High-Frequency Trading. A Workers Capital Briefing

Transcription

1 High-Frequency Trading A Workers Capital Briefing September 2012

2 About the ACTU The Australian Council of Trade Unions (ACTU) is the nation s peak body for organised labour, representing Australian workers and their families. Nearly two million workers are members of the 46 unions affiliated to the ACTU. The Working Australia Papers The Working Australia Papers are an initiative of the ACTU to give working people a stronger voice in the development of social and economic policy. Previous Working Australia Papers have addressed a broad range of issues, from taxation policy to productivity. Working Australia Paper 6/2012 D No. 6/2012 Australian Council of Trade Unions Level 6, 365 Queen Street Melbourne VIC [2]

3 Introduction In August 2012 The Australian Securities & Investments Commission (ASIC) issued a consultation paper seeking views on its draft market integrity rules and guidance on automated trading 1. A high profile and increasingly important form of automated trading is high-frequency trading (HFT): a set of practices that utilises speed to generate additional returns to HFT firms and their clients. In recent years there has been growing concern that HFT is serving to exacerbate the speculative, short-term and volatile nature of financial markets. ASIC s consultation is, in part, a response to such concerns. Australian industry and not-for-profit funds are major participants in global capital markets. Via the world s major trading exchanges they invest billions in equities, bonds, derivatives and foreign exchange. But unlike some other participants, super funds have little or no choice but to remain in these markets for the long-term. Our super funds therefore have a strong interest in financial markets that are regulated to be transparent, secure and fair. Such markets are more likely to deliver stable and reliable returns over many years rather than short-term speculative gains in the space of a few days, hours or seconds. The purpose of this ACTU paper is to highlight some of the risks that HFT may present to super funds as longterm investors in global financial markets. Those who specialise in HFT earn billions in profit every year, not from long-term productive investment in jobs and communities, but from being able to trade faster than their competitors. For this skill they are paid large fees and commissions. In the context of the global pensions industry, payments for such short-term unproductive trading represent a growing leakage from the investment chain that connects workers contributions to the sources of long-term return that will ultimately help to fund their retirement. The issues raised by HFT therefore deserve close attention by unions and super funds. This paper is intended to serve as a brief introduction to HFT and why regulatory reform is necessary. After a brief explanation of what HFT is and how it can generate profits, the paper critically discusses the arguments commonly made in support of such trading. These arguments are found to be flawed. The paper therefore ends by outlining some reforms that would help to counter the risks HFT now presents to financial markets and those long-term investors who have come to rely on them, and proposes some immediate steps that super funds in Australia should take. TIM LYONS ACTU Assistant Secretary 1 ASIC (2012) Australian market structure: draft market integrity rules and guidance on automated trading, Consultation Paper 184 [3]

4 What is High-Frequency Trading? Within the next year a new fibre-optic cable, costing $300 million, will be laid across the floor of the Atlantic Ocean between New York and London. The purpose of the new cable is not to facilitate increased telecoms or internet traffic. It will not be available for use by the public or by any government. It is a private venture with one aim: to reduce the amount of time it takes trading firms to buy and sell financial instruments. At present the fastest cable between Europe and North America allows trades to be communicated in 64.8 milliseconds. The new cable, when it begins to operate in 2013, will allow those few who can afford to purchase access to communicate a buy or sell order in 59.6 milliseconds 2. Since 2005 over twenty new data centres have been built in the US state of New Jersey, close to the computer trading engines that power the New York Stock Exchange 3. These centres have been built to house servers owned by financial institutions across the world. By being physically closer to the computers that process NYSE trades these institutions can reduce the amount of time it takes to buy and sell by milliseconds. But for some even the latest fibre-optic technology is too slow. Cables often have to be laid around buildings and other structures increasing their length and therefore slowing the flow of data. Therefore some trading firms have begun to invest in building chains of microwave dishes that can fire data-signals to each other across distances in straight lines. In the context of communicating trades between Chicago and New York fibreoptic cables can only manage a speed of 6.55 milliseconds. Microwave dishes can cover the same distance in 4.25 milliseconds. It is estimated that this difference could be worth $1,350 a day to a trader wishing to profit from price differences for the same securities on the two exchanges 4. What is occurring is a technological arms-race. In the world of HFT saving milliseconds can mean millions in additional profit. The race is on to find newer, faster ways of trading across time and space. In the words of Andrew Bach, head of network services at NYSE Euronext: The speed-of-light limitation is getting annoying 5 2 Stock trading is about to get 5.2 milliseconds faster, Businessweek, Server farms hurt by glut, Wall Street Journal, Networks built on milliseconds, Wall Street Journal, Light is not fast enough for high speed stock trading, New Scientist, [4]

