Transaction Fee Pilot for NMS Stocks

1 15 U.S.C. 78k–1(a)(1)(C)(i). See also supra
Section II.G (discussing the Commission’s authority
to conduct the Pilot). 2See Securities Exchange Act Release No. 82873 (March 14, 2018), 83 FR 13008 (March 26, 2018) (‘‘Proposing Release’’ or ‘‘Proposal’’). 3See Cboe BYX U.S. Equities Exchange Fee
Schedule (as of December 2018), available at
https://markets.cboe.com/us/equities/membership/ fee_schedule/byx/; Cboe EDGA U.S. Equities Exchange Fee Schedule (as of December 2018), available at https://markets.cboe.com/us/equities/ membership/fee_schedule/edga/; Nasdaq BX Fee
Schedule (as of December 2018), available at
https://www.nasdaqtrader.com/Trader.aspx?id=bx_ pricing; NYSE National Schedule of Fees and Rebates (as of December 2018), available at https:// www.nyse.com/publicdocs/nyse/regulation/nyse/ NYSE_National_Schedule_of_Fees.pdf. EDGA adopted a taker-maker fee schedule in July 2018. See Securities Exchange Act Release No. 83643 (July 16, 2018), 83 FR 34643 (July 20, 2018) (SR– CboeEDGA–2018–012). 4See Investors Exchange Fee Schedule (as of December 2018), available at https:// iextrading.com/trading/fees/; NYSE American Equities Trading Fees and Price List (as of December 2018), available at https:// www.nyse.com/publicdocs/nyse/markets/nyseamerican/NYSE_America_Equities_Price_List.pdf. NYSE American offers rebates to eDMMs in their assigned NYSE American-listed securities. SECURITIES AND EXCHANGE COMMISSION 17 CFR Parts 200 and 242 [Release No. 34–84875; File No. S7–05–18] RIN 3235–0761 Transaction Fee Pilot for NMS Stocks AGENCY: Securities and Exchange Commission. ACTION: Final rule. SUMMARY: The Securities and Exchange Commission (‘‘Commission’’ or ‘‘SEC’’) is adopting a new rule of Regulation National Market System (‘‘Regulation NMS’’) under the Securities and Exchange Act of 1934 (‘‘Exchange Act’’) to conduct a Transaction Fee Pilot (‘‘Pilot’’) for National Market System (‘‘NMS’’) stocks to study the effects that exchange transaction fee-and-rebate pricing models may have on order routing behavior, execution quality, and market quality. We expect the data generated by the pilot, combined with data from existing sources, will facilitate an empirical evaluation of whether the existing exchange transaction-based fee and rebate structure is operating effectively to further statutory goals. DATES:
Effective date: April 22, 2019 through December 29, 2023. Compliance date: As designated by Notice pursuant to 17 CFR 242.610T(c)(2). FOR FURTHER INFORMATION CONTACT:
Richard Holley III, Assistant Director; Johnna Dumler, Special Counsel; Erika Berg, Special Counsel; or Benjamin Bernstein, Special Counsel, each with the Division of Trading and Markets, Securities and Exchange Commission, 100 F Street NE, Washington, DC 20549, or at (202) 551–5777. SUPPLEMENTARY INFORMATION: The Commission is adopting new 17 CFR 242.610T (Rule 610T) to conduct a Transaction Fee Pilot for NMS stocks.
Table of Contents
I. Executive Summary of Rule 610T
II. Discussion of Rule 610T
A. Focus on Exchange Pricing Models and the Distortions They Can Cause

Exchange Fee Models and Regulatory
Framework
Impact of Exchange Fee Models
Focus on Exchange Fee Models
Non-Exchange Trading Centers
Options Exchanges
B. Securities
The Share Price Threshold of Pilot
Securities
The Duration of Pilot Securities
Selecting Pilot Securities From All NMS
Stocks
The Ability of Issuers to Opt Out of the
Pilot
C. Pilot Design
Need for a Pilot
Pilot Design
No Overlap With Tick Size Pilot
Stratified Selection of Pilot Securities
Number of NMS Stocks Included in Each
Test Group
Reduction to the Pilot Size
Fee Cap Test Groups
Control Group
Alternative Designs
Metrics To Assess the Pilot
D. Timing and Duration
Disclosure Initiatives and the Pilot
Automatic Sunset at Year One
Pre- and Post-Pilot Periods
Early Termination
Inclusion of a Phase-In Period
E. Data
Pilot Securities Exchange Lists and Pilot
Securities Change Lists
Exchange Transaction Fee Summary
Order Routing Data
F. Implementation
G. The Commission’s Authority To Conduct the Pilot
III. Paperwork Reduction Act
A. Summary of Collection of Information B. Proposed Use of Information
C. Respondents
D. Total Initial and Annual Reporting and Recordkeeping Burdens
Pilot Securities Exchange Lists and Pilot
Securities Change Lists
Exchange Transaction Fee Summaries
Order Routing Datasets
E. Collection of Information Is Mandatory F. Confidentiality of Responses to Collection of Information G. Retention Period for Recordkeeping Requirements
IV. Economic Analysis
A. Background and Market Failures
Market Failure at the Broker-Dealer
Level
Market Failure at the Exchange Level
B. Baseline
Current Information Baseline
Current Market Environment
C. Analysis of Benefits and Costs of
Transaction Fee Pilot
Benefits of Transaction Fee Pilot
Costs of the Pilot
D. Impact on Efficiency, Competition, and Capital Formation
Efficiency
Competition
Capital Formation
E. Alternatives
Propose Rulemaking Without
Conducting a Pilot
Expand Transaction Fee Pilot To Include
Non-Exchange Trading Centers
Trade-At Test Group
Alternative Pilot
Adjustments to the Transaction Fee Pilot
Structure
V. Regulatory Flexibility Analysis
VI. Statutory Authority and Text of the Rule
Amendments
I. Executive Summary of Rule 610T
Congress directed the Commission, through Section 11A of the Exchange Act, to facilitate the establishment of a national market system and use its broad authority to carry out the objectives of Section 11A, including, among others, to assure the economically efficient execution of securities transactions.1 In furtherance of these goals, and as part of its oversight of registered national securities exchanges, the Commission periodically undertakes reviews of various aspects of market structure and current regulations to evaluate whether, in light of changes in technology and business practices, the current regulatory framework continues to fairly, effectively, and efficiently promote fair and orderly markets, serve the public interest and the protection of investors, and promote capital formation. As discussed below, one aspect of the current regulatory framework focuses on the current pricing and fee structure for transactions in securities. As the Commission discussed in its Pilot proposal, the predominant transaction pricing structure that developed among equities exchanges to attract order flow is the ‘‘maker-taker’’ fee model.2 Specifically, out of thirteen equities exchanges, seven utilize the ‘‘makertaker’’ fee model, in which they pay a rebate to a provider of liquidity and charge a fee to a taker of liquidity. Among the remaining exchanges, four utilize a ‘‘taker-maker’’ pricing model (also called an inverted model) where they charge a fee to a provider of liquidity and pay a rebate to a taker of liquidity,3 and two have a ‘‘flat fee’’ model.4 In recent years this area has

5See Proposing Release, supra note 2. 6The Proposal was developed, in part, by reference to a recommendation for an access fee pilot submitted to the Commission by the Equity Market Structure Advisory Committee (the ‘‘EMSAC’’). See Proposing Release, supra note 2, at 13009, 13012–14. 7Under the Exchange Act, exchange fee changes are effective on the day that the exchange files them with the Commission, and neither advance notice nor Commission action is required before an exchange may implement a fee change. See 15 U.S.C. 78s(b)(3)(A)(ii). The Commission may, within 60 days after an exchange filed its fee change with the Commission, summarily suspend the new fee and institute proceedings to determine whether to disapprove it. See 15 U.S.C. 78s(b)(3)(C). 8See 15 U.S.C. 78f(b)(4)–(5). attracted considerable attention and generated significant debate, focusing on the effects, both positive and negative, that exchange transactionbased pricing models may have on market quality and execution quality, with some commenters advocating action by the Commission. The Commission is uniquely situated and vested with the responsibility under
Section 11A of the Exchange Act to
examine the impact that this aspect of our market structure has on our national market system. And, in light of the questions raised about the impact of these fee models and the amount of attention garnered, we believe this is an area ripe for Commission review. But, the Commission currently lacks the data necessary to meaningfully analyze the impact that exchange transaction feeand-rebate pricing models have on order routing behavior, market and execution quality, and our market structure generally. To address this information gap, the Commission has designed the Pilot to produce data that will facilitate a more thorough understanding of the potential issues associated with exchange transaction-based pricing models. In particular, the Commission has designed the Pilot to gather data on the effect both current regulatory fee caps and rebates have on market quality and execution quality. The data gathered will assist the Commission in determining whether any changes in the current regulatory framework are appropriate and enable the Commission to make more informed and effective policy decisions. This, in turn, enables the Commission to carry out the objectives of the national market system and oversee the national securities exchanges. As discussed fully in the proposing release, the Commission proposed a pilot to test the effect of exchange transaction fees and rebates.5 The following chart summarizes the terms of the Pilot as adopted, which are discussed in more detail below:
TRANSACTION FEE PILOT FOR NMS STOCKS Duration ................. 2 years with an automatic sunset at 1 year unless, no later than 30 days prior to that time, the Commission publishes a notice that the pilot shall continue for up to 1 additional year; plus a 6-month pre-Pilot Period and 6-month post-Pilot Period. Applicable trading centers. Equities exchanges (including maker-taker & taker-maker) but not ATSs or other non-exchange trading centers. Pilot securities ....... NMS stocks with average daily trading volumes ≥30,000 shares with a share price ≥$2 per share that do not close below $1 per share during the Pilot and that have an unlimited duration or a duration beyond the end of the post-Pilot Period. Group Number of NMS stocks Fee cap Rebates permitted? Pilot design ........... Test Group 1 ..... 730 ........................ $0.0010 fee cap for removing and providing displayed liquidity (no cap on rebates). Yes. Test Group 2 ..... 730 (plus appended Canadian interlisted stocks). The 17 CFR 242.610(c) (Rule 610(c)) $0.0030 cap continues to apply to fees for removing displayed liquidity. No. Rebates and Linked Pricing Prohibited for removing and providing displayed and undisplayed liquidity (except for specified market maker activity). Control Group .... Pilot Securities not in Test Groups 1 or 2. The Rule 610(c) cap continues to apply to fees for removing displayed liquidity (no cap on rebates). Yes. Pilot data ............... 1. Pilot Securities Exchange Lists and Pilot Securities Change Lists.
2. Exchange Transaction Fee Summary.
3. Order Routing Datasets.
II. Discussion of Rule 610T
In response to its proposal to conduct a Transaction Fee Pilot in NMS stocks (the ‘‘Pilot’’), the Commission received a number of comment letters from a diverse group of commenters, including exchanges, investment managers, broker-dealers, and other market participants, as well as academics, listed issuers, analytics firms, market observers, and industry associations.6 As discussed below, after review and consideration of the comments received, the Commission is adopting Rule 610T with certain modifications from that in the proposal. A. Focus on Exchange Pricing Models and the Effects They Can Cause

Exchange Fee Models and Regulatory
Framework
Regardless of the fee model, all fees of a registered national securities exchange ‘‘exchange’’) are subject to the standards and process requirements set forth in the federal securities laws.7 In particular, Section 6 of the Exchange Act requires, among other things, that the rules of an exchange provide for the ‘‘equitable allocation’’ of ‘‘reasonable’’ fees and that they not be ‘‘designed to permit unfair discrimination.’’ 8 Section 11A of the Exchange Act directs the Commission to use its authority to facilitate the establishment of a national market system for securities that assures economically efficient execution of securities transactions, fair competition, availability of information with respect

9See 15 U.S.C. 78k–1(a)(1).
10 17 CFR 242.610(c); Securities Exchange Act Release No. 51808 (June 9, 2005), 70 FR 37496, 37543–46 (June 29, 2005) (‘‘NMS Adopting Release’’). See also 17 CFR 242.600(b)(58) (defining ‘‘protected quotation’’); 17 CFR 242.600(b)(57) (defining ‘‘protected bid or protected offer’’); 17 CFR 242.600(b)(3) (defining ‘‘automated quotation’’). 11NMS Adopting Release, supra note 10, at 37545. 12At the time of its adoption in 2005, the fee cap codified the then-prevailing fee level set through competition among the various trading centers. See NMS Adopting Release, supra note 10, at 37545 (stating that ‘‘the $0.003 fee limitation is consistent with current business practices, as very few trading centers currently charge fees that exceed this amount’’). 13The potential distortions mentioned by the commenters (and discussed in this section) include, among others: (1) Conflicts of interest faced by routing broker-dealers; (2) excess intermediation and potential adverse selection; (3) market fragmentation; (4) exchange fee avoidance; (5) complexity; (6) transparency; and (7) elevated fees to subsidize rebates. 14Capital Group Letter, at 2. See also, e.g., ICI Letter I, at 2; Vanguard Letter, at 2; Invesco Letter, at 2; CFA Letter, at 2; Oppenheimer Letter, at 2; Spatt Letter, at 4; AJO Letter, at 1; Larry Harris Letter, at 3. 15Healthy Markets Letter I, at 5. See also, e.g., Copeland Letter, at 1; Wellington Letter, at 1; Norges Letter, at 2. 16Babelfish Letter, at 1–3 (also referencing a Clearpool Group study that found that a ‘‘fee sensitive VWAP algorithm executed during volatile times incurred seven times as much cost as a fee agnostic algorithm’’). See also T. Rowe Price Letter, at 2 (stating that ‘‘[r]etail orders . . . are generally placed on the exchange that offers the highest rebate to the broker, but show[s] lower execution quality in terms of reduced probability of execution’’); Capital Group Letter, at 2 (‘‘Our internal trade analysis suggests that execution quality may be negatively impacted when brokerdealers’ routing decisions are made to minimize access fees.’’). 17 IEX Letter I, at 6, A–1–A–2; IEX Letter II, at 7; IEX Letter IV (appending research to support these views). See also, e.g., Babelfish Letter, at 2 (stating that a ‘‘frequently realized scenario is that flow sent solely to a high rebate destination waits in queue, often winds up canceled because price moves away, and then receives an inferior price upon the eventual execution’’); Larry Harris Letter, at 1, 3; Brandes Letter, at 1–2. But see Grasso Letter, at 3 (‘‘waiting for a rebate[ ] may be fine’’ if ‘‘you have low confidence about future prices for a large order and don’t mind if the order trades slowly while you accumulate shares’’). 18See, e.g., ICI Letter I, at 2. 19Credit Suisse Commentary, at 2. See also, e.g., Larry Harris Letter, at 3 (noting that ‘‘orders standing at inverted exchanges usually execute before orders standing at the same price at makertaker
exchanges’’). 20Capital Group Letter, at 2. See also, e.g., IEX Letter I, at 3 (‘‘Excessive take fees . . . have been criticized as leading to the migration of some order flow to less-regulated non-exchange venues in search of reduced transaction costs, resulting in increased market fragmentation and market complexity.’’). 21 ICI Letter I, at 2. See also, e.g., Vanguard Letter, at 2 (indicating that the ‘‘desire to maximize rebate revenue and avoid fees created order complexity within the equity markets as traders sought profitable trading strategies’’). 22 ICI Letter I, at 2. See also, e.g., Goldman Sachs Letter, at 3; Invesco Letter, at 2; State Street Letter, to quotations for and transactions in securities, and the practicability of brokers executing investors’ orders in the best market.9 In addition, Rule 610(c) of Regulation NMS imposes upon exchanges a fee cap of $0.0030 per share for the execution of an order against its ‘‘protected quotation.’’ 10 In 2005, when it adopted the fee limitation in Rule 610(c), the Commission noted, in part:
The adopted fee limitation set forth in Rule 610(c) of Regulation NMS is designed to preclude individual trading centers from raising their fees substantially in an attempt to take improper advantage of strengthened protection against trade-throughs and the adoption of a private linkage regime. In particular, the fee limitation is necessary to address ‘outlier’ trading centers that otherwise might charge high fees to other market participants required to access their quotations by the Order Protection Rule. It also precludes a trading center from charging high fees selectively to competitors, practices that have occurred in the market for Nasdaq stocks. In the absence of a fee limitation, the adoption of the Order Protection Rule and private linkages could significantly boost the viability of the outlier business model. Outlier markets might well try to take advantage of intermarket price protection by acting essentially as a toll booth between price levels. The high fee market likely will be the last market to which orders would be routed, but prices could not move to the next level until someone routed an order to take out the displayed price at the outlier market.11 In light of the considerable debate surrounding exchange fee models that pay rebates, which is well documented in the comment letters submitted on the proposed Pilot, and the passage of time since the Commission first adopted the Rule 610(c) fee cap as part of Regulation NMS in 2005, the Commission now seeks to gather data to facilitate an empirical assessment of the effect of exchange transaction fees and rebates broadly—including the impact and continued appropriateness of the Rule 610(c) fee cap 12—by testing the effects of changes to exchange fees and rebates on the markets and market participant behavior.
2. Impact of Exchange Fee Models
In response to the Proposing Release, the Commission received a number of comment letters criticizing existing feeand-rebate pricing models, but also a number of comment letters expressing support for those same pricing regimes.13 Many commenters focused on one potential distortion—whether current pricing models ‘‘present broker-dealers with a potential conflict of interest,’’ because their ‘‘duty to pursue best execution could be compromised when their trading venue decision is driven by the economic incentive to minimize access fees paid and maximize rebates received.’’ 14 As another commenter explained, ‘‘a broker is incentivized to route an order to the venue that pays it the most (or costs the least), instead of the venue that has the highest likelihood of offering the best execution for its customers, such as the one that offers a higher probability of execution or meaningful price improvement.’’ 15 As evidence of the potential harm that can result from the conflicts presented by exchange rebates, one commenter noted that institutional investors ‘‘that specifically instruct brokers to remove rebate-driven trading behaviors from their algorithms achieve significantly lower trading costs that result in higher returns to their investors.’’ 16 One commenter attributed this harm to the tendency of rebates to ‘‘affect the length of the order queue of passive limit orders on the major maker-taker exchanges, while high take fees on these markets make them less attractive for marketable orders that cross the spread.’’ The commenter argued that the ‘‘net result of this perverse pricing dynamic is a lower likelihood of execution and a higher likelihood of adverse selection for orders in the maker-taker queues,’’ because orders at the ‘‘middle or back of the queue . . . are less likely to trade at their desired price, and when they do trade, the overall market price as reflected by the [National Best Bid and Offer (‘‘NBBO’’)] is more likely to move against them, than when trading on venues that do not pay rebates.’’ 17 A number of commenters discussed other potential effects of exchange pricing models. Some commenters believed that transaction fees and rebates contribute to market fragmentation 18 because they encourage investors to ‘‘turn to inverted markets to improve queue priority’’ 19 or to ‘‘route orders to non-exchange trading centers to avoid the higher access fees that exchanges charge to subsidize the rebates they offer.’’ 20 Likewise, one commenter thought that ‘‘transaction fees and rebates contribute to market complexity through the proliferation of new order types . . . designed to exploit different transaction pricing models.’’ 21 Other commenters believed that ‘‘[t]ransaction fees and rebates . . . undermine market transparency because the prices displayed by exchanges—and provided on trade reports—do not include fee or rebate information and therefore do not fully reflect net trade prices.’’ 22 Finally, some commenters

at 2; Wellington Letter, at 1; Oppenheimer Letter, at 2; Capital Group Letter, at 3. 23See, e.g., Clearpool Letter, at 3 (stating that ‘‘exchanges chase order flow and provide rebates and other pricing incentives to the largest trading firms at the expense of smaller market participants who cannot take advantage of such rebates and, in effect, end up subsidizing the trading of larger firms’’); IEX Letter I, at 3 (stating that transaction fees are ‘‘used in effect to subsidize the payment of rebates,’’ which ‘‘results in a substantial penalty on investors and other participants who . . . have a need for immediate liquidity’’). 24See, e.g., T. Rowe Price Letter, at 2 (stating that rebates lead to ‘‘excessive intermediation . . . benefiting short-term intermediaries at the expense of long-term investors’’); ModernIR Letter, at 3 (stating that rebates ‘‘promote[ ] arbitrage, and pricesetting as its own end,’’ leading to a ‘‘paucity of real orders’’); Larry Harris Letter, at 1, 5–6 (stating that current pricing models facilitate ‘‘the execution of various parasitic trading strategies by proprietary traders to the detriment of public investors’’); Capital Group Letter, at 3. 25State Street Letter, at 2. See also, e.g., Virtu Letter, at 3; Fidelity Letter, at 3; Nasdaq Letter I, at 9; Cboe Letter I, at 15–16. See also Nasdaq Letter III, at Exhibit A (providing graphs using data from September 2018 on average quoted spread across exchanges in S&P 500 stocks and time at the best quote across those stocks). But cf. Larry Harris Letter, at 6–9 (acknowledging that ‘‘quoted spreads are narrower under maker-taker pricing,’’ but opining that ‘‘the narrower quoted spreads do not benefit the public’’). 26Magma Letter, at 3. See also, e.g., NYSE Letter IV, at 2 (arguing that ‘‘pricing incentives enhance the quality and reliability of display markets’’); FIA Letter, at 4. 27FIA Letter, at 3–4. See also NYSE Letter I, at 6 (stating that rebates ‘‘allow liquidity providers to quote narrower spreads by providing another source of revenue’’); Grasso Letter, at 4 (‘‘the main outcome of exchange pricing seems to be that it forces exchanges to compete for customers,’’ because it ‘‘keeps their margins tight and gives them incentives to improve the quality of their offerings’’). 28See, e.g., Joint Asset Managers Letter, at 2; Brandes Letter, at 2; Themis Trading Letter I, at 3; AJO Letter, at 1–2; OMERS Letter, at 2; Copeland Letter, at 2; Virtu Letter, at 6; Nuveen Letter, at 2; BlackRock Letter, at 1; RBC Letter I, at 3; Vanguard Letter, at 2; CFA Letter, at 4; Wellington Letter, at 2; Joint Pension Plan Letter, at 2; Oppenheimer Letter, at 2; Clearpool Letter, at 5 n.8; TD Ameritrade Letter, at 4; Capital Group Letter, at 3; Healthy Markets Letter I, at 10; Morgan Stanley Letter, at 3 n.5; AGF Letter, at 1. 29AJO Letter, at 1–2. 30See RBC Letter I, at 3–4. 31Capital Group Letter, at 3. See also, e.g., Clearpool Letter, at 5 n.8; Oppenheimer Letter, at 2; Brandes Letter,
at 2; Copeland Letter, at 2. 32See, e.g., Cboe Letter I, at 28. 33See Proposing Release, supra note 2, at 13014. As discussed in the Proposing Release, the term ‘‘trading center’’ as used there and throughout this release is a collective term that refers broadly to the venues that trade NMS stocks. See id. at 13009 n.7. For purposes of this release, the term ‘‘trading center’’ includes national securities exchanges that are registered with the Commission and that trade NMS stocks (referred to herein as ‘‘equities exchanges’’ or ‘‘exchanges’’), as well as other types of ‘‘non-exchange venues’’ that trade NMS stocks, including ATSs and broker dealers that internalize orders by matching them off-exchange with reference to the national best bid and offer. 34See, e.g., Brandes Letter, at 2; AJO Letter, at 2; MFA Letter, at 2; BIDS Letter, at 1–2; BlackRock Letter, at 1; SIFMA Letter, at 5; Virtu Letter, at 6; Fidelity Letter, at 10; Citi Letter, at 2; Clearpool Letter, at 4–5; Luminex Letter, at 1; Morgan Stanley Letter, at 3 n.5. 35 Virtu Letter, at 6. See also, e.g., SIFMA Letter, at 5; Clearpool Letter, at 5. asserted that current pricing models unfairly subsidize rebates 23 or benefit sophisticated market participants like market-makers and proprietary traders at the expense of other market participants.24 Other commenters expressed support for current exchange pricing models. For example, one commenter believed that maker-taker pricing ‘‘provides important benefits to issuers and investors,’’ because exchanges ‘‘use rebates as a tool to promote displayed liquidity and price discovery, which results in competitive bid-ask spreads, saving transaction costs that investors may otherwise incur.’’ 25 Another commenter argued that rebates can promote displayed liquidity by providing ‘‘a payment in exchange for posters of liquidity giving up several valuable options,’’ including ‘‘the power to decide the time of the trade’’ and the ability to conceal trading intentions until the point of execution.26 Building on this idea, one commenter characterized ‘‘[a]ccess fee caps and related rebates’’ as features that ‘‘enable exchanges to compete with nonexchange trading venues by essentially subsidizing the posted prices . . . and narrow[ing] the NBBO, making it slightly more expensive to either match or improve upon those prices offexchange.’’ 27 As commenters fundamentally disagreed about the effect of exchange transaction fee models and whether they have a positive or a negative impact on the U.S. equities markets, commenters also held conflicting views regarding whether and how the Commission should conduct the Pilot.
3. Focus on Exchange Fee Models
Recognizing the unique regulatory framework applicable to exchange fees, and the disagreement over the impact of exchange fees and rebates on the markets and market participants, the Commission focused its proposed Pilot on studying the effect of exchange transaction fees and rebates on order routing behavior, execution quality, and market quality. Accordingly, the Commission proposed to include within the Pilot all equities exchanges regardless of fee model. A large number of commenters supported applying the Pilot to all equities exchanges.28 For example, one commenter believed that the Pilot ‘‘should include all equities exchanges . . . because rebates of any kind provide inducements to trade and distort markets.’’ 29 A different commenter thought that including taker-maker exchanges was ‘‘both logical and feasible, given that all equities exchanges assess fees that are subject to the Exchange Act and its rule filing requirements.’’ 30 Other commenters ‘‘agree[d] with the Commission’s assessment that the Pilot should apply to all equity exchanges . . . thus treating all similarly situated exchanges equally,’’ because this would be ‘‘critically important in determining what impact the reduction of access fees or the elimination of rebates will have on order routing practices.’’ 31 Some other commenters, however, opposed including taker-maker exchanges in the Pilot, noting that Rule 610(c) does not apply to taker-maker exchanges.32 After considering the comments on this issue, the Commission continues to believe that focusing the Pilot on equities exchanges regardless of fee model is appropriate because it treats alike similarly situated entities that all are subject to the same regulatory framework and thereby will allow the Commission to evaluate the effect of exchange fee-and-rebate pricing models and the continued appropriateness of the Rule 610(c) fee cap. Further, it would be incongruous to study rebates and fees offered by one type of equities exchange (maker-taker), but not another type of equities exchange (taker-maker) where the fees of both types of entities are subject to the same legal requirements and can introduce the same types of distortions that the Pilot seeks to study.
4. Non-Exchange Trading Centers
As proposed, the Pilot would exclude non-exchange trading centers such as alternative trading systems (‘‘ATSs’’).33 Several commenters opined on this aspect of the proposal. A number of commenters agreed with the Commission’s proposal to exclude nonexchange trading centers from the Pilot.34 Some of those commenters noted that exchanges are subject to various fee-related regulatory provisions that are entirely inapplicable to nonexchange trading centers. For example, one commenter noted that nonexchange trading centers are not currently subject to any access fee caps, and including such trading venues in the Pilot ‘‘would have the unintended and harmful effect of unnecessarily changing ATS business models . . . .’’ 35 In addition, several commenters emphasized the fundamental ways in which the fee structures employed by

36See, e.g., Morgan Stanley Letter, at 3 n.5 (stating that ‘‘many broker-dealer[ ] operators of ATSs generally charge clients an overall commission rate (rather than an access fee) for a bundle of services, including access to their ATSs’’); BIDS Letter, at 1–2, AJO Letter, at 2; Healthy Markets Letter I, at 10. 37AJO Letter, at 2. 38Citi Letter, at 2. See also, e.g., Fidelity Letter, at 10 (stating that ‘‘ATS’ fee structures are already subject to competitive market forces and have more complex pricing models than exchanges[,] making their participation in the Proposed Pilot less useful’’); SIFMA Letter, at 5 (opining that ‘‘competitive forces already push access fees [at ATSs] to an appropriate level . . . lower than the access fees charged by exchanges,’’ because ATS access fees ‘‘are included in the total cost consideration of trading’’). 39Luminex Letter, at 1. 40See, e.g., Nasdaq Letter I, at 2, 5–7; Cboe Letter I, at 12–13; MFS Letter, at 2; RBC Letter I, at 4; ASA Letter, at 3; ViableMkts Letter, at 2; Angel Letter II, at 2. 41See Wellington Letter, at 2 (acknowledging, however, that it is ‘‘impractical for the Commission to include off-exchange venues’’). See also, e.g., RBC Letter I, at 4; ProAssurance Letter, at 2. 42Nasdaq Letter I, at 2, 5–7. See also, e.g., NYSE Letter I, at 2. 43See Nasdaq Letter I, at 7. 44See, e.g., Cboe Letter I, at 13. 45See NYSE Letter I, at 7–8. 46See, e.g., Better Markets Letter, at 8. 47As noted by several commenters, equities exchanges and non-exchange trading centers currently employ different fee models. While equities exchanges charge transaction-based fees, non-exchange trading centers may not charge separate transaction-based fees, but instead may use bundled pricing such that a particular order is not necessarily associated with a particular fee. See, e.g., Morgan Stanley Letter, at 3 n.5 (stating that ‘‘many broker-dealer[ ] operators of ATSs generally charge clients an overall commission rate (rather than an access fee) for a bundle of services, including access to their ATSs’’); BIDS Letter, at 1– 2, AJO Letter, at 2. See also Proposing Release, supra note 2, at 13016. The Commission is not aware of any ATSs that currently pay transactionbased rebates. 48See supra notes 310–312 and accompanying text (discussing recent amendments to Regulation ATS and their relevance to the proposed Pilot). 49See, e.g., ASA Letter, at 3; Cboe Letter I, at 12, 26–27; Nasdaq Letter I, at 5–7; NYSE Letter I, at 3–
8. 50See Cboe Letter I, at 12. See also Nasdaq Letter
I, at 6; NYSE Letter I, at 3–5; NYSE Letter II, at 12. 51See, e.g., Wellington Letter, at 2; Oppenheimer Letter, at 3; Angel Letter II, at 2; Nasdaq Letter I, at 6–7; Cboe Letter I, at 12; NYSE Letter I, at 3–5; Curtiss-Wright Letter, at 1; ASA Letter, at 3. non-exchange trading centers are different from the fee models utilized by the equities exchanges and, as a result, concluded that excluding non-exchange trading centers was appropriate.36 For example, one such commenter explained that ‘‘inducements (low fees, no fees, rebates) offered by ATSs and other off-exchange venues are not universal across all broker-dealers or market participants. Instead, the fees paid (or not paid) by market participants to ATSs and other off-exchange venues are negotiated between each market participant and the trading venue,’’ such that ‘‘the number of fee permutations and inconsistencies across brokers for any single ATS could be substantial.’’ 37 Still other commenters believed that excluding non-exchange trading centers from the Pilot was appropriate because ‘‘ATSs are not protected venues, and thus free market competition among them constrains their pricing power.’’ 38 One commenter supported excluding ATSs because ‘‘there is nothing to be gained by including venues that don’t have the same underlying issues that exchanges present with their rebate and ‘maker-taker’ pricing models.’’ 39 On the other hand, other commenters expressed concerns with omitting nonexchange venues from the Pilot.40 One concern was that by excluding nonexchange venues, the Pilot data would be incomplete. For example, one commenter believed that excluding nonexchange venues ‘‘could create an imperfect picture of the overall impact of the transaction fees put in place under the Pilot program’’ and could compromise the value and utility of the data collected during the Pilot. 41 Another commenter argued that by excluding non-exchange venues, the Pilot will not return ‘‘meaningful data upon which to make informed analysis and conclusions’’ because it would ‘‘ignore off-exchange trading representing approximately 39 percent of total U.S. equities market trading.’’ 42 This commenter further believed that the Pilot would be unable to properly assess the potential conflicts of interest because it will not know ‘‘the baseline for remuneration occurring offexchange, or know what impact the Proposal has on that baseline[.]’’ 43 One commenter objected to excluding ATSs ‘‘based on the fact that the proposed Pilot is a ‘new regulatory regime’ for ATSs . . . .’’ 44 While one commenter recognized the complexity involved with subjecting non-exchange trading centers to the access fee cap under Rule 610(c), it argued that such complexity did not provide a sufficient basis to treat exchanges and non-exchange trading centers disparately.45 A few commenters recommended excluding ATSs, but requiring them to submit the required order routing data.46 The Commission believes that excluding non-exchange
venues from the Pilot should not negatively impact the Pilot’s data or impact its results. As noted above, the Pilot is designed, among other things, to assess the effects of exchange fee models. Because exchange fee models are materially different both in their structure and regulatory treatment, the potential effects that may be associated with exchange fee models are not applicable in the same manner to ATSs. Similarly, the question of whether rebates narrow the quoted spread is inapplicable to ATSs, which do not publicly display an automated quotation. Further, ATS activity is not being overlooked as increases or decreases in ATS volume during the Pilot will be reflected in other existing data sources. Accordingly, Commission researchers (hereinafter ‘‘researchers’’) will be able to assess market-wide changes in order flow during the Pilot. Further, even if non-exchange venues provided order routing data pursuant to the Pilot, researchers would be unable to meaningfully correlate changes in an ATS’s order flow with the fees of that ATS because those fees are bespoke, typically bundled, and are not as transparent as exchange fees.47 Exchange fees are not only fully transparent in published fee schedules, but exchange fee changes must be filed with the Commission and thus they have a precise effective date attached to each filing. This level of transparency for exchange fees and rebates, which is not present for ATSs,48 is an important component facilitating researchers’ ability to draw causal connections with the Pilot’s results. While obtaining order routing data from ATSs might provide interesting insight into their business, it could not be meaningfully correlated with ATS fees and fee changes and is not necessary to study the Pilot’s results. Rather, existing sources of data on ATS activity, including data published by the Financial Industry Regulatory Authority (‘‘FINRA’’), will permit researchers to observe changes in ATS activity during the Pilot. Among commenters critical of excluding non-exchange venues, some believed it could raise competitive issues to apply the Pilot’s pricing limitations to the equities exchanges, but not impose the same pricing limitations on non-exchange trading centers that trade the same equities securities.49 One exchange commenter found it ‘‘inexplicabl[e]’’ that the Pilot ‘‘focuses only on exchanges and entirely ignores off-exchange venues, which are the venues that are most likely to benefit from a pilot that pointedly decreases the incentive (i.e., rebates) to post protected quotes on-exchange.’’ 50 Several commenters suggested that the exclusion of non-exchange trading centers from the Pilot could ‘‘create incentives for market participants to move more order flow to off-exchange platforms,’’ thereby putting the national securities exchanges at a competitive disadvantage as compared to offexchange trading centers.51 However, a

52See, e.g., Citi Letter, at 2; Decimus Letter, at 5– 6. 53See, e.g., Citi Letter, at 2; Decimus Letter, at 5–
6. See also, infra Section IV.D ‘‘Impact on
Efficiency, Competition and Capital Formation’’ and note 782 infra and accompanying text. 54See Rule 610T(a)(2). 55See infra Section IV.A.2. and C.1.a.i. 56See 15 U.S.C. 78f(b)(4)–(5). 57All exchange fee changes are published for public comment and required to be publicly posted on the internet, whereas fees of non-exchange trading centers are typically bespoke. Fee changes of non-exchange trading centers are not subject to the provisions of the federal securities laws requiring that fees be an ‘‘equitable allocation’’ of ‘‘reasonable’’ fees and not ‘‘unfairly discriminatory.’’ 58See 15 U.S.C. 78f(b)(4)–(5) (requiring, among other things, that an exchange’s fees be an ‘‘equitable allocation’’ of ‘‘reasonable’’ fees and that they not be ‘‘designed to permit unfair discrimination.’’). In addition, only exchange fees are subject to the rule filing requirements under
Section 19(b) of the Exchange Act and 17 CFR
240.19b–4 (Rule 19b–4) thereunder. See also Proposing Release, supra note 2, at 13016. 59While exchange fees are filed with the Commission on Form 19b–4 and the Commission publishes notice of them for public comment and has an opportunity to summarily suspend them within 60 days, the Commission’s non-action on a fee filing within that period does not constitute an endorsement or approval of an exchange fee. Issues with fees and how they impact market participants and market structure may or may not be obvious at first and adverse effects may take time to manifest as the market adjusts to a new fee. The Commission, and the exchanges as self-regulatory organizations, must enforce their rules and the federal securities laws with the goal of protecting investors and the public interest. 60See Proposing Release, supra note 2, at 13015. 61See, e.g., MFA Letter, at 2; SIFMA Letter, at 5; Fidelity Letter, at 10. 62See Proposing Release, supra note 2, at 13015. 63See 17 CFR 242.600(b)(47) (defining ‘‘NMS stock’’). 64See Proposing Release, supra note 2, at 13017. See also Proposed Rule 610T(b)(1)(ii). 65See Proposing Release, supra note 2, at 13017; Proposed Rule 610T(b)(1)(ii). The Commission notes that the proposed language in Rule 610T(b)(1)(ii) has been modified slightly. As proposed, Rule 610T(b)(1)(ii) contained the phrase Continued commenter suggested the opposite could happen and that the Pilot might actually ‘‘encourage more order flow to gravitate to the exchanges’’ because the Pilot would reduce the access fee cap on the equities exchanges thereby making it less expensive to transact on an exchange.52 The Commission does not believe that the Pilot necessarily will put the equities exchanges at a competitive disadvantage or disproportionally harm them when competing with nonexchange trading centers for investors’ orders. Currently, only exchanges are subject to the Rule 610(c) fee cap, and Test Group 1 is designed to test a lower cap. The Commission does not believe that exchanges charging lower fees will necessarily make them less competitive with other venues for natural order flow, for example order flow that removes liquidity. Rather, it is possible that lower fees in Test Group 1 across all exchanges may actually improve their competitive position in attracting that order flow,53 particularly with respect to fee sensitive routing algorithms because, all else being equal, fee sensitive algorithms generally seek to minimize trading costs and would likely rank exchanges more favorably in their routing tables when exchanges reduce their fees to remove liquidity. In addition to testing a lower fee cap level, the Pilot also will test a prohibition on rebates and ‘‘Linked Pricing,’’ which, as discussed further below, is defined as a discount or incentive on transaction fee pricing applicable to removing (or providing) liquidity that is linked to providing (or removing) liquidity.54 The intent of this is to gather data to assess, among other
things, the effect of exchange rebates. Potential distortions, which may be caused or exacerbated by exchange rebates, may themselves be placing exchanges at a competitive disadvantage, in which case the elimination of rebates could improve the competitive position of exchanges, for example if taker fees are set at levels independent of the need to subsidize maker rebates. Once again, data is needed to empirically assess this issue, and the Commission believes that the Pilot is the best way to obtain that data.55 Further, while exchanges may compete with non-exchange trading centers for order flow, exchange fees and the fees of non-exchange trading centers are treated very differently under the federal securities laws. Indeed, one of the distinguishing features of registered national securities exchanges is that—unlike non-exchange trading centers—their fees are subject to the principles-based standards set forth in the Exchange Act, as well as the rule filing requirements thereunder. In particular, the federal securities laws require the entirety of each and every fee, due, and charge assessed by an exchange to be transparent and publicly posted for all to see, and must be an equitable allocation of reasonable dues, fees and other charges and not be unfairly discriminatory.56 On the other hand, similar requirements do not apply to the fees of non-exchange trading centers that do not provide public transparency into their full itemized fee schedules and typically are individually negotiated on a customer-by-customer basis.57 By including all equities exchanges regardless of fee model, and excluding other types of trading centers, the Pilot is designed to include all trading centers whose fees are subject to the principles-based standards set forth in the Exchange Act as well as the rule filing requirements thereunder.58 Thus, the Pilot will produce data to empirically evaluate the effects that transaction-based fees and rebates may have on, and the effects that changes to those fees and rebates may have on, order routing behavior, execution quality, and market quality more generally. The Commission believes that subjecting non-exchange trading centers to the Pilot would go beyond the scope of the current regulatory framework that applies only to exchanges and would not further the Commission’s evaluation of the impact of the existing regulatory regime, including, but not limited to, the Regulation NMS fee cap, which applies exclusively to exchange fees and rebates. In effect, the Pilot will help the Commission carry out its statutory responsibility to assess the effect of exchange fees and rebates, which do not apply to non-exchange trading centers.59
5. Options Exchanges
Finally, the Commission proposed to exclude options exchanges from the Pilot, because options and equities are materially different types of securities. In addition, the access fee cap under Rule 610(c) does not currently apply to the options exchanges.60 Several commenters agreed with the Commission’s exclusion of the options exchanges.61 No commenters suggested that the Commission include options markets in the Pilot. For the reasons noted above and discussed in the Proposing Release, the Commission is not including options markets within the scope of the Pilot.62 B. Securities As proposed, all NMS stocks 63 that meet specified initial and continuing minimum standards would be eligible for inclusion in the Pilot (collectively, ‘‘Pilot Securities’’).64 The Commission received a number of comments regarding the scope of Pilot Securities to be included in the Pilot.

The Share Price Threshold of Pilot
Securities
The Commission proposed that an NMS stock must have a minimum initial share price of $2 at the time the prePilot Period commences to be included in the Pilot and that any Pilot Securities that close below $1 at the end of a trading day during the proposed Pilot would be removed from the Pilot.65

‘‘minimum initial share price of at least $2 . . . .’’ As adopted, the clause ‘‘minimum initial share price of $2’’ is being substituted for the phrase ‘‘minimum initial share price of at least $2’’ to delete redundant text. In addition, as proposed, Rule 610T(b)(1)(ii) explained that a Pilot Security that closes below $1 would be ‘‘removed from the Test Group or the Control Group and will no longer be subject to the pricing restrictions set forth in (a)(1)–(3). . . .’’ As adopted, this language is being modified slightly to make it more concise. Accordingly, as adopted, this language provides that if the share price of a Pilot Security closes below $1 at the end of a trading day ‘‘it will be removed from the Pilot.’’ 66See Angel Letter I, at 2. 67RBC Letter I, at 5. See also, e.g., Better Markets Letter, at 6; Healthy Markets Letter I, at 11–12. 68RBC Letter I, at 5. 69Healthy Markets Letter I, at 12. 70See Proposing Release, supra note 2, at 13017. 71See id. at 13017 n.102 (noting that only 4.3% of publicly traded common stocks and ETPs with a share price above $2 during 2012–2016 dropped below $1 in that period). 72See Proposing Release, supra note 2, at 13017; Proposed Rule 610T(b)(1)(ii). 73See Proposing Release, supra note 2, at 13018 n.103. 74See id. at 13018. The EMSAC’s recommendation was to limit a pilot to stocks above $3 billion in market capitalization in order to avoid overlap with the Tick Size Pilot. See id. The Commission notes, however, that the Tick Size Pilot ended on September 28, 2018 and the Pilot Period for the Transaction Fee Pilot will not start before the post-pilot period for the Tick Size Pilot ends on April 2, 2019. See Section II.C.3. infra. 75RBC Letter I, at 6. See also, e.g., Harris Letter, at 1; T. Rowe Price Letter, at 4. 76Cboe Letter I, at 28. 77 Id. See also, e.g., Morgan Stanley Letter, at 4; Leaf Letter, at 1. 78Cboe Letter I, at 19. See also, e.g., Proposing Release, supra note 2, at 13069. 79Nasdaq Letter I, at 8–9. 80 Id. at 3, 9 (alleging that the Pilot was ‘‘arbitrary and capricious and not in accordance with law,’’ because it gave ‘‘short shrift’’ to these concerns). See also Virtu Letter, at 7 (expressing concern that the Pilot would ‘‘harm investors in . . . less liquid ETPs, which will be faced with less liquidity and wider spreads when they seek to sell their holdings’’). One commenter opposed the $2 initial minimum share price threshold as overly restrictive.66 Other commenters, however, agreed that the securities in the Pilot should have an initial minimum $2 per share price threshold at the time of the initial stock selection, because this threshold ‘‘will capture virtually all NMS stocks while minimizing the risk that securities will drop out of the Pilot . . . .’’ 67 One of these commenters believed the proposed thresholds would ‘‘help ensure consistency among the Test Groups and limit the risk of data anomalies due to changes in the composition of those groups.’’ 68 Another commenter noted
that the choice of ‘‘$2 and $1 thresholds . . . follows the reasonable parameters established during [the] . . . Tick Size Pilot’’ and asserted that the ‘‘determination to pull out securities that close at under $1 during the pilot seems appropriate, especially given the fundamentally different fee structures applicable to stocks with prices less than $1.00.’’ 69 The Commission continues to believe that the proposed share price thresholds for Pilot Securities are appropriate. The Commission notes that no commenters opposed the proposed $1 minimum continuing price threshold, which will exclude such stocks from the Pilot because stocks with quotations of less than $1 are subject to different regulatory and fee treatment.70 The Commission continues to believe that an initial $2 share price threshold will best balance the need to include a broad set of NMS stocks in the Pilot with the desire to ensure that substantially all of the securities selected at the outset of the Pilot remain part of their respective Test Groups throughout the duration of the Pilot, including during the pre- and post-Pilot periods. The Commission does not believe that the $2 threshold is overly restrictive because, as discussed in the Proposal, it is uncommon for securities priced at $2 or more to fall below $1.71 Lowering the initial stock selection threshold below $2 could increase the likelihood that securities selected for the Pilot get dropped from the Pilot if their share price closed below $1 during the Pilot. Such a result would change the composition of the Test Groups during the Pilot, which might adversely impact the quality of the data produced by the Pilot. For these reasons and the reasons discussed in the Proposing Release, the Commission adopts as proposed the share price thresholds set forth in Rule 610T(b)(1)(ii).
2. The Duration of Pilot Securities
The Commission proposed that, in order to be included in the Pilot, an NMS stock must have an unlimited duration or a duration beyond the end of the post-Pilot period in order to be included in the Pilot.72 No comments were received regarding this condition. For the reasons outlined in the Proposing Release, the Commission adopts this aspect of the Pilot as proposed.73
3. Selecting Pilot Securities From All
NMS Stocks
The Commission proposed to select Pilot Securities from among the entire universe of NMS stocks, subject to the minimum share price threshold and duration requirements. As proposed, the Pilot would include a broad and diverse cross-section of securities, including, for example, stocks of all market capitalizations as well as ETPs. The Commission received comments on the universe of Pilot Securities that generally fell into four categories: (1) The inclusion of stocks with market capitalizations below $3 billion, (2) the inclusion of ETPs, (3) the inclusion of Canadian interlisted stocks, and (4) the inclusion of NMS stocks other than stocks of operating companies and ETPs. Each of these points is discussed below. a. Market Capitalization and Liquidity The Commission proposed to select Pilot Securities from among NMS stocks of all market capitalizations.74 A few commenters recommended that the Pilot exclude securities with smaller market capitalizations and/or thinly-traded securities. One commenter suggested that the ‘‘majority of securities within the Test Groups should be more liquid’’ and that thinly-traded securities, if included, ‘‘should be a minority of all securities in the Test Groups.’’ 75 Similarly, one exchange commenter stated that the Pilot ‘‘should exclude less active stocks as the liquidity in such stocks will likely be severely and negatively impacted by this Pilot.’’ 76 This commenter asserted that ‘‘[l]ess active stocks are highly dependent on professional liquidity providers to post liquidity’’ and speculated that ‘‘[d]ecreasing incentives for liquidity providers to post liquidity in less active stocks will have a pronounced impact on liquidity . . . manifest[ing] in significantly wider spreads and significantly less depth in these securities.’’ 77 Noting that ‘‘many industry participants appear to advocate for increased incentives for liquidity provision in thinly-traded stocks,’’ the commenter did not believe that the Pilot’s goals were ‘‘worth the risk to liquidity and capital formation that the Commission itself identifie[d.]’’ 78 Another commenter was similarly concerned that the Pilot would ‘‘have a significant impact on small to medium issuers since exchanges will not be able to provide incentives to market makers to support trading in those companies’ securities.’’ 79 This commenter stated that ‘‘[l]iquidity rebates can be critical for such securities to motivate market makers to support the stock with aggressive and actionable quotations.’’ 80 Further, the commenter opined that the Pilot would ‘‘risk damaging companies’ ability to efficiently raise capital,’’ which it believed would ‘‘particularly harm small and medium sized companies, for which the current market structure is already not

81Nasdaq Letter I, at 2. See also ASA Letter, at 5. 82Nasdaq Letter III, at 1. The commenter provided a chart showing how the exchanges compare to each other with respect to maintaining a two-sided quote at least 50% of the day. In the chart, some of the exchanges with a higher percent of two-sided markets more than 50% of the day have taker-maker pricing, in which they incentivize the removal of liquidity and charge fees to the provider of liquidity. Id. at Exhibit A. But cf. NYSE Letter II, at 9–10 (arguing that rebates are necessary to promote display of liquidity). 83 Joint Asset Managers Letter, at 2. See also, e.g., Spatt Letter, at 1–2 (stating that the Pilot was a ‘‘very significant improvement over the EMSAC proposal’’ and that one of the ‘‘major improvements’’ was ‘‘the inclusion of lower market value stocks’’); Healthy Markets Letter I, at 11–12; Wellington Letter, at 2; MFA Letter, at 2; Nuveen Letter, at 2; Lipson Letter, at 1; BlackRock Letter, at 1; Vanguard Letter, at 2; CFA Letter, at 4; CIEBA Letter, at 2; Joint Pension Plan Letter, at 2; Oppenheimer Letter, at 2. 84AJO Letter, at 2. 85Babelfish Letter, at 3. 86Healthy Markets Letter I, at 13. 87Better Markets Letter, at 6. See also, e.g., Vanguard Letter, at 2 (‘‘By including all NMS stocks, the SEC will receive data to analyze the impacts of transaction fees on market quality across various types of securities.’’); TD Ameritrade Letter, at 6–7 n.11 (‘‘including securities of small, mid and large cap companies . . . will include some data on the impact that varying transaction fees will have [on] thinly traded securities’’). 88 IEX Letter II, at 7. See also Credit Suisse Commentary, at 1, 3 (stating that the Pilot ‘‘is likely to affect stocks differently depending on their liquidity profile,’’ but expecting stocks ‘‘with wider spreads’’ in Test Groups 2 and 3 ‘‘to continue to behave similarly given that their liquidity may be less driven by rebate-incentivized trading strategies to begin with’’). But cf. NYSE Letter II, at 11 (asserting that it was ‘‘untrue’’ that ‘‘spreads for less-liquid securities are not sensitive to rebate levels’’ and referring to chart showing that NYSE American-listed securities, ‘‘which are generally less-liquid securities’’ spent less average time at the NBBO compared to maker-taker venues). 89 IEX Letter II, at 7. 90 Decimus Letter, at 4–5 (citing Marios Panayides et al., Trading Fees and Intermarket Competition 26 (Charles A. Dice Ctr. for Research in Fin. Econ., Ohio State Univ., Working Paper No. 2017–3, 2017, available at, https://ssrn.com/abstract=2910438). 91 Id. at 5. 92 Id. 93 Id. 94See, e.g., Proposing Release, supra note 2, at 13065–66, and 13069. 95See, e.g., notes 88–92 supra and accompanying text. 96See supra Section II.C.6 (discussing the exclusion of securities that trade fewer than 30,000 shares per day on average from Test Groups 1 and 2). See also supra notes 88–92 and accompanying text. Accordingly, the Commission notes that many
thinly-traded securities will be excluded from the Pilot, which should assuage commenters’ concerns regarding the impact of the Pilot on less liquid or thinly-traded securities. optimized.’’ 81 The commenter further argued that ‘‘incentives (rebates) are important to creating two-sided markets across all stocks, especially thinly traded stocks.’’ 82 Many other commenters supported including a broad scope of Pilot Securities. For example, a group of twenty-one asset managers submitting a joint letter stated that ‘‘[a]s many NMS stocks as possible should be in scope, including those with market capitalizations below $3bln,’’ in order to create a ‘‘meaningful’’ dataset.83 Another commenter agreed that the Pilot ‘‘should encompass the broadest universe of securities, as is feasible, in order to maximize the sample size and provide the most robust dataset possible,’’ further arguing that ‘‘[o]mitting securities of a specific market cap seems arbitrary, would provide an incomplete view of the overall market, and runs the risk of excluding meaningful data and biasing the study.’’ 84 Building on these arguments, other commenters believed it was important to specifically ‘‘test the argument that rebates are required to promote liquidity provision in illiquid stocks.’’ 85 One commenter noted that this debate ‘‘has raged for years,’’ which is ‘‘the point of the pilot: To provide market participants and the Commission with the data needed to make those analyses.’’ 86 Another commenter similarly asserted that the Pilot should include a broad set of NMS stocks to ‘‘help settle academic debates on the relative impact of rebates on liquid vs. less-liquid stocks and other supposedly beneficial aspects of rebates.’’ 87 Notably, some of these commenters directly challenged the argument, set forth by a number of other commenters, that thinly-traded or smallercapitalization NMS stocks would be harmed by the Pilot’s pricing restrictions. One commenter explained that, ‘‘for less liquid stocks, spreads tend to be wider, and as a result rebates become less relevant as a matter of simple mathematics.’’ 88 To illustrate the point, the commenter referred to a ‘‘stock that typically trades at a five-cent quoted spread,’’ noting that a ‘‘typical .0025 per share rebate would equal onetwentieth of the quoted spread, so in these instances a market maker’s revenue from capturing the spread would far outweigh the contribution of the rebate’’ 89 (emphasis in original). Another commenter also questioned the ‘‘significance of liquidity rebates for making markets in less liquid/smallercap stocks,’’ because it believed this ‘‘marginal incentive to provide liquidity . . . is likely to be weak in the smallercap space typically characterized by wide bid-ask spreads . . . .’’ 90 To support this argument, the commenter referred to ‘‘an empirical study of changes in maker-taker arrangements on two European trading venues owned by BATS,’’ now owned by Cboe Global Markets, which suggested
that ‘‘ ‘an elimination of the make fee and a reduced take fee cap would result in worse market quality for large capitalization stocks but better market quality for small capitalization stocks’ ’’ (emphasis in original).91 For this reason, the commenter asserted that the ‘‘link articulated by the opponents of the proposed pilot is at best uncertain and that the pilot may in fact result in improved liquidity for smaller-cap stocks’’ (emphasis in original).92 The commenter therefore contended that it was ‘‘imperative to include a set of smaller-cap stocks in the pilot, as the opponents’ claims on the existence of unambiguous harm to liquidity appear to be exaggerated and driven by preconceived notions.’’ 93 The Commission believes that the many commenters have, through their analysis and ultimate disagreement on this issue, emphasized the need for the Pilot to test the effect of transaction fees and rebates on NMS stocks of all market capitalizations. It is unclear whether or not changes to fees and rebates would harm smaller capitalization or thinlytraded NMS stocks.94 As some commenters have noted, it also is possible that the Pilot may have little effect on smaller-capitalization or thinly-traded NMS stocks or that the Pilot may even improve the liquidity of such stocks.95 The Commission also notes that a pilot focused solely on large capitalization stocks may not produce sufficient data to investigate how changes to transaction fees and rebates will affect liquidity or capital formation across the market. Because including smaller-capitalization NMS stocks in the Pilot will produce a more meaningful dataset to support a broad investigation into the effect of transaction fees and rebates on the full spectrum of NMS stocks and among different segments of the securities market, the Commission adopts this aspect of the rule as proposed. As discussed further below, notwithstanding the decision to include all NMS stocks regardless of market capitalization, the Commission believes it is appropriate to exclude certain thinly-traded securities (e.g., securities that trade fewer than 30,000 shares per day), in part because rebates at that level of trading would be low enough to be unlikely to impact order routing behavior and researchers would be unlikely to get sufficient statistical power to analyze them in isolation at those volume levels.96

