15 June 212 Volume 3 Issue 14 The Blotter presents ITG s insights on complex global market structure, technology, and policy issues. CONTRIBUTORS Ian Domowitz Ian.Domowitz@itg.com 1.212.444.6279 Kumar Giritharan Kumar.Giritharan@itg.com 1.212.588.4156 Jacqueline King Jacqueline.King@itg.com 1.212.588.4977 CONTACT Asia Pacific +852.2846.35 Canada +1.416.874.9 EMEA +44.2.767 4 United States +1.212.588.4 info@itg.com www.itg.com Institutional Trading in Exchange-Traded Funds Leveraging our recently published research on liquidity and transaction costs for select US ETFs, we open up our analysis to defining the characteristics of institutional trading activity in ETFs. Between 21 and 211 the number of Exchange Traded Funds (ETFs) grew from 12 to 1134. Net assets rose from $83 billion to $1,26 billion. 1 An increasing number of liquidity providers, algorithms, and trading technologies targeting these financial products foster this growth. ITG s Financial Engineering team recently published a white paper, exploring liquidity and trading cost characteristics of a select group of widely traded ETFs. 2 The work shows that the ETFs exhibit very different liquidity and cost characteristics than common stocks, a widely-held belief looking for documentation. We broaden the scope of the investigation, while presenting a less technical view of the market. We are interested in one seemingly simple question: what are the characteristics of institutional trading activity in ETFs? Most ETF knowledge is in the context of the retail market, and this question is capable of spawning a host of others. We concentrate on the scope of the market, the nature of orders, transaction costs, and volume patterns, as well as the link of ETFs to algorithmic trading activity. We find that ETFs trade very differently from their single stock counterparts. They are characterized by smaller order sizes, narrower spreads, lower volatility levels, and lower price impact costs. Algorithmic trading dominates institutional activity. The strategies used to execute ETFs have changed overtime, however. In particular, a shift from schedule-based algorithms to liquidity seeking algorithms is emerging. Finally, the relative size of the institutional market for this derivative security is much smaller than observed retail participation, and to this we now turn. The Scope of the Market for ETFs ETFs are registered under the Investment Company Act of 194 as open-end funds or unit investment trusts (UITs). Shares of ETFs trade intraday on stock exchanges at market-determined prices. The Standard and Poor s Depository Receipt (SPDR)
THE BLOTTER 15 June 212 Volume 3 Issue 14 2 Trust and the ishares Trust are examples of unit investment trusts that issue ETFs. In 1993, the SPDR Trust introduced the first ETF a domestic equity fund that track the S&P5 Index. Exhibit 1 illustrates the rapid growth of both the number of ETFs and their total net assets. Exhibit 1: Growth of ETFs $1,4 $1,2 $1, $8 $6 $4 $2 Total Net Assets and Number of ETFs Billions of USD, 21-211 12 1 8 6 4 2 $- 21 22 23 24 25 26 27 28 29 21 211 Net Assets in Billions USD (left scale) Number of ETFs (right scale) Source: Investment Company Institute There were 12 ETFs with net assets of $83 billion in 21. By February 212, the total number of ETFs reached 1,187 with net assets of $1.19 trillion. ETFs track broad market indexes, sectors and industries, fixed income, commodities and derivatives. ETFs may also be actively managed. As of the end of 211, BlackRock/ ishares, State Street and Vanguard account for approximately 7 of the ETF market share. We compare traded dollar volume of ETFs and equities in the US markets in Exhibit 2. Exhibit 2: Equity and ETFs Traded in the US Dollar Volume ($, Billion) $3, $2,5 $2, $1,5 $1, $5 $- Dollar Volume ($, Billions) Jan-4 Apr-4 Jul-4 Oct-4 Jan-5 Apr-5 Jul-5 Oct-5 Jan-6 Apr-6 Jul-6 Oct-6 Jan-7 Apr-7 Jul-7 Oct-7 Jan-8 Apr-8 Jul-8 Oct-8 Jan-9 Apr-9 Jul-9 Oct-9 Jan-1 Apr-1 Jul-1 Oct-1 Jan-11 Apr-11 Jul-11 Oct-11 Jan-12 NYSE Listed ETF SPY 16 14 12 1 8 6 4 2 SPY Index Source: NYSE.com In the 3-year period between 24 and 27 the dollar volume traded for US equities and ETFs grew steadily. As the financial crisis unfolded in 27, there has been a steady decline in US equities traded dollar volume. Nevertheless, ETF dollar volume increased throughout 27 and for the most of 28, then reverting to levels more reminiscent of 26. 3
