What is the economic impact of the MiFID rules aimed at regulating high-frequency trading? In association with Oxera An economic impact assessment This paper has been commissioned as part of the UK Government’s  Foresight Project on The Future of Computer Trading in Financial Markets.  The views expressed are not those of the UK Government and do not represent its policies. The economic impact of MiFID rules 2 Contents 1  Introduction and executive summary.......................................................................................5  1.1  Objectives and remit.....................................................................................................................5  1.2  Approach and sources of information...........................................................................................6  1.3  Main conclusions..........................................................................................................................8  1.4  Structure of the report.................................................................................................................14  2  Approach to assessing the impact of the proposed rules....................................................14  2.1  Methodology...............................................................................................................................14  2.1.1  Practical implications for Oxera’s assessment................................................................15  2.2  What is high-frequency trading?.................................................................................................16  2.2.1  What drives high-frequency trading?...............................................................................17  2.2.2  High-frequency trading and intermediation, or how high-frequency trading can be profitable .........................................................................................................................................19  2.2.3  The role of hedging and risk transfer within (high-frequency) intermediation..................22  3  Generic trading sequences used by high-frequency traders...............................................25  3.1  Introduction.................................................................................................................................25  3.2  Pure arbitrage—same security traded on different venues.........................................................27  3.3  Linked arbitrage between related securities...............................................................................30  3.4  Pure price movement prediction (directional strategy), on the same venue...............................31  3.4.1  Using aggressive orders only (for example, with a predicted price rise).........................31  3.4.2  Using a mixture of aggressive and passive orders..........................................................32  3.4.3  Using only resting orders.................................................................................................33  3.5  Statistical arbitrage.....................................................................................................................35  3.6  Market making in a single security (using HFT techniques).......................................................35  3.7  Combinations of trading strategies.............................................................................................36 The economic impact of MiFID rules 3 3.8  Information discovery (pinging)...................................................................................................38  4  Market-making requirement.....................................................................................................39  4.1  The proposed rule.......................................................................................................................39  4.2  The Commission’s impact assessment.......................................................................................40  4.3  How would the rule work in practice?.........................................................................................40  4.4  Potential general impact of the proposed rule............................................................................42  4.5  Impact on trading sequences......................................................................................................42  4.5.1  Market-making strategies................................................................................................42  4.5.2  Directional strategies.......................................................................................................43  4.5.3  Arbitrage strategies.........................................................................................................43  5  Minimum tick size.....................................................................................................................44  5.1  Requirement...............................................................................................................................44  5.2  The Commission’s impact assessment.......................................................................................44  5.2.1  Harmonising tick sizes in Europe.....................................................................................44  5.3  General impact of the proposed rule...........................................................................................46  5.4  Impact on trading sequences......................................................................................................48  5.4.1  Market-making strategies................................................................................................48  5.4.2  Directional strategies.......................................................................................................48  5.4.3  Arbitrage strategies.........................................................................................................49  6  Minimum resting times.............................................................................................................49  6.1  The proposed rule.......................................................................................................................49  6.2  Rationale