97BlackRock Letter, at 1. See also, e.g., Fidelity Letter, at 9. 98 Vanguard Letter, at 2. 99Cboe Letter I, at 17–18. 100 Id. 101State Street Letter, at 3. 102See, e.g., id. 103See, e.g., Securities Exchange Act Release No. 75165 (June 12, 2015), 80 FR 34729, 34732 (June 17,
2015) (Request for Comment on Exchange-Traded
Products) (discussing the create-and-redeem process for ETPs); Transcript of the Division of Trading and Markets’ Roundtable on Market Structure for Thinly-Traded Securities (April 23, 2018), available at https://www.sec.gov/spotlight/ equity-market-structure-roundtables/thinly-tradedsecurities-rountable-042318-transcript.txt (Panel Three discussing ETPs). In particular, large volumes in ETPs can be transacted directly with the ETP issuer in creation units, making the trading center volume in ETPs less relevant to institutional traders that transact in large size orders. 104Morgan Stanley Letter, at 3–4. See also Nasdaq Letter I, at 8–9 (stating that the Pilot was ‘‘arbitrary and capricious and not in accordance with law,’’ in
part because the Commission had ‘‘fail[ed] to
consider’’ the competitive effects of placing ‘‘ETPs tracking similar indexes . . . in different test groups’’); Cboe Letter I, at 17. 105 ICI Letter I, at 4 n.8. 106 Id. at 4. See also, e.g., NYSE Letter I, at 7; Nasdaq Letter I, at 8. 107See, e.g., Credit Suisse Commentary, at 6 (stating that the Pilot could ‘‘unintentionally advantage ETFs in the lower fee group’’). But cf. Nasdaq Letter I, at 8 (stating that ETPs ‘‘in the lower rebate groups would find themselves at a competitive disadvantage to their competitors and may lose market share during the pilot as a result’’). 108See, e.g., SIFMA Letter, at 4–5; Invesco Letter, at 2–3; Morgan Stanley Letter, at 3–4. 109See, e.g., Cboe Letter I, at 28; Invesco Letter, at 2–3; State Street Letter, at 3; STA Letter, at 4. 110See, e.g., ICI Letter I, at 4–5, 5 n.10 (suggesting that the Commission rotate securities every three to six months); Oppenheimer Letter, at 3; Angel Letter II, at 3 (suggesting a quarterly rotation). These commenters did not believe that rotation would ‘‘adversely affect the validity of pilot data’’ or ‘‘impose more than a de minimis implementation burden or other costs on market participants.’’ ICI Letter I, at 4. See also Angel Letter II, at 3. These commenters suggested that ‘‘[a]nalysis of individual security characteristics before and after a rotation to a new group[ ] could yield relevant and important results.’’ Oppenheimer Letter, at 3. See also Angel Letter II, at 3. 111See, e.g., SIFMA Letter, at 5; State Street Letter, at 4; Healthy Markets Letter II, at 8. 112SIFMA Letter, at 4. See also, e.g., Nuveen Letter, at 2; BlackRock Letter, at 2; FIA Letter, at 4; Fidelity Letter, at 9; State Street Letter, at 4; STANY Letter, at 4; Healthy Markets Letter II, at 8. But cf. Angel Letter II, at 3 (stating that ‘‘similar ETFs are probably the best natural controls for each other, as their underlying portfolios are virtually identical,’’ such that ‘‘similar ETFs should definitely be in different treatment groups to increase the power of the pilot’’). 113Schwab Letter, at 3. 114 Invesco Letter I, at 2–3. See also, e.g., Healthy Markets Letter II, at 8 (noting that it may be ‘‘difficult to clearly and consistently define ‘similar’ ETPs’’). b. The Inclusion of ETPs The Commission proposed to select Pilot Securities from among all NMS stocks, including ETPs. A number of commenters supported including ETPs in the Pilot. Several commenters noted, for example, that including ETPs ‘‘would produce a more inclusive analysis of rebates and fees across all segments of NMS stocks.’’ 97 One such commenter believed that ‘‘the benefits from collecting data that informs longterm market structure improvements will outweigh any potential temporary disadvantage.’’ 98 On the other hand, a number of commenters expressed concern with including ETPs in the Pilot. For example, one commenter stated that ‘‘[m]any ETP issuers are . . . strongly opposed to the inclusion of ETPs in the Pilot’’ and suggested
that the Commission had not ‘‘sufficiently explained why it is appropriate to include ETPs in any Pilot.’’ 99 This commenter noted that ‘‘exchanges have implemented numerous incentive structures designed to promote liquidity and narrow spreads in ETPs’’ that could be disrupted by the Pilot, ‘‘negatively impact[ing] liquidity and spreads in ETPs to the detriment of both new and existing investors.’’ 100 Similarly, another commenter expected the Pilot to ‘‘result in spreads widening for ETPs holding pilot stocks, even if ETPs are not included in the pilot, given that fair value calculations rely on underlying constituent pricing,’’ and therefore cautioned that ‘‘any negative effects of the pilot on transaction costs could be intensified for ETP investors.’’ 101 A few commenters ‘‘believe[d] that the goals of the pilot can be achieved without having to include ETPs in the pilot,’’ because ‘‘[t]he effects of the pilot on stocks will be sufficient to draw conclusions about potential changes to access fee rules.’’ 102 The Commission continues to believe that it is important to include ETPs in the Pilot, because excluding them would hamper the Commission’s ability to gather key data that could be used to inform future regulatory action in this area. The Commission does not believe it will be able to draw meaningful conclusions about the impact of changes to transaction fees and rebates on ETPs by observing the effects of the Pilot on other securities, in part because ETPs have a unique create-and-redeem process that does not apply to other NMS stocks.103 Nevertheless, ETPs are subject to the same rules and fees that apply to all NMS stocks. To the extent that the Pilot results may inform future policymaking, Pilot data that includes all types of NMS stocks that would be impacted, including ETPs, will be more useful. Further, some commenters expressed concern regarding the potential for competitive effects among certain ETP issuers. As one commenter noted, ‘‘if two ETPs with similar underliers or that track the same index are placed in the two different [T]est [G]roups, the Pilot would inevitably determine winners and losers.’’ 104 Another commenter explained that ‘‘ETPs with similar investment strategies are more substitutable than stocks of operating companies,’’ such that ‘‘market quality metrics likely play a greater role in driving flows to ETPs.’’ 105 For that reason, ‘‘[i]f competing ETPs are in different test groups—and market quality varies among the test groups,’’ the commenter believed that ‘‘investors might migrate toward products in the test groups with better market quality,’’ thereby ‘‘tilt[ing] the playing field in favor of ETPs that happen to be assigned—at random—to test groups that perform better at the expense of other products.’’ 106 While a few commenters discussed which treatment group would be most problematic,107 many of the commenters took no position on the direction of the presumed competitive impact and did not speculate
about how (or whether) inclusion in specific Pilot Groups would help or harm ETPs.108 To address the potential competitive harm, a few of these commenters recommended that the Commission exclude ETPs from the Pilot altogether,109 while most recommended that the Commission select ETPs in a manner that may avoid any potential competitive effects among similar ETPs, by: (1) Rotating all of the Pilot Securities through the various treatment groups,110 (2) rotating only ETPs through the various treatment groups,111 or (3) placing in the same Test Group ETPs tracking similar indexes or holding similar investments.112 Other commenters criticized these proposed alternatives for selecting ETPs. One commenter, for example, questioned ‘‘whether any of the proposed remedies would address these concerns effectively or fairly.’’ 113 Another commenter expressed concern that the suggestions to place ‘‘similar’’ ETPs in the same Test Group might be too complex to implement, as determining whether ETPs are ‘‘similar’’ to one another for purposes of Pilot rotation can be extremely nuanced.114 This commenter explained that an ‘‘effective classification should take into account an ETP’s underlying index, portfolio constituents and asset class to provide an appropriate ‘apples to apples’ analysis,’’ in addition to ‘‘factors such as assets under management, spread size and daily trading volume,’’ which the commenter believed ‘‘would

115 Invesco Letter, at 2–3.
116The Commission also considered comments providing suggestions relevant to the implementation of these three alternatives. As discussed above, the Commission is not adopting the alternatives. 117See Section II.D.2 (discussing the duration of the Pilot) and Section II.C.5. through 6. (discussing the number of stocks to be included in the Pilot) infra. 118See Proposing Release, supra note 2, at 13019 (Questions #5 and 8). See also id. at 13013 n.46 (noting the receipt of a letter from the Canadian Security Traders Association proposing a crossborder study on the effect of rebates on market quality in conjunction with the Canadian Securities Administrators). 119See, e.g., Fidelity Letter, at 8; OMERS Letter, at 1; FIA Letter, at 4; Healthy Markets Letter I, at 35; STA Letter, at 5. Canadian interlisted stocks are stocks of Canada-based companies that are primarily listed on a Canadian exchange (generally the Toronto Stock Exchange), but that choose to also dually-list on a U.S. exchange. See https:// www.tsx.com/trading/toronto-stock-exchange/feeschedule/ni-23-101 (for a quarterly list of approximately 187 interlisted securities published by the Toronto Stock Exchange featuring stocks that are listed on the Toronto Stock Exchange or the TSX Venture Exchange). 120FIA Letter, at 4. See also Fidelity Letter, at 8. 121See, e.g., STA Letter, at 5. 122See CSA Letter. The preliminary details of the pilot contemplated by the CSA, as reflected in the CSA Letter, were not publicly available prior to the Proposing Release. 123 Id. at 1. 124 Id. at 2. 125 Id. at 1–2. introduce unnecessary complexity into the Proposal.’’ 115 The Commission recognizes the concern that securities placed in one treatment group could be impacted differently than similar securities placed in a different treatment group. While that effect could occur for any security (e.g., stocks of different operating companies in the same industry), it could potentially be more prominent for ETPs that may be substantially similar. Nevertheless, the Commission notes that similar ETPs are not necessarily identical and many other factors influence investor demand and trading, including expense ratios, trading commissions, and existing holdings. The Commission has carefully considered the three alternatives suggested by the commenters 116 and declines to adopt them. Rotating either (1) all Pilot Securities or (2) only ETPs would increase complexity and could increase the costs of the Pilot as the Commission, exchanges, and market participants would need to manage a pilot whose securities change treatment groups every several months. In particular, a rotation design would be considerably more complex than the proposed design by, for example, adding more treatment subgroups and requiring frequent rotation of those subgroups. Given the choice between a simple Pilot design with a short duration, on one hand, and a considerably more complex design with a longer duration, on the other hand, the
Commission prefers to adopt this aspect of the rule as proposed. Compared to the alternative designs suggested by some commenters, the proposal results in a short narrowly drawn pilot with fewer complexities and burdens, which is an outcome supported by many commenters.117 The Commission also considered the suggestion to group ETPs with similar underlying holdings into the same treatment group. While this suggestion involves slightly less ongoing complexity than rotating securities during the Pilot, the Commission declines to adopt this suggestion because it introduces its own complexity in that categorizing ETPs according to their underlying holdings (and potentially other characteristics) involves the exercise of subjective judgment. In addition, grouping similar ETPs can negatively impact the representativeness of the different treatment groups, particularly if all of the similar ETPs are similar in volume, price, and market capitalization. The Commission believes it may learn more from a study that compares how different pricing regimes affect similarly-situated ETPs, whereas keeping similar ETPs in the same treatment groups could reduce the quality and usefulness of Pilot’s results by inhibiting the ability of researchers to compare treatment groups. While the potential exists that similar ETPs in different Pilot treatment groups might trade differently during the Pilot, it is not certain—and commenters held divergent views concerning—whether and to what extent the Pilot would be a contributing factor. Whether the absence of rebates or lower fees help or hurt trading in similar ETPs is far from certain, and whether investors would base trading decisions on those distinctions is unclear. Excluding ETPs to avoid speculative harm would, however, decidedly reduce the utility of the Pilot’s results to inform future policy making. Therefore, the Commission has determined not to adopt a requirement to rotate securities or to group like ETPs. For these reasons, the Commission adopts the rule as proposed to include ETPs in the Pilot.
c. The Inclusion of Canadian Interlisted
Stocks
In the Proposal, the Commission requested comment on the selection criteria and whether the Commission should consider inclusion or exclusion of certain stocks from the Pilot sample set.118 In response, several commenters discussed the inclusion of Canadian interlisted stocks in the Pilot and recommended that the Commission coordinate with Canadian securities regulators to avoid altering the trading dynamics between Canada and the U.S. in those securities.119 For example, one commenter was ‘‘concerned that the inclusion of Canadian interlisted stocks in either one of the reduced access fee or no rebate test groups may materially impact order flow by encouraging transactions to move away from U.S. exchanges and on to Canadian exchanges.’’ 120 Other commenters suggested that the Commission coordinate with the Canadian Securities Administrators to avoid ‘‘dramatic differences in the trading economics on inter-listed stocks between Canadian and U.S. markets.’’ 121 The Commission also received a comment letter from the academics retained by the Canadian Securities Administrators (‘‘CSA’’) to assist with planning, conducting, and analyzing a Canadian transaction fee pilot (‘‘Canadian Pilot’’).122 According to the CSA researchers, the Canadian Pilot likely will propose that, for approximately 180 interlisted stocks, 90 of them would be included in a norebate test group with the remaining 90 placed in a control group.123 In their letter, the CSA researchers requested that the Commission’s Pilot treat interlisted stocks similarly to their Canadian Pilot proposal—i.e., that both pilots place the same 90 interlisted stocks into their respective no-rebate group and place the other 90 stocks into their respective control group.124 By doing so, the CSA researchers believe that both pilots will avoid confounding the analysis for each respective pilot with respect to interlisted stocks because differences in fees and rebates otherwise could incentivize shifts in cross-border routing.125 The Commission agrees with the CSA researchers and believes that it is appropriate to coordinate with the CSA on a transaction fee pilot in order to avoid the potential for distortionary effects between U.S. and Canadian markets if rebates in the ‘‘no-rebate’’ interlisted stocks continue to be allowed on one country’s exchanges but not the other. Accordingly, in the event that the CSA proceeds with the Canadian Pilot concurrently with the Commission’s Pilot, the Commission will append to the no-rebate Test Group the same Canadian interlisted stocks that the CSA selects for its no-rebate treatment group, and the remaining interlisted stocks will

126 In the event that the Canadian pilot does not go forward or does not commence simultaneously with the Commission’s Pilot, interlisted stocks will be placed at the Pilot’s outset into the Control Group. Placing interlisted stocks in the Control Group will preserve the status quo for interlisted stocks and avoid altering the trading dynamics in them between U.S. and Canadian exchanges, which will avoid adversely impacting Test Groups 1 and 2 with respect to those stocks. If the Canadian pilot does go forward, but the interlisted stocks that will be included in its no-rebate test group are not known by the Commission at the time the Commission issues the initial List of Pilot Securities, the Commission may separately issue a subsequent list identifying the interlisted stocks that will be appended to Test Group 2 or the Control Group for the remainder of the Pilot. 127See, e.g., Proposing Release, supra note 2, at 13024 (discussing the design of proposed Test Group 3 and the prohibition in Linked Pricing to support the integrity of a no-rebate test group). See also CSA Letter, at 1 (expressing concern that ‘‘the results of the Canadian Pilot may be statistically and economically inconclusive’’ without coordination with the Pilot). 128See CSA Letter, at 1. 129Oppenheimer Letter, at 3. 130TD Ameritrade Letter, at 4. 131See, e.g., Vanguard Letter, at 2; Joint Pension Plan Letter, at 2; Oppenheimer Letter, at 2. 132See Proposing Release, supra note 2, at 13019. 133 Issuer Network Letter I, at 2 (emphasis omitted) and Issuer Network Letter II. See also Cboe Letter I, at 14–15 (criticizing the Pilot as ‘‘based on recommendations made by a committee that, however well-meaning, was flawed in its construction’’ because it lacked ‘‘exchange or issuer representation’’); Home Depot Letter, at 2 (stating that the EMSAC ‘‘did not include any input from issuers or issuer advocates . . . like NYSE and Nasdaq’’ and that it was ‘‘difficult’’ for ‘‘issuers . . . to understand how this Pilot could be implemented without input from the issuers . . . it will directly impact’’); ModernIR Email, at 1 (stating that a ‘‘study . . . crafted without input or choice for issuers . . . would be an inexcusable travesty’’). 134 Issuer Network Letter I, at 2, 7 (emphasis omitted). 135ModernIR Email, at 1. See also Issuer Network Letter I, at 7 (suggesting that the Commission ‘‘[p]lace the Access Fee Pilot on hold for 90 days while [it] gathers a Blue Ribbon Panel . . . of a dozen or so NYSE and Nasdaq listed company financial executives so that we might conduct a comprehensive review’’ of the Pilot (emphasis omitted)). 136The EMSAC held meetings open to the public, which were publicly webcast, as it was developing its recommendations. To promote awareness of those meetings, the Commission issued press releases to announce those meetings, which included the agenda for those meetings. See, e.g., SEC Press Release 2015–216 (announcing the agenda for an October 27, 2015 EMSAC
meeting, highlighting the discussion of fees and rebates, and soliciting comments from the public thereon), available at https://www.sec.gov/news/pressrelease/ 2015-216.html. The Commission also published meeting minutes and transcripts of the full EMSAC meetings. Finally, the Commission provided a mechanism for the public to submit comments to the EMSAC for its consideration, and a number of people did submit comments. See https:// www.sec.gov/comments/265-29/265-29.shtml (comment file for File No. 265–29). 137See, e.g., P&G Letter, at 1; McDermott Letter, at 1; Level Brands Letter, at 1; ACCO Letter, at 1; NorthWestern Letter, at 1–2; Ethan Allen Letter, at 1; Unitil Letter, at 1; Johnson Letter, at 2; Sensient Letter, at 2; Hawaii Letter, at 1; Cott Letter, at 1; Leaf Letter, at 1–2; First Majestic Letter, at 1; SIFCO Letter, at 2; Weingarten Letter, at 1; Ennis Letter, at 2; Trex Letter, at 1; Genesis Letter, at 1; Tredegar Letter, at 1; Energizer Letter, at 1; ProAssurance be placed into the Control Group.126 Placing the same interlisted stocks into the Pilot’s no-rebate test group that the Canadian Pilot places into its no-rebate test group will avoid the potential to alter the trading dynamics between Canadian exchanges and U.S. exchanges in those stocks that otherwise could result if not all exchanges were subject to the same conditions, which should support the integrity of the no-rebate test groups in both pilots.127 Coordination also will avoid the potential for the Commission’s Pilot to interfere with the ability of Canadian securities regulators to conduct a pilot of their own on Canadian-listed stocks which could be adversely impacted in the absence of coordination.128 The Commission appreciates the interest expressed by the CSA researchers in coordinating on a pilot with respect to interlisted stocks, and looks forward to cooperating with the CSA on this important data-gathering initiative in a manner that benefits both nations’ securities markets. d. The Inclusion of Other Types of NMS Stocks A few commenters addressed the inclusion of other types of NMS stocks, such as American Depositary Receipts (‘‘ADRs’’), rights, and warrants. One commenter supported the proposed broad scope of Pilot Securities and believed that ‘‘analysis of . . . ADRs could provide additional insight into the effect rebates and fees have on liquidity, spreads and the overall trade experience.’’ 129 Another commenter objected to the Commission’s proposal to include rights and warrants in the Pilot, but did not explain the basis for its objection.130 As noted above, however, most commenters expressed general support for a Pilot that includes all NMS stocks.131 The Commission continues to believe that it is appropriate to select Pilot Securities from among the overall universe of NMS stocks. Accordingly, the Commission will include all types of NMS stocks in the Pilot, subject to the selection criteria described below. The Commission believes this is
appropriate because exchange fees and rebates apply to all NMS stocks, as does the fee cap under Rule 610(c). Aligning the scope of the Pilot with the scope of equities fees and the equities fee cap will best facilitate analysis of the impact of changes to transaction fees and rebates on different segments of the securities market. Excluding from its scope any categories of NMS stocks would deprive the Commission of data to inform future regulatory action regarding this segment of the market. For those reasons, the Commission adopts this aspect of the Pilot as proposed, subject to the selection methodology described below in Section II.C.
4. The Ability of Issuers To Opt Out of
the Pilot
The Commission solicited comment as to whether issuers should be allowed to request that their securities not be included in one of the Pilot’s Test Groups (i.e., ‘‘opt out’’) and the potential impact that such an approach might have on the extent and quality of the data collected by the Pilot.132 Several commenters argued that issuers should be permitted to opt out of participation in the Pilot based on process concerns. For example, one commenter’s ‘‘largest concern [was] that the genesis of the proposal . . . deliberately excluded issuer representation’’ by ‘‘excluding the NYSE and Nasdaq from participation on the [EMSAC].’’ 133 This commenter asserted that the ‘‘exclusion . . . from participation in the pre-proposal discussions renders the ‘Opt Out’ option absolutely essential.’’ 134 Another commenter suggested that the Commission could address such concerns by ‘‘conven[ing] a summit for issuers and perhaps [creating] a series of webcasts . . . to explain the purpose of the test,’’ as well as by ‘‘form[ing] an Issuer Advisory Committee that can weigh data and let companies opt into or out of a test.’’ 135 The Commission’s proposal was subject to a full notice-and-comment rulemaking process during which the Commission received a large number of comments from the public, including issuers and their listing exchanges. While the EMSAC recommendation was one of many inputs that informed the Commission’s development of the Pilot, the Commission’s Pilot differs substantially from EMSAC’s recommendation as numerous commenters have recognized.136 Accordingly, the Commission believes that issuers, as well as other market participants, have had ample opportunity to participate in the consideration of the Commission’s proposal for the Pilot. Other commenters supported opt out based on specific concerns surrounding the potential impact of the Pilot. A number of these commenters were listed company issuers that expressed concern about how the Pilot would affect trading in their securities.137 Commenters

Letter, at 1; Home Depot Letter, at 1; SMP Letter, at 2; Halliburton Letter, at 1; Era Letter, at 2; Natural Grocers Letter, at 2; Newpark Letter, at 2; KnightSwift Letter, at 2; Farmer Mac Letter, at 1; BancorpSouth Letter, at 1–2; Haverty Letter, at 1; Ampco-Pittsburgh Letter, at 2; Anixter Letter, at 2; Avangrid Letter, at 2; NHC Letter, at 1; HP Letter, at 2; Curtiss-Wright Letter, at 2; Murphy Letter, at 1. 138See, e.g., Cboe Letter I, at 29; ASA Letter, at 4–5. 139See Addendum to Healthy Markets Letter II, at 11 (attaching an email from NYSE to its listed companies). See also note 137 supra. 140See NYSE Letter I, at 4. In its letter, the commenter mentioned analysis it performed on NYSE-listed issuer secondary offerings in 2017 that suggested that issuers ‘‘with average spreads under 20 basis points paid an average discount to market price of 2.6%’’ and that ‘‘companies with spreads above 20 basis points had to discount their offerings nearly twice as much, to 4.9%.’’ NYSE Letter I, at 14 n.51. It is unclear, however, whether wider spreads cause larger offering discounts or whether they are simply correlated with them. For example, smaller companies that are less well capitalized may have a wider spread compared to a larger, better capitalized company, which could result in spreads being correlated with a company’s cost of capital (i.e., wider spreads could be a reflection of a company’s relative credit risk and cost of capital, not a driver of it). 141See, e.g., Joint Asset Managers Letter, at 2; Citi Letter, at 5; AJO Letter, at 2; Lipson Letter, at 1. 142See supra notes 138–140 and accompanying text. 143Themis Trading Letter II, at 3. 144 Id. at 2–3. 145 IEX Letter II, at 3. See also, e.g., Joint Pension Plan Letter, at 2 (stating that the ‘‘asset manager/ asset owner community is heavily supportive of such a pilot,’’ which should ‘‘provide the necessary confidence to all public companies to be included’’); ICI Letter II, at 2 (‘‘market structure is not a primary consideration guiding the investment decisions of long-term investors’’); Joint Asset Managers Letter, at 2; Healthy Markets Letter II, at
2. But cf. NYSE Letter II, at 4 (stating that ‘‘many
buy-side institutions’’ supporting the Pilot ‘‘are willing to experiment with real-world public companies and end investors to ‘get the data,’ even if the expected impact of limiting or eliminating rebates will be a deterioration of the public quote’’). 146 IEX Letter II, at 3–4. 147RBC Letter I, at 6. See also, e.g., LATEC Letter, at 2; Joint Pension Plan Letter, at 2; MFS Letter, at 3; Clearpool Letter, at 8. 148Spatt Letter, at 3. See also, e.g., Healthy Markets Letter I, at 12; CII Letter, at 4. 149T. Rowe Price Letter, at 4. The issuer explained that its stock, ‘‘on average, trades about
1.5 million shares daily, with an average displayed
size of 200 shares and a spread of nearly $0.07,’’ with ‘‘40% of [its] average daily volume occur[ring] as displayed on exchange volume.’’ Id. at 4–5. 150 Id. at 5. 151Better Markets Letter, at 7. 152MFS Letter, at 3. 153See, e.g., Short Sale Position and Transaction Reporting, Study by the Staff of the Division of Economic and Risk Analysis, June 5, 2014, at 66– 67 (discussing selection bias in the context of an ‘‘opt in’’ voluntary pilot design). supporting opt out emphasized the importance of giving issuers the ability to avoid potential costs and uncertainty resulting from the Pilot.138 For example, one commenter believed that the Pilot could ‘‘caus[e] spreads to widen in securities selected for the test groups,’’ such that ‘‘companies conducting a repurchase program or secondary offering would incur higher costs,’’ and the Commission received a number of comment letters from listed issuers specifically referencing that point and echoing the same concerns.139 This commenter further argued that ‘‘the Proposal would also harm the ability of issuers whose securities are subject to access fee caps to compete’’ with issuers not subject to the Pilot’s exchange fee restrictions.140 Many other commenters opposed opt out.141 Some of these commenters dismissed the concerns described above regarding the potential costs on issuers whose stock is included in the Pilot.142 For example, one commenter disagreed with the notion that ‘‘rebates are needed to incentivize market makers to quote tight spreads’’ in the stocks of certain issuers who had submitted comment letters.143 This commenter explained that the ‘‘fifth of a cent rebate is not incentivizing a tight bid-ask spread in these issuers’ stocks,’’ because that rebate represents an insignificant portion of their average spread.144 Another commenter disagreed with the suggestion that the Pilot would have a negative impact on issuers, arguing that such position ‘‘directly contradicts the public support by investors for the Pilot.’’ 145 This commenter opined that the ‘‘fundamental forces of supply and demand that affect . . . the relative attractiveness of individual public company stocks will be in no way impaired if . . . exchanges are precluded from paying a rebate, or required to accept a lower access fee.’’ 146 Other commenters asserted that opt out would ‘‘adversely affect the quality of the data and the credibility of the Pilot,’’ which could weaken the findings that could be drawn from it.147 One commenter explained that opt out ‘‘would undercut the ability of economists to draw sharp inferences based upon performance differences between the treated and control stocks’’ and that the ‘‘non-random character of ‘opt outs’ ’’ could ‘‘disproportionately reflect firms that were especially responsive to feedback from the listing exchange or could disproportionately reflect less liquid stocks, which would be especially important for the access fee pilot.’’ 148 One listed issuer, which is a large investment
manager, ‘‘welcome[d] the opportunity for [its] stock to be included in the Pilot, with the ultimate goal of improving the overall market to be one where prices can be set by long-term investors without distortion from speculative market participants.’’ 149 This issuer did not ‘‘expect that a reduction or outright removal of rebates will have any significant or harmful effects on the quality of prices displayed in the public lit market, interfere with genuine liquidity and price formation, or negatively impact [its] stock’s trading volume, spread or displayed size.’’ 150 Finally, two commenters further argued that opt out would be inconsistent with the existing market structure. One of these commenters observed that ‘‘[i]ssuers currently have no say over exchanges’ policies’’ and that ‘‘exchanges that modify their access fees dozens of times a year do not survey issuers or permit them to opt-out of these fee changes or creation of order types.’’ 151 The other commenter opined that opt out ‘‘may set an unfortunate precedent that would allow an issuer to pick and choose among those aspects of the National Market System that it likes while rejecting other aspects that it may find less attractive to it, but [which] are necessary to the smooth functioning of [the] United States public equity markets.’’ 152 After careful consideration, the Commission does not believe that issuers should be permitted to opt out of participation in the Pilot. While the Commission understands issuers’ concerns, allowing issuers to opt out could undermine the representativeness of the Pilot’s treatment groups and potentially bias the Pilot’s results, depending on the number and characteristics of issuers that opt out. In turn, researchers would be less able to rely on the data to perform analyses and draw specific conclusions about the impact of the Pilot, thereby limiting the usefulness of the Pilot’s data to the Commission and future regulatory initiatives.153 Although some commenters believe that issuers may incur potential costs or endure competitive harms depending on which of the Pilot’s treatment groups their stock is in, other commenters have argued that such effects are unlikely to manifest. The Commission does not believe it is appropriate to implement an opt out provision that could frustrate the collection of useful and representative data based solely on concerns expressed by some commenters regarding uncertain harms. It is precisely because of this uncertainty that the Commission believes it is necessary to conduct the Pilot to study these contested issues through an objective empirical review of exchange transaction fees and rebates. For those reasons, the Commission

154Cboe Letter I, at 5. See also, e.g., Virtu Letter, at 1–2; Nasdaq Letter I, at 12–13. But cf. MFA Letter, at 2 (stating that ‘‘regulators should periodically assess market practices and regulations to ensure that U.S. equity markets continue to remain efficient, liquid, fair, resilient and transparent for all market participants’’). 155See Larry Harris Letter, at 9–10. 156See, e.g., Decimus Letter, at 4 (stating that the Pilot ‘‘would be valuable in generating concrete information and more preferable to back-of-theenvelope calculations based on questionable assumptions’’); Wellington Letter, at 1 (stating that the Commission could only ‘‘draw[ ] definitive conclusions on the impact of existing pricing models . . . through an actual implementation’’ of the Pilot); Verret Letter I, at 4 (stating that the Commission ‘‘appears to have considered adoption of a mandatory rule to reshape market structure, and determined instead to take the more deliberative and less costly approach of an initial pilot program to generate more data from which it can determine a path forward on market structure reform’’); IAC Recommendation, at 2; MFA Letter, at 2; ICI Letter I, at 1–2; RBC Letter I, at 2; Joint Asset Managers Letter, at 2; Clark-Joseph Letter, at 1; Babelfish Letter, at 3; State Street Letter, at 2; Themis Trading Letter II, at 3; IEX Letter I, at 2–3. 157Fidelity Letter, at 2. See also, e.g., Brandes Letter, at 1 (expressing support for the Pilot and the ‘‘Commission’s effort to shed light into a subject of heated debate among market participants’’); Barnard Letter, at 1 (stating that the Pilot was ‘‘important, as historically there are many views on this topic, but a paucity of credible data from which to draw conclusions’’); Angel Letter II, at 1 (stating that ‘‘various commenters have wildly differing perspectives on what will happen under the pilot,’’ which is ‘‘strong evidence as to why the pilot is necessary’’). 158See, e.g., T. Rowe Price Letter, at 3; Clearpool Letter, at 2. The Commission notes that Nasdaq conducted an independent access fee experiment in 2015, but the limited nature of that experiment makes it difficult to draw conclusions from the data gathered by Nasdaq. See Proposing Release, supra note 2, at 13011–12. See also, e.g., IEX Letter III, at 6 (‘‘Nasdaq’s experiment and its outcomes aren’t a perfect proxy for what is likely to happen in the Transaction Fee Pilot. That experiment was done unilaterally and only in highly-liquid securities.’’); Larry Harris Letter, at 9 (noting that Nasdaq’s ‘‘experimental fee reduction did not occur at all trading venues that traded the subject securities,’’ demonstrating that ‘‘regulatory action is necessary to establish a common pricing standard because market forces alone will not do it’’). 159See Section II.A.2 for a discussion of these comments. 160See also Section II.A.2 for a discussion of these impacts. 161See Proposing Release, supra note 2, at 13019. The Commission notes that the
proposed language in Rule 610T(b)(2)(ii)(E) has been modified slightly. As proposed, Rule 610T(b)(2)(ii)(E) was labeled as ‘‘Test Group.’’ As adopted, the label ‘‘Pilot Group’’ is being substituted for the phrase ‘‘Test Group’’ to provide additional clarity. 162See id. adopts this aspect of the Pilot as proposed.
C. Pilot Design

Need for a Pilot
As a threshold issue, commenters disagreed about whether the Commission should conduct any kind of pilot study of transaction fees and rebates. One commenter, for example, characterized the proposed Pilot as ‘‘a solution in search of problem’’ and claimed that the Commission ‘‘has provided no evidence that existing fee practices are harming investors or interfering with fair competition.’’ 154 Another commenter believed that the Pilot was unnecessary, but for the opposite reason—namely, that there is ample evidence of the negative effects of exchange rebate pricing models, such that the Commission should instead take immediate action to ban them.155 Most commenters, however, thought a Commission-led pilot was necessary and supported the Commission’s proposal to conduct one.156 These supportive commenters observed that ‘‘market participants have heavily debated the effects that transactionbased fees, particularly access fees, and rebates may have on the equity markets’’ and ‘‘commend[ed] the SEC for advancing this discussion through a time-limited, empirical study.’’ 157 Some of those commenters thought a Commission-led pilot was necessary because competitive pressures among exchanges may serve as a barrier to market-led reforms in this area.158 The Commission agrees with the commenters that stated that the Pilot is necessary because, as reflected in the comments discussed above,159 there is strong disagreement about the impact of exchange fee-and-rebate pricing models but a lack of data to study the issue. The Commission believes it is important to further investigate these impacts.160
Pilot Design
For each NMS stock that meets the initial criteria to be a Pilot Security, discussed above, the Commission proposed to assign it to one of three Test Groups, with 1,000 NMS stocks each, or the Control Group.161 The composition of each Test Group would remain constant for the duration of the Pilot, except, as described below, to reflect changes to the composition of the groups caused by mergers, delistings, or removal from a Test Group due to the share price of a stock closing below $1.162 The Commission received a number of comments on the proposed Pilot design, discussed below, focusing mainly on the number of securities included in each Test Group. After consideration of all the comments received and for the reasons discussed below, the Commission is adopting two Test Groups that each contain 730 NMS stocks, functionally combining proposed Test Groups 1 and 2 into a new Test Group 1 with a blended fee cap of $0.0010. Accordingly, for the duration of the Pilot, the following pricing restrictions will apply to Test Groups 1 and 2, while the Control Group will remain subject to the current access fee cap in Rule 610(c):
Proposed Adopted
Fee Cap Test Group 1 .......... 1,000 NMS stocks .......................................................... 730 NMS stocks. $0.0015 fee cap for removing & providing displayed liquidity. $0.0010 fee cap for removing & providing displayed liquidity. Fee Cap Test Group 2 .......... 1,000 NMS stocks .......................................................... Not adopted. $0.0005 fee cap for removing & providing displayed liquidity. No Rebate Test Group .......... 1,000 NMS stocks .......................................................... 730 NMS stocks (plus appended Canadian interlisted stocks). Rebates and Linked Pricing Prohibited for removing & providing displayed & undisplayed liquidity (except for specified market maker activity). No change. Rule 610(c) cap applies ................................................. No change. Control Group ........................ Pilot Securities not in a Test Group ............................... No change.

163See Proposing Release, supra note 2, at 13019– 13020 n.117, 13020 (describing the proposed composition of the Tick Size Pilot overlap subgroups). In the Proposal, the Commission specifically solicited comment on whether the Pilot should overlap with the Tick Size Pilot. See id. at 13025. 164Cf., e.g., Clark-Joseph Letter, at 2 (noting that overlap ‘‘certainly would not be a serious impediment’’); SIFMA Letter, at 3 (arguing against an overlap). 165See Cboe Letter I, at 30. 166See Healthy Markets Letter I, at 14. 167See Cboe Letter I, at 29; Nasdaq Letter I, at 4. 168See Proposing Release, supra note 2, at 13019. 169See Spatt Letter, at 3. 170See, e.g., Mastercard Letter, at 2; Avangrid Letter, at 2; Energizer Letter, at 1. 171See supra Section III.C.4. 172See supra Section III.C.3.b. 173See Proposing Release, supra note 2, at 13019– 20. 174See Brandes Letter, at 2; Themis Trading Letter I, at 3; Oppenheimer Letter, at 2; Spatt Letter, at 2; IEX Letter I, at 5; Verret Letter I, at 4; AGF Letter, at 2; MFA Letter, at 3.
3. No Overlap With Tick Size Pilot
While the Commission’s proposed Pilot design took into consideration the possibility that the Pilot could have been adopted before the end of the Tick Size Pilot Program, the Commission also noted that the overlap design would not be necessary if that were not the case.163 A few commenters opined on the potential overlap between the proposed Pilot and the Tick Size Pilot, disagreeing on whether overlap would be appropriate.164 However, because the Tick Size Pilot ended on September 28, 2018, there no longer is any need for the Transaction Fee Pilot to control for potential data distortions that could have otherwise resulted from the simultaneous operation of the two pilot programs. Accordingly, the Commission is not adopting the proposed Tick Size Pilot overlap design. Relatedly, some commenters discussed whether there should be a delay between the end of the Tick Size Pilot and the start of the proposed Transaction Fee Pilot, with commenters disagreeing on that point. For example, one commenter thought a delay would be appropriate to allow markets to normalize before conducting a subsequent pilot 165 while another commenter thought markets would revert to their baseline state extremely quickly after the Tick Size Pilot ends.166 The Tick Size Pilot concluded, but post-pilot data continues to be collected until April 2, 2019. However, the Transaction Fee Pilot is subject to a onemonth implementation period followed by a six-month pre-Pilot Period. Accordingly, the core of the Transaction Fee Pilot will not commence until after the post-pilot period for the Tick Size Pilot ends. By then, the Commission believes that the markets will have had sufficient time to normalize and any overlap between the Transaction Fee Pilot’s pre-Pilot Period and the Tick Size Pilot’s post-pilot period will be minimal. In both cases, the respective pre- and post-pilot periods are collecting benchmark data on the status quo. As such, the overlap between them should not compromise either dataset. Finally, two commenters recommended that the Commission analyze the Tick Size Pilot data prior to proceeding with the Transaction Fee Pilot.167 While preliminary results from the Tick Size Pilot have been made public, the two pilots are sufficiently dissimilar that the Commission sees no reason for delay. The Tick Size Pilot tested a wider minimum increment (from one cent to five cents) for smallercapitalization stocks, whereas the Transaction Fee Pilot will test a lower rate for the Rule 610(c) fee cap and a prohibition on exchange rebates (which typically are less than one-third of a penny) for stocks of all market capitalizations. Accordingly, findings from the Tick Size Pilot are not relevant to the design of the Transaction Fee Pilot.
4. Stratified Selection of Pilot Securities
The Commission proposed to select the stocks to be included in each of the Test Groups and the Control Group through stratified sampling in a manner that permits comparisons between each Test Group and the Control Group.168 One commenter expressed support for the proposed approach to stratification and noted that it was ‘‘fundamental to the ability to undertake causal inference in this setting . . . .’’ 169 In contrast, a number of public company commenters expressed concern that stratified sampling could result in their stocks being placed in a different Test Group from other similar stocks in their ‘‘peer group,’’ which could complicate comparisons of their stock’s performance against peer-group metrics.170 As discussed above, those commenters supported allowing companies to ‘‘opt out’’ of the Pilot, which could impact the stratification.171 Further, as discussed above, some commenters recommended that the Commission select ETPs for the Pilot in a manner that may avoid any potential competitive effects among similar ETPs, either by: (1) Rotating all of the Pilot Securities through the various treatment groups, (2) rotating only ETPs through the various treatment groups, or (3) grouping ETPs with similar underlying holdings into the same treatment group.172 While the Commission understands the concerns of these commenters, as discussed above in Section II.B, allowing issuers to opt out of the Pilot could undermine the representativeness of the Pilot’s treatment groups and bias the Pilot’s results. Further, also as discussed above in Section II.B, rotating ETPs would require the Commission to implement a more complex and lengthy design in order to maintain sufficient statistical power, both of which would increase the costs and complexity of the Pilot—a result viewed unfavorably by most commenters. Finally, grouping similar ETPs also could negatively impact the stratification of the different treatment groups, particularly if all of the similar ETPs are similar in volume, price, and market capitalization. In turn, this could reduce the quality and usefulness of Pilot’s results by inhibiting the ability of researchers to compare treatment groups. In order to ensure that the Pilot Securities are selected in a way that permits researchers to investigate causal connections, it is imperative to stratify the Test Groups so that researchers can study the effects of changes in fees and rebates within each Test Group, between Test Groups, and between a Test Group and the Control Group. In permitting this type of analysis, the Pilot should be better able to inform future policy considerations to improve the operation of the national market system to the benefit of investors and issuers alike. Accordingly, the Commission is adopting the stratified sampling construct as proposed.
5. Number of NMS Stocks Included in
Each Test Group
The Commission proposed to include 1,000 Pilot Securities in each Test Group (i.e., 3,000 total across three Test Groups) with the remainder to be included in the Control Group in order to be representative of the overall population of NMS stocks and provide sufficient statistical power to identify differences between the Test Groups with respect to common stocks and ETPs.173 Several commenters supported including 1,000 stocks in each Test Group, believing that including 1,000 stocks in each Test Group would facilitate analysis of transaction fees and rebates on a broad cross section of different types of NMS stocks and generate statistically significant conclusions.174 Many commenters, however, thought that the Pilot should include fewer

175See Magma Letter, at 3; FIA Letter, at 4; SIFMA Letter, at 4; Schwab Letter, at 2; Fidelity Letter, at 8–9; Citadel Letter, at 2; State Street Letter, at 3; Citi Letter, at 5; Clearpool Letter, at 7; TD Ameritrade Letter, at 1; STA Letter, at 3–4; STANY Letter, at 3; Nasdaq Letter I, at 10; Cboe Letter I, at 27; T. Rowe Price Letter, at 4; Mastercard Letter, at 2; NorthWestern Letter, at 1; Energizer Letter, at 1; Era Letter, at 1; Knight-Swift Letter, at 2; ASA Letter, at 4–5. 176See Magma Letter, at 3; Schwab Letter, at 2; Fidelity Letter, at 8–9; Clearpool Letter, at 7; STA Letter, at 3–4; Cboe Letter I, at 27. 177See SIFMA Letter, at 4; Schwab Letter, at 2; Citadel Letter, at 6; Citi Letter, at 5. 178See Magma Letter, at 3; FIA Letter, at 4; Citi Letter, at 5; Clearpool Letter, at 7; Nasdaq Letter I, at 10. 179See Citadel Letter, at 6. See also SIFMA Letter, at 4; Citi Letter, at 5. 180See STA Letter, at 3; STANY Letter, at 3; State Street Letter, at 3; TD Ameritrade Letter, at 1, 3; Mastercard Letter, at 2. 181See FIA Letter, at 4; Schwab Letter, at 2; State Street Letter, at 3; STANY Letter, at 3; Era Letter, at 1; Cboe Letter I, at 27. 182See SIFMA Letter, at 4; Citi Letter, at 5; STA Letter, at 3. 183See T. Rowe Price Letter, at 4. 184See supra notes 175–183 and accompanying text. 185See, e.g., supra notes 75–93 and accompanying text. 186A confidence level of 95% is a standard accepted confidence level in statistical analyses. See, e.g., William H. Greene, Econometric Analysis 1033 (Appendix C.6) (6th ed. 2007) (discussing standard confidence levels in academic research). 187See also note 695 infra. 188See, e.g., Citadel Letter, at 6; TD Ameritrade Letter, at 2; Cboe Letter I, at 27. See also Securities Exchange Act Release No. 74892 (May 6, 2015), 80 FR 27514, 27517 (May 13, 2015) (File No. 4–657) (order approving the National Market System Plan to Implement a Tick Size Pilot Program). securities in each Test Group.175 Several of these commenters believed the Pilot could obtain statistically significant data even with fewer stocks in each Test Group.176 Other commenters urged the Commission to reduce the number of securities included in the Test Groups in order to reduce costs associated with the Pilot.177 Several commenters argued that the Pilot was effectively a large scale change to the current equity market structure and that it would be more appropriate for a pilot program to apply to a smaller percentage of the universe of NMS stocks.178 Further to this point, several commenters believed that a large Pilot may be difficult to unwind, with one commenter stating that an immediate return to current transaction fee and rebate dynamics for stocks included in the Test Groups ‘‘could prove to be more disruptive to market participants and overall market quality than the actual implementation of the Pilot.’’ 179 Some commenters also believed the Pilot would negatively impact trading in the stocks placed in certain Test Groups, such as by
adversely impacting spreads, and accordingly recommended including fewer stocks so as to limit potential negative consequences.180 Of the commenters that advocated for reducing the number of Pilot Securities in each Test Group, some suggested alternative amounts to be included. Several commenters recommended including 100 stocks in each Test Group.181 A few others suggested that each Test Group include 500 stocks.182 One commenter recommended ‘‘a more tailored Pilot that includes the 225 most heavily traded names, 225 mid-cap stocks, 225 small caps and 225 ETFs would provide statistically significant data without burdening a material portion of the market.’’ 183 The Commission has carefully considered the concerns expressed by commenters regarding the size of the Pilot’s Test Groups.184 As previously discussed, the Commission cannot know in advance the full effects of the Pilot, whether they be positive or negative. Indeed, commenters expressed a variety of contradicting viewpoints and estimations about the potential impacts of the Pilot on the execution quality and market quality of NMS stocks that would be included in the Test Groups.185 Given this uncertainty, it is crucial that the Pilot be able to produce results that are capable of facilitating an empirical review of the effect of the prevailing fee structures on the equities markets. To achieve this purpose, the Pilot needs to generate a sufficient number of observations over its one-year duration to obtain sufficient statistical power to identify differences among the Test Groups with respect to common stocks and ETPs, thereby permitting researchers to investigate causal connections using economic analysis capable of finding statistical significance. Statistical power refers to the ability for statistical tests to identify differences across samples when those differences are indeed significant and broadly is derived from the number of observations during a study. In other words, statistical power can be present when observing a limited number of subjects over a long period of time or a large number of subjects over a shorter period of time. Because the Commission desires a shorter duration for the Pilot, it therefore needs to have sufficient observable data points over the shorter pilot duration. Accordingly, if the Pilot does not contain enough securities, it may be incapable of producing statistically sound results and will not allow researchers to analyze differences in securities. With statistical power and a sufficiently large sample size, researchers can conduct analysis of what impact (1) reductions in fees and (2) reductions in or prohibitions on rebates might have, if any, on stocks depending on their trading volume or market capitalization. A pilot design that would not provide this meaningful data about the impact that billions of dollars of exchange fees and rebates may have on the markets and market structure, would not achieve the Commission’s goal of conducting a pilot
capable of facilitating an objective empirical view to advance that debate. To achieve these aims, using econometric methods designed to allow researchers to detect a 10% change with a standard confidence level of 95%, the Commission has determined that 730 securities in each Test Group are needed to enable the Pilot to produce statistically meaningful results capable of informing the Commission’s future policymaking efforts. The Commission believes that a 10% change in behavior represents an economically meaningful change that will facilitate analysis of the Pilot’s results, and therefore is an appropriate standard for the Pilot.186 The determination to include 730 securities in each Test Group accounted for the need to obtain statistically significant results among stocks of various liquidity profiles as well as ETPs. While the number of NMS stocks that will be included in each Test Group will be larger than what was recommended by some commenters, the Commission believes that a smaller number of stocks may not have sufficient statistical power given the Pilot’s proposed duration.187 Furthermore, in response to comments questioning why the Pilot included more securities than did the Tick Size Pilot, the Commission notes that the Tick Size Pilot featured 400 corporate stocks for each of its Test Groups.188 Importantly, the Tick Size Pilot did not contain ETPs or large-cap stocks. In comparison, the Transaction Fee Pilot will contain ETPs and largecap stocks. Accordingly, in light of the significantly higher number of securities eligible for inclusion, the Transaction Fee Pilot needs to include considerably more Pilot Securities than did the Tick Size Pilot, while continuing to achieve the same statistical power for each of those groups of securities. Moreover, while several commenters either implicitly or explicitly referenced the EMSAC recommendation to include 100 stocks in each Test Group, EMSAC’s recommendation differs substantially from the Commission’s proposal. Notably, the EMSAC recommendation was limited to common stocks with a

189See supra notes 175–180 and accompanying text. 190See Proposing Release, supra note 2, at n. 102. 191See Proposed Rule 610T(a)(1). See also Proposing Release, supra note 2, at 13021–22. 192See Proposed Rule 610T(a)(2). See also Proposing Release, supra note 2, at 13022. 193See Cboe Letter I, at 16 (stating that the Proposing Release ‘‘does nothing to justify how the $0.0015 and $0.0005 fee cap levels are appropriate’’ and that lowering the current fee cap ‘‘without meaningful discussion or justification is concerning and inappropriate’’); Morgan Stanley Letter, at 1. But cf. Healthy Markets Letter I, at 15–16 (stating that the fee caps for Test Groups 1 and 2 ‘‘appear to be well-justified’’). 194See Cboe Letter I, at 16–17. 195See Credit Suisse Commentary, at 3. 196See T. Rowe Price Letter, at 2. 197See Healthy Markets Letter I, at 15–16. 198 IAC Recommendation, at 1. 199See id. For other commenters suggesting a $0.0010 fee cap, see Goldman Sachs Letter and NYSE Letter III. market capitalization above $3 billion and did not include ETPs, mid- and small-cap stocks, or other types of NMS stocks. In order for the Pilot to permit a broader empirical review of the impact of transaction fees and rebates on order routing, execution quality, and market quality, it is critical that the sample size be representative of the population of NMS stocks for which exchange transaction fees and rebates are economically meaningful. The Pilot must contain enough securities to achieve the statistical power necessary to permit closer analysis of the Pilot’s results in order to identify differences in order routing behavior, market quality, and execution quality among subgroups of NMS stocks (e.g., ETPs, or tiers of common stock).
6. Reduction to the Pilot Size
To respond to commenters’ concerns with the size of the Pilot, including a recommendation from the SEC’s Investor Advisory Committee, the Commission has determined to eliminate one Test Group and reduce the number of stocks in each Test Group to 730. In order to materially reduce the size of the Pilot without sacrificing statistical power, the Commission has determined to: (1) Only place Pilot Securities in a Test Group if, at the time of selection, they trade 30,000 shares or more per day on average and (2) eliminate a Test Group. With respect to securities that trade fewer than 30,000 shares per day, assuming, at an extreme, that such security trades 100% of its volume on a maker-taker exchange paying a $0.0030 rebate, then it would generate $100 in rebates per day. In addition, for thinly-traded stocks with wider spreads, the rebate would be less impactful as it would represent a smaller percentage of the quoted spread. This amount of rebates would be economically insignificant and would be unlikely to impact order routing behaviors of broker-dealers. In addition, this level of trade volume makes it unlikely to produce sufficient statistical power to analyze the securities in isolation because the variability in their quoting and trading characteristics renders it unlikely the Pilot would generate a sufficient number of observations given the Pilot’s proposed duration. In addition, for commenters that believe that thinly-traded stocks need rebates to narrow their quoted spreads, excluding these securities from the Pilot will allow exchanges to continue to apply their current fee schedules to them, which will provide another point of reference to analyze when comparing these securities to those with slightly higher trading volumes. Finally, the Commission believes that eliminating one Test Group and functionally combining proposed Test Group 1 and Test Group 2 into a new Test Group with a $0.0010 cap will result in decreasing the number of NMS stocks included in a Test Group in the Pilot by one-third, which is integral in reducing the overall size of the Pilot by more than one-half. The Commission believes this material reduction directly responds to commenters’ concerns, while still providing the Pilot with a meaningful group in which to test a reduced fee cap and a prohibition on rebates and Linked Pricing. Accordingly, the Commission believes that the Pilot’s design of 730 NMS stocks per Test Group strikes an appropriate balance by reducing the number of stocks in each Test Group and thus mitigating the concerns of commenters about potential detrimental impacts that could be caused by the proposed larger size of the Pilot,189 without undermining the ability to obtain useful data to study the impact of changes to transaction fees and rebates on order routing behavior, execution quality, and market quality for a broad spectrum of stocks. It also is large enough to accommodate drop offs among Pilot Securities (e.g., due to mergers,
bankruptcies, or stocks closing below $1).190
7. Fee Cap Test Groups
The Commission proposed that for Pilot Securities in Test Group 1, equities exchanges could neither impose, nor permit to be imposed, any fee or fees for the display of, or execution against, the displayed best bid or offer of such market in NMS stocks that exceeds or accumulates to more than $0.0015 per share.191 The level proposed for Test Group 2 was $0.0005 per share.192 After careful consideration of the comments received, which are discussed below, the Commission is eliminating Test Group 2 and adopting a revised Test Group 1 with a $0.0010 cap. a. Fee Cap Level Commenters disagreed about the appropriateness or justification for the proposed fee cap levels.193 For example, one commenter stated that ‘‘exchanges currently compete on fees by offering a range of access fees and rebates within the confines of the current $0.0030 access fee cap’’ but the fee caps in Test Groups 1 and 2 ‘‘will reduce the exchanges’ ability to compete on fees by 50% in Test Group 1’’ and ‘‘83% in Test Group 2’’ which could be ‘‘to the detriment of investors and the public interest.’’ 194 In contrast, regarding proposed Test Group 1, another commenter stated that ‘‘[a]t 15 mils, there is still room for significant fee differentiation and rebates remain sizeable.’’ 195 With respect to Test Group 2, one commenter stated that ‘‘[i]f the ultimate intent of the proposal is to determine whether or not reducing access fees will have an effect on how brokers route their customers’ orders, then we fully support the notion of Test Group 2 to see if the incentive to avoid access fees is eliminated with a 5 cents per 100 share cap.’’ 196 Another commenter further stated that ‘‘to the extent that rebates have been traditionally funded by exchanges by the fees collected,’’ then Test Group 2 ‘‘may lead to rebate reductions’’ and obtaining data on this point is ‘‘part of the reason why a study is needed.’’ 197 Finally, the Investor Advisory Committee recommended that the Commission structure the Pilot’s Test Groups ‘‘as simply as possible,’’ and was not persuaded that, in addition to having the no-rebate Test Group, having two additional Test Groups with separate fee caps ‘‘will generate enough additional information to justify the additional effort.’’ 198 Accordingly, the Investor Advisory Committee recommended that the Commission consider having, in addition to the norebate Test Group, only one Test Group with a fee cap and suggested a cap of $0.0010.199 The Commission appreciates the recommendation of the Investor Advisory Committee and agrees with it. As noted above and further discussed below, eliminating Test Group 2 will

200See Citi Letter, at 1–2; Goldman Sachs Letter, at 2. 201See Proposing Release, supra note 2, at 13025. 202See Clearpool Letter, at 3 n.6. 203See Proposing Release, supra note 2, at 13023 n.136–37 and accompanying text. 204 In the Proposing Release, the Commission acknowledged that there were three exchanges that charged different fees for displayed and nondisplayed liquidity. See id. Currently, there are two, IEX and NYSE American. The Commission notes that the differences in fees are minimal and because a small portion of exchanges have chosen to adopt this fee structure to date, it is unlikely a significant portion will choose to do so. 205See Proposing Release, supra note 2, at 13022– 23. 206See Section II.C.6.d. infra. See also Proposing Release, supra note 2, at 13021–24. 207See CFA Letter, at 6; Clearpool Letter, at 2– 3; Healthy Markets Letter I at 27–29. 208See Healthy Markets Letter I, at 28. 209See id. at 16. 210See supra Section II.C.5 discussing the need to generate a sufficient number of observations over the Pilot’s duration to permit researchers to investigate causal connections using economic analysis capable of finding statistical significance. third. New Test Group 1 will have a cap of $0.0010, which adopts the Investor Advisory Committee’s recommendation and represents a blended average of the two fee caps the Commission originally proposed. The Commission believes that new Test Group 1 retains the equities exchanges’ ability to compete through differing fees and rebates, as a fee cap of $0.0010 provides exchanges with an opportunity to utilize various fee and rebate structures to compete for order flow. As some commenters noted, the current access fee cap was set thirteen years ago and may represent an outsized portion of transaction costs in light of the technological efficiencies achieved by the equities markets in the last decade.200 As revised, new Test Group 1 will facilitate an analysis of the extent to which exchanges reduce rebates from their current levels as a result of a materially reduced cap on the fees used to subsidize those rebates, and the impact of a reduced fee and rebate level on order routing behavior, execution quality, and market quality. In addition, by materially reducing the fee cap, the Commission believes that new Test Group 1 will provide useful data on the extent to which current exchange fee levels (bounded by the current access fee cap) serve as a disincentive to take liquidity on an exchange. Obtaining useful information to better understand the potential impact of a significantly reduced access fee cap will ultimately be beneficial to investors and the public interest, as it may help illuminate the extent to which the current fees and rebates effect the market and the extent to which those effects have a detrimental impact on investor transaction costs. b. Applicability to Depth-of-Book and Non-Displayed Liquidity As proposed, Test Groups 1 and 2 were designed to isolate and test a
reduction in the Rule 610(c) fee cap, with all else remaining unchanged. In the Proposing Release, the Commission asked whether commenters thought the fee caps in Test Groups 1 and 2 also should apply to depth-of-book and undisplayed liquidity.201 One commenter recommended that it should.202 In the Proposing Release, the Commission stated that it preliminarily believed it was unnecessary for the fee cap Test Groups to apply to depth-ofbook and undisplayed liquidity because it would be highly unlikely for an exchange to begin charging more to access non-displayed interest or depthof-book quotes (compared to displayed interest), as it would lead to uncertainty for market participants that remove liquidity because they typically would not be able to know in advance or control with absolute certainty whether they interact with non-displayed interest or depth-of-book quotes.203 The Commission continues to believe it would be unlikely that either makertaker or taker-maker exchanges would begin charging differing fees in such a manner.204 Furthermore, the Commission notes that the Rule 610(c) access fee cap does not currently apply to non-displayed interest or depth-ofbook quotes. Introducing a new variable into the fee cap Test Groups would make it more difficult to isolate the effects of a particular change and uncover causal connections. Accordingly, for the reasons noted above and discussed in the Proposing Release, the Commission is not adopting this suggestion.205
c. Prohibiting Rebates and Linked
Pricing in Test Groups 1 and 2 In Test Groups 1 and 2 the Commission did not propose to cap the level of rebates, prohibit rebates, or prohibit Linked Pricing, the latter two of which it proposed to do in the no-rebate Test Group as discussed below.206 In response, several commenters advocated for applying restrictions on rebates to the fee cap Test Groups, primarily in reaction to the potential for exchanges to subsidize their rebates at or near current levels from sources other than transaction fee revenue.207 For example, one commenter stated that ‘‘[t]here is already ample evidence to suggest that some exchanges currently use revenues from other sources to subsidize their order routing incentives, including rebates,’’ such that the proposed fee caps may have no impact on the level of rebates paid for Pilot Securities in the fee cap Test Groups.208 This commenter therefore suggested that the fee cap Test Groups include two subgroups, one as proposed, and a second that would prohibit rebates and Linked Pricing (and also apply to depth-of-book and nondisplayed liquidity).209 The Commission has carefully considered these comments and has determined not to adopt these additional restrictions. While adding more variables or more Test Groups to the Pilot could produce informative results, it would directly complicate the Pilot’s design thus raising the Pilot’s costs and burdens. For example, if the Commission were to add subgroups to new Test Group 1 to prohibit rebates, it likely would have to expand the number of stocks included in the treatment groups or expand the duration of the Pilot in order to achieve statistical power.210 It also would further complicate exchange fee schedules and could lead to more variability in exchange fees if exchanges customized their pricing differently for each Test Group and subgroup. Rather, the Pilot’s design represents a comparatively simple construct that is easier to implement and manage and yet should still facilitate the Commission’s ability to analyze the impact of fees and rebates on order routing behavior, execution quality, and market quality. Achieving these goals, while minimizing complexity and burdens, will also assist the Commission as it considers potential future policy initiatives informed by the results of the Pilot. In addition, the fee cap Test Groups were specifically selected to provide the exchanges with the continued ability to offer rebates, should they so choose, albeit at lower levels, without impacting an exchange’s ability to maintain its net profit on a per transaction basis. The Commission declines to prohibit rebates in new Test Group 1 as doing so would go beyond the construct and application of the Rule 610(c) fee cap by introducing additional variables, and thus would distinctly alter the status quo in that Test Group, thereby complicating the analysis in that treatment group. Lastly, the Commission continues to believe that it is unlikely that exchanges will offer rebates at
their current levels for Pilot Securities in new Test Group 1 because exchanges will need to charge lower offsetting transaction fees in that group in order to maintain a profitable