THE BLOTTER 15 June 212 Volume 3 Issue 14 3 Publicly available data include retail activity, of course. In order to say anything about the institutional market, we rely on ITG s Peer Universe. 4 These data must be considered a sample, albeit a large one. The scope of the data permits us to calculate percentages with some statistical precision and to identify quantities such as average order sizes. We begin with the size of the institutional market, illustrated in Exhibit 3. Exhibit 3: Institutional ETF Trading as a % of Total Dollar Volume % ETF tradiing 6% 4% 3% 2% 1% 8 7 6 5 4 3 2 1 Number of ETF 21Q1 21Q2 21Q3 21Q4 211Q1 211Q2 211Q3 211Q4 % ETF 4% 4% 4% 4% 4% Number of ETFs 563 636 613 674 693 719 738 714 - % ETF Number of ETFs In the fourth quarter of 211, ETFs accounted for of institutional dollar volume. 5 The underlying data include all equity types, including common stock, ADRs, GDRS, close-end funds, REITS, and preferreds. In early 21, ETF trading as a percentage of total trading dollar volume was 4%. Over the same period, the number of ETFs increased from 563 to 714. Aggregate figures do not tell the entire story. More information is available in the distribution of ETF trading volume as a percentage of total trading activity, given in Exhibit 4. Exhibit 4: Distribution of ETF Trading As a Percentage of Dollar Volume 3 2 2 1 1 1% 2% 3% 4% 7% 1 1 2 5 1 % of Clients 1 24% 1 12% 7% 8% 4% 3% 1% % of ETF trading
THE BLOTTER 15 June 212 Volume 3 Issue 14 4 The distribution of institutional ETF volume is highly skewed toward the smaller percentage categories. Only 1 of the universe does not trade ETFs, suggesting wide adoption. On the other hand, ETF trading accounts for only 1% of all equity trading for about a quarter of the population. The composition of ETF trading relative to all equity trading is 4% or less for 68% of ITG s Peer Universe. Approximately 31% of institutional asset managers trade greater than of ETFs relative to other equity security types. Adoption may be wide, but is not uniform across institutional managers. Institutional ETF turnover is dominated by 1 funds. The ten accounts for over 6 of total institutional ETF trading by dollar volume, illustrated in Exhibit 5. Exhibit 5: Top 1 ETFs Trading as a Percentage of ETF Dollar Volume 6 3 2 2 1 1 SPY IWM QQQ EFA EEM VWO XLE GLD XLF XRT 21 211 Trading in the SPY dominates, accounting for a quarter of total dollar volume in 21 and 211. The ETF that tracks the Russell 2 Index (IWM) and the ETF that tracks the NASDAQ 1 Index (QQQ) are the second and third largest traded ETF by dollar volume. TRADING CHARACTERISTICS OF ETFS We turn to an examination of the profile of institutionally traded ETFs. The focus in this section is on order size, volatility, and spreads. Order Size The average order size for ETFs is consistently smaller than that for non-etfs, illustrated in Exhibits 6 and 7.
THE BLOTTER 15 June 212 Volume 3 Issue 14 5 Exhibit 6: Time Series of Order Size for ETFs and Non-ETFs 3, 25, 2, Shares 15, 1, 5, 21Q1 21Q2 21Q3 21Q4 211Q1 211Q2 211Q3 211Q4 ETF Non ETFs From 21 to 211, average order sizes declined for all security types. The average order size for ETFs decreased by 12%, while that for common stocks fell by 16%. In 21 the average order size for ETFs (non-etfs) was 8,737 (22,557) shares. In 211, the average order size for ETF (non-etfs) was 7,674 (18,973) shares. Exhibit 7 illustrates the average order size of ETFs and non-etfs across liquidity groups as determined by average daily volume (ADV). Exhibit 7: Average Order Size for ETFs and Stocks by Liquidity Groups 7, Average Order Size (Shares) 6, 5, 4, 3, 2, 1, - A) to 1% B) 1 to C) 5 to 1 D) 1 to 2 E) 2 to 5 F) > 5 ETF NON ETF For all liquidity groups, ETFs average order size is smaller than that of non-etfs. The average order size of non-etfs exceeds those of ETFs by more than three times in the 5-1, 1-2 and 2-5 ADV category. In the 1-2 ADV category, average order size is approximately 72,427 shares for ETFs and more than 226, shares for non-etfs. Spreads and Volatility Across various liquidity groups, common stocks consistently exhibit higher spreads and experienced higher volatility levels (see Exhibit 8).