behind the rule and the Commission’s impact assessment........................................50  6.3  Reducing stress on operators’ IT systems..................................................................................51  6.4  Are HFT strategies harmful/abusive?.........................................................................................51  6.5  Why are cancellation ratios so high?..........................................................................................55 The economic impact of MiFID rules 4 6.6  General impact............................................................................................................................56  6.6.1  Impact on the costs of immediacy...................................................................................62  6.7  Impact on trading sequences......................................................................................................63  6.7.1  Market-making strategies................................................................................................63  6.7.2  Directional strategies.......................................................................................................64  6.7.3  Arbitrage strategies.........................................................................................................66  6.7.4  Pinging.............................................................................................................................66  7  Minimum execution ratios........................................................................................................67  7.1  The proposed rule.......................................................................................................................67  7.2  Rationale behind the rule and the Commission’s impact assessment........................................68  7.3  General impact............................................................................................................................68  8  Strengthening organisational requirements (circuit breakers)............................................69  8.1  The proposed rule.......................................................................................................................69  8.2  Rationale behind the rule and the Commission’s impact assessment........................................70  8.3  Effects on markets overall...........................................................................................................71  8.4  Coordination of volatility interruptions across Europe.................................................................74  8.5  Alternative implementations........................................................................................................74  9  The impact of further increases in speed on the operation of the rules that impinge on trading sequences.................................................................................................................................75  A1  Annex.........................................................................................................................................76  A1.1  Introduction......................................................................................................................76  A1.2  Arbitrage..........................................................................................................................77  A1.3  Directional........................................................................................................................78  A1.4  Market making.................................................................................................................84 The economic impact of MiFID rules 5 1 Introduction and executive summary 1.1 Objectives and remit Foresight has commissioned Oxera to conduct an economic impact assessment of the rules proposed by the European Commission in its review of MiFID aimed at regulating high- frequency trading (HFT).1 In recent years, the amount of trading done by ‘high-frequency traders’ has increased significantly, resulting in a debate about the impact on society of this type of algorithmic trading (AT). On the one hand, there may be benefits in terms of increased liquidity; on the other hand, it may facilitate the implementation of certain trading strategies that could be considered harmful or abusive. HFT may also have consequences for financial stability. Furthermore, it has been argued that the provision of additional liquidity may, in practice, be more limited than initially thought, and that high-frequency traders may take liquidity from, rather than provide it to, long-term investors, particularly at times when liquidity is already low and/or the market in a particular security is under stress. As part of the MiFID review, the Commission has proposed a number of measures to address potential concerns about the impact of AT and HFT:2 1) A series of new specific organisational requirements for market participants would be introduced with the possibility of further specification in implementing acts on each of the issues below:  authorised firms involved in automated trading would have in place robust risk controls to mitigate potential trading system errors;  firms involved in automated trading would notify their competent authority of the computer algorithm(s) they employ, including an explanation of its design, purpose and functioning;  firms who provide ‘sponsored access’ to automated traders would have in place robust risk controls and filters to detect errors or attempts to misuse facilities;  operators of trading venues would have in place proper risk controls and arrangements to mitigate the risk of errors generated by automated trading leading to disorderly trading (e.g. circuit breakers) or the breakdown of their trading systems (e.g. by stress testing to ensure resilience);  operators of trading venues would give equal and fair access to market participants to co- location services. 