5 Source: Wall Street Journal, [5]

6 HFT has its origins in the introduction of real-time quotation systems to financial markets during the 1980s and the use of algorithms to automate trading within pre-determined price, volume and time parameters. Over the past thirty years trading in highly liquid instruments such as derivatives, equities, foreign exchange and bonds has become increasingly driven by computer programmes designed by mathematicians and physicists, requiring little or no human intervention. HFT utilises the latest developments in computing, telecommunications and quantitative modelling to reduce trading latency (the time needed to execute an order), generating profit by exploiting the comparative slowness of other market participants. The graphic below illustrates how this can work. Source: The New York Times, This example illustrates one common form of HFT. An investor submits an order to a trading venue for shares in company XYZ. HFT traders are able to detect the order milliseconds before others. Sometimes, as in the above graphic, certain traders are given advance notice of an incoming order by around 0.03 seconds (these are called flash orders ). Knowing that an order is coming they buy all available shares in XYZ. The order then hits the market. The HFT traders can then sell the shares they bought milliseconds before for a higher price. Within a fraction of a second HFT traders have bought-low and sold-high without any human intervention and with negligible risk. Repeated many times a day, sometimes in connection with large volume transactions, this can generate significant profit. It has been estimated that HFT firms made profits of nearly $13 billion in 2009 and High Frequency Trading, The New York Times, [6]

7 HFT has become an increasingly significant feature of global financial trading. Approximately 60 per cent of equity trading in the US, and 40 per cent of equity trading in Europe, is the result of high-frequency trades. In the UK the figure may be as high as 77 per cent 7. In Australia around only 10 per cent of ASX trading is attributed to HFT activity. This is expected to grow 8. The graphs below illustrate the increase in algorithmic trading, much of it driven by HFT techniques, since Source: The Economist, In addition to developments in computing and telecommunications HFT has been encouraged by a number of recent changes in financial regulation. In particular, as part of its broader drive to deregulate markets, the European Union adopted the Markets in Financial Instruments Directive in Despite strong opposition from some member states the Directive abolished the concentration rule. This rule had allowed member states to require that trading in financial instruments be conducted only via a regulated exchange. The rule served to limit the extent to which such transactions could be channelled through non-regulated venues or taken completely off-market. The EU concluded that this rule limited competition resulting in high transaction costs. These costs inhibited trading and so compromised liquidity, particularly in secondary securities markets. This placed European venues at a competitive disadvantage to other trading centres. However, one consequence of rule abolition has been an increase in opportunities for inter-exchange pricing arbitrage. Such a trading landscape lends itself to HFT. This regulatory change helps to explain why the majority of specialist HFT firms have been established within the last ten years. 7 tabbgroup.com, High frequency trading is cuckoo, Business Spectator, [7]

8 The 2010 flash crash HFT became an issue for regulators in May 2010 when the Dow Jones Industrial Average suddenly fell by nearly 9 per cent in the space of a few minutes the largest intraday fall in the history of the index. Most the fall was reversed within 20 minutes. The fall had been triggered by a large mutual fund entering the market to sell an unusually large number of futures contracts. Buyers included HFT firms. However, because HFT algorithms are typically designed to avoid holding positions for long the HFT traders began selling the contracts again within minutes. The combined impact of the original sale and the high-frequency reselling drove futures prices down leading to equivalent sales in equities markets to offset the falling value of the derivatives. The spiral of selling continued for several minutes until the sharp plunge in prices triggered an automatic 5-second pause in futures trading after which demand-side interest recovered and prices began to stabilise and rise. An investigation into the May 2010 flash crash by the US Securities and Exchange Commission and the Commodity Futures Trading Commission concluded that while HFT did not initially trigger the crash it nevertheless exacerbated the massive swings in trading activity and prices. It took intervention by non-hft agents to prevent markets falling further than they did. Problems of High-Frequency Trading Since the flash crash in 2010 regulators and parts of the business media have taken an increasing interest in HFT and the risks it may pose to the stability and integrity of financial markets. The public debate has become increasingly polarised between those who view HFT as making a valuable contribution to increasing the cost and price efficiency of global finance, and those who see it as increasing systemic risk while making little or no contribution to directing financial capital into productive long-term investments. Advocates commonly make the following arguments in support of HFT and why further regulation is unnecessary: a) HFT helps to generate liquidity. HFT agents compete to buy incoming offers to trading venues and so provide an important source of demand and market-depth. They therefore play a role akin to designated market makers : registered market participants whose role is to offer firm buy/sell quotes on a range of securities in order to maintain broader market liquidity. [8]