211See Proposed Rule 610T(a)(3); Proposing Release, supra note 2, at 13022–24. 212 ‘‘Top-of-book’’ means the aggregated best bid and best offer resting on an exchange; in other words, aggregate interest that represents the highest bid (to buy) and the lowest offer (to sell). See 17 CFR 242.600(b)(7) (defining ‘‘best bid’’ and ‘‘best offer’’). 213 ‘‘Depth-of-book’’ refers to all resting bids and offers other than the best bid and best offer; in other words, all orders to buy at all price levels less aggressive than the highest priced bid (to buy) or all offers to sell at all price levels less aggressive than the lowest priced offer (to sell). See 17 CFR 242.600(b)(8) (defining ‘‘bid’’ and ‘‘offer’’). 214 ‘‘Undisplayed’’ refers to resting orders that are ‘‘hidden’’ and not displayed publicly in the consolidated market data. See 17 CFR 242.600(b)(13) (defining ‘‘consolidated display’’) and (b)(60) (defining ‘‘published bid and published offer’’). 215See, e.g., Joint Asset Managers Letter, at 1; Clark-Joseph Letter, at 2; Brandes Letter, at 1; CII Letter, at 3; Themis Trading Letter I, at 3; AJO Letter, at 3; OMERS Letter, at 2; Copeland Letter, at 2; ICI Letter I, at 3; Nuveen Letter, at 2; SIFMA Letter, at 3–4; Better Markets Letter, at 2, 5; RBC Letter I, at 3; Vanguard Letter, at 2–3; Fidelity Letter, at 9; Invesco Letter, at 2; CFA Letter, at 4; MFS Letter, at 2; Wellington Letter, at 2; Joint Pension Plan Letter, at 2; Citi Letter, at 2; Oppenheimer Letter, at 2; Clearpool Letter, at 2; Spatt Letter, at 2; Capital Group Letter, at 3; Healthy Markets Letter I, at 17; IEX Letter I, at 5; Verret Letter I, at 4; Norges Letter, at 2; AGF Letter, at 1; Decimus Letter, at 3; JPMorgan Letter, at 3. 216See Cboe Letter I, at 7, 15–16; NYSE Letter I, at 3–6; Nasdaq Letter I, at 7–8. See also, e.g., Mastercard Letter, at 1–2; Capital Group Letter, at 3; Magma Letter, at 2; FIA Letter, at 4. 217See, e.g., Cboe Letter I, at 7; Nasdaq Letter I, at 9; NYSE Letter I, at 6; Magma Letter, at 2; State Street Letter, at 3; Morgan Stanley Letter, at 4; Cboe Letter II, at 4–7. See also Nasdaq Letter III, at Exhibit A (providing graphs using data from September 2018 on average quoted spread across exchanges in S&P 500 stocks and time at the best quote across those stocks). But cf. Larry Harris Letter, at 6–9 (acknowledging that ‘‘quoted spreads are narrower under maker-taker pricing,’’ but opining that ‘‘the narrower quoted spreads do not benefit the public’’). 218See, e.g., Nasdaq Letter I, at 9; NYSE Letter II, at 11; RBC Letter I, at 5; Nasdaq Letter III. 219See, e.g., Virtu Letter, at 7; Schwab Letter, at 3; State Street Letter, at 2. 220See, e.g., Decimus Letter, at 5 (observing that ‘‘claims on the existence of unambiguous harm to liquidity appear to be exaggerated and driven by preconceived notions’’). See also Section IV infra (discussing the uncertainty of the Pilot’s outcomes). 221 IEX Letter II, at 7. 222See NYSE Letter II, at 2. One commenter questioned
NYSE’s analysis in this regard, noting that in general EDGA’s volume is limited to ‘‘the most liquid names.’’ This commenter stated that NYSE ‘‘distorts the real likely impact of the [P]ilot’’ by including spreads on less liquid securities. See Mulson Letter II, at 2. pricing model. However, the Commission also recognizes, as did commenters, that it is possible that the exchanges may choose to subsidize rebates in Test Group 1 from other sources of revenue, which could result in rebates exceeding the fee cap in that group. Whether and to what extent that would occur in practice would be an important result in new Test Group 1, and so the Commission believes the Pilot should be structured so as not to preclude that possible result. The Commission will closely monitor the fees charged by the exchanges for nontransaction services during the Pilot and will consider the Pilot’s impact on such fees. d. No-Rebate Test Group The Commission proposed that for Pilot Securities in Test Group 3, equities exchanges generally would be prohibited from offering rebates, either for removing or posting liquidity, and from offering Linked Pricing, which, as discussed further below, is defined as a discount or incentive on transaction fee pricing applicable to removing (or providing) liquidity that is linked to providing (or removing) liquidity.211 In addition, Test Group 3 would be unique in that its restrictions would apply not only to displayed top-of-book 212 liquidity, but also would apply to depth-of-book 213 and undisplayed liquidity.214 Transaction fees for securities in Test Group 3 would remain subject to the current $0.0030 access fee cap in Rule 610(c) for accessing a protected quotation. After careful consideration of the comments received on Test Group 3, discussed below, the Commission is adopting Rule 610T(a)(3) as proposed, though it is being renamed as ‘‘Test Group 2’’ since the Commission has reduced the number of Test Groups from three to two. e. Prohibiting Rebates While there was significant disagreement among commenters on this aspect of the Pilot, most commenters supported a ‘‘no rebate’’ group as they believed it was critical to fully examine the effect that transaction fees and rebates have on order routing behavior, execution quality, and market quality.215 In contrast, several commenters opposed prohibiting equities exchanges from paying rebates. Specifically, three of the four exchange commenters asserted that it would inhibit the ability of exchanges to compete with offexchange trading venues.216 In addition, these three commenters, together with other commenters, expressed concerns that prohibiting exchanges from paying rebates to liquidity providers would widen the quoted bid-ask spread on exchanges, which could raise costs on investors.217 Several of these commenters believed that eliminating rebates for ‘‘less-liquid’’ or ‘‘small and medium sized companies’’ would disproportionately impact the quoted spreads for such stocks as they
believed that rebates are a more significant incentive to provide liquidity for less actively traded securities.218 Other commenters also expressed concerns that spreads would widen for ETPs, specifically less liquid ETPs, if rebates were prohibited or significantly reduced.219 The Commission is aware of the potential for adversely impacting smaller capitalization securities, however, the Commission does not agree with the commenters that believe that the Pilot necessarily will result in such harm, or if there are adverse effects in the trading of all or some portion of smaller capitalization securities, that the net effect across securities will be negative. Rather, the Commission agrees with the many commenters who believed that it is unclear what the ultimate net impact of a no-rebate Test Group will be on quoted spreads and trading costs for NMS stocks of different market capitalizations and trading characteristics.220 The purpose of the Pilot is to generate results that can offer data-driven insight on these questions as a basis for possible future policy making in this area. As discussed elsewhere, the revised Pilot has excluded securities that trade fewer than 30,000 shares per day, as they are less likely to provide actionable data. This lack of empirical clarity is reflected in the divergent views of commenters who offered conflicting predictions of the outcome of a norebate Test Group. For example, one commenter questioned whether rebates were necessary to attract displayed liquidity, opining that ‘‘[p]ublic data shows that inverted and flat-fee exchanges often have quotes on both sides of the NBBO, which shows that market participants are willing to pay these exchanges to post quotes at the NBBO based on their intrinsic desire to trade and not just in response to an exchange rebate’’ 221 (emphasis in original). In response, one exchange commenter suggested that Cboe EDGA Exchange, which does not pay rebates, has wider spreads for displayed liquidity as compared to Cboe EDGX Exchange, which does pay rebates for posting liquidity.222 A different commenter did not ‘‘anticipate a material widening for the most liquid names (where rebates aren’t necessary to incentivize liquidity providers) or the most illiquid names (where rebates aren’t sizable enough to incentivize liquidity providers),’’ and instead anticipated ‘‘a likely outcome of increased spreads for the middle tier of

223Citi Letter, at 3–4. See also Credit Suisse Commentary, at 3. 224See Mulson Letter I, at 1. See also IEX Letter II, at 6. 225NYSE Letter II, at 11. 226See T. Rowe Price Letter, at 5. 227See Proposing Release, supra note 2, at 13022– 23. 228See, e.g., T. Rowe Price Letter, at 2; Brandes Letter, at 1–2; Babelfish Letter, at 2. 229See Cboe Letter I, at 12–13. See also Section II.G (responding to comments regarding the Commission’s legal authority to conduct the Pilot). 230For example, if take fees are set at levels to subsidize maker rebates, and if those rebates have little or no impact on quoted spreads of certain NMS stocks, then the take fees on trades in those stocks may constitute a tax on takers of liquidity without a corresponding benefit to the market. 231See, e.g., Clark-Joseph Letter, at 2; Clearpool Letter, at 3 n.6; Healthy Markets Letter I, at 18; IEX Letter I, at 7. 232NYSE Letter I, at 12. 233Nasdaq Letter I, at 6. 234See Proposing Release, supra note 2, at 13023. securities, where rebates have perhaps kept spreads artificially narrow.’’ 223 Another commenter believed that quoted prices are ‘‘almost always set by natural investors’’ and therefore, ‘‘[r]emoving rebates will not disrupt the desire of natural investors to post liquidity and tighten spreads.’’ 224 In response, one commenter was ‘‘skeptical’’ about this and stated that ‘‘it is not realistic for the buy-side to be continuously active on both sides of the market across all stocks impacted by the Transaction Fee Pilot.’’ 225 That said, another commenter, which also is a listed issuer, stated that it did not ‘‘expect that a reduction or outright removal of rebates will have any significant or harmful effects on the quality of prices displayed in the public lit market, interfere with genuine liquidity and price formation, or negatively impact [its] stock’s trading volume, spread or displayed size.’’ 226 The Commission believes that the significant disagreement among commenters on the potential impacts of prohibiting rebates demonstrates the need to include a no-rebate bucket in the Pilot. For example, it is unclear what effect—if any—the payment of a rebate has on a stock that trades over 10 million shares per day with an average natural quoted spread width constrained by the minimum trading increment of $0.01. Likewise, it is unclear what effect—if any—the payment of a rebate has on a stock that trades less than 100,000 shares per day with an average quoted spread of $0.10 or more. In either case, the absence of rebates may have little or no effect on quotes or competition for natural order flow in such securities. Data is needed to empirically evaluate commenters’ diverging views of the effect of rebates. The Pilot is designed to produce this and other data. By prohibiting rebates in one Test Group the Pilot should produce results that facilitate a direct study of the effect of rebates, including on fees, order routing, execution quality, and market quality.227 The
Commission believes that the no-rebate Test Group will provide useful information on trading in the absence of rebates that will facilitate a data-driven approach to better understand the role and effect of rebates in our current market structure. The results generated by this Test Group will allow researchers to study the relationship between rebates and quoted spreads for stocks of varying liquidity profiles and market capitalizations. It also will allow market participants to directly test with their own order flow whether, in the absence of rebates in the most actively traded stocks, they are better able to compete for queue priority and thereby capture the quoted spread when posting liquidity.228 Therefore, the Commission continues to believe that the Pilot will be substantially more informative with a no-rebate bucket and the value of generating that information to inform the Commission’s consideration of the effect of exchange transaction fee models justifies proceeding with the Pilot to better inform both sides of the rebate debate with data to test their hypotheses. In summary, the Commission has carefully considered commenters’ suggested alternatives and whether to include the no-rebate feature in the Pilot, and in light of the important regulatory purpose the Pilot is designed to achieve, the Commission has determined that, for the reasons discussed throughout, it is important to have a Test Group that specifically focuses on the removal of rebates and the corresponding impact on conflicts of interest, execution quality, and market quality. Finally, one commenter asserted that banning rebates ‘‘presents [a] misapplication of Rule 610(c)’’ because the Commission has never before banned rebates.229 While neither Rule 610(c), nor any other Commission rule, currently prohibits a national securities exchange from paying a rebate to provide or remove liquidity, the Commission does not believe that the no-rebate Test Group misapplies Rule 610(c), or any other rule. The no-rebate Test Group is not based on or related to Rule 610(c). Rule 610(c) caps fees for removing a protected quotation, whereas the no-rebate Test Group does not further limit fees and instead prohibits rebates, among other things. Indeed, the Rule 610(c) fee cap continues to apply—unchanged and in its entirety—to the no-rebate Test Group. The data generated by the Pilot will help empirically assess, in light of changing market conditions, whether the existing transaction-based fee and rebate structure continues to further the statutory goals.230 Importantly, while exchanges would retain the ability to charge transaction fees as high as the current $0.0030 cap in the no-rebate Test Group, they would no longer need to charge transaction fees at levels priced to offset the rebates they formerly paid. Accordingly, the no-rebate Test Group is intended to test, within the current Regulation NMS regulatory structure, natural equilibrium pricing for transaction fees. f. Application
to Depth-of-Book and Non-Displayed Liquidity Several commenters supported applying the prohibition on rebates in the no-rebate Test Group to depth-ofbook and non-displayed liquidity as they believed it would avoid the risk that the Pilot’s results could be subject to distortions if exchanges continue to offer rebates for depth-of-book and nondisplayed liquidity.231 In contrast, two exchange commenters opposed this aspect of the proposal. One characterized this aspect of the proposal as an ‘‘unjustified pricing restriction[ ]’’ that was part of a ‘‘new regulatory scheme . . . .’’ 232 The other argued that ‘‘[t]he Proposal lacks internal coherence’’ in that it excludes ATSs ‘‘because they do not have protected quotes, but then includ[es] unlit exchange orders that also are unprotected.’’ 233 For the reasons stated in the Proposing Release, the Commission continues to believe that allowing exchanges to continue to offer rebates in the no-rebate Test Group for depth-ofbook and non-displayed orders could substantially distort the Pilot results.234 The no-rebate Test Group is designed to test the absence of exchange transaction rebates. It would weaken the Pilot’s results to prohibit rebates on displayed orders but allow them on non-displayed orders, as the Pilot would not be able to collect data on what would happen in the absence of rebates. Only by prohibiting the payment of all rebates in one Test Group will the Commission be able to gather data on a pure ‘‘no rebate’’ environment, thereby facilitating a direct observation of the impact of rebates on order routing behavior, execution quality, and market quality

235See supra note 215.
236Cf. supra note 233.
237Price-time priority (where orders are prioritized for execution based on ranking by price and, when two orders are at the same price, by time of entry), generally does provide the ability for an incoming order to bypass non-displayed liquidity. 238See Healthy Markets Letter I, at 18; CFA Letter, at 6–7. 239See Proposing Release, supra note 2, at 13023. 240See supra notes 217–218 and accompanying text. 241See Citadel Letter, at 5. 242See id. 243See Fidelity Letter, at 9; Citadel Letter, at 5. 244Credit Suisse Commentary, at 4. 245See Proposing Release, supra note 2, at 13023. The Commission notes that most exchanges also utilize tiering in their pricing models in which they offer lower fees or larger credits in return for additional volume. See, e.g., Spatt Letter, at 4; RBC Letter II, at 4. 246See, e.g., Capital Group Letter, at 3; IEX Letter I, at 7. 247See NYSE Letter I, at 12; Cboe Letter I, at 10. when compared to the other Test Group and Control Group. As noted above, the Commission received a significant number of comments in support of directly studying the effects of prohibiting rebates.235 In order to avoid the potential distortion from a too-narrowlytailored Test Group that focuses only on one type of rebate but ignores another, the Commission believes that prohibiting rebates on all exchange volume—including depth-of-book and non-displayed liquidity—is necessary to generate the most useful Pilot results on the effect of exchange transaction rebates broadly. In addition, the Commission believes that the no-rebate Test Group’s application to depth-of-book and nondisplayed orders is consistent with the Commission’s decision to exclude ATSs, which do not have protected quotes.236 As discussed above, ATSs are excluded from the Pilot based on a number of reasons, including the materially different treatment of exchange fees under the current federal securities laws and their lack of a protected quotation. With respect to the no-rebate Test Group, it would be incoherent for the Commission to purport to test a prohibition on exchange transaction-based rebates but do so only for some rebates (i.e., on displayed interest) while ignoring the potential for exchanges to pay rebates on non-displayed liquidity and depthof-book interest.237 The possibility that an exchange could offer rebates for nondisplayed and depth-of-book quotes, while eliminating them on displayed interest, could present a loophole with the potential to undermine the design of the no-rebate Test Group and distort the Pilot results for the no-rebate Test Group, rendering the results of the Pilot’s ‘‘no-rebate’’ Test Group incapable of speaking to the impact of rebates. g. Maintaining Rule 610(c) Access Fee Cap Two commenters recommended that, unlike Rule 610(c), the no-rebate Test Group go beyond Rule 610(c) to also prohibit exchanges from charging fees in excess of $0.0030 to provide displayed liquidity.238 As noted in the Proposing Release, the no-rebate Test Group is
designed specifically to test, within the current regulatory structure, natural equilibrium pricing for transaction fees in an environment where exchange transaction-based rebates are prohibited.239 While this would theoretically allow an exchange to charge fees in excess of $0.0030 to provide liquidity, the Commission notes that several exchanges stated that one of the perceived benefits in providing rebates to liquidity providers is that it facilitates narrower spreads and therefore believes it is unlikely exchanges would charge such higher fees during the Pilot.240 One commenter expressed concerns that the no-rebate Test Group would ‘‘provide exchanges with the flexibility to propose a variety of new fee structures for liquidity-taking orders,’’ which could create new conflicts for brokers routing customer orders.241 Accordingly, this commenter believed that the no-rebate Test Group should instead impose a fee cap of $0.0002, where the expectation would be that rebates would be lowered to a de minimis amount and the Pilot would be more symmetrical and thereby more effective in analyzing broker order routing practices.242 The Commission continues to believe that in light of the current debate surrounding the potential conflict of interest posed by the payment of rebates and potential effects they may have on the markets, including the many comments received in response to the Proposal, the Pilot will be substantially more informative with a no-rebate bucket than a bucket that dramatically lowers the fee cap assuming that rebates would follow. While reducing the fee cap to $0.0002 would reduce the likelihood that an exchange would offer rebates at current levels (assuming the exchange desired to fund transaction-based rebates only through transaction-based fees), exchanges would retain the ability to pay rebates and could subsidize them from other sources of revenue leading to rebates that greatly exceed $0.0002. In contrast, only a complete prohibition on rebates will permit researchers to observe directly the impact of rebates on order routing behavior, execution quality, and market quality, and compare this Test Group to the Control Group and the other Test Group where rebates can continue to be offered. Further, imposing a fee cap of $0.0002 instead of prohibiting rebates would not allow Test Group 2 to test, within the current Regulation NMS regulatory structure, natural equilibrium pricing for transaction fees, particularly if the cap is below where the natural equilibrium price would otherwise be found. Two commenters expressed concern that because exchanges can continue to charge access fees of up to $0.0030 per share in the no-rebate Test Group, they may fail to engage in competition on fees.243 In contrast, another commenter believed that, in the no-rebate Test Group, ‘‘the fee for removing liquidity could still move closer to zero in order for exchanges to incentivize takers in the absence of rebates.’’ 244 The Commission believes
that observing price competition in the absence of any distortive effects caused by rebates is an important aspect of the Pilot. Accordingly, the no-rebate Test Group is intended to test, within our current regulatory structure, whether competitive market forces are sufficient to produce natural equilibrium pricing for transaction fees in the absence of rebates. h. Prohibiting Linked Pricing In connection with prohibiting rebates, the no-rebate Test Group also would prohibit Linked Pricing, such that an exchange would be prohibited from adopting any discounts on transaction fees to remove (i.e., ‘‘take’’) liquidity where that discount is determined based on the broker-dealer’s posted (i.e., ‘‘make’’) volume on the exchange, which would result in the broker-dealer paying a lower take fee in return for providing a certain level of liquidity on the exchange.245 Some commenters that addressed the prohibition on Linked Pricing were supportive of the proposal and generally believed that the prohibition would preserve the integrity of the Pilot and facilitate an environment where exchanges are able to set transaction fees at a natural equilibrium level.246 In contrast, two exchange commenters opposed the prohibition.247 Specifically, one commenter characterized this aspect of the proposal, in conjunction with the prohibition on rebates, as an ‘‘unjustified pricing restriction’’ that is ‘‘unrelated to Regulation NMS’s Access

248NYSE Letter I, at 12. 249See Proposing Release, supra note 2, at 13023–
24. 250See RBC Letter I, at 3; MFS Letter, at 2–3. 251As is the case for any fee or fee change an
exchange adopts, if an exchange were to propose such a fee change it would need to analyze in its Form 19b-4 filing how its fee change constitutes an ‘‘equitable allocation’’ of ‘‘reasonable’’ fees and how it is not ‘‘unfairly discriminatory.’’ See 15 U.S.C. 78s(b)(3)(A)(ii). 252See IEX Letter I, at 7–8; Norges Letter, at 2. 253 IEX Letter I, at 7. 254See Proposing Release, supra note 2, at 13024. 255See id. at 13024 n.140. 256See Clark-Joseph Letter, at 2–3. 257See, e.g., Brandes Letter, at 2; Themis Trading Letter I, at 3; CFA Letter, at 7; Clearpool Letter, at 4; Healthy Markets Letter I, at 33; Decimus Letter, at 6 n.22. 258See Proposing Release, supra note 2, at 13024. Fee Cap.’’ 248 As discussed above, the no-rebate Test Group, including the Linked Pricing prohibition, is not based exclusively on the Rule 610(c) fee cap. The Commission continues to believe that prohibiting Linked Pricing supports the objective of the no-rebate Test Group, which is to gather data on the impact of creating an environment where fee levels are not potentially distorted by the rebates they subsidize and rebates do not influence routing, particularly for customer orders.249 In the absence of a Linked Pricing prohibition, exchanges could use make (take) volume to subsidize take (make) activity, which could perpetuate the cross-subsidization of fees. For example, if an exchange adopts Linked Pricing for the no-rebate Test Group securities, it might offer a discounted transaction fee to remove liquidity only to those market participants that post a certain volume on the exchange. Perpetuating this potential distortion could cloud the Pilot results for the no-rebate Test Group if the Linked Pricing incentive interferes with the Pilot’s ability to isolate and analyze the impacts on fees and routing that the no-rebate Test Group is designed to study. Two commenters recommended that the Commission also prohibit an exchange from offering any inducement, including discounts on non-transaction fees, such as those for market data, colocation, or connectivity ports, which are linked to trading volumes in the norebate Test Group.250 The Commission is not expanding the application of the Linked Pricing prohibition in the manner suggested by these commenters. The Pilot, and the no-rebate Test Group specifically, is designed to test the extent to which transaction fees and rebates create conflicts of interest that influence order routing or introduce distortions that impede execution quality and market quality. The Pilot is not designed to eliminate or control for all potential inducements to transact on a particular market and the Commission believes that expanding the Pilot to a wider array of variables could inhibit the Pilot’s ability to isolate the impacts of exchange transaction-based rebates and the effects they may have.251 Further, two commenters requested the Commission to clarify that the Linked Pricing prohibition applies across Test Groups such that exchanges may
not tie rebates or transaction fee discounts in another Test Group to volume in the no-rebate Test Group.252 As previously stated in the Proposal, the no-rebate Test Group is designed to gather data on the impact of creating an environment where fee levels are not potentially distorted by rebates and rebates do not influence routing. In proposing the Linked Pricing prohibition, the Commission recognized that a Linked Pricing arrangement could potentially distort transaction fee pricing if fees continue to be set at a subsidy level above their natural equilibrium, and it also could perpetuate the potential conflicts of interest associated with rebates and order routing. Any Linked Pricing incentives offered by exchanges that are linked, or otherwise related to, posting or removing liquidity in Pilot Securities included in the no-rebate Test Group would contradict the Commission’s intent for the no-rebate Test Group and frustrate the ability of the Pilot to generate useful data in that group. Accordingly, the Linked Pricing prohibition in Test Group 2 prohibits exchanges from offering any discounts or incentives on transaction fees that are linked to activity, whether it be posting or removing activity, in any securities included in Test Group 2, as well as prohibits exchanges from offering Linked Pricing arrangements in Test Group 2 securities that are based on, or include, activity in any Pilot Securities. In addition, one commenter ‘‘suggest[ed] that the linked pricing prohibition should extend to auction fees or any other transaction fees charged by the exchange,’’ as ‘‘[c]losing auction fees, especially, are a significant source of listing market revenue, and . . . discounts on these fees could likewise lead to the distortions described by the Commission (or even to increases in auction fees to other participants to fund the targeted discounts).’’ 253 Because Rule 610T(a)(3) prohibits exchanges from providing a discount or incentive on transaction fees applicable to removing (providing) liquidity that is linked to providing (removing) liquidity, and auction fees are ‘‘transaction fees,’’ the Linked Pricing prohibition applies to auction fees. Exchanges will not be permitted to consider make (take) volume during intraday trading when calculating auction fees, as such an arrangement would perpetuate potential distortions associated with fee-and-rebate pricing models including the crosssubsidization of fees.
i. Linked Pricing Market Maker
Exception
The Commission proposed an exception to the Linked Pricing prohibition to permit an exchange to adopt new rules to provide non-rebate Linked Pricing to its registered market makers during the Pilot in consideration for the market maker meeting rulesbased market quality metrics.254 The Commission explained that to qualify for this limited exception, an exchange would need to propose market making standards in a rule change filing submitted pursuant to Section 19(b)(2) of the Exchange Act, and also would need to propose the fee incentive it would provide for meeting those standards.255 Several commenters requested further clarification about the market maker exception to the prohibition on Linked Pricing. Specifically, one commenter recommended that the Commission provide additional detail about the types of market quality metrics upon which access to Linked Pricing is contingent.256 Other commenters believed that it is important that any such standards adopted by exchanges be sufficiently stringent to prevent market participants from availing themselves of Linked Pricing in a manner that would jeopardize the ability of the no-rebate Test Group to provide valuable data on the impact of the absence of rebates (or a rebate-like incentive) on order routing behavior, execution quality, and market quality or that would permit market participants to unfairly exploit this aspect of the Pilot.257 The Commission continues to believe that permitting exchanges to adopt rules to offer Linked Pricing to their registered market makers for securities in the no-rebate Test Group preserves the ability of an exchange to attract market makers through non-rebate incentives and thereby helps maintain the baseline framework in which exchanges can provide incentives to their registered market makers.258 Commenters highlighted the importance of ensuring that any new rules that exchanges propose to provide Linked Pricing to registered market makers in the no-rebate Test Group be designed so as to not inhibit the Pilot’s ability to

259See id. While it will be up to each individual exchange to design market quality metrics for offering non-rebate Linked Pricing to their registered market makers, such metrics could include, for example: (1) Requirements to trade to stabilize the market; (2) requirements on consecutive price changes and price continuity; (3) material time quoting on both sides of the NBBO; (4) materially enhanced quoted depth on both sides of the NBBO; (5) frequency of setting an improved BBO on the exchange; (6) frequency of setting an improved NBBO; and (7) compliance with narrow maximum quote widths. 260See Proposing Release, supra note 2, at 13024. 261See Healthy Markets Letter I, at 19. 262See id. 263See 17 CFR 242.610(c). Consistent with Rule 610(c), the Control Group will only cap fees for taking (removing) a protected quotation; it will not apply to fees for posting liquidity or otherwise cap or prohibit rebates. See also Proposing Release, supra note 2, at 13024. 264See Proposing Release, supra note 2, at 13025. A ‘‘trade at’’ provision would require that orders be routed to a market with the best displayed price or be executed at a materially improved price. 265See e.g., Adorney Letter, at 1; Birch Bay Letter, at 1. In addition, in clarifying its position on rebates in equity market structure, NYSE stated that it could support a prohibition on rebates if ‘‘done in a measured manner that creates an offsetting incentive to display liquidity, such as a ‘Trade At’ provision[ ]’’ which the Pilot does not provide. NYSE Letter II, at 5. 266See, e.g., MFA Letter, at 3; ICI Letter I, at 3; BlackRock Letter, at 2; FIA Letter, at 5; SIFMA Letter, at 4; Fidelity Letter, at 10; Citadel Letter, at 6–7; Citi Letter, at 3. 267See SIFMA Letter, at 4; Citi Letter, at 3. 268See also infra Section IV.E (discussing tradeat). 269See, e.g., Clark-Joseph Letter, at 2; Nasdaq Letter I, at 13; Cboe Letter I, at 27–28. 270See Barnard Letter, at 1. Another commenter recommended including a Test Group that did not cap fees because it believed that the current structure encourages exchanges to charge fees for data feeds and technology services, which the commenter suggests are higher than they otherwise would be if transaction fees were not capped. See Modern IR Letter, at 3. 271NMS Adopting Release, supra note 10, at 37545. generate useful data on the impact of rebates on order routing behavior, execution quality, and market quality. The Commission agrees that if they are not narrowly tailored, these non-rebate incentive programs could continue to potentially distort transaction fee pricing, particularly if the exchange’s fees are set at a subsidy level above the natural equilibrium within the current regulatory structure to subsidize these market maker incentives. Rather, the market maker exception to Linked Pricing is intended to permit an exchange to impose rules for its registered market makers in ways that would improve its market in a meaningful way, such that it could
use the enhanced liquidity provided by its registered market makers to improve its displayed quotation and thereby attract buyers and sellers to the exchange.259 The non-rebate incentives would only apply to trading activity by a registered market maker in its capacity as a market maker (i.e., acting as principal), and would not apply to any customer activity or activity from other trading desks or business units affiliated with the market maker (and possibly using the same MPID), be it agency, principal or riskless principal trading, traded by or through such market maker. Accordingly, only a registered market maker’s principal trading activity in its capacity as a registered market maker in the no-rebate Test Group would be able to satisfy any market quality metrics, and the only trades that would be eligible to receive the non-rebate incentive pricing would be a registered market maker’s principal trades in its capacity as a registered market maker in the no-rebate Test Group securities.
8. Control Group
The Commission proposed that Pilot Securities that are not placed in one of the Test Groups would be placed in the Control Group.260 One commenter addressed the Control Group and supported the Commission’s proposed approach.261 The Commission continues to believe that a control group is vital to test the effects of fee changes in the Test Groups, as a control group subject to the current access fee cap would provide an appropriate baseline for analyzing the effects of the Pilot against the status quo.262 For these reasons and the reasons discussed in the Proposing Release, the Commission is adopting the Control Group as proposed, which will be subject to the current Rule 610(c) access fee cap.263
9. Alternative Designs
a. Include a Trade-At Requirement In the Proposing Release, the Commission asked whether the Pilot should include a ‘‘trade-at’’ provision that would restrict price matching of protected quotations.264 Several commenters supported including a trade-at requirement because they believed doing so would increase the amount of liquidity available on exchanges and thereby further price discovery.265 In contrast, other commenters opposed including a trade-at requirement as they believed doing so would increase the Pilot’s complexity; impact the ability of the data to assess the impact of transaction fees and rebates on order routing, execution quality, and market quality; be inconsistent with, or unnecessary for, a study of the issues pertinent to the Pilot; and be anti-competitive.266 In addition, two commenters noted that a trade-at requirement would not be necessary because the reduction in the fee cap ultimately could result in more volume being executed on exchanges.267 The Commission believes that adding a trade-at requirement would unnecessarily complicate the Pilot in a manner that would increase costs on market participants and potentially impact the ability of the Pilot to isolate the effects of changes in exchange transaction fees and rebates. Accordingly, the Commission is not including a trade-at requirement in the Pilot. If the Pilot were to also assess the impact of a trade-at requirement, it would need to increase the number of Test Groups, thereby increasing the number of securities included in the Pilot, to be able to isolate the effects of a trade-at requirement separately from the effects of changes in exchange transaction fees and rebates. The Commission believes any potential benefits from analyzing the impact of a trade-at requirement do not justify the additional costs that expanding the Pilot would impose. Rather than introduce another variable into the Pilot, the Commission believes that the Pilot should remain focused on permitting an analysis, in the context of our current market structure, of the effect of exchange transaction fees and rebates. Further, the Commission notes that the Tick Size Pilot featured a trade-at test group, so as that pilot’s post-pilot period concludes, the Commission will have access to current data to analyze the impact of a trade-at prohibition in the context of that pilot.268 b. No Fee Cap Test Group Several commenters advocated for including a Test Group that does not cap transaction fees, believing that it is important to test whether competition alone can constrain pricing and result in a natural equilibrium transaction fee.269 One commenter noted that currently fees tend to ‘‘cluster’’ at the access fee cap imposed by Rule 610(c) and as such recommended including an additional Test Group that does not cap fees.270 When it adopted Rule 610(c), the Commission explained that the access fee cap is necessary to, among other things, inhibit the ability of exchanges to take advantage of the Order
Protection Rule by acting as a ‘‘toll booth’’ between price levels and ensure that quotations are fair and useful by limiting the ability of high fees to distort the price of displayed limit orders.271 The Commission believes that the norebate Test Group will permit analysis of the impact of competitive forces on fees in the absence of current practices

272See, e.g., Goldman Sachs Letter, at 2 (characterizing $0.0030 as an ‘‘outdated benchmark’’ that ‘‘is too high and far from representative of true prices in the marketplace’’). 273The Commission notes that the Proposal included a question regarding whether a fee cap would continue to be necessary to constrain exchange pricing if equilibrium pricing is achieved and the Commission expects that some market participants may analyze the Pilot results for answers to this question. See Proposing Release, supra note 2, at 13025. 274See Clearpool Letter, at 3; Cboe Letter II, at 8–
9. Another commenter also stated that it thought it
was ‘‘a worthwhile exercise to explore the possibility of a move to basis points. . . . ’’ See Citi Letter, at 6. 275See Clearpool Letter, at 3. 276See Nasdaq Letter I, at 13; TD Ameritrade Letter, at 6; Angel Letter II, at 2; Cboe Letter II, at 8. 277See TD Ameritrade Letter, at 6. 278See Angel Letter II, at 2. See also Cboe Letter II, at 8. 279See Morgan Stanley Letter, at 3. 280See supra note 200 and accompanying text. 281See id. 282See Transcript of the Division of Trading and Markets’ Roundtable on Market Structure for Thinly-Traded Securities (April 23, 2018), available at https://www.sec.gov/spotlight/equity-marketstructure-roundtables/thinly-traded-securitiesrountable-042318-transcript.txt. 283See Proposing Release, supra note 2, at 13025. 284See SIFMA Letter, at 4; JPMorgan Letter, at 3; Schwab Letter, at 3 (also stating that eliminating Rule 611 for certain Pilot Securities ‘‘would significantly negatively impact retail order flow and the quality of trade execution’’). 285See T. Rowe Price Letter, at 2–4. 286See id. at 4. that use fees to subsidize those rebates. Specifically, to the extent exchanges will no longer need to charge access fees up to $0.0030 to subsidize rebates in that Test Group, the Commission believes that competitive forces among the exchanges may result in fees approaching a new equilibrium level, within the current regulatory structure, for stocks in the no-rebate Test Group.272 The Commission notes that the order protection requirements of 17 CFR
242.611 (Rule 611) will continue to
apply to all of the Pilot Securities including those in the no-rebate Test Group. As such, the basis for imposing a fee cap (summarized above) remains intact during the Pilot and the Commission believes that applying the current fee cap to the no-rebate Test Group will guard against the possibility, albeit highly unlikely, that an outlier exchange could seek to charge exorbitant fees for the no-rebate Test Group stocks that would be inconsistent with the rationale behind the Rule 610(c) fee cap.273
c. Basis Point Pricing
Two commenters recommended that, because stock prices have increased (i.e., a number of high profile stocks currently trade above $100 per share), using basis point pricing may be a better reference point than using the current access fee cap because the current access fee cap can impact stocks differently based on their price.274 Specifically, one of these commenters proposed that ‘‘Test Group 1 contain the same constraints as Test Group 3 but with an access fee limitation expressed in basis points.’’ 275 However, the Commission believes that doing so would increase the Pilot’s complexity and could interfere with the Pilot’s ability to provide useful data to assess the impact of the current exchange fee models on order routing behavior, execution quality, and market quality because exchange transaction fees and rebates are currently not assessed in basis points and thus this would introduce a new variable into the Test Group as it could raise or lower the fees depending on a stock’s share price, which can vary over time. The more variables that are introduced, the more difficult it could be to isolate the effects of a particular change and uncover causal connections. Accordingly, the Commission is not adopting a requirement that one of the Test Groups include an access fee cap expressed in basis points. d. Higher Fee Caps and Fees Based on Tick Size Four commenters addressed a question in the Proposing Release about including a Test Group that would allow for access fees higher than the current cap under Rule 610(c).276 One of these commenters specifically recommended reducing access fees to $0.0005 per share for the most liquid securities, while imposing gradually higher access fees for stocks of lower liquidity, up to a cap of potentially $0.0050 for the least liquid securities.277 Another commenter recommended including an additional Test Group with an access fee cap of $0.0040, believing this would provide data to test whether an increase in the fee cap reduces bidask spreads in light of the many comments contending that spreads will increase in conjunction with lower rebates connected to a reduced access fee cap.278 In addition, one commenter suggested that if tick sizes were set based on the characteristics of an individual stock, the transaction fee cap could then be a particular percentage of the tick size.279 Such an approach could result in an access fee cap above $0.0030 per share for certain securities. The Commission has carefully considered these suggestions. As discussed above, other commenters have noted that the current access fee cap was set thirteen years ago when markets and technology were markedly different.280 Indeed, a few commenters argued it was outdated and too high.281 Accordingly, the Commission does not believe that raising the access fee cap to levels that are above what trading centers were charging thirteen years ago necessarily is consistent with the technological efficiencies that have been realized in the
intervening years. While market-based solutions and even regulatory responses to enhance the investor experience with trading in thinly-traded securities are worthy of attention, and were the subject of a recent Division of Trading and Markets staff roundtable, the Commission does not believe that the Pilot should introduce the potential for higher rebates—and the further exacerbated distortions that would likely accompany them—when it is attempting to study the effect of the current exchange fee models and fee and rebate levels.282 Accordingly, the Commission is not adopting a higher fee cap in any of the Pilot’s Test Groups. e. Order Protection Rule The Commission solicited comment on whether it would be appropriate to suspend the Rule 611 order protection requirements in one or more Test Groups.283 In response, three commenters opposed eliminating the order protection requirements within the Pilot because doing so would increase the cost and complexity of the Pilot, and also could complicate analysis of the Pilot’s results to the extent it clouded the focus on transaction fees and rebates.284 In contrast, one commenter recommended eliminating the order protection requirements for securities in the no-rebate Test Group.285 This commenter stated that prohibiting rebates is insufficient to ‘‘remove the perceived or real conflicts on broker routing or materially address’’ various negative effects that the commenter believed Rule 611 has had on the equities markets.286 After considering the comments received, the Commission believes that the Pilot should not introduce additional variables by, in this case, removing the Rule 611 protected quotation status for automated quotations in any particular Test Group. In order to add a new variable to the Pilot, the Commission would need to include additional Test Groups and increase the number of securities in order to be able to isolate separately the effects of each variable that is included

287Pragma Letter, at 4. See also Proposing Release, supra note 2, at 13025. 288See Clearpool Letter, at 3–4. 289See Birch Bay Letter, at 2. 290Morgan Stanley Letter, at 2. 291 Id. 292 Id. 293See Goldman Sachs Letter, at 1–4. 294 Id. at 1. 295 Id. in the Pilot or else it would create an asymmetric Pilot that would make it more difficult to evaluate the data and establish causal inferences regarding the impacts of changes to exchange transaction fees and rebates. As discussed above, most commenters were critical of the Pilot’s proposed size. The Commission desires to have a narrowly tailored pilot focused on generating useful data on the impact of exchange fees and rebates on order routing behavior, execution quality, and market quality. Adding another variable to the Pilot would increase the Pilot’s complexity as well as costs of the Pilot. f. Other Ideas for Additional Test Groups and Related Questions In addition to the above questions, the Commission asked a number of other questions in the Proposing Release to solicit commenters’ opinions on equities market structure issues and whether the Pilot should be used as a vehicle to further investigate other related areas. The Commission received a few comments on these points. For example, in response to a question about whether commenters believe the minimum trading increment should be reduced for the most actively traded NMS stocks if the Pilot’s data suggests that rebates do not significantly improve market quality or execution quality for these securities, one commenter stated it ‘‘would strongly support inclusion of a halfpenny spread bucket, or consideration of a separate small-tick pilot for highly liquid stocks.’’ 287 Another commenter recommended that the Pilot test a prohibition on ‘‘tiered pricing,’’ whereby exchanges offer lower per share fees or greater per share rebates to market participants that transact in greater volumes, believing that absent such a prohibition, exchanges would continue to offer these incentives, which would serve ‘‘to potentially work around the prohibition on offering rebates.’’ 288 Further, one commenter suggested adding a new Test Group ‘‘to test an anti-fragmentation policy,’’ in which ‘‘the order that sets the SIP NBBO receives the execution in all circumstances (e.g., bypassing hidden orders). ’’289 The Commission appreciates all of these recommendations. After considering these comments, as well as other comments opposed to including more NMS stocks in the Pilot, the Commission believes that the Pilot should not introduce additional variables. In order to add a new variable to the Pilot, the Commission would need to include additional Test Groups and materially increase the number of securities (or materially increase the Pilot’s duration) to be able to isolate separately the effects of each variable that is included in the Pilot. Otherwise, adding variables would create an asymmetric Pilot that would make it more difficult to evaluate the data and
establish causal inferences regarding the impacts of changes to exchange transaction fees and rebates. As discussed above, most commenters were critical of the Pilot’s proposed size and the Commission similarly desires to have a narrowly tailored pilot focused on generating useful data on the impact of exchange fees and rebates on order routing behavior, execution quality, and market quality. Adding another variable to the Pilot would increase the Pilot’s size, complexity, and costs. g. Gradual Reduction of Current Fee Cap Across All Stocks One commenter suggested that, rather than conducting the Pilot, the Commission should instead consider imposing a ‘‘gradual reduction of the current fee cap across all stocks periodically.’’ 290 The commenter stated that this approach would facilitate data collection and an opportunity ‘‘to observe order routing behavior changes, while applying the same economics to all stocks uniformly.’’ 291 Furthermore, the commenter stated that if a control group was necessary in this scenario ‘‘for comparison purposes’’ it would recommend placing 50% of stocks in the control group and the other 50% in the Test Group subject to the gradual reductions in access fees.292 The Commission has considered this alternative but believes that the Pilot is a preferable approach because it will permit researchers to conduct differences-in-differences analysis over a much shorter time frame. By establishing stratified treatment groups and simultaneously testing different changes in the same variable, the Pilot will reduce the impact of events (economic, natural, political, etc.) across time and thereby is more conducive to an apples-to-apples comparison of the various treatment groups to one another. Pursuing a simultaneous and linear gradual reduction, such as that proposed by the commenter, could require greatly extending the Pilot’s duration depending on the number of fee cap levels to be tested. More importantly, this proposed alternative would not provide the Commission with the opportunity to directly observe the impact of rebates on order routing behavior, execution quality, and market quality because it would not necessarily include a prohibition on rebates and therefore having a no-rebate bucket will be substantially more informative. Lastly, as the Commission believes that a Control Group is necessary to ensure the usefulness of the Pilot’s data, pursuing the proposed structure would impact more NMS stocks than the Pilot (as 50% of stocks would be included in the Test Group and 50% in the control group). h. $0.0010 Access Fee One commenter recommended that rather than pursuing the Pilot, the Commission should instead amend Rule 610(c) to reduce the access fee cap to $0.0010 and also ‘‘conduct an abbreviated study of the effects of eliminating rebates similar to the criteria of Pilot Test Group Three.’’ 293 This commenter stated that ‘‘there is broad recognition’’ that the access fee cap should be reduced and the Pilot will
‘‘be lengthy, complex and costly’’ but ‘‘will not yield a different conclusion.’’ 294 The commenter stated that reducing the access fee cap to $0.0010 would be calibrated with present-day trading and execution costs, would better ensure displayed prices reflect the actual economic costs of an execution, and would allow exchanges to continue maintain their current net capture rates, while also choosing to offer rebates to incentivize liquidity provision if they chose to do so.295 The Commission believes that its revised Pilot design responds to this commenter’s core recommendation, though the Commission is instituting a $0.0010 fee cap as part of the Pilot and not as an amendment to Rule 610(c). The Commission continues to believe that a Pilot is necessary to provide data to objectively evaluate the effect of exchange fees and rebates. Ultimately, the Pilot will enable a data-driven analysis of the impact of transaction fees and rebates on order routing behavior, execution quality, and market quality, which will serve as a valuable precursor to the Commission’s consideration of future policy making in this area.
10. Metrics To Assess the Pilot
A number of commenters recommended that the Commission

296See, e.g., State Street Letter, at 2; Fidelity Letter, at 3–4; Capital Group Letter, at 4; ICI Letter I, at 5; OMERS Letter, at 2; MFS Letter, at 2; Virtu Letter, at 8; FIA Letter, at 2; SIFMA Letter, at 2– 3, 5; TD Ameritrade Letter, at 2; STANY Letter, at 3; Morgan Stanley Letter, at 3; Cboe Letter I, at 29; Nasdaq Letter I, at 7; NYSE Letter, at 2; Pragma Letter, at 2; ModernNetworks Letter; Healthy Markets Letter I, at 35. 297See, e.g., State Street Letter, at 2–3; Fidelity Letter, at 8; Vanguard Letter, at 3; ICI Letter I, at 5; T. Rowe Price Letter, at 4; MFS Letter, at 2; BlackRock Letter, at 2–3; SIFMA Letter, at 5–6; Morgan Stanley Letter, at 3; Spatt Letter, at 5; Cboe Letter I, at 29; IEX Letter I, at 2; Pragma Letter, at 2–3. 298See SIFMA Letter, at 3. Another commenter requested that the Commission clarify the role it expects the Division of Economic and Risk Analysis to play in analyzing the Pilot’s data and provide an anticipated timeline for the issuance of a report on the Pilot data. See IEX Letter I, at 2. 299See Fidelity Letter, at 3, 8; MFS Letter, at 2. 300See Virtu Letter, at 8. 301For example, institutional firms could study their ability to capture the spread when passively posting, and how that is impacted within the Pilot’s treatment groups. 302The Commission encourages market participants to disclose what sources of data they used for their analyses and describe the methodology they used, and to make those reports publicly and freely available. 303For example, the Pilot’s order routing datasets will collect aggregated data, not individual orderby-order level data, and reflects the ‘‘child’’ orders that are processed by an exchange. Thus, the order routing dataset will not capture the entire lifecycle of a ‘‘parent’’ order from its inception through to execution. Accordingly, the Pilot’s order routing datasets will not by themselves permit analyses on an order-by-order basis, and will therefore be unable to assess the execution quality of orders at the ‘‘parent’’ level. If market participants and other interested parties conduct parent order-level analyses and make their findings public, then the Commission would be able to consider them as it assesses the Pilot’s ultimate impact on order routing behavior, execution quality, and market quality. See, e.g., T. Rowe Price Letter, at 4 (recommending that the Commission view analyses of the Pilot conducted by registered investment advisers as a ‘‘key input’’). 304As noted in the Proposal, the Pilot is designed to produce an exogenous shock that simultaneously creates distinct fee environments, each of which restricts transaction-based fees or rebates differently, enabling synchronized comparisons to the current environment for purposes of inferring the existence of causal relationships. See Proposing Release, supra note 2, at 13047 and 53. An exogenous shock to a system occurs when an element of the system is changed from without the system. (i.e., the change or shock is
not under the control or influence of those within the system) but can induce endogenous (i.e., within the system) responses. In the Pilot’s context, the exogenous shock takes the form of a reduction of the maximum permissible transaction fees and a prohibition on rebates and Linked Pricing on all U.S. equities exchanges. See infra Section IV. 305See infra Section IV.B.1. 306For example, a result that shows no impact on liquidity for a Test Group may still be relevant to the Commission’s consideration of the effects of transaction fee-and-rebate pricing models on order routing behavior, execution quality, and market quality and whether the existing exchange transaction-based fee and rebate structure continues to further the statutory goals. more clearly articulate what it believes would constitute a ‘‘successful’’ Pilot and how it will judge whether the Pilot achieves that measure of success.296 Several of these commenters suggested specific metrics or criteria they thought the Commission should analyze when evaluating the impact of the Pilot, many of which were measurement criteria suggested by EMSAC.297 One commenter suggested that the Commission provide guidance about how its staff will be evaluating the metrics used to determine whether to recommend market structure changes to the Commission following the Pilot.298 In addition, two commenters suggested the Commission designate an independent third party to conduct an analysis of the Pilot data upon the Pilot’s completion.299 Another commenter stated that the ‘‘industry should be afforded the opportunity to comment’’ on the metrics and criteria used to evaluate the Pilot.300 In response to these comments, the Commission emphasizes that its staff will likely not be the sole entity analyzing data related to the Pilot. As was the case for the recent Tick Size Pilot, the Commission believes that market participants will publish their own analyses of the Pilot using data that is uniquely available to them and the metrics that they believe are most useful or relevant, and encourages market participants to do so.301 To the extent that interested parties prepare their own analyses, they may submit them to tradingandmarkets@sec.gov with the words ‘‘Transaction Fee Pilot Analysis’’ in the subject line, and the Commission will post those reports on its public website.302 The Commission encourages market participants to make public any analysis they perform on their own trading activity, such as non-proprietary transaction cost analysis (‘‘TCA’’), so that it may be publicly reviewed and used to help inform the public dialogue concerning the effect of exchange fees and rebates.303 To the extent that independent analyses are made public, they can contribute to the Commission’s consideration of any future regulatory action in this area. Given the valuable input of independent analysis, the Commission believes that the success or failure of the Pilot will be determined by whether it produces an exogenous
shock that generates measurable responses capable of providing insight into the effects of fees and rebates on the markets and market participant behavior.304 In the absence of a Commission Pilot that effects change across all equities exchanges in a coordinated manner, researchers would be unable to collect meaningful, comparative data to test the effects of such changes and perform those analyses.305 Success or failure of the Pilot is thus independent of the outcome of the Pilot. For example, a Pilot that shows, with statistical significance, that rebates narrow the quoted spread in thinlytraded stocks would be equally ‘‘successful’’ as a Pilot that shows that rebates do not narrow the spread in such stocks. In this sense, the ‘‘success’’ of the Pilot is that it created the conditions that permit measurement and analysis of that issue in a manner that helps resolve speculative assumptions among the commenters about the impact of fees and rebates. Furthermore, it is worth noting that the data collected pursuant to Rule 610T is only part of what researchers will need to conduct analysis of the impact of exchange fees and rebates on the markets. For example, the Pilot’s order routing datasets contain data to help assess order routing and certain aspects of execution quality, but will not contain any data on exchange quotations, which is available from existing sources. Consequently, researchers will need to use existing data sources to assess the impact of the Pilot on exchange quoting activity and market quality. As such, to the extent that the Pilot data produces null results, for example the Pilot’s order routing datasets do not show any change in liquidity during the Pilot, the Commission believes that independent analysis from market participants, looking at order-level data, may nevertheless detect an impact. Even if the Pilot produces a null result for some metrics, and third-party analysis is not publicly available or does not find an impact, the Commission nevertheless believes the Pilot would still be useful to inform future policymaking that is intended to benefit investors.306 In response to commenters’ requests for additional insight into the types of questions that the Commission hopes the Pilot will be able to answer, the Commission believes that the order routing datasets, as well as other data that is already readily accessible to researchers, should facilitate analysis of the impact of the Pilot through a broad spectrum of metrics. In particular, the Commission will consider, and encourages others to consider, the following questions in contemplating the impact of changes to fees and rebates across the exchanges. These questions include, but are not limited to:

307See, e.g., IEX Letter I, at 1; Joint Pension Plan Letter, at 2; Better Markets Letter, at 3; Brandes Letter, at 2; Clearpool Letter, at 7. 308See, e.g., RBC Letter I, at 4; T. Rowe Price Letter, at 4; Citi Letter, at 2. 309See, e.g., Nasdaq Letter I, at 1, 3 (‘‘[A] transaction fee experiment is inappropriate at this time because there are alternatives and prerequisites the Commission must further evaluate.’’); NYSE Letter I, at 17–19 (stating that the Commission should consider ‘‘less costly and more effective alternatives’’ to the Pilot); Cboe Letter I, at 12, 22, 27 (recommending that the Commission undertake a ‘‘holistic examination of the entire equities market framework’’ including consideration of ‘‘possible changes to the Order Protection Rule [and] the Minimum Tick Increment Rule,’’ ‘‘[s]trengthening and [a]rticulating the Duty of Best Execution,’’ providing ‘‘greater broker-dealer transparency,’’ and adopting amendments to Regulation ATS). 310See, e.g., Nasdaq Letter I, at 1, 3; NYSE Letter, at 17–18; Cboe Letter I, at 12, 22, 27, Fidelity Letter, at 4. See also Securities Exchange Act Release Nos. 83663 (July 18, 2018), 83 FR 38768 (August 7, 2018) (‘‘Regulation ATS–N’’) and 34528 (November 2, 2018), 83 FR 58338 (November 19, 2018) (‘‘Amendments to Order Handling Disclosure’’). 311 ICI Letter I, at 5–6, 6 n.12; ICI Letter II, at 3. See also, e.g., RBC Letter I, at 4; Citi Letter, at 2; Citadel Letter, at 3, 7 (stating that ‘‘it is important to first finalize and implement . . . Rule 606 enhancements before implementing the Pilot’’ to ‘‘safeguard the integrity of the Pilot by ensuring that any changes to broker-dealer order routing practices that result from the increased transparency mandated by amended Rule 606 are isolated from any similar changes that result from the design of the Pilot’’); Spatt Letter, at 4 (stating that the ‘‘the enhanced disclosures proposed would strengthen the potential causal inference that the response to [the Pilot] would allow’’). Some commenters questioned whether the Pilot should proceed, because they believed that the adoption of Regulation ATS–N and the Amendments to Order Handling Disclosure will ‘‘impact the very potential conflicts of interest the Commission aims to study . . . .’’ Nasdaq Letter II, at 2–4; see also NYSE Letter IV, at 2–3. As noted in this section, the scope of the Pilot is broader than just conflicts of interest. Therefore, those initiatives, or the impact they may have on order routing behavior, would not provide sufficient data to evaluate the effects of transaction fees and rebates on market quality and execution quality. See infra Section IV.B.1. 312Nasdaq Letter I, at 1–2, 4; Nasdaq Letter II, at 2–4 (suggesting that the adoption of these regulations ‘‘further reduce[d] the already weak need for the [Pilot]’’). See also, e.g., STANY Letter, at 2; ASA Letter, at 5–6; Era Letter, at 1. But cf. Verret Letter I, at 4 (stating that the ‘‘collection of data from broker-dealers’’
or the use of ‘‘existing data contained in [OATS]’’ were not ‘‘feasible alternatives,’’ because a ‘‘randomized trial is far superior for the purpose of generating robust statistical analysis to inform subsequent rulemaking’’); Proposing Release, supra note 2, at 13046–47 (outlining the limitations of existing data sources). 313See, e.g., STANY Letter, at 2; FIA Letter, at 3; Grasso Letter, at 2. But cf. ICI Letter II, at 3 (noting that disclosure-based rulemakings ‘‘will not directly reduce the potential for exchange transaction pricing models to create conflicts of interest for broker dealers, nor will they provide data that would allow an institutional investor to measure the impact of fee avoidance on routing decisions’’); Luminex Letter, at 1; Spatt Letter, at 4; IEX Letter II, at 9.