2) Implementing measures could further specify minimum tick sizes; 3) Market operators would be required to ensure that if a high frequency trader executes significant numbers of trades in financial instruments on the market then it would continue providing liquidity in that financial instrument on an ongoing basis subject to similar conditions that apply to market- makers; and 1 European Commission (2010), ‘Public Consultation: Review of the Markets in Financial Instruments Directive (MiFID)’, Consultation Document, European Commission, December 8th. 2 These rules would be applied to AT, which would be defined in a broad manner as ‘trading involving the use of computer algorithms to determine any or all aspects of the execution of the trade such as the timing, quantity and price.’ HFT would be considered a subcategory of AT. Furthermore, all persons involved in HFT over a specified minimum quantitative threshold would need to be authorised as investment firms. This would ensure that they are subject to organisational requirements (such as systems and risk management obligations and capital requirements) and to full regulatory oversight. The economic impact of MiFID rules 6 4) Market operators would be required to ensure that orders would rest on an order book for a minimum period before being cancelled. Alternatively they would be required to ensure that the ratio of orders to transactions executed by any given participant would not exceed a specified level. In either case, further specification would be needed on the specific period or level. During the course of Oxera’s assessment, the Commission’s final proposals were published.3 Most of the proposed rules are more or less in line with the original proposals and therefore have not affected Oxera’s analysis.4 The Commission’s own impact assessment has been incorporated, as far as possible, into Oxera’s analysis. 1.2 Approach and sources of information This report applies a conceptual framework in which the Commission’s proposals are assessed from a policy perspective. It systematically examines the potential justifications for regulation by assessing the extent to which concerns about the impact of HFT that have been raised in the literature could mean that there are specific market failures that need to be addressed. Furthermore, it assesses the impact of the proposed rules on the market and end-investors, and, importantly, the mechanisms through which the impact would arise. In order to inform its understanding of HFT and the potential impact of the proposed rules, Oxera conducted interviews with market participants, academics who have specific expertise in the area of trading and the functioning of capital markets, and other stakeholders; and reviewed the literature on HFT. The assessment presented in this report should be seen as a high-level impact assessment.5 The scope of the research was restricted to the information sources listed below—any empirical analysis was beyond the scope of the report and would require more time and resources. It is not the purpose of this report to provide answers to all questions in relation to HFT. Many questions are likely to require detailed empirical analysis. In the debate on HFT, there is some confusion and a diversity of views. To some extent, this may be due to people defining HFT in different ways (or not defining it) and not being clear about the particular concerns or market failures in relation to HFT. However, the nature of HFT also seems to be subject to considerable change, which may affect its impact, and the need for and impact of regulation. There also seems to be a difference between the nature and extent of HFT in the USA compared with Europe. For example, the ratio of messages to executed orders (in relation to HFT) seems to be higher in the USA than in Europe. HFT may have been developed more in the USA as a result of lower trading and post-trading transaction fees, and possibly as a result of the way in which different trading venues are linked together.6 The following information sources were used.  Academic literature—there is a growing body of academic literature on the impact of HFT. The purpose of this economic impact assessment is to put the findings of academic research 3 European Commission (2011), ‘Proposal for a Directive of the European Parliament and of the Council on markets in financial instruments repealing Directive 2004/39/EC of the European Parliament and the Council’, European Commission, October 20th. 4 As explained in section 4, the final proposal for a market-making requirement seems to go beyond the original proposal. 5 Similar to high-level cost–benefit analyses undertaken by the Financial Services Authority. 6 For example, Regulation National Market System (RegNMS) in the USA and the best-execution requirements on brokers in Europe. See, for example, Gomber, P., Arndt, B., Lutat, M. and Uhle, T. (2011), ‘High-Frequency Paper’, commissioned by Deutsche Börse. The economic impact of MiFID rules 7 into a policy-making context rather than summarising the literature. Useful reviews of the academic literature can be found in Gomber et al. (2011), Penalva (2010) and Biais and Woolley (2011).7 A number of books have been written on automated trading and HFT in particular.8  Interviews with stakeholders—Oxera conducted interviews with a large number of market participants (for example, fund management, hedge, brokerage firms and market-making firms, HFT traders), infrastructure providers, regulators, academics and other stakeholders. Oxera also participated in various seminars and workshops that were taking place. The interviews were used to obtain a better understanding of the rationale behind the trading strategies pursued by high-frequency traders, and the potential impact of some of the proposed rules.  