9 b) HFT reduces the cost of trading. By injecting demand into markets at low-risk HFT agents contribute to tightening spreads. c) HFT makes markets more efficient. Aggressive trading across venues helps to minimise or eliminate price discrepancies between related financial instruments. The notion that HFT generates liquidity, lower costs and market efficiencies has become the shared common sense of the HFT industry, many regulators and exponents of efficient markets theory. However, there are important weaknesses in the pro-hft case including in relation to market integrity and fairness. Firstly, as discussed above, HFT agents are able to create and cancel orders in milliseconds. The average HFT latency is 3 milliseconds. The liquidity they provide therefore tends to be a thin and volatile form of demand that involves a highly selective buying of securities that are then re-sold within fractions of a second. Liquiditymaking gives way to liquidity-taking before most other market participants are able to buy/sell at prices that resemble real levels of supply/demand and real long-term investment values. Instead, and in contrast to designated market makers, HFT agents utilise speed to generate information asymmetries that result in many institutional investors, such as pension funds, paying more for securities simply because they are too slow to pay less. In the view of a European asset manager who gave evidence to the UK government s inquiry into computerised trading: Although HFT may contribute to tighter spreads in the lit markets, and potentially higher volumes, most of the added liquidity is artificial, in that large institutional orders cannot interact with it to any great benefit. High trading volume does not necessarily mean greater market liquidity for institutions. 9 Secondly, the tight spreads generated by HFT tend to be of limited value to the broader market. They are often of limited depth, meaning that they relate only to small trading volumes which, once exhausted, are rapidly replaced by wider-spreads as a result of automatic algorithmic re-pricing. Further, HFT spreads are usually on offer for millisecond durations which means they are only of use to other HFT agents. Finally, tight spreads on particular trades do not necessarily mean lower aggregate trading costs. Evidence from asset managers to the UK government inquiry illustrates this point: Lately spreads, correlation and volatility have remained high whereas liquidity is limited (we would argue that much of the liquidity supposedly added by HFTs is not of any use to institutional investors 9 Government Office for Science (2011) End-user Perspectives on Computerised Trading, p. 41 [9]

10 and therefore is not really liquidity at all). This combination leads to an increase in overall trading costs. Overall there has been a marginal increase in the cost of trading Some direct costs, such as commission rates and head count, have decreased as a result of alternative trading venues being used, and greater efficiency in the methods of trading. Other direct costs, such as technology spend have increased as the market has become more fragmented. Indirect costs, such as market impact and opportunity costs, have risen largely because of the increased volatility in markets. Increased IT costs and the IT team have offset the savings on commissions and spreads. 10 Thirdly, HFT may contribute to levelling prices across venues. However, this is a solution to a problem that regulators, via the deregulation of trading exchanges, have contributed to creating. It is therefore a problem that could be alleviated by means other than HFT. Nor is it clear how levelling prices in 3 milliseconds rather than 1 second benefits long-term institutional investors and other non-hft market participants. In addition to questioning the arguments made for HFT some argue it can be used deliberately to manipulate markets and price formation to the advantage of those with the fastest technology and most advanced algorithms. This helps to explain the massive investments made by many HFT firms in recent years in colocating to major exchange servers, laying new fibre-optic cables and hiring the most able mathematicians and physicists. Examples of manipulative strategies that some HFT firms are suspected of using include: a) Quote stuffing: the practice of placing a large number of quotes in the market for the purpose of slowing down the processing of orders by an exchange or the activity of other traders. In both cases the causing of delays will benefit the fastest HFT agents who will able to buy/sell at prices that reflect a lag in priceformation. b) Momentum ignition: the practice of buying/selling in sufficient volume to ignite a generalised rise/fall in prices for a security that the HFT agent can then exploit. c) Spoofing: the practice of issuing waves of false orders at different prices for a security already held by an HFT agent to give the misleading impression to the wider market that demand is greater than it really is. Once orders for the security start to flow-in from other agents the lower prices initially fed into the market 10 ibid, p. 45 [10]

11 are withdrawn so that the security is eventually sold for a price higher than would have been the case before the spoofing was initiated. However, because of the speed and complexity of HFT market abuse it is often difficult to prove. Nevertheless, many of those who gave evidence to the UK government inquiry believe that market manipulation often takes place: We try to avoid venues that allow these types of strategies but it is very difficult to find evidence that market abuse has occurred. These strategies should be banned. According to the FSA quote stuffing does not take place in Europe but if 95 per cent of orders are cancelled before being executed then you would think an element of that represented orders which were never meant to be executed. Quote stuffing happened on the day of the Flash Crash. There was a single exchange that was the source of these quote stuffing events and there was possibly only a single participant. What happened was that a single stock would get 5000 quotes in a second, and then they were all cancelled instantly. We definitely see this as market abuse and it could get worse. 11 More important than particular instances of market abuse is the systemic risk and instability that HFT contributes to generating and the inability of regulators to keep pace with new developments and the new abusive strategies they make possible: The biggest risks are to financial stability, especially when computers are entrusted to perform tasks quicker than human common sense can dictate. Unless distinction is made between certain types of computer-programmed trading, and curbs are placed on the more opportunistic types, the next ten years will see an increase in instability of markets, further loss of confidence and an erosion of market integrity. There is no way that the current regulatory authorities have the expertise or capabilities to detect or enforce market abuse ibid, pp ibid, pp [11]