To what extent do access fees and
rebates impact routing decisions for liquidity-taking orders? Are orders to take liquidity more likely to be routed to an exchange (compared to an off-exchange venue or ATS) in a lower access fee environment than they are currently? To what extent are impacts or changes in routing decisions driven by potential conflicts of interest created by transaction fees and rebates rather than other factors such as fill rates and execution quality?
To what extent do access fees and
rebates impact routing decisions for liquidity-supplying orders? Are orders to provide liquidity less likely to be routed to an exchange (compared to an off-exchange venue or ATS) in a lower rebate environment than they are currently? To what extent do impacts or changes in order routing appear to be driven by potential conflicts of interest caused by rebates rather than other factors such as execution quality (e.g., fill-rates, time to fill, capturing the quoted spread, adverse selection, or reversion)?
What impact does a reduction or
elimination in rebates have on the NBBO, including spread width and the depth of interest displayed at the NBBO? To what extent does a potential decrease in depth of interest at the NBBO result in lower fill rates or smaller fill sizes for investor orders? Are natural investors better able to obtain queue priority in exchange order books, and are they more frequently able to capture the quoted spread when posting passively (e.g., buy on the bid and sell on the offer)?
Are there common characteristics for
securities (e.g., average daily trading volume, price, or market capitalization) where a reduction or elimination of rebates begins to impact quoted spread? If so, what are those common characteristics and at what level do reduced rebates begin to have an impact on quoted spread? To what extent does a change in quoted spread affect transaction costs for investor orders? If quoted spread widens in a security, to what extent is the potential spread cost offset by the reduction in the transaction fees paid, or a change in the ability to capture the quoted spread?
Are there common characteristics for
securities where a reduction or elimination of rebates does not impact quoted spread? If so, what are those common characteristics (e.g., average daily trading volume, price, or market capitalization)?
Are there common characteristics for
securities (e.g., average daily trading volume, price, or market capitalization) where a reduction or elimination of rebates begins to impact effective spread?
How can we best understand the effects
of rebates provided on inverted venues (where rebates are paid to takers of liquidity)?
What impact do lower access fees and
rebates have on the amount of displayed and non-displayed liquidity on exchanges?
In the absence of rebates, do competition
and market forces operate to produce a market equilibrium (within the current regulatory structure) that constrains transaction fees to levels at or below today’s current access fee cap? What do such market forces, and any resultant equilibrium pricing, tell us about the need to impose a cap on access fees? Does the Pilot provide any data that suggests, in the absence of rebates, an access fee cap would still be necessary as long as Rule 611 of Regulation NMS continues to impose order protection requirements on exchanges with protected quotes?
What is the impact of a lower fee cap
on trading volumes on each exchange? What is the impact of a lower fee cap on other measures of liquidity on each exchange? How should we understand the difference between volume and liquidity?
What is the impact of lower rebates on
the ability of smaller exchanges to attract liquidity-supplying orders? By providing a mechanism that is uniquely capable of facilitating an empirical review of these and similar questions, the Pilot is an essential tool that can further the understanding of an important component of equities market structure. While other market structure issues also might benefit from a pilot, exchange transaction fees currently are a prime focus for empirical study, as evidenced by, among other things, the EMSAC’s recommendation to the Commission and the number and nature of comments the Commission received on its proposal. Ultimately, the Commission desires to use the Pilot’s results to help assess whether (and, if so, in which types of NMS stocks) rebates have a positive impact on execution and market quality, or whether they have no or little effect or a negative effect. D. Timing and Duration
Disclosure Initiatives and the Pilot
While a number of commenters urged the Commission to proceed expeditiously with its proposed pilot,307 other commenters believed the Commission should pursue different market structure initiatives before conducting the Pilot 308 or in lieu of the Pilot.309 The Commission has adopted two of the market structure initiatives identified by commenters—namely, proposals to enhance the operational transparency of ATSs and to enhance disclosure of order routing behavior.310 While some commenters believed that the information and data from those new rules would complement the Pilot and ‘‘improve understanding of pilot data,’’ 311 others believed the new rules would instead allow the Commission to determine ‘‘whether a problem exists without risking the potential negative impact of a pilot’’ 312 or thought that potential conflicts of interest in order routing behavior would be better addressed through increased transparency and disclosure than by the Pilot.313 The Commission disagrees. Comments urging the Commission to pursue disclosure-based initiatives focused only on one narrow aspect of

314See, e.g., Nasdaq Letter I, at 3–4; Fidelity Letter, at 6; Cboe Letter I, at 21–22. 315See Cboe Letter I, at 12; FIA Letter, at 3; Nasdaq Letter I, at 4. 316See, e.g., IEX Letter II, at 9; Better Markets Letter, at 3; Brandes Letter, at 2; AJO Letter, at 2; OMERS Letter, at 3; Clearpool Letter, at 7. Some of these commenters suggested that the Pilot should proceed in conjunction with action on other market structure initiatives. See, e.g., SIFMA Letter, at 1; Pragma Letter, at 3–4. 317See, e.g., Verrett Letter I, at 5; Better Markets Letter, at 3. 318Brandes Letter, at 2. 319Themis Trading Letter I, at 6. See also ICI Letter I, at 3. 320See Proposing Release, supra note 2, at 13025. The Commission notes that the proposed language in Rule 610T(c)(a)(ii) has been modified slightly. As proposed, Rule 610T(c)(1)(ii) contained the phrase ‘‘shall continue for up to another year.’’ As adopted, the phrase ‘‘shall continue for up to one additional year’’ is being substituted for the phrase ‘‘shall continue for up to another year’’ to simplify the rule text without substantively changing the requirement. 321See, e.g., SIFMA Letter, at 3; AGF Letter, at 2; Wellington Letter, at 2. See also RBC Letter I, at 6 (stating ‘‘a pilot of at least one year and no more than two years will ensure that ample data is collected over time, that the restrictions of the various Pilot Test Groups cannot be evaded by delay, and that the Pilot does not exist for a period of time beyond which its data would be cumulative or of marginal significance relative to data produced earlier in the Pilot period’’). 322See Citi Letter, at 5 (stating that ‘‘[c]ostsensitive firms may be able to more quickly adapt to new pricing, while liquidity-based routers may need time to collect a new sample set to adjust their routing logic,’’ such that ‘‘data in the weeks closer to the conclusion of the Pilot may more accurately reflect the state of the market and what the implications would be if implemented long-term’’). One commenter, however, did not believe that certain ‘‘broker-dealers, proprietary traders, and algorithm vendors’’ would ‘‘incorporate the new fees into their routing systems on a timely basis, if ever,’’ because according to this commenter, ‘‘[c]hanges are costly and may prove to be ultimately unnecessary if pricing reverts following the termination of the pilot study.’’ Larry Harris Letter, at 11. 323 Joint Asset Managers Letter, at 2. See also Joint Pension Plan Letter, at 2; Brandes Letter, at 2. 324See, e.g., SIFMA Letter, at 3; CFA Letter, at 6; Fidelity Letter, at 9; IEX Letter I, at 4. 325See, e.g., Babelfish Letter, at 3; Healthy Markets Letter I, at 19. 326See TD Ameritrade, at 5; see also, e.g., NorthWestern Letter, at 1. 327See, e.g., FIA Letter, at 4; STANY Letter, at 3– 4. 328Magma Letter, at 3. See also, e.g., Apache Letter, at 1; Unitil Letter, at 2. 329See, e.g., Ethan Allen Letter, at 1; ProAssurance Letter, at 2; Knight-Swift Letter, at 2. 330NYSE Letter I,
at 16. 331See, e.g., Citi Letter, at 5; Babelfish Letter, at 3. 332See Proposing Release, supra note 2, at 13071. the Pilot—studying the conflicts of interest between brokers and their customers that are presented when exchanges pay rebates. However, such an approach does not adequately advance the Pilot’s broader purpose— obtaining a better understanding of all potential impacts from fees and rebates, and how fees and rebates may affect stocks differently depending on their liquidity. Similarly, some commenters recommended that, either before conducting the Pilot or in lieu of the Pilot, the Commission should pursue other market initiatives such as enhancing broker-dealers’ duty of best execution 314 or undertaking a ‘‘broader review of equity market structure,’’ including the consideration of possible changes to the Order Protection Rule or the Minimum Tick Increment Rule.315 Other commenters disagreed and did not believe that the Commission should delay the Pilot in order to pursue other market structure initiatives.316 For example, a few commenters advocated proceeding with the Pilot because the Pilot may help to inform future policy changes in these other areas.317 Other commenters characterized the ‘‘holistic reform’’ advocated by other commenters as ‘‘an elusive goal’’ 318 in light of market participants’ competing interests—one that has been used to ‘‘slow down market structure reform for the past decade.’’ 319 The Commission believes that there is no need to delay proceeding with the Pilot in order to pursue other potential equity market structure initiatives. The Pilot seeks to resolve several equity market structure questions that have been debated for several years. Similarly, the Commission does not believe that it needs to complete the Pilot before proceeding to consider other equity market structure initiatives. Other initiatives may implicate equity market structure questions that are narrower or broader than, or independent of, exchange fee models, such as considering innovative approaches to thinly-traded securities. The Commission expects that it will continue to evaluate the need for other changes to equity market structure during the Pilot.
2. Automatic Sunset at Year One
The Commission proposed that the Pilot have a duration of one year with a maximum period of two years. Specifically, the proposed Pilot duration featured an automatic sunset at the end of the first year unless, prior to that time, the Commission publishes a notice that the Pilot shall continue for up to one additional year.320 After careful consideration of the comments received, the Commission is adopting Rule 610T(c) as proposed. Many commenters supported the proposed duration of the Pilot.321 For example, one commenter asserted that ‘‘each pricing experiment needs to be in place for a sufficient length of time to enable the firms to adjust their routing logic.’’ 322 Others agreed that the proposed duration would reduce the ‘‘desire to ‘wait out’ the Pilot’’ and would avoid ‘‘the incentive to alter behavior in order to distort the Pilot’s results . . . .’’ 323 Several commenters supported the automatic sunset provision after one-year.324 A few commenters, however, thought the proposed duration was too short and that a minimum two-year pilot would be necessary.325 Some other commenters believed that the necessary data could be obtained within a shorter time frame. Among commenters advocating for a shorter Pilot Period, the recommended duration varied and ranged from those who felt there would be an ‘‘immediate and measurable impact upon implementation’’ 326 to those who felt the appropriate time frame should be modified to an absolute maximum of one year.327 One commenter questioned whether a ‘‘1–2 year pilot that changes fees on 3,000 names’’ was ‘‘really a ‘pilot’ or in fact a de facto imposition of a significant reduction of transaction fees[.]’’ 328 Several commenters expressed their view that the proposed length of the Pilot would ‘‘exacerbate[ ] the negative impact upon the affected issuers.’’ 329 One commenter took issue with the proposed length of the Pilot by challenging what it believed to be conflicting statements of the Commission in its original Proposal. According to the commenter, the Commission asserted, on the one hand, that the ‘‘market quickly reacts to changes in (and elimination of) pricing changes, but on the other hand, claims that the market does not react unless the changes are in effect for at least a year.’’ 330 The Commission believes both of those statements are correct and do not conflict. While many market participants will react promptly to pricing changes, particularly those with cost-based routing algorithms, others may need additional time to fine tune liquidity-based routing algorithms as order flow changes in response to fee changes.331 More importantly, however, the Pilot needs to be long enough to discourage any market participant inclined to resist adapting its behavior to the fee changes.332 A few commenters opposed the oneyear sunset provision, but for a variety of different reasons. For example, one commenter thought a full two-year pilot was necessary, another thought the Commission separately has the
authority to revise or terminate the Pilot early and does not need a sunset

333See Babelfish Letter, at 3; Healthy Markets Letter I, at 19; Cboe Letter I, at 19. 334See, e.g., Citi Letter, at 5. 335See Proposing Release, supra note 2, at 13025–
26. See also Proposed Rule 610T(c).
336Proposed Rule 610T(d) and (e). See also Proposing Release, supra note 2, at 13029, 13032. Primary listing exchanges will also be required to prepare and publicly post updated Pilot Securities Exchange Lists and Pilot Securities Change Lists for the duration of the Pilot Period and through the post-Pilot Period. Id. at 13027–28. The pre-Pilot data is intended to establish a baseline against which to assess the effects of the Pilot, while the post-Pilot Period is intended to help assess any post-Pilot effects following the conclusion of the Pilot. 337See FIF Letter, at 7, 9; Healthy Markets Letter I, at 19. 338See IEX Letter I, at 4; FIA Letter, at 4. 339See e.g., SIFMA Letter, at 3; Issuer Network Letter I, at 4; State Street Letter, at 4; Cboe Letter I, at 28. 340See Citi Letter, at 4; TD Ameritrade Letter, at 1, 5; Morgan Stanley Letter, at 4; Cboe Letter I, at 28–29; Vanguard Letter, at 3; STANY Letter, at 4; Angel Letter II, at 3. 341See Schwab Letter, at 2; Cboe Letter I, at 29. 342See TD Ameritrade Letter, at 5. See also Morgan Stanley Letter, at 4; SIFMA Letter, at 3. provision, and a third was critical of the lack of metrics that would accompany the automatic sunset.333 After careful consideration of the comments received, the Commission is adopting the Pilot’s duration as proposed. The Commission believes that the Pilot’s duration will provide an appropriate balance between providing certainty about the maximum duration for the Pilot while also allowing flexibility to conduct a Pilot for more than one year if necessary to collect representative data. Further, the Pilot’s duration should be long enough to make it economically worthwhile for market participants to adapt their behavior and not ‘‘wait out’’ the Pilot. In addition, in light of the number of Pilot Securities selected, which were selected to ensure sufficient statistical power to allow for meaningful analysis, the Pilot’s duration will allow for the collection of a robust and representative data set over a sufficiently long period of time,. The Commission considered a shorter time period for the Pilot, but is concerned that short-term or seasonal events could unduly impact the Pilot results and therefore data collected over a shorter duration may not yield a sufficiently representative dataset that would be capable of permitting analysis into the impact of transaction-based fees and rebates and the effects that changes to those fees and rebates have on order routing behavior, execution quality, and market quality. For example, a shorter pilot period could be impacted by seasonal idiosyncrasies, macroeconomic factors, or even weather events. Further, the Commission recognizes that some market participants, for example, broker-dealers whose liquidity-focused routing strategies are based on, and continually updated based on, several weeks’ worth of data, will need time to fine tune their revised routing strategies. While some market participants may adjust quickly, others, like
proprietary trading market participants, may wait to see how other market participants react before refining their own routing strategies.334 In other words, it could take a few months before some market participants finish calibrating their routing strategies to the fees and rebates that the exchanges adopt consistent with the Pilot’s requirements and adjust them as trading dynamics settle in response to those changes. The exchanges also could take a number of weeks to settle on new fee models as they see how other exchanges modify their fee models to comply with the Pilot’s requirements and then respond accordingly, which could further delay the time it takes for brokerdealers to adjust their routing and trading algorithms. Accounting for all of this, the Commission intends that the proposed duration of the Pilot be long enough to encourage wide participation by all market participants (and discourage ‘‘waiting out’’ the Pilot) and thereby help ensure that the Pilot produces results that are more reliable, robust, and useful. The Commission also considered extending the Pilot period to two-years as suggested by several commenters, and as was recommended by the EMSAC, but continues to believe that the inclusion of the automatic sunset provision at the end of the first year is preferable because it will provide flexibility in the event that the Commission believes additional time is necessary to ensure the collection of a robust dataset with adequate statistical power for analysis, but will allow the Pilot to automatically end after one-year in the event that sufficient data is collected by that point with sufficient statistical power to allow for meaningful analysis.
3. Pre- and Post-Pilot Periods
The Commission proposed a sixmonth pre-Pilot Period as well as a sixmonth post-Pilot Period.335 During those periods, the Commission proposed to require the equities exchanges to collect and make available the order routing datasets and Exchange Transaction Fee Summaries in order to provide necessary benchmark information against which researchers could assess the impact of the Pilot.336 Two commenters supported the proposed six-month pre-Pilot and postPilot data collection periods.337 In contrast, two commenters suggested adopting three-month long pre-Pilot and post-Pilot Periods.338 The Commission desires to implement the Pilot in a manner that imposes the least amount of costs on the exchanges without compromising the ability of the Pilot to obtain useful data. The Commission believes that sixmonth pre- and post-Pilot Periods are necessary to establish a baseline against which to compare the data collected during the Pilot Period and any postPilot effects following the conclusion of the proposed Pilot. Although the Commission appreciates the desire of market participants to expedite the Pilot while constraining costs, the Commission considers six months to be necessary to provide the targeted statistical power for obtaining baseline data. As discussed above, statistical power largely is a function of the number of observations over a specified period of time. In order to shorten the pre- and post-Pilot Periods (e.g., to three months instead of six months) while maintaining the same statistical power, the Commission would need to increase the number of securities in the Pilot by at least 120 securities. As discussed above and consistent with the comments it received, the Commission desires to limit, not increase, the number of securities included in the Pilot. Accordingly, the Commission is not adopting a shorter duration for the pre- and post-Pilot Periods.
4. Early Termination
Proposed Rule 610T did not contain a specific provision regarding early termination of the Pilot. Several commenters recommended that the Commission develop specific criteria for evaluating the possibility that the Pilot may need to be terminated early.339 Some recommended that the Pilot specifically include a ‘‘kill switch’’ to effectuate an early termination.340 Several commenters supported the need for the Commission to address unanticipated negative consequences quickly,341 but one commenter cautioned that the Commission would need to act in a measured manner because the industry would need time to unwind the Pilot.342 One commenter suggested that the Commission might want to terminate the Pilot early if (1) it produced a ‘‘robust statistical sample set earlier than a year, such that [the Commission could] end the Pilot and proceed to adopt permanent rule changes’’ and (2) ‘‘if there is unintended impact from the Pilot that warrants a

343See Vanguard Letter, at 3. See also Angel Letter II, at 3 (noting that the Pilot could be suspended quickly if ‘‘there is abundant evidence one way or the other about the results,’’ such as ‘‘a dramatic increase in market quality for one particular treatment group,’’ in which case ‘‘that particular group’s treatment could become the new rule,’’ or ‘‘if the pilot produces fast and unequivocal results showing harm to one particular treatment group, that treatment should be halted’’). 344See Schwab Letter, at 2; Citi Letter, at 4. 345See Schwab Letter, at 2. 346See STANY Letter, at 4. 347 Verret Letter I, at 5. 348See 15 U.S.C. 78mm (setting forth the Commission’s authority, by rule, regulation or order, to conditionally or unconditionally exempt persons, transactions or securities (or classes thereof) from any Exchange Act provision, rule or regulation if such exemption is necessary or appropriate in the public interest and is consistent with the protection of investors). 349See State Street Letter, at 4; TD Ameritrade Letter, at 4; STANY Letter, at 3. See also Recommendation for an Access Fee Pilot (July 8, 2016), available at https://www.sec.gov/spotlight/ emsac/recommendation-access-fee-pilot.pdf (‘‘EMSAC Pilot Recommendation’’). 350See AJO Letter, at 2. 351See Proposing Release, supra note 2, at 13026. 352See Proposed Rule 610T(b)(1). See also Proposing Release, supra note 2, at 13026. When the Commission publishes this list, the pre-Pilot Period will have been in place for approximately five months. 353See Proposed Rule 610T(b)(1)(iii) (defining ‘‘primary listing exchange’’ for purposes of Rule 610T). 354See Proposing Release, supra note 2, at 13027–
28. The Commission notes that the proposed
language in Rule 610T(b)(3)(i) has been modified slightly. As proposed, Rule 610T(b)(3)(i) contained the phrase ‘‘throughout the duration of the Pilot, including the post-Pilot Period.’’ As adopted, the phrase ‘‘throughout the end of the post-Pilot Period’’ is being substituted for the phrase ‘‘throughout the duration of the Pilot, including the post-Pilot Period’’ to simplify the rule text without substantively changing the applicability of the posting requirement. 355See FIF Letter, at 5. 356 Id. 357See 17 CFR 242.1001(a)(1). SCI systems include all computer, network, electronic, technical, automated or similar systems of, or operated by or on behalf of, an SCI entity that directly support activities such as trading and order routing, among other things. 17 CFR 242.1000. 358See Proposing Release, supra note 2, 13026– 28. stoppage.’’ 343 Other commenters emphasized the need for the Commission to closely monitor the impact of the Pilot on retail investors in particular.344 For example, one commenter argued that if the Pilot data suggests ‘‘clear harm to the retail investor in . . . relevant execution quality metrics’’ like ‘‘quoted spread, depth of liquidity, intraday stock volatility, and opportunities for price improvement on impacted securities,’’ then the Pilot ‘‘should be immediately suspended.’’ 345 Another commenter urged the Commission to closely monitor the Pilot’s effect on thinlytraded stocks and establish ‘‘predetermined means for discontinuing the Pilot in the event that the reviewed data shows undue harm to market or execution quality.’’ 346 However, one commenter noted that the Commission is not obligated to ‘‘cease the [P]ilot if the costs to liquidity prove significant.’’ 347 The Commission acknowledges the concerns raised by commenters regarding the potential for unintended and unanticipated consequences to the equities markets that the Pilot may have. The Commission intends to carefully monitor for any such effects during the Pilot Period. However, the Commission does not believe that it is necessary to add a ‘‘kill switch’’ to Rule 610T because the Commission already has broad exemptive authority that obviates the need for a separate kill switch. For example, if at any time the Commission believes that the protection of investors may be compromised by the Pilot, the Commission has broad authority under
Section 36 of the Exchange Act to
modify or terminate the Pilot early.348
5. Inclusion of a Phase-In Period
The Commission did not propose a phase-in period for the Pilot. Three commenters recommended a phase-in period without elaborating on its purpose, though they referenced the EMSAC’s recommendation for an initial three-month phase-in period involving 10 stocks.349 A different commenter did not believe that the EMSAC’s threemonth phase-in period was necessary.350 The Commission has considered these comments and believes a phase-in approach is not necessary and unnecessarily would add to the length of the Pilot. Although such an approach would allow the markets and market participants to implement the required fee changes in a staged manner and provide an opportunity to address unforeseen implementation issues, the Commission continues to believe that, because exchange fees can become immediately effective upon their filing with the Commission, the markets and market participants are accustomed to dealing with frequent exchange fee changes in which fees can change on all stocks at once, or only for a subset of stocks or a subset of trading mechanisms. Accordingly, exchanges and market participants should be capable of accommodating the terms of the proposed Pilot with the advance notice contemplated by the Pilot. Further, although exchanges would be required to collect and report certain data, the proposed Pilot would not necessitate changes to exchange trading systems, and therefore, the Commission continues to believe a phased implementation schedule is not necessary to test the types of changes contemplated by the Pilot. E. Data The Commission proposed that certain data be collected and made publicly available in order to facilitate the Commission’s and researchers’ ability to assess the impact of the Pilot, as well as to promote transparency about the Pilot Securities and to provide basic information about equities exchange fees and changes to those fees during the Pilot.351 The Commission is adopting the Pilot Securities Lists, the Exchange Transaction Fee Summary, and the order routing datasets subject to the modifications described below.

Pilot Securities Exchange Lists and
Pilot Securities Change Lists As proposed, the Commission would publish, approximately one month before the start of the Pilot Period, the initial List of Pilot Securities, which identifies the securities in the Pilot and their designated Test Group (or the Control Group).352 Thereafter, each primary listing exchange 353 would publish a freely and publicly available daily Pilot Securities Exchange List of the Pilot Securities that are primarily listed on its exchange and also publish a Pilot Securities Change List of the cumulative changes to that list, and keep both lists available on their websites for five years.354 The Commission received one comment that was supportive of the proposed requirements for disseminating and updating the Pilot Securities lists, including the pipedelimited ASCII file format and the five year retention period.355 This commenter also had ‘‘no objections to the proposed posting requirements, providing there is adequate data security and controlled access.’’ 356 The Commission is not adopting any new requirements for data security with respect to the Pilot Securities lists because that data is not private or otherwise sensitive in nature and because the exchanges already are subject to Regulation SCI governing access to their systems that support trading.357 For the reasons stated in the Proposing Release, the Commission is adopting as proposed the requirements in Rule 610T(b) for the primary listing exchanges to publicly post on their websites downloadable files containing the Pilot Securities Exchange Lists and the Pilot Securities Change Lists.358 The Commission is adding one additional

359The Commission is also modifying the name of the field specified in proposed Rule 610T(b)(2)(ii)(E). The Commission proposed the field be named ‘‘Test Group.’’ As adopted, the field will be named ‘‘Pilot Group’’ to provide additional clarity. 360See Proposing Release, supra note 2, 13019, 13051. 361The Commission notes that the proposed language in Rule 610T(e) has been modified slightly. As proposed, Rule 610T(e) contained the phrase ‘‘each national securities exchange that trades NMS stocks. . . .’’ As adopted, the clause ‘‘that facilitates trading in NMS stocks’’ is being substituted for the phrase ‘‘that trades NMS stocks’’ to clarify that exchanges facilitate trading by their members in NMS stocks. In addition, the Commission notes that, as proposed, Rule 610T(e) contained a parenthetical which explained that data requirements set forth in subsection (e) were ‘‘applicable to securities having a price greater than $1.’’ As adopted, that parenthetical has been modified slightly to clarify that the requirements of subsection (e) apply to ‘‘securities having a price equal to or greater than $1.’’ 362See Proposed Rule 610T(e). See also Proposing Release, supra note 2, at 13029–30. 363See Proposing Release, supra note 2, at 13030. See also Rule 610T(e). 364See, e.g., SIFMA Letter, at 6; Better Markets Letter, at 7; Spatt Letter, at 4–5; and IEX Letter I, at 9. 365See Proposing Release, supra note 2, at 13031. 366See CFA Letter, at 5; Healthy Markets Letter I, at 22. 367See IEX Letter I, at 9. 368See Proposing Release, supra note 2, at 13030. 369See Proposing Release, supra note 2, at 13031. 370See RBC Letter I, at 5. 371See Better Markets Letter, at 7; CFA Letter, at 5; Healthy Markets Letter I, at 23. field, ‘‘stratum code,’’ to both lists.359 As discussed in the Proposal and above, the Commission will stratify Pilot Securities as it assigns them to the Test Groups and Control Group to ensure that each group has a similar composition, which facilitates comparison across groups.360 As it does so, the Commission will assign a stratum code to each Pilot Security that identifies that security’s liquidity strata. The code is a static value and, as such, will remain constant throughout the Pilot. The Commission will include this stratum code on the initial List of Pilot Securities that it disseminates. To link each Pilot Security and its stratum code, the Commission is requiring the primary listing exchanges to include this data element on each Pilot Securities Exchange List and Pilot Securities Change List. Including this field on each list will clearly identify each Pilot Security’s liquidity stratum, thereby allowing researchers to control for the fact that within some liquidity strata, the ratio of Test Group stocks to Control Group stocks is lower than it is for others, which should facilitate analysis of the Pilot’s data.
2. Exchange Transaction Fee Summary
As proposed, each exchange that trades NMS stocks would be required to compile, update monthly, and make freely and publicly available a dataset using an XML schema published on the Commission’s website that contains specified information on its fees and fee changes during the Pilot.361 In particular, each exchange would identify, among other things, the ‘‘Base’’ take fee (rebate), the ‘‘Base’’ make rebate (fee), the ‘‘Top Tier’’ take fee (rebate), and the ‘‘Top Tier’’ make rebate (fee), as applicable, as well as the Pilot Group (i.e., 1, 2, or Control) that applies to the fee being reported.362 Exchanges also would calculate the ‘‘average’’ and ‘‘median’’ per share fees and rebates, which the exchange would compute as the monthly realized average or median per-share fee paid or rebate received by participants on the exchange during the prior calendar month, reported separately for each participant category (registered market makers or other market participants), Test Group, displayed/non-displayed, and top/depth of book.363 In the Proposing Release, the Commission asked several questions about the Exchange Transaction Fee Summary including questions about the proposed form, content, and posting requirements. Commenters supported requiring the equities exchanges to publicly post the Exchange Transaction Fee Summary as well as the proposed fields included in the summaries.364 Among those questions included in the Proposing Release, the Commission specifically asked commenters to suggest types of information that should be captured on the Exchange Transaction Fee Summary that would be useful to make comparisons across exchanges, and a few commenters offered specific suggestions.365 Specifically, two commenters requested that the Exchange Transaction Fee Summary include the number of pricing tiers used by the exchanges, the number of firms that were in each tier, and information on transaction costs in each tier.366 Similarly, another commenter suggested that the Exchange Transaction Fee Summary provide context on the Base and Top Tier fees by including the number of member firms, by participant type, that qualified for the Base and Top Tier fees and rebates reported on the Exchange Transaction Fee Summary.367 While the Commission appreciates these suggestions, it believes that adding more granular details about specific pricing tiers, which can vary greatly by exchange, would overcomplicate the fee summaries such that it would be difficult to standardize the information, thereby rendering the data less useful to researchers when comparing exchanges for purposes of the Pilot.368 Further, with respect to the number of members qualifying for the Base and Top Tier fees and rebates, the Commission believes that the information that exchanges will report on average and median realized fees and rebates should be sufficient for purposes of analyzing the Pilot’s results, including any changes in order routing. We believe that the disclosure of the number of
members qualifying for the Base and Top Tier fees and rebates would also require other disclosures (including, e.g., such member’s trade volume at each tier) in order to provide context to the information. Providing all of these additional data points would increase the costs and complexity of the Pilot. The Commission however, does not believe that the incremental benefit of this information justifies additional costs and complexities. Accordingly, the Commission will not be requiring the exchanges to include additional information on their pricing tiers. As part of its request for comment in the Proposal on what additional information would be helpful to include in the Exchange Transaction Fee Summary, the Commission specifically asked whether other measures beyond average and median fees should be selected.369 In response, one commenter recommended that in addition to requiring the average and median per share fees and rebates, the Commission also require the ‘‘mode’’ per share fee and rebate (i.e., the most frequently paid fee and rebate by each exchange’s members), because the commenter believed it would ‘‘enable a more accurate comparison of the fees and rebates most often applied by each exchange.’’ 370 The Commission appreciates this suggestion, but continues to believe that for purposes of this Pilot, the proposed information on mean and median realized fees and rebates will be sufficient for purposes of analyzing the results of the Pilot, including any changes in order routing. Lastly, a few commenters requested that the Exchange Transaction Fee Summary information be hosted at a central location rather than posted on the exchanges’ individual websites.371 While the Commission recognizes that it could be more convenient if the information were made available in one central location, because the data must be made available unencumbered and in a standardized XML schema format, the Commission believes that any person would readily be able to obtain and combine the summaries posted by each equities exchange with minimal effort.

372See Proposing Release, supra note 2, at 13031. 373Proposed Rule 610T(d)(1)–(2). The Commission notes that the proposed language in Rule 610T(d) has been modified slightly. As proposed, Rule 610T(d) contained the phrase ‘‘each national securities exchange that trades NMS stocks. . . .’’ As adopted, the clause ‘‘that facilitates trading in NMS stocks’’ is being substituted for the phrase ‘‘that trades NMS stocks’’ to clarify that exchanges facilitate trading by their members in NMS stocks. 374See Proposing Release, supra note 2, at 13032. 375See, e.g., Barnard Letter, at 1; CII Letter, at 2– 3; Better Markets Letter, at 7; Invesco Letter, at 1– 2; Wellington Letter, at 1; CFA Letter, at 5; Clearpool Letter, at 6; ICI Letter I, at 5; RBC Letter I, at 5; Joint Pension Plan Letter, at 2; Oppenheimer Letter, at 2; IEX Letter I, at 10; Capital Group Letter, at 4; Healthy Markets Letter I, at 24; Angel Letter, at 1; Verret Letter I, at 1, 7; Spatt Letter, at 3. 376See, e.g., Clark-Joseph Letter, at 1; Nuveen Letter, at 2; NYSTRS Letter, at 1; RBC Letter I, at 2; Invesco Letter, at 2; CFA Letter, at 1; State Street Letter, at 2; AJO Letter, at 1; Vanguard Letter, at 2. 377See, e.g., AJO Letter, at 3–4; ICI Letter I, at 5; CFA Letter, at 5. 378See infra notes 382, 386, and 395–397. 379See supra notes 41–46. 380See infra note 404. 381See Proposing Release, supra note 2, at 13031, 13033. 382See FIA Letter, at 3 fn. 8; FIF Letter, at 6; Citadel Letter, at 3 fn. 5; IEX Letter I, at 10. 383See FIA Letter, at 3 fn. 8; IEX Letter I, at 10. 384See, e.g., FIX Tag 528 (Order Capacity) under FIX 4.4 and Fix Tag 47 (Rule80A) under TIF 4.2, available at http://btobits.com/fixopaedia/ index.html. 385See Proposing Release, supra note 2, at 13033. 386See FIA Letter, at 3; SIFMA Letter, at 8; Citadel Letter, at 3; Citi Letter, at 5–6. 387SIFMA Letter, at 8. See also Cboe Letter I, at 3 fn. 8. Because of this, the Commission is not adopting a requirement on exchanges to consolidate this material and make it available in a central location.
3. Order Routing Data
To facilitate an examination of the impact of the Pilot on order routing behavior, execution quality, and market quality, the Commission proposed to require throughout the Pilot (including during the pre-Pilot Period and the postPilot Period) that each equities exchange prepare and publicly post a monthly downloadable file containing sets of anonymized order routing data in accordance with the specifications proposed in Rule 610T(d).372 Specifically, Rule 610T(d) would require exchanges to provide the order routing information in two datasets— one for liquidity-providing orders and one for liquidity-taking orders, both aggregated by day, security, and brokerdealer.373 The Commission further proposed that equities exchanges would be required to anonymize the identity of individual broker-dealers before making the order routing datasets publicly available, using an anonymization key provided by the Commission.374 A number of commenters supported the proposed requirements regarding the order routing datasets, expressing the belief that these requirements would provide researchers with useful data that would facilitate an analysis of the impact of transaction fees and rebates on order routing, execution quality, and market quality.375 Several commenters believed the data would enable the Commission to make data-driven decisions on potential future equity market structure policy initiatives.376 Others specifically supported the website posting requirement to make the data freely and publicly available.377 In addition, other comments addressed matters such as: Separating held and not-held orders in the datasets, separating principal from agency orders in the datasets, and not collecting ‘‘parent order’’ routing information.378 Other commenters expressed concern that not collecting similar data from non-exchange venues would decrease the utility of the data and provide the Commission with an incomplete picture of the Pilot’s impact.379 Other commenters were critical of the proposed order routing data requirements because they believed, despite the anonymization and aggregation requirements, that publicly available data could be reverse engineered to reveal commerciallysensitive information about individual broker-dealers.380 These concerns are discussed further, below. a. Held and Not-Held Orders The Commission proposed to require exchanges to separate out held and notheld orders in the order routing datasets and requested comment on whether orders should be separated out in that manner and whether there are certain shared characteristics of such orders that would be beneficial to assess when analyzing the Pilot data.381 In response, several commenters stated that exchanges currently do not capture whether orders are held or not held.382 Two commenters added that capturing that information would impose additional costs on market participants who would need to update their systems to include this information in the order messages they send to
exchanges, as well as impose additional costs on exchanges to capture and report whether an order is held or not held.383 The Commission has considered these comments and has determined not to require the exchanges to separate held and not-held orders in the order routing datasets. In proposing to require capture of held and not-held orders, the Commission sought to include a data field that is readily available to and currently captured by exchanges and that would provide insight into the capacity in which a broker-dealer is handling orders. In turn, that information could be useful to assess the broker-dealer’s routing of those orders. For example, orders that are ‘‘held to the market’’ may be routed differently than orders that are ‘‘not held’’ and for which the broker-dealer exercises more discretion in their execution. By separating out these orders, researchers would have access to an additional metric that potentially could be helpful in analyzing the Pilot data and parsing the results. As commenters have indicated, however, broker-dealers do not transmit this information to exchanges and exchanges thus do not capture it. The Commission does not wish to impose new data collection requirements with respect to this Pilot data field, and therefore is not adopting this element. However, as detailed below, the Commission is adopting a new requirement for exchanges to instead separate out orders based on their order capacity (e.g., principal, riskless principal, and agency), which information currently is transmitted to exchanges by broker-dealers.384 b. Principal Order Flow and Order Capacity In response to the Commission’s question in the Proposing Release about what data are necessary to facilitate an analysis of the potential conflicts of interest associated with transaction fees and rebates,385 several commenters requested that the order routing datasets exclude orders marked as principal or riskless principal because the potential conflicts of interest posed by exchange transaction fees and rebates pose a potential harm primarily when brokerdealers are routing orders for customers in an agency capacity and may be unduly influenced by exchange fees and rebates to the detriment of obtaining the best execution for the customer’s order.386 To the extent a broker-dealer is routing its own proprietary order and is unduly influenced by exchange fees and rebates, then, at worst, it would only be harming itself. In other words, as noted by one commenter, ‘‘a broker may route principal orders to maximize rebates and minimize access fees which would not be considered a conflict of interest.’’387 Without separating out orders by their order capacity, one commenter argued that the order routing datasets could generate ‘‘misleading results’’ because the trades of various market participants could be aggregated at the same broker

388See FIA Letter, at 3 fn. 9.
389See Clearpool Letter, at 6.
390 Id.
391See, e.g., Fidelity Letter, at 8; ICI Letter I, at 5; SIFMA Letter, at 5–6; IEX Letter I, at 2. See also Proposing Release, supra note 2, at 13033. 392See supra notes 224–225 and accompanying text. 393See Mulson Letter I. 394See supra Section II.C.10. due to ‘‘direct market access arrangements,’’ and these orders would be indistinguishable from customer orders routed by that broker.388 In this way, agency orders (which are subject to conflicts of interest concerns that are relevant to the Pilot) could be mixed in with principal orders (which are not subject to conflicts of interest concerns that are relevant to the Pilot) and the inability to distinguish them could cloud the results. Accordingly, one commenter recommended separating principal and agency orders in the order routing datasets, while continuing to include both types of order flow.389 The commenter believed that specifically identifying the extent to which orders are principal orders or agency orders ‘‘would further facilitate the analysis of order flow and a better understanding of the efficacy of the [P]ilot.’’ 390 After careful consideration of these comments, the Commission has determined to require exchanges to separate out orders by order capacity (e.g., principal, riskless principal, and agency). Requiring exchanges to separately aggregate orders according to their order capacity will allow researchers to more precisely parse the data as recommended by several commenters, particularly when analyzing the potential conflicts of interest in broker-dealer routing presented by exchange fee-and-rebate pricing models. For example, researchers will be able to separate out and exclude principal orders when studying conflicts of interest, as conflicts of interest do not present the potential for harmful impact with respect to such orders as they do for agency orders where the broker-dealer is routing for others. In addition, researchers will be able to include orders of any order capacity when studying other questions, such as intermediation, queue length, and time to execution, as such issues are relevant to orders of any capacity. Further, the Commission believes that principal orders should be included in the order routing datasets, as the Pilot is designed to assess more than just conflicts of interest between brokers and their customers in order routing. It also is designed to observe the impact of exogenous shocks to transaction fees and rebates on execution quality and market quality broadly. Accordingly, the Pilot will provide the opportunity to obtain useful data on matters such as intermediation, queue length, and time to execution; the impact of fees and rebates on liquidity adding and liquidity removing activity; the relationship between payment of rebates on making activity (or taking activity on an inverted exchange) and fee levels for taking activity (or making activity on an inverted exchange); and the impact of fees and rebates on order routing behavior generally.
Consideration of these issues directly implicates principal order flow and, as such, the Commission believes it is critical for the aggregated volume statistics included in the order routing datasets to include principal orders.
c. Order Designation
In response to questions in the Proposing Release on specific measures and data that would facilitate an analysis of the effects that changes to transaction fees and rebates have on order routing behavior, execution quality, and market quality, several commenters recommended that the Commission analyze the impacts of fees and rebates on various aspects of the execution quality of investors’ limit orders.391 Further, on the impact that prohibiting rebates may have on quoted spreads and displayed liquidity, commenters also disagreed about the willingness and ability for investors, other than those that are motivated by rebate capture, to post liquidity in order to capture the quoted spread.392 In addition, in attempting to utilize transaction data to analyze the impact of reduced or eliminated rebates, one commenter recommended that the dataset exclude orders that presently are not eligible for rebates, such as those designated for participation in opening and closing auctions.393 While analyzing the impact of reduced or eliminated rebates is one potential analysis for which the Pilot’s data may be useful, the Pilot’s purpose is broader in scope. As such, the Commission continues to believe that it is appropriate for the order routing datasets to capture all liquidityproviding and liquidity-taking orders. However, in response to the commenters’ recommendations discussed above and in an effort to ensure that the order routing data be as useful as possible and facilitate an analysis of the impacts of the Pilot, the Commission has determined to further refine the order routing dataset by requiring exchanges to report separately the volume statistics by ‘‘order designation,’’ which will require exchanges to separate out post-only orders as well as auction orders. Separating the volume statistics in this manner will allow isolation of the cumulative number of post-only orders, which are limit orders that include instructions to never remove liquidity, and may be more reflective of a rebatesensitive market participant. With the data further refined in this manner, the Commission believes the data will be more useful in analyzing the impacts of the Pilot both in comparing the pre-Pilot data to the Pilot data and in comparing the data across the Test Groups and Control Group during the Pilot. In particular, the further refinement will facilitate assessment of the impact of the Pilot on the willingness of investors to passively post orders and their ability to obtain queue priority (i.e., represent the best price in the exchange’s limit order book) and capture the quoted spread when doing so (i.e., buy on the bid and sell on the offer).394 Furthermore, with respect to auction orders, which are orders specifically designated for execution in either an opening or closing auction, instead of separating out auction orders, exchanges may instead elect to simply exclude them from the order routing datasets, as an alternative means of complying with the order
designation requirement. The Commission has determined to allow the exchanges to choose between these two approaches so that they may choose the option that is the least burdensome. If exchanges choose to include auction order data in the order routing datasets, they will need to comply with the requirement by separating orders by order designation, so that these orders may be separately identified and accounted for in any analyses of the Pilot’s data. The ability to isolate auction orders recognizes the uniqueness of the auction process and will facilitate separation of that data in order to study the Pilot’s impact on trading during the regular market session without potentially biasing the results by including auction activity, for which different trading rules, order types, and fees apply.

395See, e.g., Healthy Markets Letter I, at 24–25; Pragma Letter, at 3; NYSE Letter I, at 9–10; NYSE Letter II, at 12–13; Viable Mkts Letter, at 2; Babelfish Letter, at 3. 396See e.g., Healthy Markets Letter I, at 24–25; Pragma Letter, at 3; NYSE Letter I, at 9–10; NYSE Letter II, at 12; Viable Mkts Letter, at 2. 397See Babelfish Letter, at 3. 398See FINRA OTC Transparency Data, available at https://otctransparency.finra.org/. 399See supra note 302 and accompanying text. 400See Healthy Markets Letter I, at 23–24; Clearpool Letter, at 6. 401Healthy Markets Letter I, at 23. 402See Citadel Letter, at 3; TD Ameritrade Letter, at 6. In response to the Commission’s solicitation of comment in the Proposing Release on whether the CAT repository, if it were operational, would provide sufficient data to evaluate the Pilot, one commenter stated that it believed the data reported from the CAT would provide the necessary information with respect to order routing data. See FIF Letter, at 2. 403See Proposing Release, supra note 2, at 13031 n. 172 and accompanying text. 404See, e.g., FIA Letter, at 3; Virtu Letter, at 7– 8; SIFMA Letter, at 6; FIF Letter, at 2; Citadel Letter, at 4; Citi Letter, at 6; TD Ameritrade Letter, at 5; STANY Letter, at 5; IEX Letter I, at 10; Credit Suisse Commentary, at 6; Morgan Stanley Letter, at 4. 405See Clearpool Letter, at 6–7. d. Broker Routing Data Several commenters addressed the utility of obtaining order routing data from broker-dealers that route customer orders in assessing the potential conflicts of interest related to transaction fees and rebates.395 Several of these commenters explained that obtaining data from broker-dealers (in addition to or in place of obtaining such data from exchanges) would facilitate an analysis of the impact of transaction fees and rebates on order routing behavior and potential conflicts of interest from the perspective of customers, as the brokers would have information that can be used to assess the execution quality of a ‘‘parent order’’ and would provide information on the broader universe of potential routing destinations, including non-exchange trading venues.396 One commenter added that investors needed to conduct their own analyses of their orders to understand the impact of the Pilot on their brokers.397 The Commission is not requiring data collection from broker-dealers or nonexchange trading venues. The order routing datasets will include aggregated data from exchanges (as opposed to individual order level data from brokerdealers) representing the sum totals of the ‘‘child’’ orders that are processed by an exchange. While the Pilot will not capture the entire lifecycle of a ‘‘parent’’ order from its inception, the Commission believes that its approach will minimize the implementation costs on market participants while ensuring that the Commission and researchers have useful data on child orders to observe the impacts of introducing exogenous shocks to exchange transaction fees and
rebates. The order routing data provided by the exchanges represents the information that would be directly correlated to these exogenous shocks. Data that is available elsewhere 398 will provide the ability to understand any observed changes in order flows or market share to nonexchange venues during the Pilot. Further, the Commission agrees with the commenters that noted that market participants need to conduct their own analyses of their own order flow. If market participants conduct their own analyses, including parent order-level analyses, and wish to provide that information to the Commission and the public, the Commission would be able to consider the information in assessing the Pilot’s ultimate impact on order routing behavior, execution quality, and market quality. The Commission encourages market participants to conduct analyses and make the results of their analyses public. The Commission also encourages any interested party that prepares an analysis of the Pilot to submit it to the Commission for posting on the Commission’s website.399 e. Directed Orders Two commenters recommended that the order routing datasets identify whether orders are directed or nondirected.400 One of these commenters believed that directed orders do not feature ‘‘the same level of discretion and conflicts of interest that are the primary focus of the’’ Pilot.401 After careful consideration of these comments the Commission has determined not to require the order routing datasets to identify directed orders. The Commission recognizes that researchers may be interested in isolating orders directed by customers to specific exchanges because these orders may not be subject to the same potential conflicts of interest that may be present when a broker chooses where to route a customer order. However, separating out directed orders in the datasets (which report aggregated data and not order-byorder data) would require exchanges and broker-dealers to incur additional costs in preparing the Pilot’s order routing data. Further, the Pilot is designed to assess more than just conflicts of interest between brokers and their customers in order routing, and separate identification of directed and non-directed orders is not germane to the other questions the Pilot is designed to explore. Accordingly, the Commission believes that the additional implementation costs that adding such a requirement would impose are not justified by any benefits that may accrue from identifying, on an aggregated basis, directed orders in the order routing data. f. Utilizing the Consolidated Audit Trail Two commenters recommended that, instead of requiring separate order routing datasets, the Commission instead use data that the equities exchanges will report to the Consolidated Audit Trail (‘‘CAT’’).402 In the Proposing Release, the Commission stated that if the equities exchanges are reporting to the CAT at the time the Pilot commences, they would be able to compile the order routing datasets
by utilizing the data they collect pursuant to the CAT national market system plan.403 However, there have been delays in the development and building of the CAT, and the reporting required by the first phase of the CAT NMS Plan has been delayed. Although the exchanges and FINRA have recently begun to report certain data to the CAT central repository, they continue to work to fully implement the first phase of the CAT NMS Plan, including linkages between reported events and regulators’ query functionality. The Commission believes that it is important to proceed with the Pilot and not delay the Pilot until the exchanges have begun full reporting to the CAT and the CAT operates in a manner that would facilitate the data analysis contemplated by the Pilot. g. Anonymization and Public Availability Several commenters expressed concerns about having the exchanges publicly post the order routing datasets, despite the requirement that the exchanges anonymize the identities of broker-dealers before making the datasets publicly available. These commenters believed that the order routing data could potentially be ‘‘reversed engineered’’ such that market participants might be able to ascertain the identities of individual brokerdealers in some circumstances.404 In contrast, one commenter acknowledged that ensuring confidentiality is ‘‘critical’’ and was ‘‘pleased to see that the SEC has recognized this in proposing anonymizing certain of the proposed data to protect confidential information.’’ 405 Of the commenters concerned about the potential for reverse engineering, one of these commenters provided an example of how the information could

406Citadel Letter, at 4.
407See id.
408 Id.
409 Id.
410See TD Ameritrade Letter, at 5.
411See, e.g., Citadel Letter, at 4; Credit Suisse Commentary, at 6; IEX Letter I, at 10; Morgan Stanley Letter, at 4; STANY Letter, at 5; SIFMA Letter, at 6–7; TD Ameritrade Letter, at 5; FIF Letter, at 7. 412See Lipson Letter, at 1; Spatt Letter, at 3. 413See Citadel Letter, at 5; Citi Letter, at 6. 414See SIFMA Letter, at 7; STANY Letter, at 5; Fidelity Letter, at 11. 415Several commenters expressed concerns that the equities exchanges would have access to the Broker Dealer Anonymization Key. See, e.g., Virtu Letter, at 8; SIFMA Letter, at 7; FIF Letter, at 2; STANY Letter, at 5; IEX Letter I, at 10; Morgan Stanley Letter, at 4. As adopted, the exchanges would not have access to the Broker Dealer Anonymization Key, which addresses the commenters’ concerns. 416The Commission will deem broker-dealer identifying order routing data as being subject to a confidential treatment request under 17 CFR 200.83 without the need to submit a request. The Freedom of Information Act provides at least two potentially pertinent exemptions under which the Commission has authority to withhold certain information. See 5 U.S.C. 552(b)(4) and (8). 417See Proposing Release, supra note 2, at 13033 (asking whether commenters think exchanges should be required to report the datasets directly to the Commission). Further, in its Proposal, the Commission noted that it considers the order routing data to be ‘‘regulatory’’ information and proposed to prohibit exchanges from accessing or using the information for commercial purposes. See id. at 13032. The Commission is adopting as proposed the prohibition on exchange personnel accessing the data for commercial purposes, as exchanges will have access to the information. 418See id. at 13032. See also Rule 610T(d)(1)(iv) and (d)(2)(iv). be reverse engineered if ‘‘a market participant could direct a large order in a particular symbol to a specific brokerdealer, and then identify the presence of that order’’ in the order routing datasets.406 This commenter added that market participants may also be able to compare the order routing datasets with reports published pursuant to 17 CFR
242.605 (Rule 605 of Regulation NMS)
to determine the identity of brokerdealers.407 Once a broker-dealer’s identity is likely known, this commenter believed that competitors could use the order routing datasets to discern that broker’s ‘‘(a) market share and activity in a given security, (b) overall routing practices, and (c) relative aggressiveness or passiveness in specific securities.’’ 408 This commenter also believed that strategies used by institutional investors that are customers of broker-dealers ‘‘may also be susceptible to reverseengineering.’’ 409 Another commenter added that it believed ‘‘market participants and others will be able to identify certain broker-dealers routing strategies by comparing the Pilot data to publicly available 17 CFR 242.606 (Rule
606) disclosures, or by other means,’’
although it did not specify those other means.410 Several of the commenters that expressed concern about the public availability of the order routing data, despite the proposed anonymization requirements, recommended approaches to address their concerns. Some of these commenters stated that the Commission should receive order routing data at the broker-dealer level, but that the public should only have access to data that is further aggregated, such that the data would include statistics for firms of similar types or business models, or simply aggregate all orders received by the exchange.411 However, in contrast, two commenters noted that the order routing data aggregated by broker would be important to analyses undertaken by researchers and therefore should be made more broadly available.412 Two other commenters suggested that if the order routing data aggregated by broker would be helpful for researchers, the Commission should provide that data to researchers only if they sign a nondisclosure agreement.413 In addition, three commenters recommended that if order routing datasets are to be made publicly available on exchange websites, they should be subject to a 120 day delay instead of a 30 day delay.414 The Commission appreciates commenters’ concerns about the need to safeguard the confidentiality of the order routing datasets. The Commission agrees that if market participants were able to identify specific broker-dealers in the datasets, there is the potential that the data could be reverse engineered to reveal proprietary information about trading attributable to specific broker-dealers. The Commission has revised its approach to eliminate the public availability of the order routing datasets to help address these concerns, while still furthering the goals of the Pilot. More specifically, to address commenters’ concerns with the public availability of the data and the exchanges’ role in preparing it for dissemination, the Commission is not adopting the requirement for the exchanges to anonymize 415 and publicly post the order routing data. The Commission, however, believes it is important for the Commission itself to have access to the order routing dataset, so the Commission can consider the effects of rebates and transaction-based fees on order routing behavior, execution quality, and market quality. Accordingly, given the potential for reverse engineering, the Exchanges will be required to provide order routing data directly to the Commission. While the Commission anticipated benefits from market participants, researchers, and others in conducting independent analysis of the Pilot and its impacts, the Commission has carefully balanced the concerns about possible reverse engineering of the order routing data against these benefits. The Commission believes that it can assess the effects of the transaction-fee and rebate models on order routing behavior and thereby achieve this goal of the Pilot without requiring public disclosure of order
routing data attributable to a specific broker-dealer. The Commission is not adopting the requirement for exchanges to make public in an anonymized form the order routing data, but the exchanges will instead identify individual brokerdealers by MPID or CRD number in the order routing data they send to the Commission. The Commission recognizes that order routing data attributable to a specific broker-dealer is particularly sensitive and is non-public information.416 The Commission, however, intends to make public analyses, results, and studies using the order routing data. In determining whether and how to make public this or any other information, the Commission will be sensitive to the concerns articulated by commenters and will consider steps such as aggregating or anonymizing order routing data. Specifically, the Commission is adopting a requirement for each exchange to prepare and transmit directly to the Commission, in pipedelimited ASCII format, no later than the last day of each month, a file containing sets of order routing data.417 While the Commission is not requiring the exchanges to anonymize the data and thus will no longer provide exchanges with the Broker-Dealer Anonymization Key, the Commission is requiring each exchange to provide its order routing data by broker-dealers’ CRD number and MPIDs in order to provide aggregated broker-dealer level data to the Commission to facilitate its analysis of the data.418 The Commission believes that the suggested alternative to further aggregate the datasets, for example, to combine the data of several firms together or combine all firms together, would seriously compromise the ability of researchers to investigate the potential conflicts of interest in routing because researchers would not be able to see an individual broker-dealer’s orders across all exchanges and thereby would not be able to assess how any particular broker-dealer may have been influenced by fees and rebates at

419See, e.g., CFA Letter, at 5 (believing that ‘‘breaking the data out at the broker-dea[le]r level will permit a closer examination of how different broker-dealers may change their order routing behavior in response to changes in fees and rebates at each exchange.’’); Lipson Letter, at 1; Spatt Letter, at 3; Better Markets Letter, at 7; Healthy Markets Letter I, at 24 fn. 87. 420The Commission notes that the proposed language in Rule 610T(d)(1)(vi)(F), (d)(1)(xii)(H), and (d)(2)(vi)(F) has been modified slightly. As proposed, Rule 610T(d)(1)(vi)(F) and Rule 610T(d)(2)(vi)(F) both noted that the order size code at the largest share bucket was ‘‘> 10,000.’’ As adopted, the largest share bucket order size code will be reflected as ‘‘≥ 10,000 share bucket.’’ In addition, as proposed, Rule 610T(d)(1)(xii)(H) set forth a time frame of ‘‘> 30 minutes of order receipt.’’ As adopted, that time frame will be clarified to state that the time frame is ‘‘≥30 minutes of order receipt.’’ 421See Proposing Release, supra note 2, at 13033. 422See id. 423See id. at 13033–34. 424See FIF Letter, at 7–8. 425See id. at 8. 426See FIF Letter, at 8. 427See Cboe Letter I, at 21. 428See supra notes 349–350 and accompanying text. 429Although broker-dealers will need to account for different fee and rebate levels across two Test Groups and the Control Group if exchanges maintain different fee and rebate levels across the treatment groups, they will have seven months before the start of the Pilot Period to update their execution algorithms, including to accommodate the prohibition on rebates and Linked Pricing in the no-rebate Test Group. 430The fields in the Exchange Transaction Fee Summary that are calculated based on a look-back period to the prior month are: Rule 610T(e)(9) (month and year of the average and median figures); (12) average take/make; and (13) median take/make. 431See Proposed Rule 610T(d). different exchanges.419 Because brokerdealer level data already is consolidated (i.e., the data would not separate out individual customer activity), adding another level of consolidation by grouping broker-dealers together would cloud insight into the potential conflicts of interest question, rendering the data potentially useless for the purpose of studying conflicts of interest. Accordingly, the Commission continues to believe that it needs access to the order routing data in its proposed form, without further aggregation of the data.420 F. Implementation The Commission proposed to publish a notice setting forth the start and end dates of the pre-Pilot, Pilot, and postPilot Periods.421 If applicable, the Commission also would publish a notice if it determines to suspend the one-year sunset of the Pilot Period.422 As discussed in the Proposing Release, the start date of the pre-Pilot Period would be one month from the date the Commission issues the notice, and the end date of the pre-Pilot Period would be six months from the pre-Pilot Period’s start date. Thus, the
Pilot, which is to start at the conclusion of the pre-Pilot Period, would begin seven months from the date the Commission issues the notice. The post-Pilot Period would commence at the conclusion of the Pilot and would end six months from the post-Pilot Period’s start date. The Commission proposed to publish the initial notice setting forth the start date for each of the Pilot’s three periods, and do so with a one-month minimum advance notice in order to allow the equities exchanges to finalize their preparations for the Pilot’s pre-Pilot Period, as well as provide at least a seven-month advance notice to market participants of the start date on which the Pilot’s conditions would go into effect.423 One commenter agreed that a onemonth period between the Commission’s notice and the start of the pre-Pilot Period would be ‘‘sufficient provid[ed] there are no changes to the Pilot securities lists and assigned test/ control groups.’’ 424 This commenter also agreed that the proposed sevenmonth period following the Commission’s notice would be ‘‘sufficient to prepare for the Pilot.’’ 425 However, this commenter requested that ‘‘any technical specification materials required to support implementation of the Pilot be reviewed with the industry and finalized in an expeditious manner, six months prior to the launch of the pre-Pilot data gathering phase,’’ which the commenter believed would ‘‘allow[ ] necessary time for industry firms to properly scope necessary development work and assign respective resources.’’ 426 Another commenter, however, did not believe that a one month period prior to the start of the pre-Pilot period would be sufficient for the industry to prepare and instead estimated that ‘‘the implementation of the pre-Pilot processing alone [would] take between three to four months.’’ 427 As discussed and addressed above, a few commenters recommended that the Pilot begin with a limited phase-in period with a small number of securities.428 After careful consideration of the comments received, the Commission continues to believe that the proposed implementation approach should provide adequate notice and time for those impacted by the Pilot to prepare for its requirements. The Pilot will begin with a six-month pre-Pilot period during which exchanges will not need to revise their fees to comply with the Pilot. At the conclusion of the pre-Pilot Period, exchanges will be required to revise any of their fees, which will apply to the Pilot Securities, that currently exceed the terms of the Pilot’s Test Groups. While the Exchange Act allows exchanges to file their fees for immediate effectiveness, exchanges may choose to preview their Pilot-related fee changes to their membership to provide them with additional time to adjust their order routing systems in response to those changes.429 The Commission does not anticipate that technical specification materials will be required to support implementation of the Pilot by broker-dealers because the Pilot
solely concerns exchange fees which exchanges commonly adjust with little or no advance notice though immediately effective fee filings with the Commission. Therefore, brokerdealers currently are accustomed to accommodating the types of fee changes that would be required to comply with the requirements of the Pilot. Further, the Commission believes that publishing the start date for each of the Pilot’s three periods in advance, with at least one month’s advance notice, will provide the exchanges with time to prepare the three types of data required by the Pilot. First, because the Commission will determine the initial List of Pilot Securities, the exchanges will only need to perform the ministerial task of separating out their listed issuers and creating the Pilot Securities Exchange Lists and Pilot Securities Change Lists. Second, the Exchange Transaction Fee Summaries will require each exchange to summarize its own fees, for which it is solely responsible, in the specified XML format. For the initial Exchange Transaction Fee Summary, which would be posted prior to the start of trading on the first day of the pre-Pilot Period, exchanges would not need to include information that is calculated on a look-back basis, because the lookback period for that report would predate the pre-Pilot Period. Accordingly, preparation of the initial Exchange Transaction Fee Summary report should be streamlined.430 Finally, the order routing datasets, because they also are prepared on a look-back basis, will not need to be prepared until the end of the second month of the pre-Pilot Period (as it will contain data for the first month of the pre-Pilot period).431 Accordingly, the Commission continues to believe that the proposed time frames set forth in Rule 610T(c)(4) are sufficient to allow the equities exchanges and market participants to prepare for the requirements of the pre-Pilot Period, the Pilot Period, and the post-Pilot Period.