THE BLOTTER 15 June 212 Volume 3 Issue 14 6 Exhibit 8: Average Spread and Volatility by Liquidity Groups Average Spread (BPS) 1 9 8 7 6 5 4 3 2 1 - A) to 1% B) 1 to C) 5 to 1 D) 1 to 2 E) 2 to 5 F) > 5 ETF Non ETF 25 2 Average Volatility (%) 15 1 5 A) to 1% B) 1 to C) 5 to 1 D) 1 to 2 E) 2 to 5 F) > 5 ETF Non ETF It might be thought that these results stem from a mismatch of the common stocks with the ETFs in question. This is not the case. We compare the average spread and volatility of SPY and IWM, two of the most widely traded ETFs, to the average spread and volatility of their underlying constituents in Exhibit 9. Spread and volatility levels, on average, are much lower than those of the constituents. Exhibit 9: Spread and Volatility of ETFs and Their Constituents 14 12 1 8 6 4 2-12 1 8 6 4 2 S&P5: SPY and Constituents Average Spread (BPS) 1/3/8 2/17/8 4/2/8 5/17/8 7/1/8 8/15/8 9/29/8 11/13/8 12/28/8 2/11/9 3/28/9 5/12/9 6/26/9 8/1/9 9/24/9 11/8/9 12/23/9 2/6/1 3/23/1 5/7/1 6/21/1 8/5/1 9/19/1 11/3/1 12/18/1 2/1/11 3/18/11 5/2/11 6/16/11 7/31/11 9/14/11 1/29/11 12/13/11 1/27/12 3/12/12 Spread of Constituents (bps) S&P5: SPY and Constituents Average Volatility (%) SPY SPREAD (BPS) 1/3/8 2/17/8 4/2/8 5/17/8 7/1/8 8/15/8 9/29/8 11/13/8 12/28/8 2/11/9 3/28/9 5/12/9 6/26/9 8/1/9 9/24/9 11/8/9 12/23/9 2/6/1 3/23/1 5/7/1 6/21/1 8/5/1 9/19/1 11/3/1 12/18/1 2/1/11 3/18/11 5/2/11 6/16/11 7/31/11 9/14/11 1/29/11 12/13/11 1/27/12 3/12/12 Volatility of Constituents(annualized,%) SPY VOLAT (%) 1 9 8 7 6 5 4 3 2-1 16 14 12 1 8 6 4 2 Russell 2 Index: IWM and Constituents Average Spread (BPS) 1/3/8 2/17/8 4/2/8 5/17/8 7/1/8 8/15/8 9/29/8 11/13/8 12/28/8 2/11/9 3/28/9 5/12/9 6/26/9 8/1/9 9/24/9 11/8/9 12/23/9 2/6/1 3/23/1 5/7/1 6/21/1 8/5/1 9/19/1 11/3/1 12/18/1 2/1/11 3/18/11 5/2/11 6/16/11 7/31/11 9/14/11 1/29/11 12/13/11 1/27/12 3/12/12 Spread of Constituents (bps) SPY SPREAD (BPS) Russell 2 Index: IWM and Constituents Average Volatility (%) 1/3/8 2/17/8 4/2/8 5/17/8 7/1/8 8/15/8 9/29/8 11/13/8 12/28/8 2/11/9 3/28/9 5/12/9 6/26/9 8/1/9 9/24/9 11/8/9 12/23/9 2/6/1 3/23/1 5/7/1 6/21/1 8/5/1 9/19/1 11/3/1 12/18/1 2/1/11 3/18/11 5/2/11 6/16/11 7/31/11 9/14/11 1/29/11 12/13/11 1/27/12 3/12/12 Volatility of Constituents(annualized,%) IWM VOLAT (%)
THE BLOTTER 15 June 212 Volume 3 Issue 14 7 Average spreads are narrow when compared to the average spread of the constituents. The spread for SPY for 211 tends to be no greater than 1 basis point (bp); the average for the S&P5 constituents is 5 bps. The same applies to average 6-day historical volatility, in percentage terms. The volatility for SPY (IWM) was approximately 17% (2) while the average volatility of the SPY constituents was 29% (44%) for 211. The intra-day volume pattern for ETFs looks very similar to that of non-etfs (see Exhibit 1). Exhibit 1: Intra-day Trading Volume 1 9% 8% 7% 6% 4% 3% 2% 1% % trading 9:3 1: % of ETFs Trading 1:3 11: 11:3 12: 12:3 13: 13:3 14: 14:3 15: 15:3 % trading 1 9% 8% 7% 6% 4% 3% 2% 1% 9:3 1: % of Non ETFs Trading 1:3 11: 11:3 12: 12:3 13: 13:3 14: 14:3 15: 15:3 As a percentage of total volume traded, ETFs trade slightly more than non-etfs in the first half hour of the regular trading day. In 211, ETF trading in that time interval accounted for 8% of volume, while non-etf trading in the first bin accounted for 6%. Trading in the last half hour of the day for ETFs and non-etfs was approximately the same, at 9%. Trading Performance We analyze