firmly supports evidenced-based policy, he also does not discount the role of judgement or consensus, particularly when quick decisions need to be made as they did in 2008 when short sales were banned. "If regulators had a week or so to mull it over, they might have made a different decision, but there was an emergency," Fairweather says. "There was a heightened level of uncertainty in the global markets and the Australian government and the securities market regulator were concerned that offshore sellers, who had been stopped from shorting stocks in their own market, would turn to ours. It required an instant response."
Fairweather notes that controls can always be unwound – as short sale controls were – but believes it was a sensible decision in relation to financial stocks at the time. However, he objects to controls being introduced because of an unrealised fear that things are getting out of control or because implications of events may be unknown. "They're not good reasons for bans or blockages," Fairweather says. Australia's proposed regulation of dark pools is a good example of this, he suggests, noting "a proposed restriction on the amount traded could stifle that market".
Why make assumptions?
Investor confidence is affected by financial markets becoming ever more complex. Investors want to know whether dark pools are good or bad for a marketplace. Does high-frequency trading increase the quality of markets or not? Are rules needed for monitoring algorithms? Investors want regulators to adopt robust approaches to formulating new rules.
Canada has built a surveillance system that looks across multiple markets, collecting data and analysing the information to detect market issues in real time. "The collection of this data across markets is necessary to understand market trends and developing issues," says Jensen. Having this level of knowledge gives regulators enough tools to stop any market manipulation of specific stocks without putting breaks on the whole market.
Interestingly, Canada's system, completed on May 5th 2010 – one day before the flash crash – has enabled analysis of why the market plummeted so dramatically. The Canadian research shows the drama was caused by heavy automated selling – long-selling, not short-selling – driven by market sentiment. Also, market orders without limits caused a series of automated stop-loss orders to depress certain stocks to extremely low prices.
"OSC has invested in the market research framework, developed by [the Australian School of Business's Michael] Aitken, as part of our move to improve our understanding of the changes to market integrity and efficiency caused by changing market structures and trading patterns," reports Jensen.
Aitken believes Australian regulators should follow the Canadian model and resist the pressure to fall in line with the US. Regulators need a framework of evaluation and the data to address the basic regulatory mandate to "operationalise" fairness and efficiency, Aitken says. "You need this type of infrastructure in place to have any idea of what has caused an unexpected outcome. In this sense, the SEC is a striking example of what not to do. In Australia, ASIC has already invested in appropriate technology for the purpose. It might not be able to predict the next crash but we can use it to learn quickly from it." For example, how can problems such as frontrunning be measured when brokers are not required to tell the regulator whether they trade as a principal or as an agent?
Gathering evidence is expensive and time consuming. The Capital Markets Cooperative Research Centre recently estimated the cost of setting up infrastructure to facilitate evidence-based policymaking is more than A$40 million. Aside from the data collection and hardware, a framework is required for processing the data and to properly understand the regulator's mandate to ensure that market changes are both efficient and fair.
For this, measurement proxies for fairness and efficiency are needed and they should be gauged before and after market design changes. Such changes should only be accepted by the Australian Securities and Investments Commission if they enhance efficiency without detriment to integrity or vice versa. Preferably they should enhance both, asserts Aitken. "For example, an efficient market is defined as one where you can trade cheaply and be sure that traded prices reflect all available information. This means measuring efficiency before and after market design changes with transaction costs and price discovery as the variables of interest," says Aitken. "Then changes in the volumes transacted in dark pools are associated with changes in transaction costs and price discovery."
"Similarly, if fairness is defined by the extent to which market participants engage in prohibited trading behaviours, then we can estimate the impact of the growth of dark pools on fairness by measuring the relationship between their growth and proxies for insider trading, market manipulation and broker-client conflict."
It's not easy or straightforward, Aitken warns. "But unless we move in this direction we are destined for a lot more surprises and less well-functioning markets to the extent that investors lose confidence."
Fairweather highlights that "evidence-based research is fine if you can get perfect knowledge". "But, for us, perfect knowledge is an elusive goal sometimes," he says. The problem of competing models is an issue for the Australian Financial Markets Association. "The regulator might come up with a research model but our members have significant quantitative resources as well, so we would be cautious about automatically elevating one model over another," Fairweather notes, adding that markets deal with unknown events as they happen. When Australia first switched to an electronic trading system, he recalls fears that "fat fingers" would lead to market distortions, but the market coped with that.
An outstanding question remains: if market behaviour gets messy, but not catastrophic, does that warrant introducing new rules to fix the problem without proof that they will have the desired effect?