432 15 U.S.C. 78k–1(a)(1)(C), (a)(2); see also id. sec. 78k–1(c)(1) (stating that self-regulatory organizations shall not make use of the mails or any means or instrumentality of interstate commerce to collect, process, distribute, publish, or prepare for distribution or publication any information with respect to quotations to assist, participate in, or coordinate the distribution or publication of such information, or to effect any transaction in, or to induce or attempt to induce the purchase or sale of, any such security in contravention of such rules and regulations as the Commission shall prescribe to ‘‘assure the . . . fairness and usefulness of the form and content of such information’’). 433 15 U.S.C. 78f(b)(4), (b)(5), (b)(8). The Commission also has authority to adopt the Pilot pursuant to Exchange Act Sections 17(a) [15 U.S.C. 78q(a)] (requiring each exchange to make and keep’’ for prescribed periods such records, furnish such copies thereof, and make and disseminate such reports as the Commission, by rule, ‘‘prescribes as necessary or appropriate in the public interest, for the protection of investors, or otherwise in furtherance of the purposes of [the Act]’’), and 23(a) [15 U.S.C. 78w(a)] (granting the Commission the power to make such rules and regulations as may be ‘‘necessary or appropriate to implement the provisions of this chapter’’ for which the Commission is responsible or for the execution of the functions vested in the Commission by the Act). 434City of Providence, Rhode Island v. BATS Global Mkts., Inc., 878 F.3d 36, 41 (2d Cir. 2017). 435See Cboe Letter I, at 10–11. See 15 U.S.C. 78s(b)(3)(A). 436 15 U.S.C. 78s(b)(1). 437 Id. Sec. 78s(b)(3). 438 Id. Sec. 78s(b)(3)(B), (C). 439 Id. Sec. 78s(c). 440 5 U.S.C. 500, et seq. 441A few commenters suggested that the Pilot ‘‘would not withstand judicial scrutiny’’ because certain aspects of the Pilot were ‘‘arbitrary and capricious and not in accordance with law.’’ See, e.g., Nasdaq Letter I, at 3. Specifically, these commenters challenged the sufficiency of the Commission’s economic analysis, the exclusion of non-exchange trading centers from the Pilot, the inclusion of ETPs in the Pilot, the ability of the Pilot to provide the Commission with usable data, and the Commission’s decision to pursue a Pilot instead of other market structure initiatives. See, e.g., Nasdaq Letter I, at 1–4, 8–9, 11; Cboe Letter I, at 12; NYSE Letter, at 2–3, 7. These specific concerns are addressed in Section IV (discussing the Commission’s economic analysis), Section II.A.4 (discussing the exclusion of non-exchange trading centers from the Pilot), Section II.B.3 (discussing the inclusion of ETPs in the Pilot),
Section II.E (discussing the ability of the Pilot to
provide the Commission with usable data), notes 307–319 supra (discussing the Commission’s decision to pursue a Pilot in conjunction with other market structure initiatives). 442Nasdaq Letter I, at 11 (quoting Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983)). 443 Id. at 11–12. 444 Verret Letter I, at 5–6. 445 Id. at 6. No comments were received regarding the required notice to suspend the automatic sunset provision. Accordingly, the Commission adopts this aspect of the Pilot for the reasons outlined in the Proposing Release. G. The Commission’s Authority To Conduct the Pilot The Commission is adopting the Pilot in furtherance of its statutory responsibilities. In 1975, Congress directed the Commission, through enactment of Section 11A of the Exchange Act, to use its authority under the Exchange Act to facilitate the establishment of a national market system to link together the multiple individual markets that trade securities. Congress intended the Commission to take advantage of opportunities created by new data processing and communications technologies to preserve and strengthen the securities markets. Congress also directed the Commission to exercise this authority ‘‘to carry out’’ certain ‘‘objectives,’’ which include assuring: ‘‘economically efficient execution of securities transactions’’; ‘‘fair competition among brokers and dealers, among exchange markets, and between exchange markets and markets other than exchange markets’’; the ‘‘availability . . . of information with respect to quotations for and transactions in securities’’; and ’’ an opportunity . . . for investors’ orders to be executed without the participation of a dealer.’’ 432 In addition, the Exchange Act elsewhere requires that the rules of national securities exchanges (i) ‘‘provide for the equitable allocation of reasonable dues, fees, and other charges among its members and issuers and other persons using its facilities,’’ (ii) not be designed to ‘‘permit unfair discrimination between customers, issuers, brokers, or dealers,’’ and (iii) not ‘‘impose any burden on competition not necessary or appropriate in furtherance of the purposes of [the Act].’’ 433 Through these provisions Congress conferred on the Commission ‘‘broad authority to oversee the SROs’ ‘. . . operation . . .’ of the national market system.’’ 434 And it is pursuant to this authority that the Commission originally adopted Rule 610(c). The Pilot reflects the Commission’s efforts to evaluate, in light of changing market conditions, whether the existing transaction-based fee and rebate structure continues to further the statutory goals. In that sense, the Pilot follows as an appropriate progression from Rule 610, and it represents an important step in the Commission’s continuing obligation to implement Congress’s objectives for the national market system. The Commission disagrees with the suggestion by one commenter that the Pilot is inconsistent with Exchange
Act
Section 19(b)(3)(A), which sets out part
of the process by which proposed rule changes by self-regulatory organizations may become effective.435 Contrary to the commenter’s suggestion, nothing in
Section 19 interferes with the
Commission’s authority described elsewhere in the Exchange Act. Indeed,
Section 19 itself makes clear that the
Commission retains ultimate authority over the rules of registered exchanges, providing that ‘‘[n]o proposed rule change [by a self-regulatory organization] shall take effect unless approved by the Commission or otherwise permitted in accordance with [Section 19(b)]’’ 436 and making clear that the Commission retains authority to suspend and institute proceedings to approve or disapprove even those exchange rules that are permitted to take effect upon filing with the Commission.437 Moreover, Section 19 explicitly permits the Commission to summarily implement or suspend any such proposed rule changes if, in the Commission’s view, doing so would serve the public interest, protect investors, or assist in maintaining fair and orderly markets.438 And it makes clear that the Commission retains authority to amend exchanges’ rules on its own initiative.439 Commenters also disagreed about whether the Pilot complied with the Commission’s statutory obligations under the Administrative Procedure Act 440 (‘‘APA’’) and whether the Pilot is consistent with the Exchange Act.441 For example, working from the premise that the APA requires the Commission to ‘‘ ‘examine[ ] the relevant data and articulate[ ] a satisfactory explanation for its action including a rational connection between the facts found and the choices made,’ ’’ 442 one commenter believed that the Commission ‘‘lacks the administrative record,’’ ‘‘evidence’’ and ‘‘analysis’’ that would be ‘‘needed to justify such drastic government intrusion into free markets.’’ 443 Another commenter, however, disputed that notion and observed that the Commission had developed the Pilot, in
part, by relying on ‘‘empirical
literature’’ that ‘‘is directly on point and speaks to the potential distortionary effects that the pilot program is designed to study’’ and that ‘‘certainly provides strong empirical support for further analysis by way of data generated through a pilot study.’’ 444 The responding commenter also found it significant that the Commission was ‘‘presently in the midst of a formal notice and comment process . . . which was informed by years of discussion at, and a proposal from, the [EMSAC]’’ and that the Commission ‘‘had chosen to act via a pilot program rather than a proposal for a long-term rule.’’ 445 The commenter therefore believed the

446 Id. at 5.
447See, e.g., Swan Letter; IEX Letter I; NYSE Letter I. 448See Proposing Release, supra note 2, at 13009 n.6, 13012–14. 449See id. at 13009–14. 450See IAC Recommendation. 451See, e.g., Section II.A.2. supra for a discussion of comments regarding the impact of current pricing models on market quality, execution quality, and order routing. 452Nasdaq Letter I, at 2. See also Cboe Letter I, at 1. 453Nasdaq Letter I, at 5, 11–12. See also Cboe Letter I, at 11; Mexco Letter, at 1. One commenter agreed that ‘‘price controls on access fees indicate something is broken in market structure,’’ but observed that ‘‘there has been no serious economic analysis, let alone a cost-benefit analysis, of what the optimal fee cap (if any) should be’’ and that the Pilot would ‘‘provide solid evidence that can be used to determine the optimal fee cap.’’ Angel Letter I, at 1–2; Angel Letter II, at 2. 454Cboe Letter I, at 10 (citing 15 U.S.C. 78s(b)(3)(A)(ii), which provides that ‘‘a proposed rule change shall take effect upon filing with the Commission if designated by the self-regulatory organization as. . . establishing or changing a due, fee, or other charge imposed by the self-regulatory organization on any person, whether or not the person is a member of the self-regulatory organization’’). This commenter also noted that ‘‘every single exchange transaction fee in place today was filed with, and processed by, the Commission’’ and that any fees that were inconsistent with the Exchange Act ‘‘could have been suspended or abrogated by the Commission if that were deemed necessary.’’ Id. at 6. 455NYSE Letter I, at 11. This commenter identified the relevant ‘‘objectives’’ of Rule 610(c) as preventing the exchanges from ‘‘undermining Regulation NMS’s price protection and linkage requirements.’’ Id. Another commenter similarly characterized the ‘‘justification for the fee cap under Rule 610(c)’’ as ‘‘the existence of sustained market power created by the requirement of best execution and the prohibition against trading through,’’ which would permit exchanges to ‘‘charge high access fees thereby undermining Regulation NMS’s price protection and linkage requirements.’’ This commenter believed that the Commission had wrongfully assumed ‘‘that the market power presumably wielded by equities exchanges is so great that they may charge excessive fees now and in the future’’ unless ‘‘artificial government price constraints’’ are imposed. Nasdaq Letter I, at 12–13, 12 n.38. The third commenter stated that the ‘‘original fee cap rationale’’ was to ‘‘address predatory outlier pricing.’’ Cboe Letter I, at 14. 456Cboe Letter I, at 14. 457NYSE Letter I, at 12. See also Cboe Letter I, at 10 (stating that it was a ‘‘conflict[ ] with the purposes of the Exchange Act and [a] depart[ure] from Commission precedent’’ to ‘‘cap fees for transactions that do not implicate intermarket price protection’’ and ‘‘ban[ ] linked pricing,’’ which has been ‘‘utilized by exchanges with SEC consent for
years’’). 458See Securities Exchange Act Release No. 51808 (June 9, 2005), 70 FR 37496, 37545 (June 29,
2005) (File No. S7–10–04).
Commission had fulfilled its statutory obligations in ‘‘determin[ing] that, given existing evidence suggesting the distortive effect of practices in the market tied to rebates or access fees, a pilot program will provide sufficient information to inform potential future rulemaking.’’ 446 The Commission agrees and notes that it has carefully examined available data on this issue, engaged in a lengthy and deliberative process, and taken into account the recommendations of two independent advisory bodies (EMSAC and the Investor Advisory Committee). The Commission developed the Pilot through a thorough review of the empirical literature, which was cited and discussed in the Proposing Release, as well as submitted as comments in response to this proposal.447 Moreover, as discussed in the Proposal, the EMSAC conducted a thorough process to consider, and ultimately formally recommend, that a pilot be conducted.448 The EMSAC reflected a broad and diverse set of perspectives. In addition, EMSAC heard testimony from experts during its open meetings (which included as panelists senior executives from exchanges) regarding exchange fee models, the appropriateness of a transaction fee pilot, and the shape that such a pilot should take.449 In addition to EMSAC, the independent Investor Advisory Committee also submitted a recommendation in support of the Pilot.450 After considering all of the available information, the Commission has identified a fundamental disagreement among exchanges, market participants, academics, and industry experts regarding the impact of such fees and rebates on the markets.451 This disagreement is further exacerbated by the lack of data to evaluate these competing claims. The Commission believes that the Pilot is necessary to study the impact of exchange fees and rebates to determine whether a regulatory response is needed to mitigate the potential distortions that current exchange pricing models introduce to order routing behavior, market quality, and execution quality. Some commenters argued that the Pilot’s imposition of new fee caps constituted ‘‘impermissible government rate-making.’’ 452 For example, one exchange commenter stated that ‘‘[g]overnment-imposed price controls’’ ‘‘reduce choices for market participants,’’ ‘‘distort competition between over-the-counter venues and exchanges,’’ and are ‘‘costly to administer and lacking in an incentive to be efficient,’’ such that ‘‘they are only indicated where they overcome severe market imperfection such as monopoly ownership of a critical resource.’’ 453 As discussed above, another commenter asserted that the Exchange Act ‘‘plainly contemplates that exchanges, rather than the SEC, will make an initial determination as to the price of a particular product or service,’’ and indicating that ‘‘fee setting is the province of each exchange, subject to the competitive forces that naturally control fees’’ and ‘‘subject to oversight only in particular situations.’’ 454 Commenters expanded on
this argument by stating that the Commission had not sufficiently ‘‘evaluate[d] whether there is any evidence that the Commission’s objectives in adopting the cap on access fees . . . are not being met.’’ 455 One commenter, for example, found it ‘‘concerning that the fee caps in the proposed Pilot do absolutely nothing to further the justification of the original cap and, unlike the original access fee cap, are set at levels that completely undercut existing rates.’’ 456 Exchange commenters further contended that the Pilot imposes ‘‘completely new limitations on exchanges’ business’’ that were ‘‘unrelated to Regulation NMS’s Access Fee Cap,’’ because the Pilot would ‘‘expand[ ] the cap on fees that exchanges may charge for execution not only against a protected quote, but for execution against any quote on an exchange, including depth-of-book and non-displayed orders,’’ as well as ‘‘limit . . . the rebates that an exchange pays’’ and ‘‘pricing that is linked to providing or removing liquidity on an exchange.’’ 457 The Pilot has two Test Groups, one of which does not cap fees at all, but rather leaves in place the current Rule 610(c) fee cap and simply prohibits exchanges from paying rebates or offering Linked Pricing. The other Test Group does impose a lower fee cap for a small portion of NMS stocks (730 out of over 8,000 NMS stocks) for a limited period of time, but is doing so to study the effects of exchange fee-and-rebate pricing models and to gather data to assess the impact on the markets and market participants of a revised and lowered cap compared to the current cap. Further, the Commission selected an amount for that cap that was recommended by commenters, including the Investor Advisory Committee. As explained above, the existing fee cap was designed, in part, to prevent trading centers from charging unreasonably high fees to market participants required to honor their quotations by the Order Protection Rule.458 Because ‘‘[a]ccess fees tend to be highest when markets use them to fund substantial rebates to liquidity providers, rather than merely to compensate for agency services,’’ the Commission was concerned that ‘‘the published quotations of [outlier]

459 Id.
460 Id.
461 Id.
462Nasdaq Letter I, at 13.
463See, e.g., BlackRock Letter, at 1 (‘‘[T]he existing access fee cap is outdated and permits market forces to drive fees and rebates to excessive levels relative to the current magnitude of commissions and bid-ask spreads.’’); Goldman Sachs Letter, at 2 (identifying a ‘‘well-developed, general consensus amongst market participants that a $0.0030 per share Fee Cap is an outdated benchmark for execution costs in today’s trading environment . . . and far from representative of true prices in the marketplace’’); Citi Letter, at 1– 2 (stating that ‘‘today’s 30-mil cap on access fees that the exchanges can charge to access liquidity on their venues represents a more significant percentage of the economics of each trade’’). 464See, e.g., ICI Letter I, at 2 (‘‘Transaction fees and rebates also undermine market transparency because the prices displayed by exchanges—and provided on trade reports—do not include fee or rebate information and therefore do not fully reflect net trade prices.’’); Goldman Sachs Letter, at 3 (stating that ‘‘displayed prices do not reflect the actual economic costs because exchange fees and rebates are not reflected in those prices’’); Oppenheimer Letter, at 2 (‘‘[T]o the extent that transaction fees and rebates obfuscate the actual price bid or offered for a security, the ‘maker-taker’ pricing model has the potential to undermine price transparency . . . .’’). 465 In response to commenters who complained that the Pilot’s fee cap Test Group applies to fees to provide liquidity, instead of being limited to fees to remove liquidity as is the case for Rule 610(c), and therefore it is ‘‘unrelated’’ to the existing fee regime and the Rule 610(c) construct, the Commission notes that when it adopted the Rule 610(c) fee cap it expressly noted that it would ‘‘monitor the operation of these rules to assess whether in practice . . . broader coverage of the rule is necessary.’’ See NMS Adopting Release, supra note 10, at 37546. 466 IEX Letter I, at 6. 467 Verret Letter I, at 2. See also IAC Recommendation, at 1 (‘‘[T]he purpose of the Pilot is not to consider imposing price controls, but instead to consider requiring fees (of whatever size) to be structured so as to minimize complexity and agency costs.’’). 468 IEX Letter I, at 6–7 (‘‘The fact that the SEC has not previously chosen to use its authority to prohibit rebates, or test their elimination through a pilot, does not mean it lacks authority. . . .’’); see also Verret Letter I, at 3. 469 IEX Letter II, at 9. See also Verret Letter I, at 2 (stating that ‘‘one might properly describe the Reg NMS regime as itself a decade-long experiment in price controls’’). 470 IEX Letter I, at 7; IEX Letter II, at 9 (‘‘NYSE seems to be saying, ‘We are fine with the current fee regulation, because we have been able to operate very profitably under it, but it would be illegal to even test different fee restrictions unless you impose them on ATSs.’’). See also, e.g., Verret Letter I, at 2 (‘‘Exchanges
appear comfortable when price controls on the liquidity taking side benefit their business models, but challenge the Commission’s authority to implement what they describe as price controls when their own business models are negatively impacted.’’); Larry Harris Letter, at 6 (noting that ‘‘exchange holding companies have a strong interest in maintaining the current system’’ and that the ‘‘SEC may reasonably consider these interests when evaluating comments submitted by the exchanges); Themis Trading Letter II, at 3 (stating that the Commission should not be ‘‘distracted. . . by conflicted stock exchanges desperately fearful that their business models might come crashing down’’). 471See, e.g., Cboe Letter I, at 9. 472NYSE Letter I, at 12. 473 Id. See also Cboe Letter I, at 11. For example, the commenter noted that the Commission had ‘‘provided no analysis or discussion demonstrating its reasoned decision-making of how the specific fee structures to be mandated in the Proposal would be equitably allocated or reasonable’’ under Section 6 of the Exchange Act. NYSE Letter I, at 12. See also 15 U.S.C. 78f(b)(4) (requiring the rules of an exchange to ‘‘provide for the equitable allocation of reasonable dues, fees, and other charges among its members and issuers and other persons using its facilities’’). 474 If any of the provisions of these amendments, or the application thereof to any person or circumstance, is held to be invalid, such invalidity shall not affect other provisions or application of such provisions to other persons or circumstances that can be given effect without the invalid provision or application. 475 44 U.S.C. 3501 et seq. markets would not reliably indicate the true price that is actually available to investors or that would be realized by liquidity providers.’’ 459 The Commission explained that the fee cap helped assure the fairness and usefulness of quotation information; limit the extent to which the true price for those who access quotations can vary from the displayed price; permit broker-dealers to route orders in a manner consistent with the operation of a national market system; and protect limit orders and promote best-priced quotations.460 Accordingly, the Commission imposed a $0.0030 fee cap, which it believed reflected a competitive rate that was consistent with current business practices at the time (i.e., in 2005).461 In establishing the Rule 610(c) fee cap, the Commission did not, however, cede its responsibility to ensure that markets continue to function in a fair, transparent, and efficient manner; nor did it state that the $0.0030 fee cap could not be revisited if market conditions changed. The Pilot is designed to determine, among other things, whether such a change has occurred. Despite assertions by one commenter that ‘‘powerful competitive forces are clearly present that discourage exchanges from exercising unabated pricing power,’’ 462 a $0.0030 fee is still consistently charged by many exchanges, raising
concerns among other commenters that the fee cap is stuck at a non-competitive and, perhaps, an artificially high rate.463 Several commenters have also indicated that current pricing models have resulted in the kind of distortive pricing that Rule 610(c) was designed to prevent.464 Testing lower fee levels, and a no-rebate fee regime,465 will help the Commission to determine whether further regulatory action is needed to achieve the objectives of Rule 610(c) as well as the Commission’s statutory mandate to oversee the equities markets. The Commission’s position is echoed by other commenters that found the ‘‘suggest[ion] that the Commission lacks the authority to implement the Pilot, or that testing a rebate ban or alternative access fee caps would constitute an impermissible form of price control . . . meritless’’ 466 or ‘‘entirely inaccurate.’’ 467 One commenter, for example, noted that the Exchange Act ‘‘provides very broad authority for the Commission to regulate all aspects of exchange operation, including fee schedules . . . .’’ 468 This commenter further observed that ‘‘it makes no sense to attack the Commission’s proposal as an impermissible form of ‘rate setting’ when the markets have been operating with exchange fee limits for more than 10 years.’’ 469 Moreover, this commenter asserted that ‘‘exchange criticisms’’ regarding ‘‘price control[s]’’ are ‘‘contradicted by their acceptance of th[e] existing price regulation’’ in Rule 610(c), which ‘‘may better serve their interests than the alternative caps and rebate prohibition included in the Pilot.’’ 470 A few other commenters believed that the Commission had not sufficiently identified or discussed the statutory authority to conduct the Pilot.471 One commenter stated that the Proposing Release did not contain an ‘‘explanation as to how those specific statutory sections [cited by the Commission], either individually or collectively, provide the Commission with the authority to carry out the Proposal’s broad rate-setting requirements’’ or a ‘‘discussion of the Commission’s statutory authority at all . . . .’’ 472 This commenter asserted that the Commission ‘‘cannot simply skip this analysis or assume it has unrestricted authority to conduct pilots on the basis that the Proposal is intended to be temporary.’’ 473 The Commission notes that it followed its standard practice in the Proposing Release to identify the statutory authority under which it promulgated its Proposal. The Commission has complied with its statutory obligations in promulgating the Pilot and has clear statutory authority to adopt the Pilot, which the Commission believes furthers the purposes of the Exchange Act.474
III. Paperwork Reduction Act
Certain provisions that the Commission is adopting today contain ‘‘collection of information requirements’’ within the meaning of the Paperwork Reduction Act of 1995 (‘‘PRA’’).475 The Commission published

476See Proposing Release, supra note 2, at 13038– 39. 477 44 U.S.C. 3507; 5 CFR 1320.11. 478See NYSE Letter I, at 15; Cboe Letter I, at 21. 479See supra Section II.E.0. 480See supra Section II.E.3. 481See supra Section II.E.0. 482See Proposing Release, supra note 2, at 13036. The Commission based this estimate on a full-time Compliance Manager and Programmer Analyst each spending approximately 4 hours, for a combined total of approximately 8 hours, to compile and publicly post to an exchange’s website a downloadable file containing the initial Pilot Securities Exchange List. See id. at 13036 n.186. 483See NYSE Letter I, at 15. 484See id. a notice requesting comment on the collection of information requirements in the Proposing Release 476 and submitted relevant information to the Office of Management and Budget (‘‘OMB’’) for review in accordance with the PRA and its implementing regulations.477 The title of the new collection of information for Rule 610T is ‘‘Transaction Fee Pilot Data.’’ Compliance with these collections of information requirements is mandatory. An agency may not conduct or sponsor, and a person is not required to respond to, a collection of information unless the agency displays a currently valid control number. We have applied for an OMB Control Number for this collection of information. The Commission requested comment on the collection of information requirements in the Proposing Release. The Commission received two comment letters on the estimates for the collection of information requirements included in the Proposing Release, which are discussed below.478 A. Summary of Collection of Information The Pilot requires the equities exchanges to prepare four sets of data that constitute a collection of information within the meaning of the PRA. First, pursuant to Rule 610T(b), the primary listing exchanges will be required to prepare and publicly post two sets of data on the Pilot Securities listed on their markets—the Pilot Securities Exchange Lists and the Pilot Securities Change Lists.479 In addition, pursuant to Rule 610T(d), all equities exchanges will be required to provide to the Commission monthly order routing datasets.480 Lastly, pursuant to Rule 610T(e), all equities exchanges will be required to prepare and publicly post the Exchange Transaction Fee Summaries, which are monthly summaries of information concerning fees assessed and rebates paid to market participants transacting on the exchange.481 B. Proposed Use of Information The data collected during the Pilot, including the Pilot Securities Exchange Lists, Pilot Securities Change Lists, Exchange Transaction Fee Summaries, and order routing datasets, will allow researchers and market participants to have ready access to information that will facilitate the study of the impact of an exogenous shock to transaction fees and rebates on order routing behavior, execution quality, and market quality. In turn, this information should facilitate a data-driven evaluation of
future policy choices. In addition, by publishing and maintaining a Pilot Securities Exchange List and a Pilot Securities Change List, each primary listing exchange would help ensure that the Commission, market participants, researchers, and the public have up-to-date information on corporate changes to listed issuers that impact the list of Pilot Securities, as well as changes to the composition of any of the Test Groups during the Pilot.
C. Respondents
The respondents to this collection of information will be the equities exchanges, which are registered national securities exchanges that trade NMS stocks. Specifically, Rule 610T(b), which covers the Pilot Securities Exchange Lists and Pilot Securities Change Lists, will apply to the six primary listing exchanges for NMS stocks. Rule 610T(d), which requires datasets on order routing, will apply to all thirteen equities exchanges that are currently registered with the Commission. Rule 610T(e), which requires datasets on fees (rebates) and fee (rebate) changes, will apply to all thirteen equities exchanges currently registered with the Commission. D. Total Initial and Annual Reporting and Recordkeeping Burdens The burdens associated with the Pilot are described fully below, but the below
table briefly summarizes the relevant
burdens set forth in the Proposing Release and in this release. Category Release Annual burdens (hours/ exchange) One-time burdens (hours/ exchange) Pilot Securities Exchange Lists ...................................................... Proposing Release ...................................... N/A 8 Adopting Release ........................................ N/A 44 Pilot Securities Change Lists ......................................................... Proposing Release ...................................... 126 12 Adopting Release ........................................ 126 12 Exchange Transaction Fee Summaries ......................................... Proposing Release ...................................... 64 86 Adopting Release ........................................ 64 86 Order Routing Datasets ................................................................. Proposing Release ...................................... 112 80 Adopting Release ........................................ 124 80

Pilot Securities Exchange Lists and
Pilot Securities Change Lists Upon publication of the initial List of Pilot Securities by the Commission, the primary listing exchanges would be required to determine which Pilot Securities are listed on their market and compile and publicly post downloadable files containing a list of those securities, including all data fields specified in Rule 610T(b)(2)(i) on their websites in pipe-delimited ASCII format. The Commission preliminarily estimated that each primary listing exchange would incur, on average, a one-time burden of approximately 8 burden hours per primary listing exchange to compile and publicly post its initial Pilot Securities Exchange List.482 One commenter stated that it ‘‘anticipates it could take as many as 44 hours’’ to compile the initial Pilot Securities Exchange List.483 The commenter stated that its estimates of the costs associated with the Pilot are based on its ‘‘prior experience implementing the Tick Size Pilot, and other similar initiatives . . . .’’ 484 In light of this comment, the Commission is increasing its estimate. While, unlike for the Tick Size Pilot, the Commission will prepare the Initial List of Pilot Securities and assign them to their

485The Commission continues to believe that this will require the services a full-time Compliance Manager and Programmer Analyst. The Commission estimates that each Compliance Manager and Programmer Analyst will each spend approximately 22 hours, for a combined total of approximately 44 hours, to compile and publicly post to an exchange’s website a downloadable file containing the initial Pilot Securities Exchange List. 486 44 burden hours per primary listing exchange × 6 primary listing exchanges = 264 burden hours. 487The Commission derived the total estimated burdens from the following estimates: (Attorney at 4 hours) + (Compliance Manager at 4 hours) + (Programmer Analyst at 4 hours) = 12 burden hours. See Proposing Release, supra note 2, at 13036. 488The Commission based this estimate on a fulltime Compliance Manager and Programmer Analyst together spending approximately 30 minutes per trading day updating and posting the required lists (approximately 252 trading days × 30 minutes per trading day = 7,560 minutes (126 hours)). See id. 489NYSE Letter I, at 15. 490 Id. See also Cboe Letter I, at 21 (stating that the ‘‘implementation and ongoing costs of the Pilot will be significantly larger in terms of burden hours and expenditures than the Commission estimates,’’ but providing no specific analysis or alternative estimates). 491See Proposing Release, supra note 2, at 13036. The Commission notes that it has revised its aggregate burden estimate upwards to 371 hours for each exchange to address commenter concerns that the estimated burden associated with compiling and publicly posting the initial Pilot Securities Exchange List was too low. 492See Proposing Release, supra note 2, at 13027 n.153 and accompanying text; note 740 infra. 493 126 burden hours per primary listing exchange × 6 primary listing exchanges = 756 burden hours. 494See Proposing Release, supra note 2, at 13037. The Commission preliminarily estimated that an equities exchange would assign responsibilities for review and potential modification of its systems and technology to an Attorney, a Compliance Manager, a Programmer Analyst and a Senior Business Analyst. The Commission estimated the burden of reviewing and potentially modifying its systems and technology to be as follows: (Attorney at 20 hours) + (Compliance Manager at 20 hours)

(Programmer Analyst at 20 hours) + (Business
Analyst at 20 hours) = 80 burden hours per equities exchange. See id. at 13037 n.194. respective treatment groups, and therefore the exchanges will only need to separate out their listed securities into a separate list, the Commission nevertheless will increase its estimate as the commenter suggested. Accordingly, the Commission estimates that each primary listing exchange would incur, on average, a one-time burden of approximately 44 burden hours per primary listing exchange to compile and publicly post their initial Pilot Securities Exchange List.485 Accordingly, the Commission believes that the aggregate one-time burden associated with the initial Pilot Securities Exchange Lists would be 264 burden hours.486 After posting its initial Pilot Securities Exchange List, each equities exchange will be required to keep current that list to reflect any changes, and to also prepare and publicly post on its website until the end of the postPilot Period the Pilot Securities Change List prior to the beginning of trading each trading day. The Commission preliminarily estimated that each primary listing market would incur a one-time burden of approximately 12 burden hours of internal legal, compliance, and information technology operations to develop appropriate systems to track and compile changes relevant to Pilot Securities listed on its market.487 The Commission also preliminarily estimated that, once the primary listing exchanges have established these systems, on average, each primary listing exchange would incur 0.5 burden hours daily, or 126 burden hours annually to compile any changes related to Pilot Securities, such as name changes or mergers, and to publicly post the updated Pilot Securities Exchange Lists and Pilot Securities Change Lists on its website prior to the start of each trading day.488 One exchange commenter stated that ‘‘the Commission predicts it that would take only 12.5 hours to develop and maintain systems to comply’’ with the requirements to update prior to the start of each trading day the Pilot Securities Exchange Lists and Pilot Securities Change Lists.489 Based on ‘‘its prior experience implementing the Tick Size Pilot, and other similar initiatives,’’ this commenter further stated that it believed ‘‘it could take as many as 300.5 hours to develop and maintain those systems.’’ 490 While the commenter did not elaborate on how it computed its estimate or whether it represents an aggregate burden estimate or an annualized estimate, the commenter appears to have misunderstood the burden estimates contained in the Proposing Release because the Commission’s estimate greatly exceeded
12.5 hours. Specifically, the
Commission’s preliminary estimates included a one-time burden of 8 hours for primary listing exchanges to compile and publicly post the initial Pilot Securities Exchange List, a one-time burden of 12 hours for primary listing exchanges to develop appropriate systems to track and compile changes to Pilot Securities, and an ongoing burden of 126 hours annually to compile any such changes and publicly post the updated Pilot Securities Exchange Lists and Pilot Securities Change Lists, for an aggregate burden estimate of 335 hours per exchange for the entire Pilot.491 Assuming that the commenter’s estimate of 300.5 hours is meant to be an aggregate burden estimate, the Commission notes that its revised aggregate burden estimate of 371 hours exceeds the commenter’s estimate. The Commission’s estimates are averages that take into account the diverse set of six primary listing exchanges and the expected burdens that they would collectively experience as a result of the Pilot. Moreover, the Commission expects that the primary listing exchanges will be able to leverage their experience and resources from the recent Tick Size Pilot to meet the requirements of the Pilot. As noted above, unlike for the Tick Size Pilot, the Commission will set the initial List of Pilot Securities and the primary listing exchanges only need to keep those lists up to date if their listed issuers experience any relevant change. Accordingly, the burdens on the primary listing exchanges with respect to the lists of Pilot Securities should be less than those incurred during the Tick Size Pilot.492 For those reasons, the Commission continues to believe its estimate of the aggregate one-time burden for primary listing exchanges to develop appropriate systems to track and compile changes relevant to Pilot Securities listed on their markets will be approximately 12 burden hours for each primary listing exchange, or 72 total burden hours, and the average, aggregate annual burden to update and publicly post the lists of Pilot Securities will be approximately 126 burdens hours for each primary listing exchange, or 756 total burden hours for all 6 exchanges.493

Exchange Transaction Fee Summaries
The Commission is requiring that each equities exchange publicly post on its websites the Exchange Transaction Fee Summary each month, using an XML schema published on the Commission’s website. The Commission believes that all the data necessary to complete the summary are currently maintained by the equities exchanges. However, the equities exchanges will be required to compute the monthly realized average and median per share fees and rebates, each by participant type, that qualified for the Base and Top Tier fees and rebates, using fee and volume information that the equities exchanges maintain. The Commission preliminarily estimated that each equities exchange would incur a one-time burden of approximately 80 burden hours of internal legal, compliance, information technology, and business operations to develop appropriate systems for tracking fee changes, computing the monthly averages, and formatting the data and posting it on its website.494 One commenter objected generally to the Commission’s burden estimates, but

495Cboe Letter I, at 21 (stating that the ‘‘implementation and ongoing costs of the Pilot will be significantly larger in terms of burden hours and expenditures than the Commission estimates,’’ but providing no specific analysis or alternative estimates). But cf. Better Markets Letter, at 2 (‘‘All of the data-fields are thoughtfully proposed, and the cost of producing them is minimal and certainly acceptable given the enormity of the benefits.’’ The commenter did not provide specific burden hour or cost estimates.). 496 (Attorney at 20 hours) + (Compliance Manager at 20 hours) + (Programmer Analyst at 20 hours) + (Business Analyst at 20 hours) = 80 burden hours per equities exchange. 497 80 burden hours per equities exchange × 13 equities exchanges = 1,040 burden hours. 498See Proposing Release, supra note 2, at 13037. (Attorney at 10 hours) + (Compliance Manager at 10 hours) + (Programmer Analyst at 10 hours) + (Senior Business Analyst at 10 hours) = 40 burden hours. 499See note 495 supra. 500 (Attorney at 10.5 hours) + (Compliance Manager at 10.5 hours) + (Programmer Analyst at 11 hours) + (Senior Business Analyst at 10.5 hours) = 40 burden hours. 501 40 burden hours per equities exchange × 13 equities exchanges = 520 burden hours. 502See Proposing Release, supra note 2, at 13037. The Commission derived the total estimated burden from the following estimates: (Compliance Manager at 2 hours) + (Senior Business Analyst at 2 hours) = 4 burden hours per equities exchange. See id. at 13037 n.198. 503See id. at 13037 n.199 and accompanying text. 504See id. at 13037. The Commission derived the total estimated burden from the following estimates, which reflect the Commission’s preliminary belief that the equities exchanges have experience posting information in an XML format on publicly-available websites: (Compliance Manager at 1 hour) + (Programmer Analyst at 1 hour) = 2 burden hours per equities exchange. See id. at fn. 200. 505See note 495 supra. 506 (Compliance Manager at 2 hours) + (Senior Business Analyst at 2 hours) = 4 burden hours per equities exchange. 507 4 burden hours per equities exchange × 13 equities exchanges = 52 burden hours. 508 2 burden hours per equities exchange × 13 equities exchanges = 26 burden hours. 509See Proposing Release, supra note 2, at 13037. 510See id. The Commission derived the total estimated burden from the following estimates:
(Compliance Manager at 1 hour) + (Programmer Analyst at 1 hour) = 2 burden hours per equities exchange per month. 2 burden hours per equities exchange per month × 12 months per year = 24 burden hours per equities exchange per year. See id. at 13037 n.203. 511See note 495 supra. 512 (Compliance Manager at 1 hours) + (Programmer Analyst at 1 hours) = 2 burden hours per equities exchange per month. 2 burden hours per equities exchange per month × 12 months per year = 24 burden hours per equities exchange per year. did not provide its own estimates of specific burden hours or costs.495 The Commission continues to estimate that each equities exchange will incur a onetime burden of approximately 80 burden hours of internal legal, compliance, information technology, and business operations to develop appropriate systems for tracking fee changes, computing the monthly averages, and formatting the data and posting it on its website.496 Accordingly, the one-time initial aggregate burden for all equities exchanges necessary for the development and implementation of the systems needed to capture the transaction fee information and post it on their websites in the specified format in compliance with Rule 610T(e) will be 1,040 hours.497 The Commission also preliminarily estimated that, on average, an equities exchange would incur an ongoing burden of approximately 40 burden hours per year, approximately half the estimated burden to develop appropriate systems, to monitor and, if necessary, update its systems used for compiling, formatting and publicly posting the Exchange Transaction Fee Summaries.498 One commenter objected generally to the Commission’s burden estimates, but did not specifically explain whether or how this burden estimate was incorrect.499 The Commission continues to estimate that the annual ongoing burdens associated with monitoring and, if necessary, updating these systems would be approximately half the burdens of initially developing the systems. Accordingly, the Commission continues to estimate that an equities exchange will incur an ongoing burden of approximately 40 burden hours per year to monitor, and if necessary, update its systems used for compiling, formatting and publicly posting the Exchange Transaction Fee Summaries.500 The average aggregate, ongoing, annual burden for all equities exchanges to monitor their systems will be 520 hours.501 The equities exchanges will be required to format, calculate certain figures, and post their initial Exchange Transaction Fee Summary at the outset of the pre-Pilot Period. As this would be the first time an equities exchange would be required to produce and post on its website such a summary, the Commission preliminarily estimated that it would require approximately 4 burden hours for each equities exchange to complete the initial Exchange Transaction Fee Summary and perform the necessary calculations.502 In addition, each equities exchange will be required to make its summary publicly
available on its website using an XML schema to be published on the Commission’s website. As the Commission preliminarily believed that the equities exchanges had experience applying the XML format to market data,503 the Commission estimated that initially each equities exchange would incur a burden of 2 burden hours specific to the initial Exchange Transaction Fee Summary to ensure that it has properly implemented the XML schema.504 One commenter objected generally to the Commission’s burden estimates, but did not specifically explain whether or how this burden estimate was incorrect.505 The Commission continues to estimate that each equities exchange will require approximately 4 burden hours to complete the initial Exchange Transaction Fee Summary,506 for an aggregate, initial burden of 52 hours to complete its initial Exchange Transaction Fee Summary.507 The Commission also continues to estimate that each equities exchange will incur an initial burden of approximately 2 burdens hours for an aggregate, initial burden of 26 hours to post that dataset publicly on its website using an XML schema to be published on the Commission’s website. The total aggregate, initial burden to complete the initial Exchange Transaction Fee Summary will therefore be 78 burden hours.508 Each equities exchange will be required to update the Exchange Transaction Fee Summary on a monthly basis to account for changes from the prior month, if any, and to report monthly fee and rebate information. The Commission preliminarily believed that such updates would require fewer burden hours than the initial Exchange Transaction Fee Summary, as the equities exchanges would have experience calculating necessary data and formatting the reports as required by the Rule.509 Accordingly, the Commission preliminarily estimated that it would require approximately 2 burden hours each month, or 24 burden hours on an annualized basis, for each equities exchange to update the Exchange Transaction Fee Summary.510 This estimate contemplated the impact of publicly posting the summary using the XML schema to be published on the Commission’s website. One commenter objected generally to the Commission’s burden estimates, but did not specifically explain whether or how this burden estimate was incorrect.511 The Commission continues to estimate that it will require approximately 2 burden hours each month, or 24 burden hours on an annualized basis, for each equities exchange to update the Exchange Transaction Fee Summary.512 As such, the equities exchanges will incur an aggregate, annual burden of 312 burden hours to update and publicly post on

513 2 burden hours per equities exchange × 13 equities exchanges × 12 monthly updates = 312 burden hours per year. 514See Proposing Release, supra note 2, at 13038. 515See id. The Commission preliminarily estimated that an equities exchange will assign responsibilities for review and potential modification of its systems and technology to an Attorney, a Compliance Manager, a Programmer Analyst and a Senior Business Analyst. The Commission estimated the burden of reviewing and potentially modifying its systems and technology to be as follows: (Attorney at 20 hours) + (Compliance Manager at 20 hours) + (Programmer Analyst at 20 hours) + (Senior Business Analyst at 20 hours) = 80 burden hours per equities exchange. See id. at 13038 n.207. 516 80 burden hours per equities exchange × 13 equities exchanges = 1,040 burden hours. 517See Proposing Release, supra note 2, at 13038. The Commission derived the total estimated burdens from the following estimates, which reflected the Commission’s preliminary view that annual ongoing burdens would be approximately half the burdens of initially ensuring an exchange has the appropriate systems to capture the required information in the required format: (Attorney at 10 hours) + (Compliance Analyst at 10 hours) + (Programmer Analyst at 10 hours) + (Business Analyst at 10 hours) = 40 burden hours per equities exchange. See id. at 13038 n.209. 518 40 burden hours per equities exchange × 13 equities exchanges = 520 burden hours. 519See Proposing Release, supra note 2, at 13038. 520See id. See also FR Doc. 2016–08552, 81 FR 22143 (April 14, 2016) (‘‘Request to OMB for Extension of Rule 605 of Regulation NMS’’). 521Compliance Manager at 3 hours + Programmer Analyst at 4 hours = 7 burden hours per month, per equities exchange. 7 burden hours per month × 12 months = 84 burden hours per year, per equities exchange. 522 84 burden hours per year × 13 equities exchanges = 1,092 burden hours. 523NYSE Letter I, at 15. See also Cboe Letter I, at 21 (stating that the ‘‘implementation and ongoing costs of the Pilot will be significantly larger in terms of burden hours and expenditures than the Commission estimates,’’ but providing no specific analysis or alternative estimates). 524The Commission notes that it has revised this estimate upwards to 124 burden hours annually. 525See Proposing Release, supra note 2, at 13038. The Commission notes that it has revised this estimate upwards to 452 burden hours per exchange for the entire Pilot. their websites the Exchange Transaction Fee Summaries.513
3. Order Routing Datasets
The Commission preliminarily estimated that, on average, there would be no paperwork burden to the equities exchanges to capture the order routing data required pursuant to Rule 610T(d) to be included in the order routing datasets, as the Commission expected that the equities exchanges would collect the required data to create the order routing datasets by leveraging existing systems and technology already in place for the collection and reporting of data.514 The Commission believes this continues to be true with the changes to the order routing datasets, which also involve data elements currently captured by existing systems. The Commission preliminarily believed, however, that the equities exchanges would incur an initial onetime burden of 80 burden hours per equities exchange to ensure that their systems and technology are able to accommodate the proposed requirements to aggregate, anonymize, and publicly post the order routing information.515 While the exchanges will still need to aggregate the data, they no longer will need to anonymize and publicly post it and instead will transmit the information to the Commission. The Commission continues to believe that each equities exchange would incur an initial onetime burden of 80 burden hours to ensure that its systems and technology are able to accommodate the requirements to aggregate and provide to the Commission the order routing information. Accordingly, the Commission estimates that the aggregate one-time initial burden for ensuring an exchange’s systems and technology are able to aggregate and provide to the Commission the required order routing data in compliance with Rule 610T(d) will be 1,040 burden hours.516 The Commission also preliminarily estimated that, on average, it would take an equities exchange approximately 40 burden hours per year to ensure that the systems and technology are up to date so as to facilitate compliance with the Rule.517 The Commission continues to estimate that, on average, it would take an equities exchange approximately 40 burden hours per year to ensure that the systems and technology are up to date so as to facilitate compliance with the Rule. Therefore, the Commission estimates that the aggregate annual burden to maintain the systems necessary to aggregate and provide to the Commission the required order routing information is approximately 520 burden hours per year.518 Each equities exchange would incur an ongoing burden associated with creating and formatting the order routing datasets each month. The Commission noted that the equities exchanges have experience with creating similar datasets in accordance with their obligations under Rule 605 of Regulation NMS.519 The Commission preliminarily believed that each equities exchange would incur burdens similar to those associated with preparing Rule 605 reports.520 Accordingly, the Commission preliminarily believed that each equities exchange would incur a burden of six burden hours per month, or 72 burden
hours per year, to prepare and publicly post on its website the order routing datasets.521 While the order routing datasets will not be publicly posted but will instead be provided to the Commission, the Commission is requiring the equities exchanges to separate out post-only orders and auction-only orders (or exclude auction-only orders if they so choose). The Commission estimates that separating out these orders will require approximately 1 additional burden hour per month. As such, the Commission estimates that each equities exchange will incur a burden of approximately seven burden hours per month, or 84 burden hours per year, to prepare and provide to the Commission the order routing datasets. Therefore, the aggregate, annual burden to prepare and provide to the Commission order routing datasets in accordance with Rule 610T(d) will be approximately 1,092 burden hours.522 One exchange commenter stated that ‘‘the Commission allocates 160 hours associated with producing order routing data,’’ but estimated that it ‘‘would actually require over 400 hours,’’ based on ‘‘its prior experience implementing the Tick Size Pilot, and other similar initiatives . . . .’’ 523 While the commenter did not elaborate on how it computed its estimate or whether it represents an aggregate burden estimate or an annualized estimate, the commenter appears to have misunderstood the burden estimates contained in the Proposing Release because the Commission’s estimate exceeds the 160 hours cited by the commenter. Specifically, the Commission’s preliminary estimate included a one-time burden of 80 hours and an ongoing burden of 112 hours annually,524 for an aggregate burden estimate of 416 hours per exchange for the entire Pilot.525 Second, the commenter does not explain how it calculated its estimate of ‘‘over 400 hours,’’ break down the costs included in this estimate, or specify whether this number is an aggregate burden estimate or an annualized estimate. Assuming that the commenter’s estimate of over 400 hours is meant to be an aggregate burden estimate, the Commission notes that its revised aggregate burden estimate of 452 hours is substantially similar. The Commission notes that exchanges will no longer be required to publicly post this data, but will instead transmit the datasets directly to the Commission. Moreover, the Commission expects that the exchanges will be able to leverage their experience and

526See Section C.2.a.iii. infra. See also, e.g., Better Markets Letter, at 2. 527See, e.g., 5 U.S.C. 552 et seq.; 15 U.S.C. 78x (governing the public availability of information obtained by the Commission). 528 17 CFR 240.17a–1. 529Execution quality generally refers to how favorably customer orders are executed. Execution quality measures are similar to liquidity measures and tend to include transaction costs, the speed of execution, the probability that the trade will be executed, and the price impact of the trade. See NMS Adopting Release, supra note 10, at 37513– 15, 37537–38. Market quality encompasses execution quality but also relates more generally to how well the markets function. Market quality measures include liquidity, price discovery, and volatility in prices. See, e.g., Henrik Bessembinder, Trade Execution Costs and Market Quality after Decimalization, 38 J. Fin. & Quantitative Analysis 747–77 (2003), https://doi.org/10.2307/ 4126742https://doi.org/10.2307/4126742; Maureen O’Hara & Mao Ye, Is Market Fragmentation Harming Market Quality? 100 J. Fin. Econ. 459–74 (2011), https://doi.org/10.1016/ j.jfineco.2011.02.006. 530See, e.g., James Angel, Lawrence Harris & Chester Spatt, Equity Trading in the 21st Century, 1 Q. J. Fin. (2011), https://doi.org/10.1142/ S2010139211000067 (hereinafter ‘‘Angel, Harris, & Spatt’’); Robert H. Battalio, Shane A. Corwin, & Robert H. Jennings, Can Brokers Have It All? On the Relation Between Make-Take Fees and Limit Order Execution Quality, 71 J. Fin. 2193–237 (2016), http://onlinelibrary.wiley.com/doi/10.1111/ jofi.12422/full (hereinafter ‘‘Battalio Equity Market Study’’); Larry Harris, Maker-Taker Pricing Effects on Market Quotations 24–25 (USC Marshall Sch. Bus., Draft No. 0.91, 2013), http://bschool.huji.ac.il/ .upload/hujibusiness/Maker-taker.pdf (hereinafter ‘‘Harris’’). 531For commenters concurring with this assessment, see, e.g., Barnard Letter, at 1 (stating the Pilot ‘‘should provide credible analyses of the effects—both positive and negative—of exchange fees and rebates on the quality and efficiency of trading.’’); Better Markets Letter at 2 (stating that the Commission ‘‘lacks sufficient data to outlaw rebates’’ and believed that the Pilot ‘‘should fill this data and knowledge gap.’’). 532Many commenters expressed support for the Pilot and the utility of the information that may be gained from it. See AJO Letter, at 1, CII Letter, at 3, NYSTRS Letter, at 1, ICI Letter I, at 1–2, MFS Letter, at 1, Nuveen Letter, at 2, Clark-Joseph Letter, at 1, RBC Letter I, at 2, Invesco Letter, at 2, CFA Letter, at 1, State Street Letter, at 2, Wellington Letter, at 1, Joint Pension Plan Letter, at 2, Oppenheimer Letter, at 2, Angel Letter I, at 1, Vanguard Letter, at 2, Verret Letter I, at 1, T. Rowe Price Letter, at 1. 533See, e.g., Babelfish Letter, at 3; Clearpool Letter, at 2; T. Rowe Price Letter, at 1, 3, and ClarkJoseph Letter, at 1. 534See infra Section V.C.1.a.ii, for further discussion of the
benefits of studying other economic effects of transaction fees and rebates. 535See infra Section V., for discussion of existing studies related to these topics and their limitations. See also supra Section II.B (discussing the Nasdaq study, which examined a change in the access fees and rebates charged by Nasdaq for 14 stocks over a four-month period). resources from the Tick Size Pilot to meet the requirements of the Pilot.526 For those reasons, the Commission believes its estimate of the one-time burden for exchanges to develop and implement appropriate systems to aggregate the order routing data will be, on average, 80 burden hours for each exchange, and the ongoing annual burden to update these systems and to gather and to transmit the relevant data to the Commission will be, on average, 124 burden hours for each exchange. E. Collection of Information Is Mandatory All of the collections of information pursuant to Rule 610T would be mandatory. F. Confidentiality of Responses to Collection of Information The Commission believes that the broker-dealer specific order routing data should be protected from disclosure subject to the provisions of applicable law.527 The Commission will deem broker-dealer identifying order routing data as being subject to a confidential treatment request under 17 CFR 200.83 without the need to submit a request. The Pilot Securities Exchange List, Pilot Securities Change List, and the Exchange Transaction Fee Summary would not be confidential. Rather, each would be publicly posted by the exchanges. G. Retention Period for Recordkeeping Requirements National securities exchanges would be required to retain records and information pursuant to 17 CFR 240.17a–1 (Rule 17a–1 under the Exchange Act).528
IV. Economic Analysis
As discussed above, the Pilot is designed to produce information on the impact of transaction fee-and-rebate pricing models on order routing decisions by broker-dealers, as well as their impact on execution and market quality.529 In recent years, a number of academics and market participants have expressed concern that the structure of exchange transaction-based fee pricing may lead, for example, to potential conflicts of interest between brokerdealers and their customers when brokers-dealers route customer orders to trading centers offering rebates so that the broker-dealer can capture the rebates, even when these venues do not offer high execution quality.530 However, as discussed in more detail below, the Commission cannot determine from existing empirical evidence the impact, if any, of exchange transaction fee models on order routing decisions by broker-dealers or on market and execution quality.531 Specifically, determining whether a causal relationship between exchanges’ transaction fee-and-rebate pricing models and broker-dealers’ behavior is complicated because, for example, such pricing models and order routing decisions could be jointly determined and order routing decisions could influence fees just as fees could influence order routing decisions. Currently available data do not permit researchers to isolate these factors and thus identify the existence or direction of such a causal relationship, which in turn impedes researchers’ ability to determine the extent to which conflicts may exist and any potential negative impacts may manifest.532 Because of the existing lack of empirical evidence regarding the potential conflicts of interest and potential effects of exchange fee models, additional information would assist the Commission in making future regulatory decisions. To remedy the insufficiency of existing empirical evidence, the Commission is adopting the Pilot to generate data that is otherwise unavailable to study fees and rebates that exchanges assess to broker-dealers and observe the impacts of those fees and rebates on the markets and market participants. Specifically, the Commission expects that the data collected is likely to shed light on the extent, if any, to which broker-dealers route orders in ways that benefit the broker-dealer but may not be optimal for customers, and the extent to which exchange pricing models create distortions that may have adverse impacts. The data obtained from the Pilot will inform future regulatory initiatives to the ultimate benefit of investors.533 In addition, the Pilot will provide information about other potential economic effects of reducing access fee caps or prohibiting rebates and Linked Pricing. For example, the Pilot could offer information on whether prohibiting rebates and Linked Pricing alters broker-dealer behavior in a manner that affects market quality, such as by impacting quoted spreads across NMS stocks.534 The Pilot is uniquely capable of generating empirical evidence
that is currently lacking because it is designed to provide an exogenous shock to transaction fee-and-rebate pricing models across all exchanges simultaneously and facilitate the collection of representative data across a broad range of securities.535 An exogenous shock to a system occurs when an element of the system is changed from without the system. (i.e., the change or shock is not under the control or influence of those within the system) but can induce endogenous (i.e., within the system) responses. In the Pilot’s context, the exogenous shock