historical trading performance measured by two benchmarks, implementation shortfall (IS) and a price weighted participation rate of 1 of median daily volume (MDV) or PWP15 (see Exhibit 11). Exhibit 11: ETF and Non-ETF Trading Costs by Quarter and Order Size Trading Performance (bps) (1) (2) (3) (4) (5) (6) 21Q1 21Q2 ETFs Trading Costs: By Quarter 21Q3 21Q4 211Q1 211Q2 211Q3 211Q4 5, 4, 3, 2, 1, Order Size (shares) Trading Performance (bps) 1 (1) (2) (3) (4) (5) (6) (7) A < 1K ETFs Trading Costs: By Order Size Group B < 3K C < 5K D < 75K E < 1K F < 15K G < 25K H < 35K I < 5K J < 1M K >= 1M 16% 14% 12% 1 8% 6% 4% 2% % ETF Trading Avg. Order Size IS PWP15 % trading IS PWP15 Trading Performance (bps) (2) (4) (6) (8) (1) (12) (14) Non ETFs Trading Costs: By Quarter 1, 9, 8, 7, 6, 5, 4, 3, 2, 1, - Order Size (shares) Trading Performance (bps) 1 (1) (2) (3) (4) (5) (6) (7) NON ETFs Trading Costs: By Orde Size Group 16% 14% 12% 1 8% 6% 4% 2% % NON ETF Trading 21Q1 21Q2 21Q3 21Q4 211Q1 211Q2 211Q3 211Q4 Avg. Order Size IS WtdAvPWPCostBps A < 1K B < 3K C < 5K D < 75K E < 1K F < 15K G < 25K H < 35K I < 5K J < 1M K >= 1M % trading IS PWP15
THE BLOTTER 15 June 212 Volume 3 Issue 14 8 ETFs have lower average trading costs than non-etfs. The range of trading costs for ETFs across order sizes is between 1 to 3 bps for PWP15 and 1 to 6 bps for IS. IS trading costs for non-etfs ranges from 6 to 6 bps. Trading costs for ETFs were lower due to their relatively better liquidity characteristics, most particularly in the top ten ETFs by volume. Order size differences cannot account for these results. ETF order sizes of fewer than 1, shares and fewer than 3, shares account for 13% and 12% of total institutional ETF trading, respectively. Similarly, order sizes for non-etfs of fewer than 1, shares and fewer than 3, shares account for 11 percent and 12 percent of non-etf trading. In the larger order size category, greater than 5, shares, 1 (11%) of all ETF (non-etf) trading fall in the 5, to 1,, shares category and 1 (1) are in the greater than 1,, shares category. Algorithmic Trading Activity Over 5 of ETF trades are executed via algorithms, about the same as for common stocks in our sample. Exhibit 12: Algorithmic Trading of ETFs 6 5 % trading 4 3 2 1 ALGO Others ETF 53% 47% Non-ETF 5 4 ETF Non-ETF Institutions execute 3 of ETFs via scheduled algorithms, which include volume weighted average price (VWAP), time weighted average price (TWAP) and volume participation (VP) (see Exhibit 13). Exhibit 13: ETF Trading Across Strategies in 211 4 3 3 2 2 1 1 SCHEDULED LIQUIDITY SEEKING IS DMA DARK CLOSE % Trading 3 24% 19% 13% 7% 1%
THE BLOTTER 15 June 212 Volume 3 Issue 14 9 Liquidity seeking algorithms account for 24% of all ETF trading via algorithms. Implementation shortfall (IS), direct market access (DMA), dark and close algorithms account for 19%, 13 %, 7% and 1%, respectively. Scheduled algorithms have been algorithm of choice for the most popular and liquid ETFs, such as SPY and IWM. In Exhibit 14 we compare algorithm segmentation for ETFs and common stocks. Exhibit 14: Segmentation of Various Algorithms for ETFs and Non-ETFs 1 9 8 7 6 5 4 3 2 1 ETFs Trading Jan-1 Feb-1 Mar-1 Apr-1 May-1 Jun-1 Jul-1 Aug-1 Sep-1 Oct-1 Nov-1 Dec-1 Jan-11 Feb-11 Mar-11 Apr-11 May-11 Jun-11 Jul-11 Aug-11 Sep-11 Oct-11 Nov-11 Dec-11 1 9 8 7 6 5 4 3 2 1 NON ETFs Trading Jan-1 Feb-1 Mar-1 Apr-1 May-1 Jun-1 Jul-1 Aug-1 Sep-1 Oct-1 Nov-1 Dec-1 Jan-11 Feb-11 