Studies and consultations by regulatory authorities—in addition to the rules proposed by the MiFID review, there are a number of other studies and consultations published by authorities (eg, a consultation paper by ESMA)9, reports by national financial regulatory authorities in Europe (eg, a paper by the Netherlands Authority for Financial Markets)10 and outside Europe (eg, studies by the SEC and CFTC in the USA),11 and a consultation document by IOSCO.12  Responses to consultations and other papers by market participants—a selection of responses to the MiFID review, the SEC/CFTC study and the IOSCO consultation were reviewed.13 Some market participants have published their own position paper or presentation on the topic of HFT (eg, Optiver, Nanex and Tradeworx).14 This report does not provide a quantification of the impact of the proposed rules. Such quantification would require an empirical analysis, which would have necessitated more time and resources than were available. During the course of the research, it became clear that a quantification is not necessarily required for a high-level impact assessment of the appropriateness of the rules. Applying economic logic allows for an assessment of how the proposed rules could affect the market. The fact that no quantification is provided (of the positive or negative impact) does not mean that the impact is likely to be small. On the contrary, any small changes in the regulation of capital markets can have significant consequences for the efficient functioning of these markets 7 Gomber et al. (2011), op. cit. Penalva, J. (2011), ‘High Frequency Trading: An Overview’, July, available at: http://www.josepenalva.com. Biais, B. and Woolley, P. (2011), ‘High Frequency Trading’, Working paper IDEI, March. 8 See, for example. Narang, R. (2009), Inside the Black Box: The Simple Truth about Quantitative Trading, Wiley Finance, September. Aldridge, I. (2010), High Frequency Trading: A Practical Guide to Algorithmic Strategies and Trading Systems, Wiley Trading. Perez, E. (2011), The Speed Traders: An Insider’s Look at the New High-Frequency Trading Phenomenon that is Transforming the Investment World, McGraw-Hill, April. 9 ESMA (2011), ‘Consultation Paper: Guidelines on Systems and Controls in a Highly Automated Trading Environment for Trading Platforms, Investment Firms and Competent Authorities’, Consultation Paper, European Securities and Markets Authority, July 20th. 10 AFM (2010), ‘High Frequency Trading: the application of advanced trading technology in the European market place’, report, the Netherlands Authority for Financial Markets. 11 SEC/CFTC (2010), ‘Findings regarding the market events of May 6, 2010’, report, US Securities & Exchange Committee and US Commodity Futures Trading Commission, September 30th. 12 IOSCO (2011), ‘Regulatory Issues Raised by the Impact of Technological Changes on Market Integrity and Efficiency’, Consultation report, International Organization of Securities Commissions, July. 13 At the time of producing the report, responses to the ESMA consultation were not yet available. 14 Optiver (2010), ‘High Frequency Trading’, Position Paper, December, www.optiver.com/corporate/hft.pdf. Tradeworx (2010), ‘Public Commentary of SEC Market Structure Concept Release’, April, available at: www.tradeworx.com/TWX-SEC-2010.pdf. Nanex (2010) ‘Nanex Flash Crash Summary Report’, September 27th, available at: www.nanex.net/FlashCrashFinal/FlashCrashSummary.html. The economic impact of MiFID rules 8 and may affect end-users such as pension fund holders and firms that raise capital, and, ultimately, may affect the real economy. 1.3 Main conclusions Market-making requirement The MiFID consultation paper proposed that: Market operators would be required to ensure that if a high frequency trader executes significant numbers of trades in financial instruments on the market then it would continue providing liquidity in that financial instrument on an ongoing basis subject to similar conditions that apply to market makers;15 In the Commission’s final version of its proposals, this has been reworded as follows: An algorithmic trading strategy shall be in continuous operation during the trading hours of the trading venue to which it sends orders or through the systems of which it executes transactions. The trading parameters or limits of an algorithmic trading strategy shall ensure that the strategy posts firm quotes at competitive prices with the result of providing liquidity on a regular and ongoing basis to these trading venues at all times, regardless of prevailing market conditions.16 One of the differences between the original proposal and the final version is that there is no longer a reference to the fact that these high-frequency traders would be subject to ‘similar conditions that apply to market makers’. According to the Commission, the rationale behind the proposed rule is to ensure that high- frequency traders provide meaningful liquidity at all times and would contribute to more orderly and liquid markets and mitigate episodes of high uncertainty and volatility. In other words, the ultimate aim is to prevent market panics and crashes from happening. It is useful to bear in mind that market-making requirements were never introduced with the intention to prevent panics or market crashes, nor to maintain artificially the old price of a security once additional information on that security has become available and it now has a new equilibrium fair price. They were introduced with the aim of providing immediacy—ie, ensuring that a buyer in the morning does not need to wait until the afternoon to find a seller. The way the proposal is now worded, however, means that all AT will be required to use an algorithm that ‘posts firm quotes at competitive prices ... at all times’. Although all HFT can be seen as being algorithmic, not all AT is high-frequency. Discussion with market participants suggests that there is very little trading that is purely manual. Agency orders that result in long- term changes in net positions are often executed using a (non-high-frequency) algorithm. Non- high-frequency market-making may also use algorithms to maintain the market-makers’ market position, rather than relying on direct human intervention to update those positions as their resting orders execute and additional information is incorporated into the security’s price. As a result, the majority of market participants would appear to be caught by this requirement to post firm quotes at competitive prices at all times. The impact of such a requirement would, 15 European Commission (2010), ‘Review of MiFID’, December 8th, p. 16. 