536See, e.g., CII Letter, at 3, NYSTRS Letter, at 1, RBC Letter I, at 2, Joint Pension Plan Letter, at 2, Oppenheimer Letter, at 2 537See, e.g., Clearpool Letter, at 2. As discussed above, Nasdaq conducted its own fee experiment, but other exchanges did not conduct similar experiments simultaneous with Nasdaq. 538See 15 U.S.C. 77b(b); 15 U.S.C. 78c(f). 539See 15 U.S.C. 78w(a)(2). 540 Id. 541See Nasdaq Letter I, at 3, 11. 542NYSE Letter I, at 3, 12. See also, e.g., Level Brands Letter, at 1; Johnson Letter, at 1; Sensient Letter; Tredegar Letter, at 1; Halliburton Letter, at 1. 543See ASA Letter, at 5. See also T.D. Ameritrade Letter, at 3 (estimating costs of widening spreads to its clients at $24,000,000). 544NYSE Letter I, at 13. See also TD Ameritrade Letter, at 3 (estimating costs of widening spreads to its clients at $24,000,000 annually) and Energizer Letter, at 1. 545See STANY Letter, at 2; Cboe Letter I, at 21; Nasdaq Letter I, at 10; FIA Letter, at 3; Citi Letter, at 5. 546See, e.g., Vanguard Letter, at 3; Better Markets Letter, at 2; Healthy Markets Letter I, at 34; Angel Letter II, at 3. 547See, e.g., Cboe Letter, at 7 n.14 and 20 (noting failure to adequately address lost revenue to exchanges); NYSE Letter I, at 3; NYSE Letter I, at 3 and 13 (addressing impact on small businesses and issuers); Apache Letter, at 2 (noting potential negative cost impacts to issuers engaged in secondary offerings or conducting share repurchasing programs); Nasdaq Letter I, at 8 (noting potential added cost to market makers when pricing arbitrage opportunities because of additional complexity in exchange pricing models under the Pilot). 548 Verret Letter I, at 3–4. See also Verret Letter I, at 7 (asserting that ‘‘[a]rguments by the Exchanges concerning the pilot proposal’s failure to quantify costs are irrelevant, in so far as the proposal properly identifies where they might at present be unquantifiable and particularly where those unquantifiable costs relate to the data the pilot is intended to generate.’’). 549 Verret Letter I, at 3–4. See also IEX Letter III, at 8, 10 (arguing that these commenters ignore ‘‘the full range of benefits that investors could realize if rebates were banned entirely’’). 550 Verret Letter I, at 4–5. takes the form of a reduction of the maximum permissible transaction fees and a prohibition on rebates and Linked Pricing on all U.S. equities exchanges. This shock will allow researchers to explore how changes to fees and rebates could lead to changes in broker-dealer order routing and market and execution quality for a broad sample of NMS securities.536 Specifically, the reduction in fees or the elimination of rebates and Linked Pricing, as required in specific Test Groups of the Pilot, may reduce the magnitude or eliminate the potential conflict of interest between brokerdealers and their clients and the potential distortions introduced by exchange transaction-based fees and rebates. These effects would, in turn, be reflected
in measurable changes to the order routing and execution quality of stocks in the Pilot’s Test Groups. The terms of the Pilot are discussed in Section II above. Exchanges will continue to be permitted to have varying fees within each Test Group, and will be permitted to change their fees at their discretion, subject to the proposed rule change filing requirements of Section 19 of the Exchange Act, during the Pilot for securities within each Test Group, so long as they comply with the conditions applicable to that Test Group. In the absence of the Pilot, the Commission believes it is unlikely that exchanges would collectively undertake a similar pilot and voluntarily coordinate the exogenous shock to fees and rebates across a broad set of securities, broker-dealers, and exchanges that would be required to analyze the effects of changes to fees and rebates.537 By imposing the same modifications to fees and rebates on all U.S. equities exchanges, the Pilot will allow researchers to obtain data that will permit them to examine the impact of changes to fees and rebates on the order routing decisions of brokerdealers. If all exchanges were not subject to the pilot terms, the pilot data would be limited because broker-dealers could redirect their order flow to the non-participating exchanges. Accordingly, the Commission believes that the Pilot will enable the collection of valuable data that would otherwise be unavailable. The Commission is mindful of the costs imposed by, and the benefits obtained from, the rules it promulgates. Whenever the Commission engages in rulemaking and is required to consider or determine whether an action is necessary or appropriate in the public interest, Section 3(f) of the Exchange Act requires the Commission to consider whether the action would promote efficiency, competition, and capital formation, in addition to the protection of investors.538 Further, when making rules under the Exchange Act, Section 23(a)(2) of the Exchange Act requires the Commission to consider the impact such rules would have on competition.539 Section 23(a)(2) of the Exchange Act also prohibits the Commission from adopting any rule that would impose a burden on competition not necessary or appropriate in furtherance of the purposes of the Exchange Act.540 A few commenters challenged the sufficiency of the economic analysis contained in the Proposal. For example, one commenter argued the proposal was ‘‘arbitrary and capricious’’ because the Commission failed to consider the economic consequences of its proposal and only partially framed the costs and benefits of the Proposal, ignoring important and significant factors and costs.541 Similarly, another commenter believed that the ‘‘cost-benefit analysis contain[ed] numerous flaws that are inconsistent with the Commission’s obligation to provide a ‘reasoned basis’ for its regulations,’’ namely that the Commission had ‘‘substantially underestimated the costs of the Proposal’’ and ‘‘fail[ed] to
identify any countervailing market benefit that justifies imposing . . . harms on . . . exchanges and issuers.’’ 542 Another commenter thought the Commission understated the potential costs of the Pilot while overstating the benefits.543 For example, some commenters noted that they anticipated the Pilot would result in wider spreads, increased transaction costs, and increased broker commissions, all of which would result in added costs to investors.544 Several commenters thought the Commission failed to consider or underestimated the implementation costs of the Pilot,545 while other commenters challenged these assertions and instead believed the Pilot would impose minimal costs on exchanges and broker-dealers, particularly in light of the existing processes and technology that currently support immediately effective fee changes from the exchanges.546 Finally, other commenters felt that the economic analysis failed to adequately account for the projected costs to particular categories of market participant.547 Other commenters supported the Commission’s analysis. For example, one commenter argued that ‘‘differing estimates of costs is not a sufficient basis alone to challenge Commission action.’’ 548 This commenter argued that the commenters ‘‘tend to ignore the benefit side of cost-benefit analysis’’ and believed that ‘‘the most significant benefit of the pilot is its potential to inform subsequent rulemaking,’’ such that the ‘‘mere presence of uncertainty in the Commission’s estimates of potential costs and benefits does not by itself open the pilot program to challenge.’’ 549 While acknowledging the potential for ‘‘liquidity effects,’’ this commenter further noted that the Commission ‘‘is merely held to make a reasonable estimate of those costs before adopting a pilot program,’’ not to ‘‘make a perfect estimate’’ or ‘‘cease the pilot if the costs to liquidity prove significant.’’ 550 The economic analysis provided in the Proposing Release thoroughly described the potential economic effects of the Transaction Fee Pilot, including the benefits, costs, and alternatives and

551See Proposing Release, supra note 2. 552See Section infra IV.B.2.a. 553See, e.g., Better Market Letter at 1 (stating ‘‘[p]ayments by the exchanges that incentivize and induce routing decisions by broker-dealers at the expense of best execution and market quality is one of the most entrenched and insidious market practices today, and requires forceful and independent intervention by the SEC.’’). See also Themis Trading Letter at 4–5; Larry Harris Letter at 9; Clearpool at 2. 554See Themis Trading Letter I, at 5. 555See Larry Harris Letter, at 3 (noting that most brokerage customers do not know about potential broker agency problems and so do not know that their brokers may not be representing their orders as best they might). 556While consolidated revenues may be available from Form 10–K filings for broker-dealers that are public reporting companies, broker-dealers do not report revenues attributable to specific sources, such as rebates from a particular exchange or payments for order flow from a particular venue. For instance, revenues derived from commissions and fees are often just reported in aggregate as ‘‘Commissions and Fees.’’ Therefore, even though aggregate revenues for some broker-dealers are publicly available, customers do not have access to the information on individual sources of revenue that could reveal potential conflicts of interest. 557These switching costs may be monetary, but may also have a time and effort component. 558Collective action occurs when a number of individuals or entities work together to achieve a common objective, such as investors acting to reduce the potential conflicts of interest in order routing decisions by broker-dealers. the potential effect on efficiency, competition, and capital formation. Like the Proposing Release, where possible, the Commission has quantified below the likely economic effects of the Pilot; however, as explained further below, the Commission is unable to quantify all of the economic effects because it lacks the information necessary to provide reasonable estimates. In some cases, quantification depends heavily on factors outside of the control of the Commission, which makes it difficult to predict how market participants would act under the conditions of the Pilot. For example, because of the flexibility that market participants have with respect to the choice of trading center for execution of transactions and because those choices can be influenced by factors outside of the scope of the Pilot, such as volume discounts, the Commission cannot quantify, ahead of the Pilot, the economic impact of any changes in order routing decisions by brokerdealers that may result from the Pilot. Nevertheless, as described more fully below, the Commission provides both a qualitative assessment of the potential effects and a quantified estimate of the potential aggregate initial and aggregate ongoing costs, where feasible. A. Background and Market Failures The Commission’s Proposal provided a
review of transaction-based fee models, including a discussion of the history and mechanics of transactionbased pricing and an overview of the concerns about potential conflicts of interest between broker-dealers and their customers attributed to access fees and rebates assessed by exchanges as well as the potential distortions that exchange fee models can introduce into market structure.551 The Commission considered whether competition within the broker-dealer industry, as well as competition among the equities exchanges, is sufficient to alleviate potential conflicts of interest presented by exchange fees and rebates and also the potential distortions such fee-and-rebate models may introduce. The Commission believes that competition between broker-dealers may not be capable of addressing these potential conflicts and distortions for three reasons: asymmetric information, switching costs, and a lack of collective action, each of which is discussed below. Further, competition between broker-dealers is not readily capable of independently resolving the other potential concerns presented by exchange fee models, such as excessive intermediation, fragmentation, complexity, and cross-subsidization because those issues are within the exclusive control of the exchanges. The limitations of competition among the equities exchanges is discussed in detail below.

Market Failure at the Broker-Dealer
Level
The Commission considered whether competition could alleviate potential conflicts of interest between investors and broker-dealers, as investors choose broker-dealers to place orders on their behalf.552 To the extent that investors are able to identify broker-dealers that do not act on potential conflicts of interest in a manner inconsistent with the interests of their customers, investors could discourage brokerdealers from acting on such conflicts of interest and avoid doing business with those broker-dealers that do not offer such assurances. However, several commenters opined that competition and deference to market forces alone would not be sufficient to challenge the ‘‘deeply rooted conflicts of interest’’ that they believe are present in today’s market structure.553 For example, one commenter noted that many institutional clients are tied to large broker/dealers because of the multitude of services that their brokers provide, so they cannot simply ‘‘fire their brokers’’ if they are unhappy with their routing decisions.554 Further, the Commission does not believe that competition among broker-dealers alone will be sufficient to address potential conflicts of interest in order routing decisions because of three conditions that are present in today’s markets: asymmetric information, switching costs, and a lack of collective action. First, asymmetric information between broker-dealers and their customers limits the ability of customers to identify broker-dealers that do not act on potential conflicts of interest.555 For example, customers do not generally have access to information about brokerdealers’ individual sources of revenue.556 As discussed below in more detail, although disclosures required pursuant to Rule 606 provide information about material conflicts of interest related to payment for order flow, these disclosures do not provide information on the effect of transaction fee-and-rebate pricing models on order routing decisions. Moreover, while under Rule 606, a customer may request certain information about how her broker routed certain orders on her behalf, a customer cannot necessarily use this information to compare how these orders would have been treated by broker-dealers other than her own. Further while recent amendments to Rule 606 would provide customers with limited information about transaction fees paid and transaction rebates received by the broker, the disclosure would not provide data to enable the customer to assess the impact of exchange transaction fees and rebates on market quality and execution quality. Second, even if investors had sufficient information to conclude they would be better served by a different broker-dealer, investors may face costs in switching broker-dealers.557 If these switching costs are high relative to the costs that investors anticipate may arise from potential conflicts of interest, investors may not switch broker-dealers even if it appears that their brokerdealer may have acted on
conflicts of interest. The presence of switching costs also may exacerbate a collective action problem among investors.558 Investors could provide incentives to brokerdealers to eliminate potential conflicts of interest by threatening to move accounts away from broker-dealers known to act on conflicts of interest. The collective action problem arises because, although each customer individually bears a cost to switch accounts, the benefits of a successful threat are available to all customers whether they would switch or not. If the

559Clearpool Letter, at 2. See also T. Rowe Price Letter, at 1 (‘‘enthusiastically agree[ing] with the Commission that a pilot is necessary to gather data,’’ in part because ‘‘exchanges have little incentive to reduce the fee cap on their own’’). See also Larry Harris Letter, at 9 (noting that ‘‘regulatory action is necessary to establish a common pricing standard because market forces alone will not do it’’). Larry Harris Letter, at 6 (noting that ‘‘exchange holding companies have a strong interest in maintaining the current system’’ and that the ‘‘SEC may reasonably consider these interests when evaluating comments submitted by the exchanges’’); Themis Trading Letter II, at 3 (stating that the Commission should not be ‘‘distracted . . . by conflicted stock exchanges desperately fearful that their business models might come crashing down’’). 560See Larry Harris Letter, at 9. See also Themis Trading Letter I, at 5 (noting that several exchanges oppose the pilot because they are motivated by their ‘‘own profit incentives and not what is best for the market’’). 561See Proposing Release, supra note 2, at 13011–
12. See also Section IV.B.1.a.ii, infra discussing the
Nasdaq Experiment in greater detail.
562See, e.g., Themis Trading Letter I, at 3 (stating that a ‘‘more comprehensive multilateral marketwide approach would be needed to yield usable data that could be used to test how lower access fees, and a lack of rebates, would impact market quality and marketplace behavior’’ (emphasis omitted)); IEX Letter III, at 6 (‘‘Nasdaq’s experiment and its outcomes aren’t a perfect proxy for what is likely to happen in the Transaction Fee Pilot. That experiment was done unilaterally and only in highly-liquid securities.’’); Larry Harris Letter, at 9 (noting that Nasdaq’s ‘‘experimental fee reduction did not occur at all trading venues that traded the subject securities,’’ demonstrating that ‘‘regulatory action is necessary to establish a common pricing standard because market forces alone will not do it’’). 563See supra Section IV.E.1 for the discussion of the alternative that the Commission proceed with rulemaking initiatives without first conducting the Pilot. That alternative differs from the baseline presented here because it directly presumes regulatory changes whereas the baseline for the Economic Analysis does not presume regulatory changes resulting from the Pilot. 564See, e.g., Proposing Release supra note 6 at
Section IV.A. and C.
565Several commenters supported the Pilot as a necessary step to produce data to inform the heavily contested debate surrounding the impact of exchange fees and rebates on order routing, market quality, and execution quality. See, e.g., Barnard Letter, at 1 (‘‘historically there are many views on this topic, but a paucity of credible data from which to draw conclusions’’); Wellington Letter, at 1, and Clark-Joseph Letter, at 1. 566See Nuveen Letter, at 2 567See e.g., Spatt Letter, at 3 568See e.g., Larry Harris Letter, at 10 569See Babelfish Letter, at 2–3 570See CFA Letter, at 2 571See Healthy Markets Letter I, at 2, 6 switching costs are high relative to the proportion of customer defections necessary to threaten a broker-dealer, customers are unlikely to generate enough of a threat to alter brokerdealers’ behavior.
2. Market Failure at the Exchange Level
Several commenters considered whether existing market forces, including competition among the equities exchanges, are sufficient to address the potential distortions caused by exchange pricing models. Some commenters felt that ‘‘some regulatory solution,’’ like the Pilot, ‘‘may be necessary to force market participants, particularly exchanges, to change the manner in which they conduct business’’ because competitive pressures on exchanges may serve as a barrier to market-led reforms in this area.559 Further, one commenter noted that ‘‘market forces cause the exchanges to choose maker-taker and inverted fee models to the detriment of the public interest’’ and therefore regulatory action is necessary to address market distortions caused by the maker-taker and taker-maker fee models.560 Further, the Commission notes that one market conducted a limited unilateral access fee experiment in 2015 to test the impact of reductions to its fees and rebates on 14 securities traded on its market. 561 Several commenters noted the limited utility of that study given its narrow scope and applicability to one market.562 The fact that no other exchange joined in the 2015 access fee experiment, or independently undertook a similar study thereafter, supports the view that it is unlikely that competition among the exchanges alone would compel the exchanges to study, let alone address, potential distortions that may result from their fee and rebate models. B. Baseline We compare the economic effects of the rule, including benefits, costs, and effects on efficiency, competition, and capital formation, to a baseline that consists of the existing regulatory framework and market structure. As explained above, by temporarily altering the fee and rebate structure for certain NMS stocks (including ETPs), the Pilot is designed to produce information on order routing behavior that would not otherwise be available. The baseline, therefore, includes the existing information available to the Commission in the absence of a pilot, which the Commission could use to inform future regulatory action.563 The baseline also sets out the exchanges’ current practices with respect to fees and rebates and the regulations governing those fees and rebates.

Current Information Baseline
While the theoretical studies referenced in the Proposing Release suggest that transaction-based fee models create potential issues for investors,564 limited empirical evidence exists to date about the extent that potential conflicts of interest arise from maker-taker and taker-maker pricing models and how exchange transactionbased fees and rebates impact market and execution quality and affect the integrity and structure of the U.S. equity markets.565 Consequently, the relation between transaction-based pricing and conflicts of interest is not well understood.566 Additionally, commenters are divided as to how to interpret existing knowledge. One Commenter stated that we had ‘‘much to learn’’ 567 while other commenters felt that there was sufficient existing knowledge to move directly to rule making without a Pilot.568 Below, we discuss the existing information currently available to the Commission or the public that concerns the relationship between transactionbased fee-and-rebate pricing models and order routing decisions and we describe the limitations of this information for use in policy discussions regarding transaction-based fees and rebates and the potential conflicts of interest and potential distortions that may accompany them. We then discuss the potential to produce additional information regarding the impact of exchange fees and rebates absent the Pilot. While a number of studies attempt to document the relation between transaction-based fees, order routing decisions, and execution quality, these studies and available data sources are limited in ways that are likely to reduce the strength of conclusions that relate to the impact of transaction-based fees and rebates on order routing decisions and the existence or magnitude of potential conflicts of interest between brokerdealers and their customers. This
section details these limitations.
a. Limitations of Existing Studies Multiple commenters submitted empirical evidence that they argued was consistent with conflicts of interest. For example, one Commenter cited evidence that trade execution algorithms that are fee sensitive tend to have lower execution quality than algorithms that are not fee sensitive.569 Another Commenter cited existing academic, industry, and government sources suggesting the existence of conflicts of interest, or of the investing public’s perception that there exist conflicts of interest.570 Another commenter suggested evidence existed that routing decisions were not always in the best interest of investors by arguing that adverse selection differs by exchange, and that this difference can be observed using TAQ data.571 Another commenter presented their study arguing that

572See IEX Letter IV, at 9 573See AGF Letter, at 1 574See CBOE Letter I, at 5 See also Nasdaq Letter I, at II–12 575Over the last five years, the exchanges, on average, have made 34 revisions, or approximately
6.7 revisions per year, to their transaction-based
fees and rebates. See infra Section IV.B.2.b. 576The method of instrumental variables is used to estimate causal relationships when controlled experiments or exogenous shocks are not feasible. An ‘‘instrument’’ changes the explanatory variable but has no independent effect on the dependent variable, allowing a researcher to uncover the causal effect of the explanatory variables on the dependent variable of interest. 577See Harris Letter, at 10. 578The Battalio Equity Market Study’s abstract of the paper states: ‘‘We identify retail brokers that seemingly route orders to maximize order flow payments by selling market orders and sending limit order to venues paying large liquidity rebates. . . . [W]e document a negative relation between limit order execution quality and rebate/ fee level. This finding suggests that order routing designed to maximize liquidity rebates does not maximize limit order execution quality. . . .’’ See Battalio Equity Market Study, supra note 5307, at 2193. 579See Battalio Equity Market Study, supra note
5307. See also supra Section IV.A.2, for an
overview of the potential conflicts of interest that emerge. 580Rule 606 requires broker-dealers to provide quarterly reports that provide an overview of their routing practices. See Securities Exchange Act Release No. 51808 (November 27, 2000), 65 FR 75414, (December 1, 2000) (hereinafter ‘‘Disclosure of Order Execution and Routing Practices’’). See also supra note 310 and accompanying text and infra Section IV.B.1.b.i, ‘‘Rule 606 Data.’’ 581The Battalio Equity Market Study, however, does not specify whether the limit orders are marketable or non-marketable limit orders, as Rule 606 disclosures do not segment these orders. See Battalio Equity Market Study, supra note 530. 582Estimates based on data from Form 1 of the X–17A–5 filings. As of December 31, 2017, 3,860 broker-dealers that filed form X–17A–5. See infra
Section IV.B.2.a.
longer queues lead to increased transaction costs, and connected longer queues with the practice of paying rebates.572 Another Commenter referenced a study suggesting that trading costs vary across exchanges.573 Although the above listed commenters all felt that the evidence did suggest that fees and rebates led to conflicts of interest, other commenters did not come to the same conclusion. One commenter felt that there was ‘‘no evidence that fee practices are harming investors or interfering with fair competition’’ and consequently felt that a Pilot was not justified.574 The studies and analysis presented by Commenters and the studies discussed below have significant limitations with regard to establishing causal links between fees and rebates and order routing decisions. These limitations fall primarily into two categories: (1) The results of the studies may not be representative, and (2) the results of the studies cannot make a causal connection needed to inform on potential conflicts of interest. When a study’s results are representative, the results can be applied across a broadly defined group. Drawing broad inferences from limited samples could be problematic because the results might be specific to specific securities, broker-dealers, or trading venues. In the context of regulatory decision-making, representative results should inform on the potential effects over the scope of the market covered by the decision. When results are not representative of the full scope of a regulatory decision, that regulatory decision may have an unpredictable effect over the part not represented by the results. For example, if the results of a study cover only certain types of issuers, the results may not apply to all types of issuers and therefore, any regulatory changes based on such studies may have unanticipated effects on the types of issuers not included in the study. In addition to limitations in how representative results may be, existing studies cannot test for causal relationships between transaction fees and order routing decisions, even around fee revisions. Because transaction-based fees and order routing decisions could be jointly determined, researchers cannot readily disentangle the direction of causality, and therefore cannot determine the extent that potential conflicts exist. The identification of causal relations between fees and order routing decisions becomes increasingly complex because exchanges frequently modify their fees.575 In practice, researchers attempt to identify and measure causal relations in two ways: (1) Exogenous shocks and (2) econometric techniques, such as an instrumental variables approach.576 The Commission disagrees with one commenter who felt that sufficient data existed to move forward with regulation prohibiting rebates because ‘‘the theory is well-accepted, and no prior evidence contradicts it.’’ 577 In the absence of causal data, regulators can use theory— and their best judgment based on their expertise—to
guide their decision making. However, in this case, for the reasons discussed throughout this release, the Commission believes that empirically assessing the various theories, causal impacts, and effects of the transaction fee-and rebate pricing model is appropriate.
i. Battalio Equity Market Study
According to the Battalio Equity Market Study, broker-dealers appear to trade execution quality of customer orders, as measured by the likelihood of and time to execution (and not price), for the rebates obtained by providing liquidity to maker-taker venues.578 By routing orders to exchanges that pay high rebates, broker-dealers may engage in rebate capture at the expense of client execution.579 Using data obtained from mandatory Rule 606 disclosures over a two-month window,580 the Battalio Equity Market Study also identified that four of the ten broker-dealers included in the analysis route limit orders exclusively to market makers or to exchanges that offered the largest liquidity rebates (and charged the highest access fees). A number of tests in the Battalio Equity Market Study also show that low-fee venues provide better execution quality for limit orders, as measured by the likelihood of an order fill, the speed of execution, and realized spreads, relative to high-fee venues, suggesting that order routing decisions to high rebate venues are likely to be suboptimal from a customer’s perspective, and may be indicative of potential conflicts of interest. Although the Battalio study provides evidence suggestive of conflicts of interest, the study has a number of limitations which render the Commission unable to use this study to robustly determine that rebates cause costly conflicts of interest for brokerdealers. First, the Battalio Equity Market Study uses order level data from a single broker-dealer to determine the relation between maker-taker fees and limit order execution quality.581 Analysis based on observation of a single brokerdealer may not provide representative results because the relation between transaction-based fees and potential conflicts of interest may not be generalizable to other broker-dealers. For example, over 400 broker-dealers maintain membership with at least one U.S. equities exchange.582 If the single broker-dealer examined in the Battalio Equity Market Study has significantly different order routing behavior than the average broker-dealer that routes orders to exchanges, the information obtained from examining the relation between transaction-based fees and order routing decisions of that broker-dealer would not be representative of the entire market and therefore would provide an incomplete representation of potential conflicts of interest. The Battalio Equity Market Study also relies on a sample of Rule 606 order routing reports obtained directly from the reporting entities’ websites from a limited sample of ten well-known national retail brokers from a single quarterly reporting cycle (October and

583The Nasdaq study lowered access fees to $0.0005 and rebates to $0.0004 simultaneously for a set of 14 securities, half of which identified Nasdaq as the primary listing exchange, the other half which identified the NYSE as the primary listing exchange. Nasdaq released two reports see infra note 584 (examining the changes to a number of metrics related to market quality). 584The first report provided by Nasdaq can be found on their webpage http://qnasdaqomx.com/ AccessFeeExperiment (‘‘Nasdaq’s first report’’, or the ‘‘first Nasdaq report’’). The second report provided by Nasdaq can be found at http:// people.stern.nyu.edu/jhasbrou/SternMicroMtg/ SternMicroMtg2015/Supplemental/ (‘‘Nasdaq’s second report’’, or the ‘‘Second Nasdaq report’’). 585See Swan Letter which submitted the paper:
Yiping Lin, Peter Swan, & Frederick Harris, Why Maker-Taker Fees Improve Exchange Quality:
Theory and Natural Experiment Evidence, Working Paper, (2018), https://papers.ssrn.com/sol3/ papers.cfm?abstract_id=3034901 (hereinafter ‘‘Swan study’’); Nasdaq’s Second Report, at 1. 586Cum fee indicates that the computation of spreads included the fee or rebate charged. It is a measure of the total cost of transacting. 587The effective spread is the cost to transact and is defined as two times the absolute difference between the price of a trade and the prevailing midpoint at the time of trade. The effective spread can be decomposed into two components, the realized spread and price impact of the trade. The price impact is generally viewed as the portion of the effective spread that compensates market makers for adverse selection losses. The realized spread is the portion of the spread that market makers ‘realize’ after adverse selection costs are taken into account. Raw realized spreads are realized spreads that do not take into account the all-in cost of trading, i.e., they exclude rebates from the calculations 588Raw effective spreads are effective spreads that do not take into account the all-in cost of trading, i.e., they exclude fees from the calculations. 589The interpretation of the price efficiency results is difficult because it is unclear what price efficiency on one exchange means in the absence of the other exchanges. Usually price efficiency is measured across all exchanges trading a given security. November 2012). As discussed above, approximately 400 broker-dealers are members of at least one national securities exchange. The ten retail brokers analyzed in the Battalio Equity Market Study make up approximately 2.1% of the broker-dealers with exchange memberships, and less than 0.3% of broker-dealers overall. Although these are well-known retail brokers, due to the lack of representativeness of the sample (e.g., the majority of the broker-dealers represented in the Battalio Equity Market Study are online broker-dealers), these broker-dealers may be more (or less) likely than the average brokerdealer to route customer orders in ways that benefit themselves at the expense of their customers. The findings in the Battalio Equity Market Study, therefore, may not be representative of a broader sample of broker-dealers. Moreover, the Commission is unable to determine if the Battalio Equity Market Study’s analyses of the Rule 606 disclosure data has statistical power because the authors did not provide any statistical analyses beyond the percentage of market or limit orders routed to a particular exchange. In sum, the absence of an exogenous shock to access fee caps or rebates outside the control of exchanges leaves the authors unable to definitively determine the causes of broker-dealers’ order routing decisions. Consequently, the authors are unable to disentangle whether fees and rebates drive brokerdealer order routing decisions or order routing decisions determine fees and rebates chosen by exchanges.
ii. The Nasdaq Experiment
Nasdaq independently conducted an experiment, whereby it lowered access fees and rebates for a sample of 14 stocks over a period of four months in 2015, providing an exogenous shock to the transaction-based pricing model on the exchange. The Nasdaq experiment lowered both the access fees charged and the liquidity rebates paid on the securities included in their study.583 Nasdaq produced two reports on the experiment 584 and an academic study examining the experiment was submitted as a comment.585 Both Nasdaq’s first study report and the Swan study indicate that when Nasdaq lowered fees and rebates they lost market share in the stocks with lower fees and rebates. According to analysis in the Swan study, the market share that Nasdaq lost appeared to migrate to other make-take venues with higher fees and rebates. Additionally, both Nasdaq’s analysis as well as the Swan study find that the experiment led to a decrease in the fraction of time that Nasdaq quoted at the NBBO. The Swan study also estimated a variety of additional tests to measure the impact of the experiment on various aspects of market quality. The results of these tests are mixed. The Swan study found that the Nasdaq experiment improved market quality on Nasdaq in terms of improved fill rates and fill times as well as narrower cumfee effective spreads and cum-fee realized spreads.586 While cum-rebate effective spreads, fill rates and fill times improve on the Nasdaq during the experiment, the Swan study finds that the experiment diminished market quality in terms of quoted spreads and raw realized spreads which both increase during the experiment.587 Additionally, the Swan study shows that some measures of market quality were unchanged by the Nasdaq experiment, namely, the Swan study finds no change in raw effective spread.588 In Nasdaq’s second report on the experiment they examine various market quality measures and find no impact on effective spread, relative effective spread, quoted spread, relative quoted spread, displayed dollar depth at the NBBO, time between quote updates on the consolidated tape, and time between price changes in the NBBO on the consolidated tape. In examining the impact of the experiment on price efficiency, the Swan study finds mixed evidence that prices quoted on Nasdaq become less efficient during the experiment. First, the Swan study finds that global price impact declines during the experiment. Price impact is commonly employed as a measure of the informativeness of trades. The Swan study explains the decline in price impact with a theoretical model which suggests that rebates subsidize market makers for the adverse selection costs that they bearthereby allowing them the ability to bear additional adverse selection which induces informed traders to trade more aggressively in the presence of rebates. Consequently, their model predicts that informed trades will congregate on exchanges with high rebates. Additionally, the Swan study finds using variance
ratios that price efficiency declines on Nasdaq during the experiment. However when using autocorrelation of trades as a measure of price efficiency, the tests indicate a decrease in autocorrelation—suggesting more efficient stock prices on Nasdaq.589 Additional analysis on price efficiency comes from Nasdaq’s second report which explores the impact of their experiment on market wide price efficiency and finds no change in price impact, autocorrelation of trades, or variance ratios. The Swan study also empirically examines how the Nasdaq experiment impacted the trading behavior of high frequency traders (‘‘HFTs’’) and nonHFTs and finds that as a result of the experiment HFTs added liquidity less often and took liquidity more often while non-HFTs did the opposite. Nasdaq also examined trading behavior and found that there was a shift in the composition of the top five liquidity providers for the securities that occurred as a result of the experiment. The top five liquidity providers prior to the start of the pilot significantly reduced their liquidity provision from 44.5% of the liquidity provided prepilot to 28.7% in the pilot period. However, the top five liquidity providers from the pilot period had a significant increase in their liquidity provision from 29.7% pre-pilot to 41.5% in the pilot period.

590See Nasdaq Letter I, at 10.
591Only common stocks were included in the Nasdaq study, while the proposed Pilot will include NMS stocks, which includes common stocks as well as ETPs. 592Market capitalizations are computed from CRSP shares outstanding and stock price, as of December 31, 2014. See also Themis Trading Letter I, at 2; NorthWestern Letter, at 1 and IEX Letter III, at 6. 593Nasdaq acknowledges this limitation in their second report analyzing the experiment. See supra note 584 See also Themis Trading Letter I, at 2 and IEX Letter III, at 6 for commenters expressing similar concerns about the representativeness of Nasdaq’s sample. 594See Swan Letter, at 3; Themis Trading Letter I, at 2; Credit Suisse Commentary, at 2, Larry Harris Letter, at 9, and IEX Letter III, at 6. 595This point was acknowledged in Nasdaq’s second report. See supra note 584. This point was also brought up by multiple commenters. See Swan Letter, at 3; Credit Suisse Commentary, at 2; and Larry Harris Letter, at 9. 596See Amber Anand, Jian Hua, & Tim McCormick, Make-Take Structure and Market Quality: Evidence from the U.S. Options Markets, 62 Mgmt. Sci. 3085, 3217–90 (2016), https:// pubsonline.informs.org/doi/abs/10.1287/ mnsc.2015.2274 (hereinafter ‘‘Anand, Hua, & McCormick’’); Robert Battalio, Todd Griffith, & Robert Van Ness, Make-Take Fees versus Order Flow Inducements: Evidence from the NASDAQ OMX PHLX Exchange, 12th Ann. Mid-Atlantic Res. Conf. in Fin. (2017), http://www1.villanova.edu/ content/dam/villanova/VSB/assets/marc/ marc2017/SSRN-id2870000.pdf (hereinafter ‘‘Battalio, Griffith, and Van Ness’’); Robert Battalio, Andriy Shkilko, & Robert Van Ness, To Pay or Be Paid? The Impact of Taker Fees and Order Flow Inducements on Trading Costs in U.S. Options Markets, 51 J. Fin. & Quantitative Analysis (Oct.
2016) 1637, 1637–62 https://www.cambridge.org/
core/journals/journal-of-financial-and-quantitativeanalysis/article/div-classtitleto-pay-or-be-paid-theimpact-of-taker-fees-and-order-flow-inducementson-trading-costs-in-us-options-marketsdiv/ 0782CE3E9679C29BB910A66192D27201 (hereinafter ‘‘Battalio, Shkilko & Van Ness’’). Anand, Hua, & McCormick explores the transition from a payment for order flow model to a makertaker model on NYSE ARCA, while Battalio, Griffith, and Van Ness as well as Battalio, Shkilko, & Van Ness examine the shift on NASDAQ OMX PHLX (‘‘PHLX’’) from a maker-taker model to a payment for order flow model. 597 Id. 598 Id. 599See Battalio, Shkilko & Van Ness, supra note 596, at 1637–62. 600See STANY Letter, at 2; Nasdaq Letter I, at 11– 12; NYSE Letter I, at 17; and Nasdaq Letter III, at 1–2. 601See Era Letter, at 1. But cf. IEX Letter II, at 9. 602Battalio Equity Market Study, supra note 530, relies on Rule 606 disclosures to identify order routing for a small sample of broker-dealers, proprietary broker-dealer data from a single smartorder routing system to capture limit order execution quality for this broker-dealer’s orders, and the TAQ data to measure execution quality as a function of each venue’s taker fee or rebate. While Nasdaq believes that the results from their study do not support the need for a pilot,590 the Commission disagrees because the Nasdaq experiment and the subsequent analysis suffers from the following limitations. First, the Nasdaq experiment may not be representative of the broader market. Nasdaq selected 14 stocks to be part of the analysis, which represent 0.3% of all NMS stocks. The sample is unlikely to be representative of the universe of NMS securities for two reasons: (1) The sample included a small number of stocks (and no ETPs),591 and (2) less than one-third of these stocks were small or mid-capitalization at the time of the analysis, although most had market capitalizations close to $3 billion immediately prior to the study.592 The small number of stocks makes interpretation of the results more difficult because a change to such a small number of stocks may not be significant enough for traders to alter their behavior.593 Additionally, the Commission is not able to make inference about the effect of a market wide change to fees and or rebates from the Nasdaq experiment because, as noted by multiple commenters, the effects of the experiment apply to a single exchange:
Nasdaq.594 As the other equities exchanges did not have similar changes to transaction-based fees and rebates, any inferences drawn from the Nasdaq study may not be valid under different circumstances in which all equities exchanges were subject to consistent revisions to transaction-based fees. 595 Lastly, none of the analysis of the Nasdaq study analyzes the impact of potential conflicts of interest on order routing decisions. Further, even if the Nasdaq study had analyzed a causal relationship between transaction-based fees and rebates and potential conflicts of interest, the limited representativeness of the Nasdaq sample would limit the generality of the study.
iii. Options Market Studies
Three studies have examined exogenous shifts between maker-taker and payment for order flow pricing models on U.S. options exchanges.596 These studies found that the movement from a payment for order flow model to a maker-taker model led to a decrease in execution costs for option classes affected by the shift, improved quoted spreads, and altered broker-dealer order routing behavior to account for the fees.597 However, the change to a payment for order flow model from a maker-taker model yielded better execution quality, but a reduction in the number of orders and order volume.598 With respect to the transition between forms of pricing models that occurred on the option exchanges, discussed above, the key limitation is the comparison of maker-taker pricing models with payment for order flow pricing models. For example, studies that explore these regime shifts between maker-taker to payment for order flow models are not comparing situations in which one regime could theoretically have lower conflicts of interest than the other. Each of these models is likely to create potential conflicts of interest that could affect how broker-dealers route their customer orders,599 although evidence does not suggest that one form of pricing model is more or less prone to conflicts than another. Moreover, the change from one form of pricing model to another could introduce new conflicts of interest or impacts on market and execution quality that did not previously exist. Therefore, the Commission believes that exchangedriven transitions between maker-taker and payment for order flow pricing models are not likely to provide information about potential conflicts of interest and impacts on market and execution quality driven by the makertaker and taker-maker models or to inform the Commission about future regulatory decisions regarding transaction-based fee models. Additionally, these studies lack causality. Specifically the decision to invert an exchange from a taker/maker to a maker/taker exchange, which these studies are based on, is an endogenous decision, and therefore these studies lack the ability to make causal inference further hindering the Commission’s ability to draw inference from these studies. b. Limitations of Existing and Anticipated Data Some Commenters suggested that existing data sources could be employed in lieu of a Pilot to study the Commissions objectives.600 Another Commenter argued that enhancing existing data would be sufficient.601 To this end, the Commission considered whether a number of existing data sources could be used independently or in combination to relate transactionbased fees to order routing and execution quality. This section discusses these data sources. For instance, in the Battalio Equity Market Study and the Nasdaq study discussed above, the authors employed some combination of Rule 606 data, proprietary broker-dealer data, the Trade and Quote (TAQ) database,602 and proprietary exchange data. In addition, while not
employed in previous studies, CAT data, OATS data, Rule 605 data, Form ATS–N data, and exchanges’ Form 19b–4 fee filings and fee schedules available from each exchange’s website, could provide insights into the relation between transaction-based fees, order routing, and execution quality, and fees

603See Section III.B.3 of Amendments to Order Handling Disclosure, supra note 310. 604See Cboe letter I, at 26; FIA Letter, at 3; STANY Letter, at 2; Nasdaq Letter I, at 1–2, 4; NYSE Letter I, at 18. 605See STANY Letter, at 2; Nasdaq Letter I, at 11– 12; ERA Letter, at 1. 606See also IEX Letter I, at 9. 607See, e.g., ICI Letter II at 3. 608Not every fee schedule revision pertains to transaction fees or rebates. To focus only on these revisions, each Form 19b–4 fee filing was evaluated to determine that revisions to fees or rebates were pertinent to this baseline. 609Multiple commenters expressed views similar to this and urged the Commission to adopt 606 amendments prior to the adoption of the Pilot. See, e.g., Citadel Letter, at 3; OMERS Letter, at 3; ICI Letter I, at 5–6; Fidelity Letter, at 2; IEX Letter II, at 9. 610See supra Section II.E.3.vi. 611See Nasdaq Letter I, at 4. 612See Verret Letter I, at 4. and other arrangements. As noted above, several data sources provide information on order routing and execution quality. While researchers could use these data sources to produce some representative results regarding the relation between transaction-based fees, order routing, and execution quality, the Commission believes that available data has several limitations, which include: Granularity, completeness, periodicity, format, and availability.
i. Rule 606 Data
Rule 606 requires broker-dealers to make publicly available quarterly reports that provide an overview of their routing practices on certain orders in NMS securities. As amended, brokerdealers must provide information for the ten venues to which the largest number of total non-directed orders were routed for execution and for any venue to which five percent or more of nondirected orders were routed for execution. Rule 606 disclosures also require broker-dealers to disclose in a standardized format material aspects of their relationships with trading venues to which they route orders, including a description of, among other things, the payment for order flow and any profit sharing relationships, which, like rebates, could influence the brokerdealer’s order routing decision and potentially lead to potential conflicts of interest for broker-dealers when routing orders.603 Researchers and other analysts interested in order routing data can download these forms quarterly directly from broker-dealer websites. Some commenters believed that the amendments to the Rule 606 data would render the Pilot unnecessary.604 Indeed, a few commenters suggested that Rule 606 data, perhaps combined with other existing data, would be sufficient to study conflicts of interest among brokerdealers.605 The Commission disagrees that this type of analysis would serve the purposes of the Pilot.606 Such an approach would not adequately advance the Pilot’s broader purpose to study the effects that exchange transaction feeand-rebate pricing models may have on order routing behavior, execution quality, and market quality, in addition to conflicts of interest between brokers and their customers that are presented when exchanges pay rebates.607 Further, disclosure alone would not provide an exogenous shock that generates measurable responses capable of providing insight into the effects of fees and rebates on the markets and market participant behavior. In addition, the quarterly frequency of the public Rule 606 reports by brokerdealers is different from the frequency of changes in fee schedules by exchanges (e.g., as presented in Table 2, over a recent five-year measurement period, the average exchange updated its fees schedule approximately 6.7 times per year).608 Further, while the Rule 606 data provides order routing at the broker-dealer level, such information is not granular enough to thoroughly study potential conflicts of interest. Specifically, the 606 data is aggregated at the quarterly level. This frequency will not enable researchers to look at the full picture of how a brokerdealer responds to fees because exchanges on average revise their fee schedules 6.7 times per year. With 13 exchanges this amounts to 87 fee changes per year. Consequently, the fees that exchanges charge in a given quarter relative to the other exchanges will likely change multiple times within a quarter. Consequently, Rule 606 data is limited in how it can be employed to evaluate
comprehensively the impact of order flow responding to fees and rebates. The value of Rule 606 disclosures for identifying possible conflicts of interest resulting from transaction-based fees would be limited for a number of additional reasons. First, each brokerdealer discloses data for only its top ten order routing venues. Second, because broker-dealers disclose data at a quarterly frequency, a five-year sample of Rule 606 data for a single brokerdealer, would include only 20 observations, limiting statistical power. In sum, the Commission believes that the amended Rule 606 data will provide useful information to complement the Pilot; however it is insufficient by itself to determine the impact of exchange transaction fees and rebates on brokerdealer order routing decisions, or inform the Commission of the impact of exchange pricing on market and execution quality.609
ii. CAT Data and OATS Data
Once the CAT Phase 1 becomes operational,610 the Commission and SROs will have information on all exchange routing and exchange executions for all NMS securities. In CAT Phase 1, exchanges would record and report order events on every order they receive for NMS securities. Order events include order receipt, order routes, order modifications, order cancellations, and order executions. Likewise, the Order Audit Trail Systems (OATS) data could inform on order routing decisions.611 The OATS data tracks customer orders from the receipt of the order through execution or cancellation. Information in the OATS data reflects the terms of the order, including the security, price, shares, account type, handling instructions, and side of the market for which the order was placed; where the order was routed for execution; modifications to the order; and execution information, including the capacity in which the firm acted in the trade. Although the CAT and OATS data could feasibly be used to produce order routing data similar to that required by the Pilot, as indicated by one commenter, without the corresponding ‘‘randomized trial,’’ the use of OATS data alone would be insufficient to determine causality in the effect of fees and rebates on order routing decisions because it would not be possible to determine from the data whether fees respond to changes in order routing decisions or whether order routing decisions respond to changes in fees. Consequently, in the absence of an exogenous shock to fees, CAT and OATS data cannot provide the Commission with robust evidence about how access fees impact order routing decisions.612
iii. Proprietary Broker-Dealer Data
Proprietary data from broker-dealers or exchanges could also provide information about order routing decisions. Broker-dealer data include information on the orders received and routed by that broker-dealer, including where the broker-dealer routed orders, whether the orders execute, and the price, size, and time of execution. Exchange data include information on the orders received by an exchange, including which members routed orders to the exchange, whether the orders execute, and the price, size, and time of

613See Nasdaq Letter I, at 11–12 and FIA Letter, at 3. 614See STANY Letter, at 2 and Nasdaq Letter I, at 11–12. 615See Disclosure of Order Execution and Routing Practices, supra note 580, at 75417–25. 616See Securities Exchange Act Release No. 61358 (January 14, 2010), 75 FR 3594 (January 21,
2010) (hereinafter ‘‘Concept Release’’).
617See, e.g., Jennifer Conrad & Sunil Wahal, The Term Structure of Liquidity Provision (August 8,
2018) (unpublished manuscript) https://
papers.ssrn.com/sol3/papers.cfm?abstract_ id=2837111. 618See Securities Exchange Act Release Nos. 83663 (July 18, 2018), 83 FR 38768 (August 7, 2018). execution. Indeed, several commenters stated that if 606 data were not sufficient to answer the Commission’s questions about broker dealer routing decisions, then the Commission could request routing tables and information directly from broker-dealers or request other Transaction Cost Analysis (TCA) data to supplement the 606 data.613 While these data would provide potentially more granular data about order routing, as proprietary datasets, there is no standard format that exchanges or broker-dealers use to aggregate this data, which makes cross broker-dealer or cross exchange comparison difficult. Even if a dataset of proprietary data could be produced from data obtained directly from exchanges or broker-dealers, the data would still lack an exogenous shock to fees which is necessary to determine a causal link between order routing decisions and exchange fees.
iv. Rule 605 Data
A few commenters suggested that Rule 605 data used in conjunction with other data such as Rule 606 data, could provide information about broker dealer conflicts of interest.614 Rule 605 data provides information about execution quality by market center, including exchanges, ATSs, and broker-dealers that execute orders, by requiring standardized reports of statistical information regarding order execution, and was designed to improve the public disclosure of order execution practices by exchanges.615 These data are available monthly from market center websites or data vendors, and provide information on execution quality statistics such as transaction costs, execution speed, and fill rates reported separately for marketable and nonmarketable orders. While Rule 605 data is available to researchers and may provide information about execution quality, it too has a number of limitations. For example, Rule 605 data provides execution quality information for both marketable and non-marketable orders; however, the methodologies for estimating measures of the speed of execution of non-marketable orders are outdated.616 For instance, Rule 605 measures realized spreads based on quotations five minutes after the time of order execution and recent research suggests using quotations that more closely follow a trade, because any temporary price impact of a trade goes away within seconds, not minutes, of the trade.617 Finally, Rule 605 data is limited in that it covers only held orders and orders of less than 10,000 shares.
v. TAQ Data
Beyond Rule 605 data, researchers could also use the TAQ database as a means of measuring order execution quality or estimating market share to use as a measure for order routing decisions. The TAQ database is publicly available (for a fee) from NYSE and provides access to all trades and top of the book quotes for NMS securities, from which researchers and other analysts can estimate trade-based measures of execution quality such as effective spreads. While TAQ data are available to academic researchers, TAQ has a number of limitations in its precision in the measurement of order routing and execution quality. An exchange’s market share can differ significantly from its share of orders received because exchanges reroute orders they cannot execute at the best prices and some exchanges reroute more orders than others. In addition, TAQ doesn’t provide information on the brokers or dealers underlying the trades or quotes, so TAQ cannot tell us about the decisions of individual brokers. While TAQ facilitates the estimation of trade and quote-based measures of execution and market quality, it does not facilitate the estimation of order-based measures of execution quality, which are more precise than trade-based measures. In particular, order-based measures allow for the consideration of order size, which can be different and often larger than trade size. Order-based measures also consider the costs of latency whereas trade-based measures do not. Additionally, since TAQ only provides data on trades it does not provide a means of estimating execution quality for limit orders.
vi. Information From Exchange 19b–4
Filings
Finally, researchers both in and outside of the Commission, who wish to link fees and rebates to various outcomes, can manually create datasets of exchange fees and rebates from the information that exchanges provide on their websites and release in their Notice of Filing of Proposed Rule Changes, which would capture information contained in exchanges’ Form 19b–4 fee filings. The Form 19b– 4 fee filings record changes to the existing exchange fee schedules with the Commission. At any point that an exchange chooses to make a change to any aspect of its fees and rebates, the exchange must provide notice to the Commission that it is filing a proposed rule change to amend its existing fee and rebate schedule. Exchanges may file their revisions to fees and rebates for immediate effectiveness upon submitting the Form 19b–4 fee filings with the Commission. A key limitation to this data, particularly for researchers outside the Commission, is that exchanges use bespoke terminology to classify their fees and rebates. Consequently, identifying comparable fees across exchanges is difficult. For example, identifying the base or top-tier fees across exchanges could be difficult for researchers. As shown in Table 2 below, the average exchange has 24 different access fee categories and 21 different rebate categories. Further, exchanges do not disclose per share average or median fees charged and rebates earned on any report or filing, so such information is unavailable to the public. To add to the impediments to fee data aggregation and comparison, Form 19b–4 fee filings are available only as PDF files downloadable from the Commission’s website, thereby increasing the costs of aggregation across exchanges over time by researchers. Lastly, even if a comprehensive dataset of fee changes were created, it would not be sufficient by itself to study the link between order routing decisions and fees because the dataset can only tell when and how an exchange revised fees, and not why the fee changed or if the fee change affected order routing behavior. In essence the data still lacks the ability to establish a causal connection between fee changes and order routing decisions.
vii. Form ATS–N
Following implementation in January 2019, the public will have more information on ATS conflicts of interest and fees. In particular, in June 2018, the Commission adopted amendments to 17 CFR 242.300 through 242.303 (Regulation ATS) and 17 CFR 240.3a1– 1 (Rule 3a1–1 under the Exchange Act).618 As part of these amendments, NMS Stock ATSs will be required to

619 IEX charges a flat fee of $0.0009 for trades against non-displayed liquidity on both sides of the market, and charges $0.0003 for trade execution against displayed liquidity. See IEX, Investors Exchange Fee Schedule (August 1, 2018), https:// iextrading.com/trading/fees (last visited September 18, 2018). As of March 2018, EDGA is no longer operating as a taker-maker market, but is also operating as a flat-fee venue. See Cboe, Cboe US Equities (2018) http://markets.cboe.com/us/ equities/ (last visited September 18, 2018). 620While the number of exchanges was eight, there were other non-exchange trading venues in 2005 (i.e., ECNs), which were displayed markets that utilized a standard price-time-priority market model similar to exchanges. Although 13 U.S. equities exchanges currently operate as of March 2018, the majority of these exchanges are part of exchange families. For instance, NYSE, NYSE Arca, NYSE American, and NYSE National, are all part of the NYSE Group, which is wholly owned by the Intercontinental Exchange (ICE), while Nasdaq, Phlx, and BX, are owned by Nasdaq. BATS, BATS– Y, EDGA, and EDGX, which all operated as ATSs in 2005, are all subsidiaries of Cboe Global Market, Inc. IEX became a registered exchange in 2016. Further, NSX (NYSE National) existed as an exchange in 2005, but halted operations in 2016. It was acquired by NYSE/ICE in January 2017 and was re-opened for trading in May 2018. See Intercontinental Exchange, NYSE Finalizes Acquisition of National Stock Exchange, Bus. Wire:
News (Jan. 31, 2017, 6:28 p.m.) https:// www.businesswire.com/news/home/ 20170131006474/en/. Researchers can adequately control for exchanges that are subsidiaries of the same parent when conducting analyses of the effect of changes in transaction-based fees on order routes. 621See, e.g., Angel, Harris, & Spatt, supra note 530; Harris, supra note 530. publicly report on new Form ATS–N information about the manner in which the ATS operates and activities of the broker-dealer operator and its affiliates, as well as potential conflicts of interest within the NMS Stock ATSs. While Form ATS–N will contain high level information on operations and affiliates of the ATS, it will not contain detailed information, such as ATS routing tables. Therefore, it would not contain detailed information on how fees and rebates affect the order routing decisions of the ATS. Form ATS–N also will require ATSs to provide public disclosures about the different types of fees they charge, along with the ranges of those fees and whether they are bundled with any other services. However this information would not be nearly as granular as the exact fee disclosures that would be required by the Pilot. Nor do they provide as much information as the fee disclosures that exchanges are currently required to disclose. These limitations make it difficult to use the ATS–N data to make causal inference about the impact of fees and rebates on order routing decisions.
c. The Potential To Study the Causal
Link Between Fees, Rebates, and Conflicts of Interest Absent a Pilot Absent a Pilot, the Commission does not believe it would have comprehensive, empirical evidence to study the effects on the market that the Pilot is intended to study. In particular, as indicated above, the Commission does not believe the theoretical evidence on incentives and potential other effects are indicative of brokerdealers actually acting on those incentives. Further, even if the data sources above did not suffer from their limitations, researchers would struggle to identify the causality necessary to robustly link fee and rebate effects on order routing to order execution quality. Indeed, this link requires two steps:
First establishing a causal link between fees and rebates and order routing and then between fee-based order routing and order execution quality. Even with perfect data, any study linking fees and rebates to order routing would suffer from an inability to draw conclusions about causality. While such a study might find a correlation between fees/ rebates and order routing decisions, the researchers would be unable to conclude which event was driving which. In particular, since exchanges compete for market share, it is reasonable to expect that exchanges change their fees and rebates in response to changes in order routing decisions by broker dealers. If this is the case it would be the order routing decisions that drive the exchange fees. The data alone do not allow researchers to distinguish whether order routing determines fees or whether fees determine order routing. Similarly, existing data, even if it didn’t have the limitations above, would not enable researchers to infer the causal impact of fee-based order routing on order execution quality. If fees and execution quality are linked, then exchanges may change their fees in response to changes in execution quality. For example, raising rebates might attract more liquidity providers and induce additional order flow to the exchange. An exchange that is experiencing low execution quality might raise rebates to address this problem. Under these circumstances, an empirical analysis that lacks an exogenous shock to fees/rebates might erroneously conclude that increased rebates cause a conflict of interest because they are correlated with low execution quality and increased order flow. Such a conclusion might lead the Commission to draw incorrect conclusions.
2. Current Market Environment
This section provides an overview of the competitive landscape that could be affected as a result of revisions to the transaction-based fee structure required by the Pilot. Where information is currently available to the Commission, a description of the current practices of exchanges along dimensions that are relevant to the Pilot (e.g., summary information on their current fee
schedule or the frequency of fee
revisions) are included. The Commission requested that commenters provide additional information to inform the baseline as part of the proposal. Where available, the baseline has been supplemented to reflect additional baseline information that was received from commenters. a. Market for Trading Services The market for trading services, which is served by exchanges, ATSs, and other liquidity providers (internalizers and others), relies on competition to supply investors with execution services at efficient prices. These trading venues, which compete to match traders with counterparties, provide platforms for price negotiation and the dissemination of trading information. The market for trading services in NMS stocks currently consists of 13 national equity market exchanges and 34 operational ATSs. Other off-exchange venues include broker-dealer internalizers and wholesalers, which execute a substantial volume of retail order flow. The remainder of this section discusses the current competitive landscape for exchanges, ATSs, and others relevant to our economic analysis of the Pilot. Since the adoption of Regulation NMS in 2005, the market for trading services has become more fragmented and competitive. Of the 13 exchanges, seven are maker-taker exchanges and four are taker-maker pricing exchanges, as shown in Table 1; the NYSE American and IEX operate as flat-fee exchanges.619 Since Regulation NMS was adopted in 2005, the market for trading services has become significantly more competitive as measured by the decline in market share of individual exchanges, discussed in more detail below. The number of U.S. equities exchanges has increased by over 60%, as the number of exchanges increased from eight exchanges in 2005 to 13 exchanges operating today.620 Several studies have suggested that transaction-based fee pricing partially drove the increase in the number of U.S. equities exchanges since 2005.621 Execution services are a lucrative business, which encourages new trading centers to enter the market in the hopes of capturing rents associated with order