Mar-11 Apr-11 May-11 Jun-11 Jul-11 Aug-11 Sep-11 Oct-11 Nov-11 Dec-11 CLOSE DARK DMA IS LIQUIDITY SEEKING SCHEDULED The composition of algorithms used to trade ETFs changed overtime. In 21, scheduled algorithms dominated ETF trading; on average, they accounted for 51% of all algorithm usage. By 211, this number dropped to 3. ETFs experienced an increase in the usage of liquidity seeking algorithms, from an average of 16% in 21 to 24% in 211. For the stocks, scheduled algorithms dominate, on average, accounting for 4 of all algorithmic trading. The distribution of algorithm usage for non-etfs has been quite stable over time. 7 Liquidity seeking and dark algorithms have historically delivered the best performance when trading ETFs.
THE BLOTTER 15 June 212 Volume 3 Issue 14 1 Exhibit 15: Trading costs across different algorithms % of Trading 4 3 3 2 2 1 1 SCHEDULED LIQUIDITY SEEKING IS DMA DARK CLOSE % Trading 3 24% 19% 13% 7% 1% IS (1.91) (.83) (1.38) (2.6) (.88) (6.1) (1) (2) (3) (4) (5) (6) (7) Trading Performance (BPS) % Trading IS The cost of trading ETFs via liquidity seeking and dark algorithms is.83 bps and.88 bps, respectively. The two algorithms account for 31% of all algorithmic trading of ETFs in ITG s Peer Universe. Execution costs incurred by scheduled strategies are 1.91 bps. Trading at the close accounts for only 1% of ETF trading; however, costs are high at 6 bps. The qualitative nature of performance comparisons between closing algorithms and other strategies echoes previous results for single stocks. 8 CONCLUSION We began with the question, what are the characteristics of institutional trading activity in ETFs? Although a simple answer is complicated by a variety of factors, our data provide hitherto unavailable information about the nature of the institutional market. The relative size of the institutional market for this derivative security is much smaller than observed retail participation, by several orders of magnitude. In the aggregate, of institutional trading activity in 211 is ETF-based, rising from 4% in 21. The aggregate number masks some differentiating features. Fifteen percent of institutions show no ETF trading whatsoever. Further, ETF trading accounts for only 1% of all equity trading for about a quarter of the population. In fact, the composition of ETF trading relative to all equity trading is 4% or less for almost 7 of ITG s Peer Universe. The real action is in the 31% of institutional asset managers, who trade greater than of ETFs relative to other equity securities. Asset managers are adopting ETFs selectively and in accordance with different strategic portfolio imperatives. The institutional ETF market is characterized by narrower spreads and lower volatility levels than for common stocks. These in turn influence and are influenced by smaller order sizes than observed for common stocks traded by the same market participants as the ETFs in this sample. Algorithmic trading dominates institutional activity. The numbers suggest, however, that behavior in this respect is little different from the use of automated strategies for single stocks and baskets. The strategies used to execute ETFs have changed overtime, and a shift from schedule-based algorithms to liquidity seeking algorithms is emerging. This does differ from the stability of strategy usage observed in the case of single stocks. Aggregate trading patterns over the day more or less echo those for single stocks.