12 These risks and instabilities are exacerbated by Direct Access Trading, where non-members of a trading venue can participate directly in the market using the electronic systems and algorithms of a member institution. Such participants are less likely to understand and comply with market rules, and may not fully comprehend the nature and implications of the systems and algorithms they are given access to. This is likely to disrupt markets and lower confidence in their quality and integrity. A further consequence of HFT has been an increase in dark trading. In contrast to lit trading where buying/selling takes place on exchanges that impose strict obligations in relation to pre and post-trade transparency, dark trading in so-called dark pools involves conducting transactions in private venues where prices and volumes need not be publically disclosed. This form of trading has become increasingly popular in recent years partly because many market agents have lost confidence in the capacity of public exchanges where HFT is common to offer a fair and stable trading environment. However, while conducting certain transactions away from lit exchanges may make sense for particular buyers, sellers and brokers, it comprises broader public transparency and price formation. It therefore risks making financial markets and capital allocation less rather than more effective. HFT or Socially-Useful Financial Markets? The primary role of financial markets should be to generate and allocate capital for the purpose of facilitating long-term investment in productive and sustainable economic activities that help meet the needs of workers, citizens, consumers and communities. This requires a financial system whose regulatory structure and system of incentives operates to discourage short-term, unproductive speculation and related abuses. The Global Financial Crisis demonstrated that too many of the world s financial institutions had become pre-occupied with securing short-term speculative gains, generating massive systemic risks and instabilities that have since cost millions of jobs and the destruction of public services in many countries. Unfortunately, despite the events of recent years it is far from clear that the lessons of this experience are being learnt. A financial system that helps to deliver good jobs, steady growth and a sustainable environment is one that is, at the very least, transparent, accessible, stable and free from abuse. All participants must be able to enter and participate in markets on a fair and equal basis. In the absence of such conditions prices are more likely to deviate sharply from fundamental asset values, leading to speculation, volatility and market manipulation. There is strong reason to believe that many contemporary forms of HFT act to encourage speculative behaviour across the financial system. [12]

13 Recommendations for Reform There are a number of reforms that regulators and policymakers must consider implementing. These include the following: 1. Impose a moratorium on HFT until regulators have completed a detailed assessment of the role it plays our financial system and the costs/benefits it generates for system-users. 2. Impose a minimum resting time of 1 second for trades. This would help to promote genuine liquidity and radically reduce opportunities for abuses such as spoofing. The onus should be placed on the HFT industry to make a social utility case for being allowed to trade in milliseconds. 3. Outlaw flash orders. Trading venues should not be allowed to offer advance notice of incoming orders to any market participant, including HFT firms. The practice encourages a two-tier market that disadvantages the large majority of participants who cannot co-locate or buy access to the fastest fibreoptic cables. 4. Mandate fees on those who cancel orders above a certain order/trade ratio (such as 3:1). Regulation must ensure that such fees fall on those responsible for implementing the cancellations and who seek to benefit from them, rather than being passed-on to other parties. 5. Mandate a clear and consistent system of circuit-breakers within and between all trading venues. By suspending trading when market indicators breach certain pre-set limits circuit-breakers would help to minimise contagion-risk and stem volatility. 6. Regulators should aim to bring all trading venues under a uniform set of regulations that promote consistent pre and post-trade transparency and equality of market access. Encouraging more competition between greater numbers of exchanges is a recipe for instability, market abuse and the weak policing of behaviour. 7. Outlaw Direct Access Trading. Only those who fully understand market rules and the technologies at their disposal should be allowed to participate directly in financial markets. 8. Impose an appropriate set of charges on HFT traders to pay for the additional costs incurred by government and regulators resulting from the increased regulation and policing of HFT behaviour. [13]

14 Next Steps for Australian Super Funds There is reason to believe that HFT represents a risk to long-term capital market participants such as Australian industry and not-for-profit super funds. But the extent of that risk, and the associated costs to members, is presently unknown. It is therefore important that funds, individually and collectively, take the following steps: a) Each fund board should recognise HFT as an investment risk. The fund should commission a report from its investment team or consultants that details how and to what extent HFT is built into their current portfolio, and what costs/benefits HFT delivers to members. This cost/benefit analysis should include an assessment of systemic risks in addition to fund-level risk. In light of this analysis the fund should then consider the extent to which portfolio reliance on HFT should be reduced. b) Via their collective bodies and associations funds should call for and support further detailed analysis and policy development in relation to measuring collective HFT risk and what steps Australian regulators should take to mitigate it. [14]