16 European Commission (2011), ‘Proposal for a Directive of the European Parliament and of the Council on markets in financial instruments’, p. 70. The economic impact of MiFID rules 9 therefore, extend significantly beyond high-frequency traders. This does not seem to have been taken into account in the Commission’s impact assessment. The impact of this market-making requirement will depend critically on how the requirement to ‘continually post firm quotes at competitive prices’ is interpreted. At one extreme, there will be significant additional risk applied to AT if the requirement is interpreted to mean that at all times the market is open, and that every algorithmic trader has to offer to buy and sell a security across a spread that reflects the usual spread for that security. At times of increased uncertainty those required to maintain bid and offers across the usual spread are likely to find that they enter into a significant number of trades where the subsequent track of prices will have moved against them and which they will not be able to unwind without loss. This could have the effect of making AT (including buy-side trading using computer- controlled execution) non-viable because of the requirement to provide firm bid and offers at competitive prices being too risky. All trading could, therefore, return to manual trading. If, as seems likely, the return to manual trading increases the total costs of intermediaries (because they have to substitute people for computers) this increase in intermediaries’ costs will be paid for by investors (which will see a lower net return) or companies (which will see an increase in the cost of capital which will then be passed on in the form of higher prices in the end product market). If, on the other hand, the interpretation of the requirement is designed to mimic the requirements applied currently to official market-makers, so that there is still the possibility of market-making using computer-driven trading sequences being profitable, then, the impact on HFT sequences is likely to be minimal. In general, this arises because any HFT sequence algorithm is likely to be able to have a market-making module bolted on to it that will display firm bid and offers at competitive prices under normal market conditions, and to widen the spread offered or withdraw from the market under those market conditions where bid and offers across a narrow spread become highly risky (as traditional market-makers are generally currently allowed to do). Such a module could be designed to actually execute very rarely, if at all, by the repositioning of these (resting) orders before they reach the head of the queue. In addition, if the definition of ‘competitive’ price includes prices one or more ticks away from the touch, then ensuring that actual execution is limited, will be easier. The re-positioning of orders (before they search the head of the queue) could result in an increase in the number cancellations relative to executions. Unlike those intermediaries which wish to make money out of market-making, those which simply wish to meet this requirement in order to pursue other strategies will generally only wish to avoid losing money, and therefore will not be concerned with executing trades, at some point, for the purpose of market-making. The specific implications for the different trading sequences are analysed in section 4.5. In sum, the impact of the proposed market-making rule will depend on the precise interpretation. In one scenario, the proposed rule may have a limited impact but at the same time is unlikely to achieve the Commission’s objective of improving liquidity and reducing volatility. In another scenario, the proposed rule would have the potentially severely negative unintended consequence of all trading returning to manual trading resulting in an increase in costs of intermediaries. The economic impact of MiFID rules 10 Minimum tick size The MiFID review proposes that implementing measures could further specify minimum tick sizes—it does not prescribe specific minimum tick sizes at this stage. Needless to say, the minimum tick sizes would be applied to all trading and not simply automated trading or HFT. The MiFID review consultation document does not explain the objective of this proposed rule, so it is not possible to analyse directly what market failure this proposed regulation is aimed at, nor is it therefore possible to directly evaluate if an alternative intervention would be available. However, given the form of the rule (minimum tick size) the general objective would seem to be that in the absence of the rule tick sizes would become (or currently are) ‘too small’. If the rule is to have an impact it would seem to need to either:  increase the tick size that would otherwise occur; or (as a side effect)  coordinate and standardise tick sizes for the same security trading in different venues. It is possible that the Commission considers that there is a need to impose a minimum tick size in Europe if it considers that tick sizes have become (or could become) too small, and that this small size itself causes direct harm to the capital market. However, given the context of this proposal, it is also possible that the proposed rule is aimed more generally at curtailing HFT, or at least certain trading strategies that are implemented using HFT. Oxera’s analysis suggests that limited increases in the minimum tick size would be unlikely to have a significant impact on either the volume or location of (high-frequency) trading. At the margin, some HFT sequences will become slightly more profitable, but may occur slightly less frequently. For example, in the case of arbitrage strategies, an increase in tick size will tend to increase the value of fleeting mispricing as the price in each venue momentarily moves apart. Where the mispricing is one tick (for example, when the price in the leader venue has moved, but the price in the follower venue has not) a bigger tick will increase the value of that mispricing (for any given volume of security). However, the slightly larger tick may also slightly reduce the frequency with which these pricing anomalies occur. There is some empirical evidence (referred to in section 5) that very large increases in tick size could move trading away from lit venues. Minimum resting times The MiFID consultation paper proposed that: Market operators would be required to ensure that orders would rest on an order book for a minimum period before being cancelled.17 In the Commission’s near final version of its proposals, this had been reworded as follows: Impose minimum latency period of orders in the order book—Under this option an obligation would be implemented according to which orders on electronic platforms would need to rest on an order book for a minimum period of time before they can be withdrawn.18 17 European Commission (2010), ‘Review of MiFID’, December 8th, p. 16.