622See, e.g., Angel, Harris, & Spatt, supra note 530; Harris, supra note 530. 623See Proposing Release, supra note 2, at 13042. See also ICI Letter II at 4. 624See supra notes 3 and 4. 625Shares are computed based on trading volume in August 2018. Market shares for the exchanges reported do not add up to 100%, because approximately 34% of trading volume is executed off-exchange on over-the-counter venues. These market share figures differ slightly from the ones in footnote 9 of Cboe Letter I, which provided market share for May 2018. While these differences could result from the focus on more recent data, the Commission is not sure if the differences could also be driven by differing methodologies. Nonetheless, the figures in Cboe Letter I are consistent with the conclusions in this release. Also, the off-exchange share differs slightly from the 39% share in Nasdaq Letter I, at 2. Note that the off-exchange share in the Proposing Release was 40%. 626Since July 2017, NYSE American has not been a purely maker-taker market as only certain types of market participants (electronic Designated Market Makers) are eligible for rebates. See NYSE American Equities Price List (July 26, 2018), https:// www.nyse.com/publicdocs/nyse/markets/nyseamerican/NYSE_America_Equities_Price_List.pdf. 627NYSE acquired NSX in January 2017, and the exchange is now known as NYSE National. The exchange was re-opened for trading in May 2018 as taker-maker exchange. 628See Angel, Harris, & Spatt, supra note 5307, Figures 2.17 and 2.18. Although less evident than for NYSE-listed securities, the effect is similar for the Nasdaq market. 629See, e.g., CBOE Letter I, at 2; NASDAQ Letter I, at 11–13. One commenter suggested that in the Proposal, the Commission made the assumption that exchange groups had market power without providing evidence to support the assumption. This commenter also argued that no exchange group controls even 25 percent of market share and that competition is robust between and among equities exchanges. See NASDAQ Letter I, at 12–13. 630One commenter suggested that in the Proposal, the Commission failed to account for the two-sided nature of exchange platforms when assessing the competitive impact of the Proposal. See NASDAQ Letter II, at 4–5. In the Proposal, the Commission separately discussed the potential impact of the Pilot on the competition for trading volume, see the Proposal, supra note 2, at 13068. The Commission also discussed some ways the Pilot could potentially impact marketable and nonmarketable order flow, see the Proposal, supra note 2, at 13057. Additionally, in Section IV.D.2.a, the Commission separately discusses the potential effects of the Pilot on marketable and nonmarketable order flow. 631A number of commenters said exchanges use rebates to compete to attract limit orders to supply liquidity. See, e.g. NASDAQ Letter I, at 12; Virtu Letter, at 3; State Street Letter, at 2. 632One commenter also suggested that exchanges also compete to
attract to liquidity using many costly features, including rebates, incentive programs, superior execution systems, regulatory quality, and customer service. See NASDAQ Letter I, at 12. execution.622 As discussed in the proposing release, liquidity externalities, where the more liquid venues attract more interest and therefore more liquidity, could result in a single venue (or very limited number of venues) being the preferred trading location for any given stock because all traders could optimally route orders to the venue with the highest liquidity for a given stock.623 But if rebates offered by exchanges are large enough, they provide incentives for market participants to route orders to those venues, in order to capture the rebates, and possibly despite the liquidity profile or execution quality of those venues. Rebates offered by exchanges, therefore, may ‘‘break’’ the liquidity externality.
TABLE 1—U.S. NATIONAL EQUITIES EXCHANGES
Exchange Market fee type 624 Exchange in 2005? Market share 625 (%) Cboe BZX: https://markets.cboe.com ..................................................... Maker-Taker ................................... 6.14 Cboe BYX: https://markets.cboe.com ..................................................... Taker-Maker ................................... 4.86 Cboe EDGA: https://markets.cboe.com .................................................. Taker-Maker ................................... 1.26 Cboe EDGX: https://markets.cboe.com .................................................. Maker-Taker ................................... 5.58 BX: www.nasdaqtrader.com .................................................................... Taker-Maker ................................... ✓ 3.00 Phlx (PSX): www.nasdaqtrader.com ....................................................... Maker-Taker ................................... ✓ 0.70 Nasdaq: www.nasdaqtrader.com ............................................................ Maker-Taker ................................... ✓ 15.76 NYSE Arca: https://www.nyse.com/markets ........................................... Maker-Taker ................................... ✓ 8.87 NYSE American 626: https://www.nyse.com/markets .............................. Flat Fee .......................................... ✓ 0.32 NYSE: https://www.nyse.com/markets .................................................... Maker-Taker ................................... ✓ 12.16 NYSE National 627: https://www.nyse.com/markets ................................ Taker-Maker ................................... ✓ 0.64 CHX: www.chx.com ................................................................................ Maker-Taker ................................... ✓ 0.57 IEX: www.iextrading.com ........................................................................ Flat-fee ........................................... 6.14 Total ................................................................................................. ........................................................ ........................ 66.00
Table 1 highlights that the market
share of trading volume among exchanges is not very concentrated. Although NYSE and Nasdaq have the largest market share of approximately 12% and 16%, respectively, among the exchanges, as of July 2018, these two exchanges collectively account for less than 30% of the total market share of trading volume for NMS stocks, indicating that the market for trading services has become decentralized, and has become more so over time. For instance, between 2004 and 2013, the NYSE’s market share of NYSE-listed stocks declined from approximately 80% to 20%, while market share of other exchanges and off-exchange trading centers has increased.628 This decentralization provides market participants with a choice among venues when they route orders, and may also encourage exchanges to compete to attract order flow. A number of commenters suggested that exchanges compete intensely with each other to attract order flow.629 Transaction-based fees represent one means by which national securities exchanges may compete for order flow, and exchanges may adopt business models that focus on attracting order flow by offering large rebates or charging competitive fees. Exchanges may also develop different business models to attract different types of order flow.630 For example, maker-taker venues may offer large rebates to attract liquidity supplying orders.631 They may then rely on this liquidity to attract marketable orders, to which they charge a high transaction fee in order to both offset the cost of the large rebates and to ensure a profitable transaction pricing model. Alternatively, inverted exchanges offer higher rebates to compete to attract marketable orders. Exchanges may also compete for order flow on other dimensions as well, by offering better execution quality, better technology, and innovations in order types and other trading mechanisms.632 In addition to competing with other U.S. equities exchanges, exchanges also compete for order flow with offexchange trading centers, including ATSs, internalizers, and others. One way exchanges compete with offexchange trading venues is through the use of rebates. For example, a number

633See e.g., FIA Letter, at 3–4; NYSE Letter I, at 6; Grasso Letter, at 4. 634See e.g., IEX Letter I, at 3; Credit Suisse Commentary, at 2. 635See NYSE Letter I, at 6. 636 Data on off-exchange market share are available from Cboe http://markets.cboe.com/us/ equities/market_share/ (last visited November 8, 2018). 637The estimates of ATSs that trade NMS stocks and ATS trade volume share was developed using weekly summaries of trade volume collected from ATSs pursuant to FINRA Rule 4552. See also Securities Exchange Act Release No. 76474 (November 18, 2015), 80 FR 80998, 81109 (December 28, 2015) (hereinafter ‘‘Regulation of NMS Stock Alternative Trading Systems’’). The estimates in this release were calculated in the same manner as in the cited release. See also OTC (ATS & Non-ATS) Transparency, FINRA: Reg. Filing & Reporting (2018), http://www.finra.org/Industry/ Compliance/MarketTransparency/ATS/. 638Total market share is collected from Cboe http://markets.cboe.com/us/equities/market_share/ (last visited November 8, 2018). ATS weekly market share is collected from FINRA, https:// otctransparency.finra.org (last visited November 8, 2018). 639See Angel, Harris, & Spatt, supra note 530. 640See, e.g., Cboe Letter I; NYSE Letter I. 641See NYSE Letter II at 8 for more concrete factors considered by liquidity providers. 642See Securities Exchange Act Release No. 63241 (November 3, 2010), 75 FR 69791, 69822 (November 15, 2010) (hereinafter ‘‘Risk Continued of commenters argued that one way exchanges compete with off-exchange trading venues is by using liquidity rebates to attract liquidity and narrow the displayed spread, which makes it more expensive for off-exchange trading venues to either match or improve upon the NBBO.633 Exchange transaction fees may also affect competition between exchanges and off-exchange trading venues. For example, commenters suggested that broker-dealers may opt to route order flow off-exchange in order to avoid higher transaction fees charged by exchanges.634 Off-exchange trading venues may also compete with exchanges to attract order flow by offering more flexibility in how they execute orders. One commenter noted that ‘‘market participants choose to send orders to off-exchange venues for reasons other than avoiding fees,’’ including ‘‘investors anonymity, the ability to trade in more granular tick sizes, the flexibility to segment the treatment of different types of clients, the ability to choose trading counterparties, and the ability to accommodate customer errors.’’ 635 Off-exchange trading makes up a substantial fraction of total volume, as approximately 34% of all transaction reports are routed using the NYSE and Nasdaq Trade Reporting Facilities as of August 2018.636 Of that off-exchange NMS share volume, approximately 14% was attributable to ATSs, of which 34 traded NMS securities as of August
2018.637 The remaining 21% of offexchange share volume is routed to
other off-exchange trading centers, such as internalizers.638 In aggregate, broker-dealers and other market participants have a large and varied set of options as to where they route orders, whether to exchanges or to off-exchange trading centers. Moreover, empirical evidence suggests that traditional exchanges, such as NYSE and Nasdaq, are losing market share to off-exchange trading centers and newer exchanges,639 which may provide different incentives to broker-dealers in order to attract this order flow, including transaction fees and rebates. We discuss the current levels of transaction-based fees in Section IV.B.2.e below. b. Market for Liquidity Provision Several commenters discussed the importance of liquidity providers to an exchange’s ability to compete in the market for trading services.640 Within the exchange framework, liquidity providers, such as market makers, other proprietary traders, and investors, compete to supply liquidity to liquidity demanders. They compete by posting displayed limit orders on exchanges, or by posting undisplayed limit orders on exchanges or ATSs. Liquidity providers profit by buying at a price lower than the price at which they sell and/or by collecting rebates that are greater than the fees they pay. Hence, an execution is a necessary means of profiting from liquidity provision, whether the liquidity provider seeks to profit from price changes or rebates. Liquidity providers, and traders more generally, seek to manage their trading profits by managing the tradeoff between the price they get in an execution, the certainty of execution, and any adverse selection resulting from the execution.641 When a liquidity supplier more aggressively prices their limit order, they increase the chance that their order will execute, but they trade this off against their order executing at a worse price and increased chance of their order being adversely selected if it does execute. To get an execution, the limit orders need to be at the top of a queue at a given price and venue and placed on a venue able to attract liquidity demanders. Displaying a limit order attracts liquidity demanders to the venue displaying the limit order, and thus improves the probability of execution, but could also increase the risk of being adversely selected, which reduces profits. For example, an algorithm that is skilled at identifying short-term price movements may be programmed to hit displayed limit buy orders at a price following a signal that the price is about to go down. In such a situation, the liquidity provider is unlikely to quickly sell at a price higher than the recent purchase, and therefore, these situations are costly for the liquidity provider. One way to attempt to reduce adverse selection costs is to not display the limit order. When the order is not displayed, the traders with the price signals may not see it and, as a result, would be less likely to pick it off. On the other hand, an undisplayed limit order also risks not
getting executed when an execution would be profitable. For example, an undisplayed limit buy order is less likely to execute than a displayed limit buy order just prior to an increase in the price because marketable sell orders that do not anticipate the price increase are likely to route to venues with competitively priced limit buy orders and would not be able to identify which venues have undisplayed limit buy orders. Rebates and fees can also affect where liquidity providers choose to supply liquidity. Maker-taker exchanges, which pay rebates to liquidity suppliers, provide them with extra revenue when trades are executed. This could encourage liquidity suppliers to post at more aggressive prices for some securities, subject to the fact that displayed quotes on stock exchanges must be priced in one-cent increments. However, competition among liquidity suppliers to earn rebates could lead to longer queues in an order book, which could decrease the chance that a liquidity supplier’s order executes unless they are at or near the front of the queue. In contrast, inverted exchanges, which charge liquidity suppliers a fee when they supply liquidity and offer a rebate to takers of liquidity, usually have shorter queue lengths and an economic incentive to take liquidity, which increases the chance that a liquidity supplier’s order executes.
c. Market for Broker-Dealer Services
The Commission considered the potential for the Pilot to affect competition among broker-dealers that route institutional and retail orders. These broker-dealers compete in a segment of the market for broker-dealer services. The market for broker-dealer services is highly competitive, with most business concentrated among a small set of large broker-dealers and thousands of small broker-dealers competing in niche or regional segments of the market.642 Large broker-dealers

Management Controls for Brokers or Dealers with Market Access’’). 643See id. Larger brokers, or those with more order flow, also benefit from the economies of scale that accompany the tiering structure typically provided by exchanges. Accordingly, the brokers with the most liquidity-providing orders may benefit disproportionately from rebates because they generally receive higher rebates within the various tiered pricing models of exchanges. 644Not all ETPs are pooled investment vehicles. For example, exchange traded notes (‘‘ETNs’’), which are a subset of ETPs, are unsecured, unsubordinated debt securities that trade in the secondary market on exchanges. 645 Investment companies can also earn revenue from other activities such as lending securities. 646Actively managed investment vehicles also rely on their historical performance when competing to attract investor funds. 647See, e.g. JPMorgan Letter, at 4; STANY Letter, at 4; Healthy Markets Letter II, at 8; Morgan Stanley Letter, at 3–4. 648The results are based on data collected from Bloomberg and Morningstar as of September 30, 2018 for US-domiciled ETPs. 649ETFs operate under exemptive orders that allow them to register as investment companies under the Investment Company Act. See 15 U.S.C. 80a–3(a)(1). Non-ETF ETPs are other ETPs that are not registered under the Investment Company Act. Some are pooled investment vehicles with shares that trade on a securities exchange, but they are not ‘‘investment companies’’ under Investment Company Act because they do not invest primarily in securities. Such ETPs may invest primarily in assets other than securities, such as futures, currencies, or physical commodities (e.g., precious metals). Others are not pooled investment vehicles. For example, ETNs are senior, unsecured, unsubordinated debt securities that are linked to the performance of a market index and trade on securities exchanges. See fn. 10 and accompanying text in the ETF Proposal, infra note 651, at 37333. ETMFs are exchange traded managed funds. ETMFs also operate under exemptive orders that allow them to register as investment companies under the Investment Company Act, but they have different disclosure requirements than ETFs. See, e.g. Eaton Vance Management, et al., Investment Company Act Rel. Nos. 31333 (Nov. 6, 2014) (notice) and 31361 (Dec. 2, 2014) (order). 650The Investment Company Act defines ‘‘redeemable security’’ as any security that allows the holder to receive his or her proportionate share of the issuer’s current net assets upon presentation to the issuer. See 15 U.S.C. 80a–2(a)(32). While closed-end fund shares are not redeemable, certain closed-end funds may elect to repurchase their shares at periodic intervals pursuant to rule 23c– 3 under the Investment Company Act (‘‘interval funds’’). Other closed-end funds may repurchase their shares in tender offers pursuant to rule 13e– 4 under the Exchange Act. 651The Commission’s exemptive orders typically contain a
representation by the applicant that an authorized participant will be either: (a) A broker or other participant in the continuous net settlement system of the National Securities Clearing Corporation, a clearing agency registered with the Commission and affiliated with the Depository Trust Company (‘‘DTC’’), or (b) a DTC participant, which has executed a participant agreement with the ETF’s distributor and transfer agent with respect to the creation and redemption of creation units. See, e.g., Emerging Global Advisors, LLC, et al., Investment Company Act Release Nos. 30382 (February 13, 2013), 78 FR 11909 (February 20, 2013) (notice) and 30423 (March 12, 2013) (order) and related application. In June 2018, the Commission proposed a new rule under the Investment Company Act of 1940 that would permit ETFs that satisfy certain conditions to operate without obtaining an exemptive order. In connection with the proposed exemptive rule, the Commission proposed to rescind certain exemptive orders that have been granted to ETFs and their sponsors. See Securities Exchange Act Release No. 10515 (June 28, 2018), 83 FR 37332 (July 31, 2018) (‘‘ETF Proposal’’). typically enjoy economies of scale over small broker-dealers and compete with each other to service the smaller brokerdealers, who are both their competitors and their customers.643 As of December 31, 2017, approximately 3,860 brokerdealers filed Form X–17a–5. These firms vary in size, with median assets of approximately $800,000, average assets of nearly $1 billion, and total assets across all broker-dealers of approximately $4 trillion. The twenty largest broker-dealers held approximately 72% of the assets of broker-dealers overall, with total assets of $2.89 trillion, indicating the high degree of concentration in the industry. Of the 3,860 broker-dealers that filed Form X–17a–5, 397 are members of U.S. equities exchanges. Broker-dealers that are members of equities exchanges had, on average, higher total assets than other broker-dealers, with median assets of $25.5 million, average assets of $9.2 billion, and total assets across all broker-dealers that are members of exchanges of $3.65 trillion. d. Market for Assets Under Management Many commenters expressed concern about the impact of the Pilot on the market for assets under management, particularly on exchange-traded products (‘‘ETPs’’). Asset management firms compete with each other in a segment of the market for assets under management. They offer different types of investment vehicles, such as mutual funds, close-end funds, and ETPs,644 which compete with each other to attract investor funds. Investor funds in an investment vehicle are pooled together and invested in financial assets, with investors sharing any profits or losses incurred by the investment vehicle according to each investor’s interest in the vehicle. Asset management firms generally earn revenue by charging fees based on the value of the assets they manage on behalf of
investors in their investment vehicles.645 Investment vehicles compete with other investment vehicles that follow similar investment strategies to attract investor funds. They often rely on differences in expense ratios, tracking error, and redemption and trading characteristics when competing to attract investor funds.646 One subset of investment vehicles are ETPs. ETPs differ from other investment vehicles in their trading and redemption characteristics. ETPs are investment vehicles that issue shares that can be bought or sold throughout the day on securities exchanges in the secondary market at a market-determined price. ETPs provide investors with a diverse set of investment options. While the first ETPs held portfolios of securities that replicated the component securities of broad-based domestic stock market indexes, some ETPs now track more specialized indexes, including international equity indexes, fixedincome indexes, or indexes focused on particular industry sectors such as telecommunications or healthcare. Some ETPs seek to track highly customized or bespoke indexes, while others seek to provide a level of leveraged or inverse exposure to an index over a fixed period of time. Investors also have the ability to invest in ETPs that do not track a particular index and are actively managed. A number of commenters noted that ETP issuers face strong competition within similar investment strategies and that small differences in fees and trading characteristics, such as spreads, daily volume, and intraday volatility, may be meaningful to market participants when deciding which ETPs to trade or invest in.647 As of September 2018,648 there were 2,003 ETPs categorized as exchange traded funds (‘‘ETFs’’), 223 ETPs categorized as non-ETF ETPs, and 18 ETPs categorized as ETMFs.649 As of this date, ETPs had total net assets of $3.74 trillion. The ten largest ETPs accounted for 28.0% of total ETP net assets and 27.8% of the average dollar trading volume on secondary markets. As the statistics above indicate, ETFs represent the majority of ETPs, they possess characteristics of both mutual funds, which issue redeemable securities, and closed-end funds, which generally issue shares that trade at market-determined prices on a national securities exchange and are not redeemable.650 Similar to mutual funds, ETFs continuously offer their shares for sale. Unlike mutual funds, however, ETFs do not sell or redeem individual shares. Instead, ‘‘authorized participants’’ that have contractual arrangements with the ETF (or its distributor) purchase and redeem ETF shares directly from the ETF in blocks called ‘‘creation units.’’ 651 An authorized participant that purchases a creation unit of ETF shares directly from the ETF deposits with the ETF a ‘‘basket’’ of securities and other assets identified by the ETF that day, and then receives the creation unit of

652An ETF may impose fees in connection with the purchase or redemption of creation units that are intended to defray operational processing and brokerage costs to prevent possible shareholder dilution (‘‘transaction fees’’). 653The basket might not reflect a pro rata slice of an ETF’s portfolio holdings. Subject to the terms of the applicable exemptive relief, an ETF may substitute other securities or cash in the basket for some (or all) of the ETF’s portfolio holdings. Restrictions related to flexibility in baskets have varied over time. See the ETF Proposal, supra note 651, at 37354–58. 654Non-ETF ETPs also offer creation and redemption processes. Some Non-ETF ETPs that are organized as pooled investment vehicles may offer creation and redemption processes similar to ETFs. Other Non-ETF ETPs may offer creations or redemptions on a less frequent basis. For example, some ETNs may only be redeemed weekly. 655An authorized participant may act as a principal for its own account when purchasing or redeeming creation units from the ETF. Authorized participants also may act as agent for others, such as market makers, proprietary trading firms, hedge funds or other institutional investors, and receive fees for processing creation units on their behalf. See Abner, D.J. The ETF Handbook: How to Value and Trade Exchange Traded Funds, 2nd ed., Wiley Finance (2016) (‘‘ETF Handbook’’). 656See, e.g., Antoniewicz, R. & Heinrichs, J. (2014). ‘‘Understanding Exchange-Traded Funds:
How ETFs Work.’’ ICI Research Perspective, Vol. 20(5) (available at: https://www.ici.org/pdf/per20- 05.pdf) (‘‘Antoniewicz’’). 657NSCC is the sole provider of clearing services for ETF primary market transactions. Whether a creation or redemption order is eligible to be processed through NSCC depends on the eligibility for NSCC processing of the securities in the ETF’s basket. See id. 658 It is possible for both the ETF’s NAV per share and its market price to deviate from the intrinsic value of the ETF’s underlying portfolio. In addition, there may be cases in which the ETF’s market price is closer to the intrinsic value of the ETF’s portfolio than its NAV per share. See, e.g., Madhavan, A. & Sobczyk, A. (2016) ‘‘Price Discovery and Liquidity of Exchange-Traded Funds.’’ Journal of Investment Management, Vol 14(2) (available at: https:// www.joim.com/price-dynamics-and-liquidity-ofexchange-traded-funds-2/). 659See the ETF Proposal, supra note 651, at 37384. 660ETFs also operate under several conditions designed to facilitate an efficient arbitrage mechanism. For example, ETFs are required to provide some degree of transparency regarding their portfolio holdings by disclosing their holdings prior to the commencement of trading each business day (i.e., portfolio transparency). 661This redemption would also cause the ETF’s assets under management to decline. 662As part of this arbitrage process, authorized participants are likely to hedge their intraday risk. For example, when ETF shares are trading at a discount to an estimated intraday NAV per share of the ETF, an authorized participant may short the securities composing the ETF’s redemption basket. After the authorized participant returns a creation unit of ETF shares to the ETF in exchange for the ETF’s baskets, the authorized participant can then use the basket assets to cover its short positions. 663Market participants also can engage in arbitrage activity without using the creation or redemption processes by buying/shorting shares in the ETF while simultaneously shorting/buying the ETF’s underlying assets. 664As discussed above, authorized participants can also hedge the intraday risk associated with the arbitrage process. See supra note 662. 665See, e.g., Virtu Letter, at 7; Cboe Letter I, at 17–18. 666See 17 CFR 240.19b–4(m)(1), which requires each SRO to post and maintain a current and complete version of its rules, including those related to transaction-based fees and rebates, on its website. 667As discussed supra Section IV.B.1.b.vi, fee information, such as that included in exchange fee schedules or Form 19b–4 fee filings, does not have standardization or formatting requirements. 668The transaction fee and rebate ranges in Table 2 are collected from recent fee schedules (as of July 31, 2018) available from each individual exchange’s website (listed in Table 1). Table 2 provides the date from which these fee schedules were reported. Continued ETF shares in return for those assets.652 The
basket is generally representative of the ETF’s portfolio 653 and, together with a cash balancing amount, equal in value to the aggregate NAV of the ETF shares in the creation unit.654 After purchasing a creation unit, the authorized participant may hold the individual ETF shares, or sell some or all of them in secondary market transactions. The redemption process is the reverse of the purchase process: The authorized participant redeems a creation unit of ETF shares for a basket of securities and other assets.655 While the Commission currently lacks data on authorized participants, a 2015 surveybased study of fifteen fund sponsors, which together offer two-thirds of all existing ETFs (covering 90% of all ETF assets), finds that the average ETF has 34 authorized participant agreements.656 The study further reports that creation and redemption transactions occurred only on between 10% to 20% of trading days and that only 10% of the daily activity in all ETF shares (by volume) are creations or redemptions.657 Investors can purchase individual ETF shares in the secondary market at prices that may deviate from the ETF’s NAV. As a result, ETF investors may trade shares at prices that do not necessarily reflect the intrinsic value of the underlying ETF assets.658 As discussed in the ETF Proposal,659 the combination of the creation and redemption process with secondary market trading in ETF shares provides arbitrage opportunities that are designed to help keep the market price of ETF shares at or close to the NAV per share of the ETF.660 For example, if ETF shares are trading on national securities exchanges at a ‘‘discount’’ (a price below the NAV per share of the ETF), an authorized participant can purchase ETF shares in secondary market transactions and, after accumulating enough shares to compose a creation unit, redeem them from the ETF in exchange for the more valuable securities in the ETF’s redemption basket.661 The authorized participant’s purchase of an ETF’s shares on the secondary market, combined with the sale of the ETF’s basket assets, may create upward pressure on the price of the ETF shares, downward pressure on the price of the basket assets, or both, bringing the market price of ETF shares and the value of the ETF’s portfolio holdings closer together.662 Alternatively, if ETF shares are trading at a ‘‘premium’’ (i.e., a price above the NAV per share of the ETF), the transactions in the arbitrage process are reversed and, when arbitrage is working effectively, keep the market price of the ETF’s shares close to its NAV.663 However, authorized participants, other market participants, and arbitrageurs acting in secondary markets may incur costs and be exposed to risk when engaging in arbitrage. The costs include bid-ask spreads and transaction fees associated with the arbitrage trades. In addition, during the time it takes arbitrageurs to execute these trades, they are exposed to the risk that the prices of the basket assets and the ETF
shares change. As a consequence, arbitrageurs may decide to wait for any mispricing between the market price of ETF shares and NAV per share to widen until the expected profit from arbitrage is large enough to compensate for any additional costs and risks associated with engaging in the transaction.664 A number of commenters noted that, in order to promote liquidity in thinlytraded ETPs, exchanges offer market makers who meet certain quoting requirements enhanced rebates when supplying liquidity in certain less actively traded ETPs.665 e. Transaction-Based Fees and Rebates Exchanges are required to disclose their current fee schedules, which include transaction-based fees and rebates, connectivity fees, membership fees, among others.666 When exchanges update their fees, they are required to file Form 19b–4 with the Commission; fee changes are permitted to take effect upon filing under Section 19(b)(3)(A) of the Exchange Act.667 Although these fee schedules and Form 19b–4 fee filings contain information about fees beyond transaction-based fees and rebates, in this baseline, the discussion is limited to only transaction-based fees and rebates and any changes thereto.
Table 2 reports the range of minimum
and maximum transaction fees and rebates, as well as the number of categories for each (in parentheses below the fee ranges), by exchange, as reported by each exchange on their recent fee schedules.668 On average,

The ranges in fees are the minimum and maximum fees and rebates reported by each exchange. 669This average does not include the IEX exchange as the fee structure is a flat one. See also, e.g., RBC Letter II (attaching a report titled ‘‘Complexity of Exchange Pricing and Corresponding Challenges to Transparency and Routing’’ in which they identify ‘‘1,023 separate pricing ‘paths’—i.e., separate fees or rebates—across these exchanges.’’). 670See, e.g., CFA Letter, at 6. 671The median number of revisions to fee and rebate schedules by exchanges is 41 over the fiveyear period. 672NYSE acquired NSX in January 2017, and the exchange is now known as NYSE National. As of May 2018, the exchange re-opened for trading and began submitting new fee schedules periodically. 673See Nasdaq Form 10–K filings (2017), available at https://www.sec.gov/Archives/edgar/ data/1120193/000112019318000003/ ndaq1231201710-k.htm. Transaction-based revenues for equity securities accounted for approximately 59% of total operating income net of rebates and 25% net of rebates and brokerage, clearing, and exchange fees. The Commission has revised the revenue number for Nasdaq for 2017 revenue per the correction provided in the Nasdaq Letter II. 674See Intercontinental Exchange Form 10–K filings (2017), available at https://www.sec.gov/ Archives/edgar/data/1571949/0001571949 18000003/ice2017123110k.htm. For the Intercontinental Exchange, net cash equity transaction-based revenues were approximately 8.2% of operating income for 2016. 675See Cboe Form 10–K filings (2017), available at https://www.sec.gov/Archives/edgar/data/ 1374310/000155837018000953/cboe20171231x10k.htm. Cboe’s acquisition of BATS Global Markets became effective on March 1, 2017. For the year ending December 31, 2017, the net transaction-based revenues were 41% of Cboe operating profits. U.S. exchanges have 18 access fee categories and 21 rebate categories associated with these fee schedules.669 For the maker-taker exchanges, access fees do not exceed the Rule 610(c) cap at $0.0030, but are as little as zero in some fee categories for some exchanges; taker-maker exchanges, because they are not restricted in the amount they can charge to non-marketable limit orders, have fees that range as high as $0.0033. Seven exchanges have some categories of rebates that exceed the maximum access fees charged by exchanges.670
Table 2 also provides the number of
fee revisions for the exchanges as reported in their Form 19b–4 fee filings to the Commission in the last five years (August 1, 2013–July 31, 2018). Exchanges, on average, have changed their fee schedules 34 times in the last five years,671 indicating that the average exchange revises its transaction-based fee schedules about seven times per year (approximately every 7.4 weeks).
TABLE 2—SUMMARY OF TRANSACTION-BASED FEE SCHEDULES FOR U.S. NATIONAL EQUITIES EXCHANGES AS OF JULY
2018
Exchange Fee model
Number of revisions
(5 years)
Date of fee
schedule
Fees
(# of categories)
Rebates
(# of categories)
Cboe BZX ................................................ Maker-Taker ........... 54 8/16/2018 $0.0000–$0.0033 ....
(29) .........................
($0.0010)–($0.0032)
(18)
Cboe BYX ................................................ Taker-Maker ........... 51 8/9/2018 $0.0000–$0.0033 ....
(40) .........................
($0.0005)–($0.0022)
(12)
Cboe EDGA ............................................. Taker-Maker ........... 41 8/1/2018 $0.0000–$0.0032 ....
(48) .........................
($0.0004)–($0.0027)
(14)
Cboe EDGX ............................................. Maker-Taker ........... 53 8/16/2018 $0.0000–$0.0032 ....
(37) .........................
$0.000–($0.0032)
(20)
BX ............................................................ Taker-Maker ........... 29 7/20/2018 $0.0005–$0.0030 ....
(13) .........................
$0.0000–($0.0021)
(12)
Phlx (PSX) ............................................... Maker-Taker ........... 24 5/21/2018 $0.0028–$0.0030 ....
(3) ...........................
$0.00–($0.0030)
(8)
Nasdaq ..................................................... Maker-Taker ........... 54 7/25/2018 $0.0000–$0.0030 ....
(4) ...........................
$0.0000–($0.00325)
(36)
NYSE Arca ............................................... Maker-Taker ........... 51 8/1/2018 $0.0005–$0.0035 ....
(68) .........................
($0.0002)–($0.0035)
(65)
NYSE American ....................................... Flat .......................... 9 7/26/2018 $0.0002–$0.0030 ....
(12) .........................
$0.0000–($0.0045)
(4)
NYSE ....................................................... Maker-Taker ........... 42 8/10/2018 $0.0003–$0.0030 ....
(20) .........................
$0.0000–($0.0045)
(41)
NYSE National 672 .................................... Taker-Maker ........... 11 7/26/2018 $0.0003–$0.0025 ....
(16) .........................
($0.0002)–($0.0020)
(2)
CHX .......................................................... Maker-Taker ........... 8 4/26/2018 $0.0007–$0.0040 ....
(5) ...........................
($0.0009)–($0.0020)
(2)
IEX ........................................................... Flat .......................... 10 8/1/2018 $0.0009 ................... $0.0009 For several of the exchange families, information about revenues and costs attributed to transaction-based fees and rebates is available in aggregate from Form 10–K filings. Using the statements of income from Form 10–K filings for 2017 capturing the net (of rebates) transactions-based fee revenues, the Nasdaq exchanges (Nasdaq, BX, and PSX) earned $253 million.673 Based on the same measure the NYSE-affiliated exchanges (NYSE, NYSE Arca, NYSE American, and NYSE National) earned $196 million in transaction-based fees net of rebates,674 while the BATS Global Markets (now, Cboe BZX, Cboe BYX, Cboe EDGA, and Cboe EGDX), for the year ended December 31, 2017, earned $153 million in transaction-based fees net of rebates.675 Neither CHX (which became a NYSE-affiliated exchange in
2018) nor IEX or their affiliates are
publicly traded, meaning that these exchanges do not file an annual Form 10–K with the Commission. As a result, public information regarding the revenues or profits associated with transaction-based fees does not exist for these exchanges. Information on the net transactionsbased revenues for each individual exchange, as opposed to the amounts reported for exchange groups in Form 10–K filings, is not currently publicly available, making it difficult to analyze the fees and rebates for an individual exchange. To estimate the net transactions-based revenues for each individual exchange, Table 3 reports the maximum and median net transactionbased fees based on each exchange’s

676The share volume is obtained from Cboe, available at http://markets.cboe.com/us/equities/ market_share/ (last visited September 18, 2018). To compute the maximum profit attainable, staff took the difference between the highest possible transaction fee and the lowest possible rebate and multiplied it by the monthly share volume. For a midpoint profit, the median of the transaction fees less the median of the rebates is computed and multiplied it by share volume. In order to make the results comparable to those reported above from Form 10–K filings, the monthly profits are annualized by multiplying each monthly profit amount by 12. 677Monthly share volume obtained from Cboe for July 25, 2017 through August 24, 2017, Cboe, U.S. Equities Market Volume Summary, available at http://markets.cboe.com/us/equities/market_share/ (last visited September 18, 2018). 678 In contrast, one commenter opined that the primary benefit of the Pilot would be ‘‘the ‘better fills’ that institutional investors will get after the pilot is introduced.’’ Nasdaq Letter III, at 9. This commenter asserted that the Commission had not done a proper cost benefit analysis because its analysis of benefits did not account for the ‘‘opportunity costs’’ inherent in order routing decisions, or other factors which would impact institutional orders. See id. Consequently, the commenter asserted that the Commission overstated the benefit of the Pilot. See id. However, the Commission does not agree that ‘‘better fills’’ will be a certain result of the Pilot. Furthermore, the Commission disagrees that it has not adequately analyzed the costs of the Pilot. As noted above, the Commission has quantified the likely economic effects of the Pilot where possible; however, the Commission is unable to quantify all of the economic effects because it lacks the information necessary to provide reasonable estimates. The Commission agrees with the commenter that quantifying benefits using existing data is difficult, thus underscoring the need for a Pilot. A more detailed analysis of the Pilot’s impact on trading costs can be found in Section IV.C.2.b. 679See Proposing Release, supra note 2, at
Section V.
most recently reported fee schedule and the share volume of each exchange for July 25, 2018 through August 24,
2018.676 As evidenced by the significant
differences between the sum of net of rebate revenues for entities reporting to the same exchange group obtained from
Table 3 and the total net of rebate
revenues for each exchange family reported on the Form 10–K or 10–Q filings, this approach does not yield reliable results, highlighting the limitations on the data currently available to researchers.
TABLE 3—ESTIMATES OF ANNUALIZED PER-EXCHANGE NET TRANSACTION-BASED FEE REVENUES FROM TRANSACTIONBASED FEES AND MONTHLY EXCHANGE SHARE VOLUME
[For July 25, 2018–August 24, 2018] [In millions] Exchanges Share volume (millions) 677 Annualized midpoint difference Per share profit (median) Annualized maximum difference Per share profit (maximum) Cboe BZX ............................................................................ 9,014 ($486.75) ($0.00450) $248.78 $0.0023 Cboe BYX ............................................................................ 7,136 25.69 0.00030 239.76 0.0028 Cboe EDGA ......................................................................... 1,853 1.11 0.00005 62.26 0.0028 Cboe EDGX ......................................................................... 8,165 (4.90) (0.00005) 205.77 0.0021 BX ........................................................................................ 4,389 36.87 0.00070 158.02 0.0030 Phlx (PSX) ........................................................................... 1,035 16.15 0.00130 37.27 0.0030 Nasdaq ................................................................................. 23,087 (34.63) (0.00013) 831.14 0.0030 NYSE Arca ........................................................................... 13,024 23.44 0.00015 515.75 0.0033 NYSE American ................................................................... 473 (3.69) (0.00065) 17.03 0.0030 NYSE ................................................................................... 17,823 (128.33) (0.00060) 641.63 0.0030 NYSE National ..................................................................... 920 3.31 0.00030 25.40 0.0023 CHX ...................................................................................... 830 8.96 0.00090 30.87 0.0031 IEX ....................................................................................... 3,757 N/A 0.00000 N/A 0.0000
C. Analysis of Benefits and Costs of
Transaction Fee Pilot

Benefits of Transaction Fee Pilot
The Commission expects that the benefits of the Pilot will fall into two categories: (1) More informed policy decisions, including more information about the economic impact of transaction-based fees and rebates, and (2) other benefits that may accrue to market participants for the duration of the Pilot. In this section we discuss each of the categories of benefits as well as potential limitations to the applicability of information to be drawn from the Pilot. a. Benefits of More Informed Policy Decisions The Commission expects that the primary benefit of the Pilot will be to inform the Commission and public of the economic impact of exchange transaction-based fees and rebates.678 As a result, the Commission will have data to better inform its regulatory consideration of exchange transactionbased fee-and-rebate pricing models and fee changes, and the potential effects of changes to its regulatory approach concerning the same. In general, more informed regulatory decisions are more likely to result in regulatory approaches that better balance costs and benefits relative to regulatory decisions based on less precise information. In other words, many of the economic benefits derive from subsequent decisions that the Commission can neither predict nor commit to at this time. Indeed, the Commission cannot predict at this time whether the results of the Pilot will suggest any particular policy direction and recognizes that the results could suggest that existing exchange transaction-based fee caps and related rebates may be more beneficial to investors than the policy alternatives examined in the Pilot.
i. Expected Analysis From the Pilot
The Proposing Release discussed the theoretical impact of exchange transaction-based fees and rebates on several potential effects such as conflicts of interest, fragmentation, complexity, liquidity, and off-exchange competition and explained that certain components of the Proposed Pilot would facilitate the study of these effects.679 As noted above, the Commission believes that little empirical evidence currently exists regarding these effects. More specifically, the Pilot will provide information on the direct effects of exchange transaction fee and rebate levels on execution quality and market quality and will facilitate studies of the impact of fees and rebate levels on market participant behavior and competition, including potential conflicts of interest. Sections IV.C.2 and IV.D. discuss many potential economic effects for which this economic analysis

680See e.g., Cboe Letter I, at 12; NYSE Letter I, at 9; Nasdaq Letter I, at 2, 5 (suggesting that the Pilot may impact competitive dynamics between exchanges and ATSs). 681See, e.g., Nasdaq Letter III at 1. 682As discussed in the baseline, the number of exchanges has increased since 2005, and market share has become less concentrated over the same time period. The majority of the U.S. equities exchanges belong to three exchange groups. The Commission believes that any analyses of the effects of transaction-based fees on order routing decisions can appropriately control for exchange groups. 683See, e.g., ICI Letter II at 2. 684See, e.g., Fidelity Letter, at 9. 685Equilibrium refers to conditions of a system in which all competing influences are balanced. For instance, with respect to the Test Group 2, this could be the level of transaction fees charged by exchanges from which no exchange has any incentive to increase or decrease that fee outside of a constrained competitive margin. This will be the equilibrium transaction fee. See also, discussion in
Section II.C.7.e. An important potential benefit of
the Pilot could result if the no rebate Test Group were able to demonstrate a set of conditions wherein regulatory fee caps might not be necessitated in an environment in which natural competitive forces could effectively cap access fees. This would occur if in the no-rebate Test Group the equilibrium fee charged during the Pilot was lower than the fee cap—implying that the fee cap was not binding in this situation. 686 If Linked Pricing were not prohibited, market participants could potentially circumvent the prohibition on rebates through Linked Pricing mechanisms. Therefore, including prohibitions on rebates or Linked Pricing could provide information to the Commission and the public about potential conflicts of interest associated with rebates or substitutes for rebates, such as Linked Pricing, as well as the equilibrium fee that emerges in the absence of rebates or Linked Pricing. 687See, e.g., Fidelity Letter, at 9; Citadel Letter, at 5. In addition to removing rebates or Linked Pricing in Test Group 2, the Commission could also temporarily suspend limitations on access fee caps imposed by Rule 610(c). Implementing multiple changes within a single test group, however, could prevent researchers and others from clearly determining the effect of the prohibition of rebates on order routing decisions of broker-dealers from the effect resulting from the removal of access fee caps if Rule 610(c) restricted access fees during the Pilot. 688See, e.g., ICI Letter II at 3 (noting that the Pilot could facilitate the study of how access fees and rebates affect liquidity, including quoted spreads). 689See. Academic studies suggest that the majority of retail orders are executed off-exchange at prices based on the NBBO, thereby providing retail investors with better prices in the presence of rebates. If, however, large rebates provide incentives for broker-dealers to route retail orders to these exchanges instead of to off-exchange venues, retail customers may not be fully aware of the total cost associated with their orders. See, e.g., Angel, Harris, & Spatt, supra note 530. 690See, e.g., Nasdaq Letter I at 9. 691See supra Section II.C.5 (discussing statistical power) and infra note 695. is unable to draw unambiguous conclusions. For example, many commenters disagreed on how reducing exchange fees and rebates affects the competitive landscape between exchanges and off-exchange venues in the market for trading services,680 and the analysis here recognizes that many competitive forces can drive order flow in either direction. The Pilot will provide insight into the impact of transaction fees and rebates on this competitive landscape and can perhaps even shed light into the mechanism behind any observed changes. Further, one commenter argued that this economic analysis ‘‘does not accurately account for the actual level of orders impacted by conflicted broker routing.’’ 681 The Commission believes that it cannot establish the actual level of orders impacted
by potentially conflicted broker routing with current data and has designed the Pilot in part to gather more data on the extent to which rebates impact order routing decisions, as explained in the section that follows. The Commission also notes that the Pilot seeks to study the effects of exchange pricing models on market quality and execution quality, which could affect all orders. The Pilot will facilitate the study of order flow among different venues, which could provide insights into whether changes in exchange transaction-based fees and rebates affect, for example, the level of fragmentation. Existing literature suggests that transaction-based pricing has contributed to an increase in the number of venues competing for order flow over time.682 By offering rebates or Linked Pricing, start-up maker-taker and taker-maker exchanges have been able to attract order flow from exchanges such as NYSE and Nasdaq, thereby reducing liquidity externalities, or concentration of order flow to a preferred venue, and leading to increased fragmentation of the market for trading services.683 By altering the access fee and rebate structures for exchanges, researchers may be able to identify whether these changes lead to more (or less) concentration of liquidity and how they affect competition for order flow among exchanges, which could lead to less (or more) market fragmentation.684 Test Group 2 will provide insight into the natural equilibrium level of access fees, within the current regulatory structure, in the absence of rebates and Linked Pricing.685 As discussed above, prohibiting exchanges from offering Linked Pricing in Test Group 2 is intended to complement and reinforce the prohibition on rebates.686 Although Rule 610(c) caps the maximum access fee for exchanges at $0.0030, in the absence of rebates and Linked Pricing, competition among exchanges could drive the average access fee to an amount substantially below $0.0030.687 As noted in Section IV.A.2, exchanges have a reduced competitive incentive to reduce fees because doing so would require reducing the rebates that attract order flow to the exchange. Test Group 2 will allow competition among exchanges, in the absence of pressure to offer high rebates or Linked Pricing, to determine the level of access fees, which the Commission and others can observe during the Pilot. Like the other examinations the Pilot can facilitate, the results of an analysis of the equilibrium access fees are not currently predictable with much certainty. The Pilot will facilitate studies of the impact of exchange transaction-based fees and rebates on liquidity by studying metrics such as the quoted spreads.688 The width of the quoted spread is considered to be an indicator of a stock’s liquidity, with narrower spreads generally indicating more liquid securities. The analysis below is Section IV.C.2.b.iv identifies several reasons that reducing fees and rebates could increase or decrease quoted spreads. The Pilot could
provide information on whether exchange fees and rebates affect the liquidity of securities, as measured by the quoted spreads, across different test groups.689 The Commission disagrees with commenters who said that the Pilot was inappropriate because of a one-size-fitsall approach.690 In selecting the number of securities for each Test Group, the Commission staff divided NMS securities into three common stock strata and three ETP strata by liquidity to determine how many stocks each stratum requires to achieve statistical power.691 The staff also separately examined ETPs to determine how many ETPs would be required to achieve statistical power. Having statistical power within each Test Group, and within each Test Group by liquidity strata, helps to ensure that researchers will be able to use Pilot data to inform the Commission regarding the issue of whether different securities should have the same regulatory treatment.
ii. How the Pilot Facilitates Study
The Pilot will simultaneously create different fee environments, each of which restricts transaction-based fees differently to allow for the comparison of securities that are simultaneously in different regulatory regimes. The study of these comparisons will inform the Commission about economic distortions that may arise as a result of transactionbased fees. Because of the size and length of the Pilot, the Commission believes that the different fee environments over representative subsamples of NMS securities, even though implemented temporarily, will produce effects on market participant behavior that are identical or similar to

692As designed, the Pilot will exclude NMS securities that have prices below $2.00 per share as of the date of pilot selection and NMS securities with average daily volume of less than 30,000 shares. As detailed above, the data will also be produced for a six-month pre-Pilot Period and a sixmonth post-Pilot Period. 693See, e.g., Brandes Letter, at 2 (supporting applying the Pilot to the ‘‘widest range of stocks possible’’); Spatt Letter, at 1–2. 694See, e.g., Battalio Equity Market Study, supra note 5307. 695The supplemental analysis made several improvements over the analysis used to identify the proposed test group sizes in an attempt to refine the analysis to respond to commenters’ desire for smaller test groups while preserving statistical power. First, the supplemental analysis used more refined methodology that more directly controlled for time series and cross-sectional dependencies. Second, the supplemental analysis considered three quoted spread strata instead of two market capitalization stratum. The market capitalization strata was originally necessary to control for any overlap with the Tick Size Pilot, but quoted spread strata more directly align with the potential economic significance of fees and rebates relative to anticipated transaction costs and the Tick Size Pilot has ended. Third, the supplemental analysis eliminated stocks that trade below 30,000 shares per day. See also supra Section II.C.6. See supra note 175 (citing commenters that favored smaller test groups). 696See Proposing Release, supra note 2, at 13019. 697See, e.g., Battalio Equity Market Study, supra note 5307; Harris, supra note 5307; RBC Letter I, at 5–6. those that would arise under a similar permanent change. As explained below, three distinct features of the Pilot’s design will facilitate analyses of the relationship, if any, between fees and potential economic distortions. Specifically, the Pilot is designed to provide (1) representative results; (2) more direct access to data that is currently unavailable or requires lengthy and labor-intensive effort to compile and process; and (3) sufficient information to determine causality. The following sections discuss in detail how each of these aspects of the Pilot could facilitate studies of the issues described above. (1) Representative Results In the context of the Pilot, representative results mean that the impact of the Pilot’s terms on a Test Group during the Pilot Period is likely to be consistent with the impact of the results on the Test Group if the Pilot’s terms were permanent (as opposed to temporary). Representative results are desirable for researchers and policy makers because it ensures that inferences drawn from the results of analysis of Pilot data are likely to be similar to those that would emerge if the terms were permanent. As discussed in the baseline, current analyses are limited by some combination of the following: Data from a single brokerdealer, a small sample of securities, a single
exchange, or a short sample period. By contrast, the Commission believes that the Pilot, as designed, will produce more representative results. Specifically, as discussed in detail below, the Pilot will cover a large stratified sample of NMS stocks (including ETPs), both maker-taker and taker-maker exchanges, and transaction fee caps as well as a prohibition on rebates and Linked Pricing, and will have a two-year duration with an automatic sunset at the end of the first year unless the Commission determines, at its discretion, that the Pilot shall continue for up to one additional year.692 The Commission believes that the Pilot will produce representative results, presenting a significant improvement on existing studies, because the Pilot applies to a large stratified sample of NMS stocks (including ETPs) with prices of at least $2.00 per share at the date of the Pilot Securities selection, with average daily volume of 30,000 shares or more, and with no restrictions on market capitalization.693 In particular, the Commission recognizes that any possible conflicts of interest related to transaction-based fees could vary across securities such that the results of a pilot focused only on large capitalization stocks may not provide information relevant to small capitalization stocks or ETPs.694 Including a broad sample of NMS stocks allows the results to inform policy choices across subsets of these securities. The stratification of the stocks selected for each Test Group is designed to ensure that each Test Group and the control group have a similar composition within a given stratum, facilitating a comparison of Test Groups and the Control Group, which further supports the representativeness of results. If, for instance, the Test Groups and Control Group had a different composition within strata, researchers outside the Commission might not be able to distinguish whether differences across Test Groups and the Control Group stem from different fee environments or different sample composition, rendering the results less representative. In addition, the Commission believes that the sample sizes in the Test Groups are sufficient to provide the statistical power necessary to identify differences across the samples, even within strata. The Commission notes that while the adopted Pilot will be able to provide representativeness within strata, changes since the Proposal affect the representativeness of the Test Groups as a whole. In particular, in response to commenters who called for fewer stocks to be included in the Pilot, Commission staff conducted a supplemental analysis of Test Group sizes needed to achieve statistical power.695 In contrast to its analysis in the Proposal,696 the Commission analyzed the sample sizes in each stratum rather than using the lowest power stratum to determine the ratio of test group stocks to control stocks. As a result, the ratio of test group stocks to control group stocks is lower for some strata. In other words, the
Commission was able to reduce the number of securities in test groups by weighting the composition of the test groups relative to the control group more heavily toward securities in certain strata in which more data would be needed to achieve statistical power. While analyses of the Pilot that do not consider the strata may fail to provide representative results, the addition of the stratum identifier to the Exchange Lists will allow researchers in and outside the Commission to consider the strata in their analyses. The Commission believes that the inclusion of a broad sample of NMS stocks, including small and midcapitalization stocks, ensures representative results from the Pilot. Although previous studies, as discussed above, suggest that any possible conflicts of interest are likely to be the greatest for small-capitalization securities,697 the Commission believes that it is important to the design of the Pilot to include these small and midcapitalization stocks (including ETPs). In particular, including these securities in the Pilot will allow the results of the Pilot to inform policy choices across any subset of these securities. Representativeness of results of the Pilot will also be promoted by the choice of the Pilot Security selection date. Rule 610T(b) and (c) contemplate that the Commission will select and announce the Pilot Securities prior to the Pilot start date. As noted in the Proposal, the Commission anticipates that it will assign and designate by notice each Pilot Security to one Test Group or the Control Group approximately one month prior to the start of the Pilot. By assigning securities close to the start of the Pilot, each Test Group and the Control Group are likely to be more comparable during the Pilot. Because stratification criteria (e.g., market capitalization and liquidity) vary naturally over time, the closer the assignments occurs to the Pilot start date, the more comparable the Test Groups will be during the Pilot. Selection of securities close to the start of the Pilot also will be more likely to

698See, e.g., Larry Harris Letter, at 9. 699See, e.g., Joint Asset Managers Letter, at 2. Other commenters agreed that the Pilot duration will be sufficient but for other reasons. See, e.g., Fidelity Letter at 9 and CFA Letter at 6. 700See, e.g., Citi Letter, at 5 and Larry Harris Letter, at 11. Also, see infra Section IV.C.2.b.ii for a discussion of the costs broker-dealers could incur during the Pilot. 701See, e.g., Babelfish Letter at 3 and Healthy Markets Letter at 19. 702See Babelfish Letter at 3. 703See Healthy Markets Letter I at 19. 704Specifically, the supplemental analysis compared the distributional characteristics of all US listed stocks (including Canadian interlisted stocks) to the distributional characteristics of the subset of US listed stocks that excludes Canadian interlisted stocks to determine whether distributional characteristics of the subset differs statistically significantly from the distributional characteristics of all US listed stocks. The analysis finds that the distribution of the subset that excludes Canadian interlisted securities is statistically similar to the distribution of all US listed stocks. include the intended universe of securities, by avoiding securities that exit between the adoption of the Rule and the start of the Pilot, while also capturing new securities that enter the market during this period. Further, to the extent that market participants would change their behavior in anticipation of the Pilot, setting the selection period close to the Pilot effective date could reduce the effect of such behavior on pre-Pilot data. The results of the Pilot will be further representative because the Pilot applies to all U.S. equities exchanges regardless of fee structure. Broker-dealers potentially face transaction-fee related conflicts of interest regardless of whether those fees are on maker-taker exchanges or taker-maker exchanges, and rebates on either the make or take side can both impact market quality and execution quality. Further, a pilot that addresses only a single fee structure would not produce results relevant for policy choices that also would apply to another fee structure. Applying the Pilot to all exchanges also improves upon the existing analysis of the limited fee experiment conducted by Nasdaq,698 which only covered a single exchange, as explained in Section IV.B.1.a.ii. While the results from that study are suggestive that broker-dealers routed customer orders to other exchanges that did not change their transaction-based fees and rebates, reasons other than potential conflicts of interest could have impacted the changes in order routing decisions. The Commission believes that the Pilot will achieve representativeness by requiring transaction-fee changes for all U.S. equities exchanges, which will allow researchers in and outside the Commission to identify how these revisions affect order routing decisions across exchanges. As discussed above, excluding non-exchange trading centers does
not forfeit the representativeness of the results to be obtained from the Pilot, as including them would expand the Pilot to dissimilarly situated trading centers whose fee models and regulatory treatment are incomparable to exchanges. Further, the Pilot will require that changes to fees or rebates are applied at the security level, which means that for any given security, the limitation on access fees or rebates is ubiquitous across all exchanges. In addition, the Pilot achieves representativeness by imposing a fee cap and a prohibition on rebates and Linked Pricing. The existing literature suggests that the potential distortive effects arising from access fees could induce behavior that would be different from the distortions arising from rebates or Linked Pricing. Therefore, the inclusion of caps on both fees and rebates or Linked Pricing allows for a more comprehensive analysis of any possible conflicts of interest than could be achieved by focusing solely on fees or rebates. The Commission further believes that the duration of the Pilot will produce sufficiently representative results. If broker-dealers incorporate transaction fees and rebates into their order routing decisions, a two-year duration for the Pilot, with an automatic sunset at the end of the first year, unless the Commission publishes a notice determining that the Pilot shall continue for up to a second year, would likely make it economically worthwhile for broker-dealers to change their routing behavior during the Pilot by making it costly to avoid the Pilot.699 Specifically, as discussed below, the Commission recognizes that broker-dealers will incur costs to incorporate new fee schedules that are consistent with the Pilot’s requirements into their order routing decisions.700 Broker-dealers could ignore the Pilot to avoid these costs. If enough broker-dealers ignore the Pilot, the Pilot might not produce results that provide the Commission a sense of the likely impact of permanent changes to fee caps or rebates. However, to the extent that broker-dealers incorporate transaction-based fees and rebates into their order routing decisions, ignoring the Pilot will also impose costs on broker-dealers, and these costs increase with the duration of the Pilot. The Commission believes that the Pilot duration, even with a one-year sunset, is long enough to produce representative results because, as discussed below in
Section IV.C.2.b.ii, broker-dealers that
incorporate transaction-based fees and rebates into their routing decisions will find it economically worthwhile to adapt their behavior in response to the Pilot. Further, the provision to suspend the automatic sunset facilitates representative results because it provides the Commission with flexibility as the data from the Pilot develops. For example, the Commission could suspend the sunset if, for example, it believed that additional time would help ensure that market developments are fully reflected in the data with sufficient statistical power for analysis, recognizing that such market developments are uncertain. Therefore, the sunset provides flexibility to the Commission to observe developments during the Pilot to determine whether to allow the sunset to occur. Some commenters disagreed that one year will be sufficient to achieve a representative sample.701 One commenter said that ‘‘robust data . . . should take two years’’ and that ‘‘technological changes . . . to routing and algorithmic logic for some firms are a hurdle that could require significant time to implement.’’ 702 Another commenter noted the ‘‘complexities of the pilot and the opportunities for significant market evolutions.’’ 703 The Commission notes that it will consider these and other concerns, as noted above, in deciding whether or not to suspend the automatic sunset. The Commission believes that the Pilot will produce representative results despite the Pilot’s treatment of stocks cross-listed on Canadian exchanges during the Pilot and the exclusion of stocks with average daily volume of less than 30,000 shares. A supplemental staff analysis found that the exclusion of the interlisted Canadian stocks from the selection of securities for test group inclusion would not materially impact the representativeness of the remaining sample.704 Second, the exclusion of securities with average trading volume of less than 30,000 shares per day should not materially affect the representativeness of the results because the trading in these stocks generates less than $100 per day in fees or rebates. Additionally, low trading volume stocks tend to have wider spreads rendering a rebate of $.0030 a significantly smaller incentive relative to the size of the spread than it would be for higher volume tighter spread securities. Because of these two factors, the Commission believes fees and rebates are economically much less meaningful inducements to provide liquidity for these stocks. Because of the diminished economic significance of rebates in