THE BLOTTER 15 June 212 Volume 3 Issue 14 11 We include information with respect to the cost of trading, but are cautious with respect to interpretation and conclusions. The numbers suggest that ETFs are cheaper to trade by a wide margin, relative to single stocks. We can rule out simple things like order size as a driver of such results, but issues surrounding liquidity, risk, and associated trading costs have been shown to be quite complex. 9 In the context of practical ETF trading implementations, this area deserves further scrutiny. 212 Investment Technology Group, Inc. All rights reserved. Not to be reproduced or retransmitted without permission. 61212-21594 These materials are for informational purposes only, and are not intended to be used for trading or investment purposes. The information contained herein has been taken from trade and statistical services and other sources we deem reliable but we do not represent that such information is accurate or complete and it should not be relied upon as such. No guarantee or warranty is made as to the reasonableness of the assumptions or the accuracy of the models or market data used by ITG or the actual results that may be achieved. These materials do not provide any form of advice (investment, tax or legal). ITG Inc. is not a registered investment adviser and does not provide investment advice or recommendations to buy or sell securities, to hire any investment adviser or to pursue any investment or trading strategy. Broker-dealer products and services are offered by: in the U.S., ITG Inc., member FINRA, SIPC; in Canada, ITG Canada Corp., member Canadian Investor Protection Fund ( CIPF ) and Investment Industry Regulatory Organization of Canada ( IIROC ); in Europe, Investment Technology Group Limited, registered in Ireland No. 28394 ( ITGL ) and/or Investment Technology Group Europe Limited, registered in Ireland No. 283939 ( ITGEL ) (the registered office of ITGL and ITGEL is Georges Court, 54-62 Townsend Street, Dublin 2, Ireland and ITGL is a member of the London Stock Exchange, Euronext and Deutsche Börse). ITGL and ITGEL are authorised and regulated by the Central Bank of Ireland; in Asia, ITG Hong Kong Limited, licensed with the SFC (License No. AHD81), ITG Singapore Pte Limited, licensed with the MAS (CMS Licence No. 1138-1), and ITG Australia Limited (ACN 3 67 49), a market participant of the ASX and Chi-X Australia (AFS License No. 219582). All of the above entities are subsidiaries of Investment Technology Group, Inc. MATCH NowSM is a product offering of TriAct Canada Marketplace LP ( TriAct ), member CIPF and IIROC. TriAct is a wholly owned subsidiary of ITG Canada Corp
THE BLOTTER 15 June 212 Volume 3 Issue 14 12 1 Investment Company Institute, various publications. 2 Milan Borkovec and Vitaly Serbin, Create or Buy: A Comparative Analysis of Liquidity and Transaction Costs for Selected U.S. ETFs, ITG, May 212. 3 In 28, The Securities Exchange Commissions (SEC) granted exemptive relief to sponsors of actively managed ETFs that meet certain requirements, one of which requires sponsors to publish their portfolio holdings daily. The exemptive relief serve as an impetus for the launch of numerous ETFs, including leveraged and inverse ETFs. From 27 to 28, the number of ETFs increased from 629 to 728. 4 ITG Peer Universe tracked approximately $5.81 trillion of institutional dollar volume traded in 211. The data are dominated by larger institutions, with low hedge fund participation. For North America, there are 168 institutions represented in the data base. 5 The most recent Greenwich Associates survey of US institutional activity notes that 7 percent of institutional commissions are paid for ETF trades. Given the wide variation in the use and associated amount of commission payments, this number is in line with the institutional dollar volume reported here. 6 QQQQ changed symbol to QQQ on 3/23/211, and is included here as QQQ. 7 Further information on this phenomenon is contained in Ian Domowitz and Henry Yegerman, Algorithmic Trading Usage Patterns and their Costs, Journal of Trading, Summer 211, volume 6. 8 See Ian Domowitz and Henry Yegerman, Algorithmic Trading Usage Patterns and their Costs, Journal of Trading, Summer 211, volume 6, and references to other work on the close therein. 9 See Milan Borkovec and Vitaly Serbin, Create or Buy: A Comparative Analysis of Liquidity and Transaction Costs for Selected U.S. ETFs, ITG, May 212, and Milan Borkovec, Ian Domowitz, Vitaly Serbin, and Henry Yegerman, Liquidity and Price Discovery in Exchange Traded Funds, Journal of Index Investing, Fall 21, volume 1.