705See, e.g., Nasdaq Letter I, at 2, 5 and RBC Letter I, at 4. 706See Section IV.C.1.a. For commenter statements supporting the usefulness of the data to be obtained from the Pilot See, e.g., Clark-Joseph Letter, at 1, AJO Letter, at 1, CII Letter, at 3, NYSTRS Letter, at 1, Barnard Letter, at 1; ICI Letter I, at 1–2; MFS Letter, at 1; Nuveen Letter, at 2; Better Markets Letter, at 2; RBC Letter I, at 2; Invesco Letter, at 2; CFA Letter, at 1; State Street Letter, at 2; Wellington Letter, at 1; Joint Pension Plan Letter, at 2; Oppenheimer Letter, at 2; Angel Letter I, at 1. Some commenters suggest that the data will not be useful because it excludes data from ATSs and doesn’t account for other forms of remuneration that broker dealers receive which may also impact order routing decisions. See NYSE Letter 1 at 1,8–10; ProAssurance Letter, at 2; Cboe Letter I, at 15; Nasdaq Letter I, at 2. 707The aggregation and availability of the data gathered by the Pilot is one of the primary benefits of the Pilot and provides much of the value of the data collected. See, e.g., NYSE Letter II, at 13; Cboe Letter I, at 3. 708See, e.g., Battalio Equity Market Study, supra note 530. 709See NYSE Letter II, at 13 suggesting that the proposing release did not provide an illustration for how the data could be used to study the Commission’s objectives. these extremely low volume stocks, the Commission believes that there is a lower risk of applying a suboptimal transaction-based fee regulatory regime in these stocks. In other words, because rebates are economically less meaningful for these securities the benefits of the Pilot in informing policy decisions regarding transaction-based fees in these securities are likely low. In addition, the supplemental staff analysis found that excluding these securities increased the potential statistical power of the Pilot. Some commenters suggested that the exclusion of ATSs from the data gathering hinders the representativeness of the data obtained from the Pilot.705 The Commission understands that ATSs often negotiate bespoke agreements with individual subscribers for a bundle of services for which rebates may or may not play a significant role. Even if the Commission obtained detailed information on all of these agreements, it may not be possible to identify the fees or rebates they pay for order flow from the fees for the other bundled services the ATS offers the subscriber in a manner sufficient for inclusion in the Exchange Transaction Fee Data. Also, as discussed in section IV.D.2.a it is uncertain whether the Pilot will lead to exchanges to be in an improved or diminished competitive position with ATSs. Further, without including ATSs in the Pilot, ample public data exists to assess the market share of ATSs relative to exchange market share to observe and measure off-exchange order flow changes. (2) Expansion of Readily Available Data The Commission also expects the Pilot to provide data that would otherwise require lengthy
and laborintensive collection. Having a representative source of data is critical for the production of research and analyses about the impact of transaction-based fees on potential distortions. If more data becomes available, that data will assist the Commission in analyzing potential conflicts of interest. The Commission believes that the data produced by the Pilot will improve upon existing data,706 as is discussed in more detail below. The ready availability of the Exchange Transaction Fee Summaries will facilitate the study of distortions and the equilibrium level of fees and rebates by reducing the cumbersome nature of collecting fee data. Further, the Pilot will make information on order routing decisions available to the Commission on a more granular level than current readily available data and will improve the feasibility of Commission staff analysis of order routing data during the Pilot.707 The Pilot will enable Commission staff to gain improved access to order routing data and will provide access to fee data in a simplified and standardized form, which will improve the quality of the analyses produced as a result of the Pilot. Although certain order routing data and exchange fee schedules are publicly available through a combination of Rule 606 disclosures and exchange websites, respectively, the Pilot will resolve a number of limitations associated with using currently available data to study the effect of transaction-based fees on potential conflicts of interest and their impact on market quality and execution quality. The order routing data that Commission staff will obtain as a result of the Pilot will provide superior information to that readily available today. Data will be available for a representative sample of NMS stocks, across all broker-dealers, and exchanges, at the daily frequency, which will provide sufficient data for analyses, while providing more statistical power than the Rule 606(a) public reports can provide. Relative to the data that some studies have acquired from brokerdealers and exchanges,708 the order routing data will also allow Commission staff to observe a time series of order routing data across broker-dealers and exchanges. Further, more granular order routing data (e.g., daily order routing statistics that separate principal and agency trading as well as auction, post only, and other orders) than that available publicly will facilitate more targeted analysis. Together, these characteristics of the data will facilitate Commission staff research on issues such as potential conflicts of interest, which will improve the quality of the information available to the Commission for policy decisions. The following discussion illustrates how the data obtained from the Pilot could be used to study the Commission’s objectives.709 The key components in the order routing data that facilitate studies of the impact of transaction-based fees and rebates on order routing and execution quality are daily volume
information at the exchange, stock, and broker-dealer level, the separation of liquidity taking and liquidity making orders, the Order Capacity, the Order Designation, the time to execution for liquidity-providing orders, and the ability to estimate fill rates. The routing volume allows Commission researchers to measure how much volume each broker-dealer sends to each exchange each day in individual securities, which can be combined with the Pilot Securities Exchange List and the Exchange Transaction Fee Summary to observe patterns in routing and correlate those patterns with fees and Test or Control Group membership. The exchange level is required to match the order routing data with the fee data; the broker level is required to allow for different routing strategies across broker-dealers; and the daily level in the data facilitates statistical power. The separation of liquidity taking and liquidity making orders allows researchers to match the order routing volume to the potential fee or rebates in the Exchange Transaction Fee Summary. Order Capacity allows Commission researchers to compare order routing and execution quality statistics for Agency Orders to Principal Orders, which are less subject to conflicts of interest concerns than Agency Orders and, thus, provides an added means of obtaining causal identification. Order Designation allows researchers to exclude auction orders and to separately analyze Post Only orders because these orders types are subject to different fee structures (auction orders do not get rebates) or exist for the purpose of capturing rebates (Post Only). Excluding or separately analyzing these orders types provides for cleaner tests that are better able to measure the impacts consistent with the objectives of the Pilot. Finally, the time to execution and ability to estimate fill rates (using orders received, executed, canceled or rerouted)

710The Commission recognizes that many tradebased execution quality statistics are readily estimated from publicly available data. See Section IV.E.5.g infra for a discussion of an alternative to require order-based execution quality statistics during the Pilot. 711The standardized fee data, as would be required by the proposed Pilot, is discussed supra
Section III.E.2. 712Rule 610(T) requires each exchange to
publicly post on its website downloadable files containing the Exchange Transaction Fee Summary and update them on a monthly basis. Similarly, each exchange will be required to publicly post on its website downloadable files containing daily aggregated and anonymized order routing statistics, updated monthly. Each exchange will also be required to provide daily on its website downloadable files containing the List of Pilot Securities and the Pilot Securities Change List. 713As an open standard, XML is widely available to the public at no cost. As an open standard, XML is maintained by an industry consensus-based organization, rather than the Commission, and undergoes constant review. As updates to XML or industry practice develop, the Commission’s XML schema may also have to be updated to reflect the updates in technology. In those cases, the supported version of the XML schema will be published on the Commission’s website and the outdated version of the schema will be removed in order to maintain data quality and consistency with the XML standard. The Commission’s XML schema will also incorporate certain validations to help ensure data quality. 714As discussed in the baseline, analysis of causality can be accomplished through either exogenous shocks or econometric methods, such as instrumental variable analysis. 715Other econometric techniques, such as instrumental variables methodology, are used when an exogenous shock (or other controlled experiment) cannot be established. provides Commission researchers with execution quality information not readily available for liquidity providing or liquidity taking orders.710 An additional requirement of the Pilot is that the exchanges will be required to provide a standardized dataset of fees, the Exchange Transaction Fee Summary, to the public. In particular, this information will allow researchers in and outside the Commission to create proxies for which exchanges are likely to be more or less expensive and which offer the highest rebates. For instance, within Test Group 1, the maximum allowable access fee is $0.0010; however, each exchange may have different base and top-tier fees. Thus, only knowing that a security is in Test Group 1 will be incomplete information about the impact of transaction-based fees and rebates. Moreover, the Exchange Transaction Fee Summary will provide researchers in and outside the Commission with historical (realized) average and median per share fees and rebates to enable an ex post analysis of how actual fees affected past order routing decisions, which is not available from any data source today. Exchanges will construct Exchange Transaction Fee Summaries according to an XML schema to be published on the Commission’s website, and exchanges will update this information monthly.711 These data will be standardized and consistently formatted, which will ease the use of these data for researchers in and outside the Commission, as each exchange will have to report the Base,
Top Tier, average and median fees, as detailed above in Section III.E. Each month, exchanges will be required to report realized average and median per share fees, as well as any ‘‘spot’’ revisions to fees associated with Form 19b–4 fee filings to the Commission. These fee data will be publicly posted on each exchange’s website.712 The Exchange Transaction Fee Summary released during the Pilot will:
(1) Ease aggregation across exchanges, which affords researchers in and outside the Commission an opportunity to obtain representative results; (2) replicate across studies, which will provide validation of findings; and (3) reduce burdens associated with fee data collection, which could encourage more research on the impact of fees and rebates on routing behavior. Thus, the Commission believes that a standardized reporting of summary data on fees by the exchanges will facilitate analysis of the effect of transactionbased fees. The rule will require that the Exchange Transaction Fee Summary be structured using an XML schema to be published on the Commission’s website.713 Data that are structured in a standard format can result in lower costs to analysts and higher quality data. An additional key benefit of structured data is increased usability. If, for instance, the Exchange Transaction Fee Summary were not standardized across the exchanges, researchers would have to manually rekey the data, a timeconsuming process which has the potential to introduce a variety of errors, such as inadvertently keying in the wrong data or interpreting the filings inconsistently, thereby reducing comparability. With the data in the reports structured in XML, researchers in and outside the Commission could immediately download the information directly into databases and use various software packages for viewing, manipulation, aggregation, comparison, and analysis. This will enhance their ability to conduct large-scale analysis and immediate comparison of the fee structures of exchanges. The Commission believes that requiring these reports to be made available in an XML format will provide flexibility to researchers in and outside the Commission and will facilitate statistical and comparative analyses across exchanges, test groups, and date ranges. (3) Causality In addition to providing representative results, the Commission expects the Pilot to achieve the benefits identified above because it will, among other things, provide insight into the degree to which exchange transactionbased fees and rebates cause economic distortions that either harm or benefit investors. Such causal information is especially useful when considering policy choices aimed at reducing any possible harmful distortions. As detailed in the baseline, exogenous shocks are a means by which researchers may analyze a causal relationship between changes to transaction-based fees and rebates and changes to order routing decisions of broker-dealers.714 This Pilot facilitates the analysis of causality through an exogenous shock that simultaneously creates several distinct fee environments, each of which restricts transaction-based fees or rebates differently, enabling synchronized comparisons to the current environment. The Commission believes that the Pilot is able to facilitate the examination of causality because the Pilot will produce a single exogenous shock that differentially impacts either
fees or rebates on both maker-taker and takermaker exchanges. Exogenous shocks, such as those in the Pilot provide researchers with data to analyze the direction of causality.715 For example, a researcher seeking to study the impact of the rebates on transaction costs could estimate a difference-in-differences test that compares transaction costs during the Pilot to the transaction costs before the Pilot and then compare the changes in Test Group securities to the changes in Control Group securities. It also will allow investors who receive 606(b)(3) data from their broker-dealers to directly test with their own 606(b)(3) data whether, in the absence of rebates in the most actively traded stocks, they are better able to compete for queue priority and thereby capture the quoted spread when posting liquidity. More generally, the Pilot will allow researchers, including Commission staff and others, to run difference-in-difference tests on many measures of execution quality and market quality based on publicly available data to examine the causal impact of transaction-based fees and rebates on execution and market quality. As discussed above, the Pilot will produce a single exogenous shock that

716See Healthy Markets Letter I, at 19. 717See supra. Section II.D.3. 718See, e.g., Cboe 15–16; NYSE at 9–10; Nasdaq I at 7; RBC Letter I, at 4; ProAssurance at 2. 719For instance, a pilot could be designed where the information obtained from the Pilot would only be valuable if certain market conditions, such as high market volatility or a recessionary period occurred. If, however, markets experience low volatility or are in an expansionary period, the Pilot may either not be sufficiently long enough to capture the events that it requires to be useful or would have to be extended to ensure that those market conditions could occur. 720For example, one study provided evidence suggesting that trading behavior may not have completely adjusted to the Regulation SHO pilot. See Ekkehart Boehmer, Charles Jones, & Xiaoyun Zhang, Unshackling Short Sellers: The Repeal of the Uptick Rule, Colum. U. (2008), https:// www0.gsb.columbia.edu/mygsb/faculty/research/ pubfiles/3231/UptickRepealDec11.pdf. Despite this effect, the study found evidence consistent with the evidence gathered from the Regulation SHO pilot. See Securities Exchange Act Release No. 50103 (July 28, 2004), 69 FR 48008 (Aug. 6, 2004) (hereinafter ‘‘Regulation SHO’’). 721 If broker-dealers have smart order routing systems that use algorithms that maximize rebate capture, as suggested in the Battalio Equity Market Study, supra note 530, then for at least some subset of securities, broker-dealers would not be able to pursue rebates from those exchanges, so it would be suboptimal for broker-dealers to not reconsider their order routing choices. If broker-dealers, however, already have order routing decisions that are optimal from a customer’s perspective (e.g., based on execution quality) and are not driven by potential conflicts of interest (e.g., maximizing rebates), then for at least some broker-dealers, their order routing decision process may be unchanged. It is also possible that for broker-dealers with algorithms that dynamically route based upon realtime market metrics, including liquidity metrics, expected fill rates, and current queue length, routing logic may not change, however, routing choices may dynamically adjust based upon changes in those variables that result from altered fee schedules that broker-dealers may implement in conjunction with the Pilot. differentially affects multiple Test Groups at the same time. The simultaneity of the exogenous shock across Test Groups facilitates examinations of causality, particularly in the presence of any confounding effects. For instance, if some marketwide event were to result in deviations in order routing behavior during the Pilot, the event would likely affect stocks in each Test Group as well as the Control Group. The simultaneity allows researchers in and outside the Commission to control for the impact of the market-wide event, because the impact would likely affect the Test Groups and the Control Group similarly. For example, in the
difference-indifferences test of transaction costs mentioned above, any market-wide effect would result in changes to fill rates in both the Control Group and Test Group 2. Therefore, the comparison of the changes in Test Group 2 to the changes in the Control Group subtracts the market-wide effect from the total effect, thus isolating the effect of the Pilot. In addition, to facilitate causal analysis of data during the Pilot Period, the Commission believes that it is important to collect sufficient data during a pre-Pilot Period.716 The prePilot data can then be compared with the data that will be produced during the Pilot Period, which will permit analysis of any changes to order routing behavior, execution quality, and market quality between the two for the Pilot Securities in each of the Test Groups. To make this comparison informative, the length of the pre-Pilot Period needs to be long enough to obtain sufficient statistical power to permit analysis of the stocks and ETP Pilot Securities. In turn, sufficient statistical power in tests that compare the pre-Pilot data to the Pilot data would allow all researchers to more easily use the information obtained from the Pilot to inform future regulatory consideration of exchange transaction fees and rebates and their impact on the markets. The Commission believes that at least six months of prePilot data may be required to obtain the necessary statistical power to permit analysis of the Pilot Securities during the Pilot, particularly ETPs.717 The Commission further believes that the combination of the representative sample, data from the Pilot, and the exogenous shock will facilitate analysis by Commission staff (or institutions who receive 606(b)(3) reports from their broker-dealers) of the degree to which transaction-based fee- and rebatemotivated order routing harms order execution quality. In particular, with the exogenous shock, the Order Routing Data, the Exchange Transaction Fee Summaries, and publicly available data, researchers at the Commission can identify both the degree to which transaction fees and rebates impact order routing and, the impact of transaction fee- and rebate-motivated order routing impacts execution quality. Several commenters seemed to state that the Pilot would produce flawed causal results because the Pilot does not include all forms of remuneration.718 While the Commission acknowledges that other forms of remuneration may impact routing decisions, the results will still be informative. Even if some order flow migrates between exchanges and off-exchange venues, Commission staff should still be able to identify the impact of exchange fees and rebates on exchange routing. The Pilot Securities Exchange List and the Pilot Securities Change List further enhance the ability for researchers both inside and outside of the Commission to analyze the effects of transaction-based fees on order routing decisions. By requiring daily updates to the Pilot Securities Change
List, the Pilot will provide broker-dealers with the information they need to track the exact securities in each Test Group in real-time and when securities exit the Pilot. This information will be crucial for broker-dealers that choose to adjust their routing behavior during the Pilot. If broker-dealers are unable to track which securities are in which Test Groups, the Pilot results could provide misleading causal information.
iii. Potential Limitations on the Benefits
The Commission recognizes that pilots are unpredictable and as such considered whether possible limitations associated with pilots generally, as well as certain issues presented by the design of this Pilot in particular, would limit the benefits of the Pilot. This section discusses, in greater detail below, issues associated with pilots in general and the potential concerns with resultant research and analyses. Pilots may face limitations related to the unpredictable nature of market conditions and confounding events. Even if a pilot lasted several years, not all of the market conditions of interest could be experienced. Depending on the requirements of pilots, such limitations might reduce the usefulness of the information obtained.719 The Commission believes, however, that the value of the information obtained from the Pilot is not dependent upon having variation in market conditions over time, and that the duration of the Pilot will provide sufficient information to inform policy decisions. In addition, pilots also face the limitation that market participants, knowing that a pilot is underway, may not act as they would in a permanent regime.720 In the context of this pilot, broker-dealers could choose to retain their current order-routing decisions for the duration of the Pilot, which could be costly to such broker-dealers.721 Brokerdealers, when deciding whether to adjust any order routing behavior that currently depends on fees and rebates, would likely trade off the costs of retaining strategies that are no longer profitable because of the restrictions imposed by the Pilot against the costs of adjusting the algorithms for their smart order routing systems. Alternatively, broker-dealers could substantially change their business model in order to

722 It could be costly for broker-dealers to completely alter their business models because they may not find it worthwhile to do so for a temporary pilot. 723See, e.g., FIA Letter, at 3; Citadel Letter, at 4. 724The Commission does not believe that the Hawthorne effect will cause ‘‘good’’ behavior in the baseline because broker-dealers would need to implement system changes similar to those described in Section IV.B.2.c prior to the pre-Pilot. 725As noted above, the Commission encourages market participants to disclose what sources of data they used for their analyses and describe the methodology they used, and to make those reports publicly and freely available. 726See James Angel Letter I, at 2. 727See supra Section IV.A.1 (Market Failure at the Broker Dealer Level) and Section IV.A.2 (Market Failure at the Exchange Level). 728See, e.g., Nasdaq I, at 1. 729See infra Section IV.C.2.b.iv avoid the Pilot.722 In addition, the Commission recognizes that the anticipated analysis of order routing data from the Pilot could cause brokerdealers to improve execution quality. This could reflect the ‘‘Hawthorne effect,’’ which refers to the idea that people will often improve their behavior if they believe that they are observed. These outcomes could lead to results that would not represent the effects of a permanent rule change. If that were to occur, a few commenters suggested that this could lead the potential benefits of the Pilot to not justify the costs or risks that the pilot imposes.723 The Commission believes that the Pilot is designed to obtain empirical information about how fees and rebates affect order routing decisions because the size and length of the Pilot render it unlikely that broker-dealers that currently focus their routing on rebates would maintain existing order routing decisions or alter their business models to avoid the Pilot as suggested by some commenters. In particular, the Commission believes that the Pilot duration is likely to make it economically worthwhile for brokerdealers to adjust their order routing behavior. The costs of ‘‘waiting out’’ the Pilot increase with the duration of the pilot, whereas the costs of adjusting the algorithms of the smart order routers, discussed below in Section IV.C.2.b.ii do not. In addition, the potential compromise of the data due to the Hawthorne effect is limited by at least two factors. First, this is not the Commission’s first pilot study. Market participants are relatively accustomed to the Commission collecting data for analysis. Second, the analysis of pre-Pilot data will allow for a baseline observation of unaffected broker-dealer order routing activity. If broker-dealers do not act on conflicts of interest during the baseline period, the Hawthorne effect is irrelevant unless it causes that good baseline behavior.724 If, on the other hand, broker-dealers do act on conflicts of interest during the baseline and the Hawthorne effect results in good behavior during the Pilot, the
Pilot should facilitate the measurement of the conflicts. As a result, the Commission believes that the Pilot will produce useful data despite the possible influence of the Hawthorne effect. The Commission recognizes that not all objectives of the Pilot would be straightforward to study. For example, the changes in fees or rebates imposed by the Pilot may change transaction costs in a way that results in changes to order routing decisions by brokerdealers, even absent potential conflicts of interest. Studying how order routing changes during the Pilot, without jointly studying why it changes, would not be sufficient to understand any possible conflicts of interest. Researchers can carefully study the data to distinguish the proportion of changes in order routing decisions resulting from execution quality considerations from those resulting from potential conflicts of interest. Nonetheless, this complication could reduce the number and/or quality of studies of the Pilot. Another limitation on the benefits from the Pilot is that the Pilot will not require that the order routing data be released to the public. As a result, fewer independent analyses of the Pilot’s order routing datasets are likely to be performed, compared to the analysis that might have been obtained if the data were publicly released. However, the Commission believes that sufficient analysis will be produced to yield credible and reliable results without public dissemination of the order routing data. In addition, institutions, including broker dealers, asset managers, and transaction cost analysis (TCA) providers, may produce their own analyses using proprietary data and information. To the extent that interested parties prepare their own analyses, they may submit them to tradingandmarkets@sec.gov with the words ‘‘Transaction Fee Pilot Analysis’’ in the subject line, and the Commission will post those reports on its public website.725 Additionally, only NMS stocks with prices of at least $2 prior to the start of the Pilot are eligible for inclusion in the Pilot. One commenter suggested that NMS stocks with prices between $1 and $2 also be included in the Pilot, as the commenter believed that the impact of fees and rebates are likely to be greatest for these securities.726 The Commission agrees with the commenter who stated that the initial Test and Control Groups in the Pilot would be more representative if they contained securities with prices below $2. However, excluding securities with prices below $2 helps to keep the sample of stocks more stable across the Pilot. This occurs because if a stock’s price falls below $1 it is subject to different regulations, such as a different tick size, and thus would be excluded from the Pilot. By excluding stocks below $2 the Pilot mitigates the risk that the representativeness of the sample may diminish over time as Pilot stocks are removed due to their stock prices falling below $1. The Commission believes that the data obtained from the
Pilot will be sufficient to obtain data on the effects to changes in fees and rebates on small, low-priced securities (those with prices close to $2, or any Pilot security that drops below $2 per share, but exceeds $1 per share, after the start of the Pilot). b. Other Benefits of the Transaction Fee Pilot Other benefits may emerge that could affect markets and market participants for the duration of the Pilot, such as potentially reduced conflicts of interest for some Test Groups, lower all-in costs of trading, or improved market quality. The Commission believes that many of the benefits discussed below will be temporary in nature and affect markets and market participants only for the duration of the Pilot. Because the Commission lacks information on the extent to which the impact of exchange fee-and-rebate pricing models affect investors,727 the Commission is unable to quantify many of the temporary benefits of the Pilot discussed below. Some commenters stated their belief that the Pilot would not help investors and issuers.728 As discussed in Sections IV.C.2.b and IV.D.1 the Commission acknowledges that the Pilot could harm execution quality and/or market quality, but the impacts of the Pilot are uncertain. The Pilot could also improve execution quality and/or market quality for the reasons explained in those same sections. For example, as discussed in detail below,729 the Commission is uncertain about whether, or among which securities, the Pilot will result in increases or decreases in quoted spreads and investor transaction costs. A decrease in quoted spreads and/or investor transaction costs during the Pilot in some or all stocks in test groups would benefit investors. Likewise, the Commission is uncertain about how the Pilot will affect price efficiency—the

730See infra Section IV.D.1.
731 17 CFR 242.612 (Rule 612 of Regulation NMS) prohibits traders from submitting sub-penny quotations on securities trading at prices over $1.00. The purpose of the sub-penny quotation prohibition was two-fold: (1) To prevent high frequency traders from front-running standing non-marketable limit orders and (2) to reduce the complexity of trading systems. See NMS Adopting Release, supra note 10, at 37550–57. 732See, e.g., Angel, Harris, & Spatt, supra note 530; Harris, supra note 530. One study noted that as a result of the Tick Size Pilot test group with the trade-at provision, taker-maker markets have seen a significant increase in market share, in part due to this quotation issue. See Carole Comerton-Forde, Vincent Gregoire, & Zhuo Zhong, Inverted Fee Venues and Market Quality 1 (August 10, 2018) (unpublished manuscript) (forthcoming J. Fin. Econ.) https://papers.ssrn.com/sol3/ papers.cfm?abstract_id=2939012&download=yes. 733See infra Section IV.C.2.a i. 734See infra Section IV.C.2.a ii. 735See infra Section IV.C.2.a iii. 736See infra Section IV.C.2.a iv. 737The primary listing exchanges are NYSE, Nasdaq, NYSE American, NYSE ARCA, BATS and IEX. 738See NYSE Letter I, at 15. 739See id. 740The Commission notes that the Tick Size Pilot required the exchanges and FINRA to also select the Pilot securities whereas the Transaction Fee Pilot does not. Therefore, the Transaction Fee Pilot could result in lower costs than the Tick Size Pilot. Nonetheless, the Commission believes it is reasonable to rely on this commenter’s estimate because this commenter has expertise on these costs likely to result in incorporating relatively precise information into the cost estimates. Pilot could plausibly improve or degrade price efficiency in certain test group stocks.730 Any improvements would benefit issuers and investors. The Commission believes that another temporary benefit of the rule will be that the Pilot could prevent some traders from indirectly quoting in subpennies.731 Rebates have the practical effect of reducing the minimum tick size by the size of the rebate, and in effect allow trading centers to offer quotations superior to the existing quote. Several studies suggested that the use of exchange fees and rebates to effectively undercut quotations by sub-pennies is particularly severe in taker-maker markets.732 The Pilot would, in some test groups, reduce or eliminate rebates, which could stem this indirect reduction of tick sizes, and could provide the Commission and the public with information currently unavailable about this issue.
2. Costs of the Pilot
This section describes the compliance costs associated with the Pilot, followed by the additional costs, some of which are temporary, that could affect issuers, investors, broker-dealers, exchanges, and other market participants resulting from the Pilot. a. Exchange Compliance Costs of the Pilot The Pilot will impose costs on exchanges to comply with the Pilot’s requirements to collect, calculate, and publicly post data certain required by the Pilot on their websites, transmit the order routing datasets to the Commission, as well as to implement fee changes, if required in order to comply with the Pilot’s restrictions.
Table 4 provides a summary of the costs
discussed in this section.
TABLE 4—SUMMARY OF COMPLIANCE COSTS FOR EXCHANGES
Pilot securities exchange list 733 Exchange transaction fee summary 734 Order routing data 735 Fee filings 736 Total Exchange type Listing All Listing Non-listing Per exchange Implementation .......................................................................... 15,400 26,100 24,000 96,800 162,000 147,000 Periodic:
—2-yr Pilot ......................................................................... 83,500 55,000 103,800 148,400 391,000 307,000 —1-yr Pilot ......................................................................... 50,100 36,600 69,200 74,200 230,000 180,000 Total (implementation + periodic):
—2-yr Pilot ......................................................................... 98,900 81,000 127,800 245,200 553,000 454,000 —1-yr Pilot ......................................................................... 65,500 62,700 93,200 171,000 392,000 327,000 Total across exchanges Implementation .......................................................................... 92,000 339,000 311,000 1,258,000 2,001,000 Periodic:
—2-yr Pilot ......................................................................... 501,000 714,000 1,350,000 1,929,000 4,494,000 —1-yr Pilot ......................................................................... 301,000 476,000 900,000 964,000 2,641,000 Total (implementation + periodic):
—2-yr Pilot ......................................................................... 593,000 1,054,000 1,661,000 3,187,000 6,495,000 —1-yr Pilot ......................................................................... 393,000 815,000 1,211,000 2,223,000 4,642,000
i. Updating the Pilot Securities
Exchange List and Pilot Securities ChangeList During the Pilot, the primary listing exchanges will maintain and make public prior to the start of each trading day the Pilot Securities Exchange List of the securities included in each test or control group on its website. Further, each primary listing exchange will publicly post on its website the updated Pilot Securities Change List prior to the start of each trading day, which will list, separately, changes to applicable Pilot Securities. Additional details of what will be included in each list are provided in Section II.E.1. Upon the initial publication of the List of Pilot Securities by notice by the Commission, the primary listing exchanges 737 will need to determine which of those securities are listed on their market, and then compile a list of those securities and publicly post on their websites that list as a downloadable file in pipe-delimited ASCII format. The Commission initially estimated that the costs associated with the initial compilation of the Pilot Securities Exchange List would cost $2,060 per exchange based on an estimated burden of 8 hours. However, one commenter stated that it ‘‘anticipates it could take as many as 44 hours’’ to compile the initial Pilot Securities Exchange List.738 The commenter stated that its estimates of the costs associated with the Pilot are based on its ‘‘prior experience implementing the Tick Size Pilot.’’ 739 In light of this comment, the Commission is increasing its estimate.740 Accordingly, the Commission estimates that each primary listing exchange would incur, on average, a one-time

741This estimate is based on the following:
[(Compliance Manager (22 hours) × $298) + (Programmer Analyst (22 hours) × $232)] = $11,660 ≈11,700 per exchange, or $11,660 × 6 primary listing exchanges = $69,960 ≈ 70,000 in aggregate. The burden hours are obtained from supra Section III.D.1. The Commission estimates the wage rate associated with these burden hours based on salary information for the securities industry compiled by the Securities Industry and Financial Markets Association (SIFMA). The estimated wage figure for attorneys, for example, is based on published rates for attorneys, modified to account for a 1,800- hour work-year and multiplied by 5.35 to account for bonuses, firm size, employee benefits, and overhead, yielding an effective hourly rate for 2013 of $380 for attorneys. See Securities Industry and Financial Markets Association [SIFMA], Management & Professional Earnings in the Securities Industry—2013 (October 7, 2013), available at https://www.sifma.org/resources/ research/management-and-professional-earningsin-the-securities-industry-2013/. These estimates are adjusted for inflation based on Bureau of Labor Statistics data on CPI–U between January 2013 (230.280) and January 2017 (242.839). Therefore, the 2017 inflation-adjusted effective hourly wage rates for attorneys are estimated at $401 ($380 × 242.839/230.280). The Commission discusses other costs of compliance with the rule below. 742The Commission notes that the primary listing exchanges maintained public web pages containing similar lists with respect to the recently concluded Tick Size Pilot. The systems to produce lists for the Tick Size pilot should be adaptable to meet the requirements of the Transaction Fee Pilot. 743This estimate is based on the following:
[(Attorney (4 hours) × $401) + (Compliance Manager (4 hours) × $298) + (Programmer Analyst (4 hours) × $232)] = $3,724 per exchange, or $3,724 × 6 exchanges = $22,344 ≈ 22,300 in aggregate. The burden hours are obtained from supra Section III.D.1. 744 If the Pilot were to automatically sunset at the end of the first year, the total number of days that the exchanges would need to provide the Pilot Securities Exchange List and the Pilot Securities Change Lists would be up to 630 business days (504 business days for the two-year Pilot horizon (252 business days per year × 2 years), and up to 126 business days for the six-month post-Pilot Period). The cost estimate for providing these lists for the entire period is based on the following:
[(Compliance Manager (0.25 hour × 630 trading days) × $298) + (Programmer Analyst (0.25 hour × 630 trading days) × $232)] = $83,475 ≈ 83,500, or $83,475 × 6 exchanges = $500,850 ≈ 501,000, in aggregate. The burden hours are obtained from supra Section III.D.1. One commenter provided an estimate of 300.5 burden hours for providing these lists, but the Commission continues to believe its own higher burden estimates are reasonable. See
Section III.D.6, supra.
745 If the Pilot were to automatically sunset at the end of the first year, the total number of days that the exchanges would need to provide the Pilot Securities Exchange List and the Pilot Securities Change Lists would be up to 378 business days (252 business days for the one-year Pilot horizon, and 126 business days for the six-month post-Pilot Period). The cost estimate for providing these Lists for the entire period is based on the following:
[(Compliance Manager (0.25 hour × 378 trading days) × $298) + (Programmer Analyst (0.25 hour × 378 trading days) × $232)] = $50,085 ≈ $50,100, or $50,085 × 6 exchanges = $300,510 ≈ $301,000, in aggregate. The burden hours are obtained from supra Section III.D.1. 746See, e.g., NYSE Letter I, at 15. 747The estimate is based on the following:
(Compliance Manager (2 hours) × $298) + (Senior Business Analyst (2 hours) × $265) = 1,126 ≈ $1,130, or $1,126 × 13 equities exchanges = $14,638 ≈ 14,600 in aggregate. The burden hours are obtained from Section III.D.2, supra. 748This estimate is based on the following:
(Compliance Manager (1 hours) × $298) + (Programmer Analyst (1 hours) × $232) = $530 per exchange, or $530 × 13 U.S. equities exchanges = $6,890 ≈ 7,000 in aggregate. The burden hours are obtained from supra Section III.D.2. burden of approximately 44 burden hours per primary listing exchange to compile and publicly post their initial Pilot Securities Exchange List. Consequently, the Commission now estimates a cost of approximately $11,700 per listing exchange to compile the initial list of securities.741 The Commission understands that each primary listing exchange has existing systems to monitor and maintain the Pilot Securities Exchange List and the Pilot Securities Change List as a result of certain corporate actions.742 While these systems can be used to collect the data required to be made public for the Pilot Securities Exchange List and the Pilot Securities Change List, these systems would have to be adapted to conform to the requirements of the Pilot. The Commission estimates that it would cost each primary listing exchange approximately $3,720 to develop appropriate systems for the Pilot, or about $22,300 in aggregate across the six U.S. primary listing exchanges.743 Once these systems are established, the Commission estimates that it would cost each listing exchange approximately $83,500 for the entire duration of the Pilot, or approximately $501,000 across the six primarily listing exchanges,744 to publicly post on each exchange’s website the Pilot Securities Exchange List and Pilot Securities Change List prior to the start of each trading day in pipe-delimited ASCII format. If the Commission determined that the Pilot shall automatically sunset at the end of the first year, the Commission estimates that the costs to each exchange would be $50,100 for a one-year Pilot duration and the six-month post-Pilot Period, or approximately $301,000 across the six primarily listing exchanges.745 In sum, the Commission estimates a total cost for each listing exchange of approximately $98,900, or $593,000 in aggregate across exchanges, to comply with the requirement to update and post on its website at the beginning of each trading day the list of its listed securities in each of the Test Groups. This includes an estimated $15,400 in one-time implementation costs and $83,500 in ongoing costs. This estimate is based on one provided by a commenter who, based on their experience with the Tick Size Pilot, estimated that it would take up to 44 hours to compile. 746 Accordingly, the Commission continues to believe its burden estimates are reasonable.
iii. Producing the Exchange Transaction
Fee Summary in XML Format In addition to the Pilot Securities Exchange List provided by the primarily listing exchanges, all U.S. equities exchanges would also need to publicly post on their websites the Exchange Transaction Fee Summary, which are downloadable files containing the initial set of fees at the outset of the Transaction Fee Pilot as well as monthly updates to include both changes to fees and rebates reported in Form 19b–4 fee filings and realized average and median per share fees and rebates, as discussed in Section II.E.2. The Exchange Transaction Fee Summary would need to be updated in response to any changes to its fee schedule following the beginning of each calendar month from the pre-Pilot Period through the postPilot Period. The exchanges would be required to provide information on any transaction-based fee and rebate changes, according to Rule 610T(e), that they make during the Pilot, including the effective dates of fee revisions. The rule also requires that each exchange calculate numerous statistics relating to their fees as discussed in more detail in
Section II.E.2.
A requirement at the outset of the Pilot is that exchanges would need to report their base and top-tier fees and rebates, which the Commission estimates would cost each exchange $1,130, or about $14,700, in aggregate across the 13 U.S. equities exchanges.747 The reported base and top-tier fees and rebates would be mandatory elements of the Exchange Transaction Fee Summary. Concurrent with the submission of the Form 19b–4 fee filings to the Commission at the outset of the Transaction Fee Pilot, the exchanges also would be required to publicly post on their websites downloadable files containing the initial Exchange Transaction Fee Summary, using an XML schema to be published on the Commission’s website. The Commission estimates that it will cost exchanges $530 each to post this summary dataset to their websites.748 The rule would also require that exchanges compute the monthly average and median realized per share fees and rebates, as detailed in Section II.E.2. These data will provide the Commission and the public with aggregated data on the actual per share levels of fees and rebates assessed in the prior month, which the Commission believes is critical for estimating the effects of fees and rebates on order routing decisions. The Commission believes that the costs

749This estimate is based on the following, which reflects the Commission’s experience with and burden estimates for SRO systems changes:
[(Attorney (20 hours) × $401) + (Compliance Manager (20 hours) × $298) + (Programmer Analyst (20 hours) × $232) + (Senior Business Analyst (20 hours) × $265] = 23,920 ≈ $24,000 per exchange, or $23,920 × 13 exchanges = $310,960 ≈ 311,000 in aggregate. The burden hours are obtained from supra Section III. D.2. 750This estimate is based on the following, which reflects the Commission’s experience with and burden estimates for SRO systems changes:
[(Attorney (10 hours) × $401) + (Compliance Manager (10 hours) × $298) + (Programmer Analyst (10 hours) × $232) + (Senior Business Analyst (10 hours) × $265] = $11,960 ≈ $12,000 per exchange, or $11,960 × 13 exchanges = $155,480 ≈ $155,000 in aggregate. The burden hours are obtained from supra Section III.D.2. 751This estimate of updates to the Exchange Transaction Fee Summary is the aggregation of updates from the pre-Pilot Period (6), the two-year pilot period if the Commission determines that an extension of up to an additional year was needed (24), and the post-pilot period (6), for a total number of 36 updates. 752This estimate is based on the following:
[(Compliance Manager (1 hours) × $298) + (Programmer Analyst (1 hours) × $232)] = $530 per exchange, or $530 × 36 fee changes per exchange = $19,080 ≈ $19,000. The 36 fee changes for the exchange encompass six updates during the sixmonth pre-Pilot Period, 24 updates during the twoyear Pilot Period, assuming that the Commission determines that the additional year is required, and six updates during the six-month post-Pilot Period. In aggregate, updates to the Exchange Transaction Fee Summary are estimated to cost $19,080 × 13 U.S. equities exchanges = $248,040 ≈ $247,000. The burden hours are obtained from supra Section III. D.2. 753This estimate of updates to the Exchange Transaction Fee Summary is the aggregation of updates from the pre-Pilot Period (6), the one-year pilot period with an automatic sunset at the end of the first year (12), and the post-pilot period (6), for a total number of 24 updates. 754This estimate is based on the following:
[(Compliance Manager (1 hours) × $298) + (Programmer Analyst (1 hours) × $232)] = $530 per exchange, or $530 × 24 fee changes per exchange = $12,720 ≈ $12,700. The 24 fee changes for the exchange encompass six updates during the sixmonth pre-Pilot Period, 12 updates during the oneyear Pilot Period, assuming that the Commission determines that the additional year is not required and the Pilot is automatically sunset at the end of the first year, and six updates during the six-month post-Pilot Period. In aggregate, updates to the Exchange Transaction Fee Summary are estimated to cost $12,720 × 13 U.S. equities exchanges = $165,360 ≈ $165,000. The burden hours are obtained from supra Section III.D.2. 755The Financial Information eXchange (FIX) protocol is an electronic communications protocol that provides a non-proprietary, free and open XML standard for international real-time exchange of information related to the securities transactions and markets. See Fix Trading Community, available at https://www.fixtrading.org/. 756FpML (Financial products Markup Language) is an open source XML standard for electronic dealing and processing of OTC derivatives. It establishes the industry protocol for sharing information on, and dealing in, financial derivatives and structured products. See Financial products Markup Language [FpML], available at http:// www.fpml.org/. 757Most of the exchanges have at least some portion of their data available through XML formats. For instance, the NYSE Group of exchanges provides daily closing prices, among other data, in XML, Excel, and pipe-delimited ASCII, while the Nasdaq exchanges (Nasdaq, BX, and PHLX) and Cboe exchanges (Cboe BZX, Cboe BYX, Cboe EDGA, and Cboe EDGX), provide daily share volume data, among other data, in XML. Information on the use of XML by exchanges is available for the NYSE, www.nyse.com, Nasdaq, www.nasdaqomx.com, and Cboe, www.cboe.com, exchange groups, respectively, and was obtained from a staff review of information on publicly available exchange websites. The Commission was unable to obtain information from CHX or IEX on their use of XML from information available on their publicly available websites. 758This estimate is based on the following, which reflects the Commission’s experience with and burden estimates for systems changes to map to an XML schema: [(Programmer Analyst (1 hours) × $232) + (Senior Business Analyst (1 hours) × $265] = $497 ≈ $500 per exchange, or $500 × 13 exchanges = $6,461 ≈ $6,500 in aggregate. See Securities Continued associated with computing these summary data on fees and rebates are likely to be larger than the costs associated with updating the Exchange Transaction Fee Summary, discussed in detail below, and would likely require new systems by the exchanges to track the average and median fees. The Commission estimates that each exchange would have a one-time cost of about $24,000, or approximately $311,000 in aggregate across the 13 U.S. equities exchanges,
associated with the development and implementation of systems tracking realized monthly average and median share fees pursuant to the rule.749 The Commission further anticipates that it would cost an additional $12,000 annually, or $155,000, in aggregate, per year, to ensure that the system technology is up to date and remains in compliance with the rule.750 Moreover, as discussed above, exchanges would be required to produce monthly updates to the Exchange Transaction Fee Summary to capture realized average and median per share fees as well as any revisions to fee schedules made by the exchanges, which would be reflected in changes to Base or Top-Tier fees and rebates, detailed in Section II.E.2. The Commission estimates that each month it would cost each exchange $530 to update the dataset of summary fees to reflect the updates to historical realized average and median per share fees and changes to the Base and Top-Tier fees. This would require each exchange to make a total of 36 updates to the Exchange Transaction Fee Summary from the pre-Pilot Period through the post-Pilot Period, if the Commission determined that the Pilot should continue for up to a second year and not automatically sunset at the end of the first year.751 Each exchange would have total costs of updates to the Exchange Transaction Fee Summary of approximately $19,000 per exchange, or $248,000 among the 13 exchanges over the pilot duration, including pre- and post-periods.752 If the Pilot were to automatically sunset at the end of the first year, without the Commission determining that an extension for up to an additional year was needed, this would decrease the total number of updates to the Exchange Transaction Fee Summary to 24.753 Under an automatic sunset at the end of the first year, each exchange would have total costs of updates to the Exchange Transaction Fee Summary of approximately $12,700 per exchange, or $169,000 among the 13 exchanges over the pilot duration, including pre- and post-periods.754 As detailed above, the Commission estimates that the costs associated with the monthly updates to the Exchange Transaction Fee Summary would be a small fraction of the costs associated with the initial allocation of fees required at the outset of the Pilot. As discussed in Section II, the rule will require that the Exchange Transaction Fee Summary be published on the exchanges’ websites using an XML schema to be published on the Commission’s website. The Commission understands that there are varying costs associated with varying degrees of structuring. The Commission believes that most of the exchanges already have experience applying the XML format to market data. For example, the exchanges and market participants regularly use the FIX protocol 755 and FpML 756 to exchange information on highly structured financial instruments and related market data.757 The Commission anticipates that implementation of the Pilot’s XML schema would draw upon exchange resources and
experiences previously used to implement other supply chain information standards, like those discussed above, that were developed by industry consensus-based organizations. Costs generally associated with the implementation may include those for: Identifying the data required by the Pilot within the exchange source systems; mapping the relevant fields in the exchanges’ data source systems to the Commission’s XML schema; implementing, testing and executing the validation rules; and developing the website posting processes as required by the rule. The initial costs to exchanges of complying with the Commission’s XML schema in order to publicly post the Exchange Transaction Fee Summary in this format would be $500 per exchange, or $6,500 in aggregate across the 13 exchanges.758 For all updates to

Exchange Act Release No. 78309 (July 13, 2016), 81 FR 49431, 49475 (July 27, 2016) (‘‘Disclosure of Order Handling Information’’). The estimate is lower than that for proposed Rule 606 disclosures because the costs for those disclosures encompassed many additional requirements beyond the mapping to an XML schema. 759This estimate is based on the following, which reflects the Commission’s experience with and burden estimates for SRO systems changes:
[(Attorney (20 hours) × $401) + (Compliance Manager (20 hours) × $298) + (Programmer Analyst (20 hours) × $232) + (Senior Business Analyst (20 hours) × $265] = $23,920 ≈ $23,900 per exchange, or $23,920 × 13 exchanges = $310,960 ≈ $311,000 in aggregate. The burden hours are obtained from supra Section III.D.3. 760This estimate is based on the following, which reflects the Commission’s experience with and burden estimates for SRO systems changes:
[(Attorney (10 hours) × $401) + (Compliance Manager (10 hours) × $298) + (Programmer Analyst (10 hours) × $232) + (Senior Business Analyst (10 hours) × $265] = $11,960 ≈ $12,000 per exchange, or $11,960 × 13 exchanges = $155,480 ≈ $156,000 in aggregate. The burden hours are obtained from supra Section III.D.3. 761This estimate is based on the following:
[(Compliance Manager (4 hours) × $298) + (Programmer Analyst (3 hours) × $232)] = $1,888 per exchange, or $1,888 × 36 fee changes per exchange = $67,968 ≈ $68,000. The burden hours are obtained from supra Section III.D.3. The 36 updates to the order routing data for each exchange encompass six updates during the six-month prePilot Period, 24 updates during the two-year Pilot Period, assuming that the Commission determines at the end of the first year that it shall continue the proposed Pilot for up to an additional year, and six updates during the six-month post-pilot period. In aggregate, updates to the order routing data are estimated to cost $67,968 × 13 U.S. equities exchanges = $883,584 ≈ $884,000. 762This estimate of updates to the order routing data is the aggregation of updates from the pre-Pilot Period (6), the one-year Pilot Period assuming that the Commission allows the Pilot to automatically sunset at the end of the first year (12), and the postPilot Period (6), for a total number of 24 updates. 763This estimate is based on the following:
[(Compliance Manager (4 hours) × $298) + (Programmer Analyst (3 hours) × $232)] ≈ $1,888 per exchange, or $1,888 × 24 fee changes per exchange = $45,312 ≈ $45,300. The burden hours are obtained from supra Section III.D.3. The 24 updates to the order routing data for each exchange encompass six updates during the six-month prePilot Period, 12 updates during the first year of the Pilot Period, assuming that the Commission determines at the end of the first year that it shall automatically sunset the proposed Pilot, and six updates during the six-month post-pilot period. In aggregate, updates to the order routing data are estimated to cost $45,312 × 13 U.S. equities exchanges = $589,056 ≈ $589,000. 764See NYSE Letter I, at 15. 765However, it is unclear exactly how the commenter aggregated the data in the Proposing Release to arrive at 160 hours because they did not provide details of their calculation. 766See Proposing Release, supra note 2, at 13061. Discussion of comments on these estimates is presented in Section III.D.3 supra. the Exchange Transaction Fee Summary, the Commission estimates that any burden associated with making those available using the XML schema is included in the costs of the updates discussed above. In sum, the Commission estimates the total cost of the pilot associated with producing the exchange transaction fee summary in XML format to be approximately $81,000 per exchange if the Pilot runs for 2 years and $62,700 per exchange if the Pilot sunsets at the end of the first year. These costs comprise of approximately $26,100 in one time implementation costs and $55,000 in ongoing costs if the Pilot runs for two years, or $36,600 if the Pilot sunsets at the end of the first year. These costs aggregate to approximately $1,054,000 in total costs across all exchanges if the Pilot runs for the entire two years, and $815,000 if the Pilot sunsets at the conclusion of the first year.
iv. Producing the Order Routing Data
The rule also will require as part of the Pilot that exchanges prepare, in pipe-delimited ASCII format, and transmit to the Commission, order routing data, updated monthly, containing aggregated broker-dealer order routing information. As discussed in Rule 610T(d) and in Section II.E.3, the datasets would contain separate order routing data for liquidityproviding and liquidity-taking orders aggregated by day, by security, by broker-dealer, and by exchange. The Commission believes that as long as the CAT Phase 1 data are available at the implementation of the Pilot, the exchanges would be able to use that data to construct the order routing data required by the rule. In particular, the CAT data will include records for every order received by an exchange that indicate the member routing the order to the exchange and details regarding the type of security. The CAT data will also include other information necessary to create the order routing data such as order type information, special handling instructions, and execution information. In the event that the CAT Phase 1 data were not available, the exchanges would have to use existing systems to collect the required order routing data. Regardless of which system exchanges use for the order routing data, the Commission anticipates they would incur costs in producing the downloadable files containing aggregated monthly order routing data to be transmitted to the Commission. The Commission estimates that each exchange would have a one-time cost of approximately $23,900, or approximately $311,000 in aggregate across the 13 exchanges, associated with the development and implementation of systems needed to aggregate the order routing information, as well as store the data, in the pipe-delimited ASCII format specified by the rule and as detailed in Rule 610T(d).759 The Commission anticipates that it will cost each exchange an additional $12,000 per year, or approximately $156,000 in aggregate per year, to ensure that the system and storage technology is up to date and remains in compliance with the rule.760 The rule will require that exchanges produce monthly updates of the order routing data, and transmit them to the Commission in pipe-delimited ASCII format by the end of the month, as detailed in Section II.E.3 and Rule 610T(d). The Commission estimates that the transmittal and updates of the order routing datasets would cost $1,888 each month. This will require each exchange to make a total of 36 updates to the order routing data from the pre-Pilot Period through the post-Pilot Period (if the core Pilot lasts for a full two years). Each exchange would have recurring costs of updates to the order routing data of approximately $68,000 per exchange, or $884,000 among the 13 exchanges over the entire duration of the Pilot, and the pre-Pilot and postPilot periods.761 If the Commission were to allow the Pilot to automatically sunset at the end of the first year, this would decrease the total number of
monthly updates to the order routing data by 12 to 24.762 Under the automatic sunset, each exchange would have recurring costs of updates to the order routing data of approximately $45,300 per exchange, or $589,000 among the 13 exchanges over a one-year Pilot, and the pre-Pilot and post-Pilot periods.763 One commenter stated that the Commission underestimated the number of burden hours required to produce the order routing data required by the Pilot.764 This commenter indicated that the Commission allocated 160 burden hours to compile and produce the order routing data, while the commenter estimates that it would take approximately 400 burden hours.765 Over the entire Pilot duration, including the six-month pre and postPilot periods, the Commission estimates that exchanges would have initial systems burden hours of 80 hours, an additional annual burden of 40 hours to update and maintain those systems, plus 84 burden hours per year to produce and publicly post order routing data monthly.766 In sum the Commission estimates the costs of producing the order routing data to include a one-time cost of approximately $23,900 per exchange to set up the data gathering process, $12,000 per year to maintain the data gathering systems, and $1,888 per

767The estimate is based on the following:
[(Attorney (40 hours) × $401) + (Compliance Attorney (40 hours) × $352) + (Assistant General Counsel (25 hours) × $449) + (Director of Compliance (15 hours) × $470)] = $48,395 ≈ $48,400, or $48,395 × 13 equities exchanges = $629,135 ≈ $629,100 in aggregate. See OMB Control No. 3235–0045 (August 19, 2016), 81 FR 57946 (August 24, 2016) (Request to OMB for Extension of Rule 19b–4 and Form 19b–4 Filings). 768The Commission believes that the inclusion of Linked Pricing prohibitions for Test Group 2 should not increase the complexity of Form 19b–4 filings for exchanges because many exchanges already report non-cash incentives, such as tiered pricing or volume discounts, as part of their standard filings. Further, the Commission does not believe that many exchanges currently use Linked Pricing mechanisms and instead most rely on rebates. 769Maintaining the current average frequency of 7 19b–4 filings per year would mean that the average exchange would file a total of 14 19b–4 filings during the two-year pilot (7 filings × 2 year duration). If the Commission were to allow the Pilot to automatically sunset at the end of the first year, then the total number of 19b–4 filings could decrease by 7 filings. Annually, across all 13 exchanges, the Commission estimates that there will be 91 19b–4 filings (7 filings × 13 exchanges). If the Commission determines that the Pilot shall continue for a second year, in aggregate, the 13 exchanges could file a total of 182 19b–4 filings (91 × two-year Pilot duration). 770The estimate is based on the following:
[(Attorney (8 hours) × $401) + (Compliance Attorney (8 hours) × $352) + (Assistant General Counsel (6 hours) × $449) + (Director of Compliance (4 hours) × $470)] =$10,598 ≈ $10,600, or $10,598 × 182 fee changes in aggregate across 13 exchanges over the two-year pilot duration = $1,928,836 ≈ $1,929,000 in aggregate, assuming that the Commission determines that the Pilot shall continue for up to an additional year. If the Pilot were to automatically sunset after the first year, the Commission believes that the costs associated with 91 19b–4 filings (13 exchanges × 7 filings) would be approximately $964,000 ( ∼$10,598 × 91 filings). See Request to OMB for Extension of Rule 19b–4 and Form 19b–4 Filings, supra note 767. month to publish the data. Specifically, the Commission estimates initial onetime implementation costs of approximately $23,900 and ongoing costs of approximately $103,800 per exchange if the Pilot lasts two years or $69,200 if the Pilot lasts one year. These costs total approximately $127,800 per exchange if the Pilot lasts two years— or approximately $1,661,000 in aggregate. These costs decline to approximately $93,200 per exchange— or $1,211,000 in aggregate—if the Pilot sunsets after one year.
v. Fee-Related Costs to Exchanges
When exchanges alter their fees they are required to submit a Form 19b–4 filing with the Commission. Consequently, the Commission expects most exchanges to file two 19b–4 Forms that they would not have otherwise done. Additionally, the Commission expects that the pilot may increase the complexity of these filings. This section provides estimates for the costs associated with the submission of 19b– 4 Forms by the exchanges during the Pilot. At the outset of the Pilot, each equities exchange if their fees do not at that time comply with the Pilot’s pricing restrictions, would need to file with the Commission a comprehensive Form 19b–4 fee filing reflecting all of the applicable fees and rebates applicable to each of the Pilot Groups, as well as the Control Group—to reflect the temporary changes to transaction-based fees and rebates as a result of the Pilot. The Commission anticipates that exchanges will incur costs associated with and devote time to optimally assign fees and rebates across Test Groups, within the parameters allowed by the Pilot, including any incentives, tiers, caps, and discounts available. The Commission estimates that it would cost $48,400 per-exchange for the initial Form 19b–4 fee filing or $629,100 in aggregate.767 The Commission further anticipates that exchanges would bear similar costs upon the completion of the Pilot to prepare Form 19b–4 fee filings for filing with the Commission to reflect changes in fees at the conclusion of the Pilot, should they wish to change their fees or revert to their former pricing models after the Pilot concludes. In addition to the initial production of the Form 19b–4 fee filing at the outset of the Pilot, exchanges may also choose to make periodic updates to their fee and rebate schedules, and file Form 19b–4 fee filings to effectuate those changes and thereby notify the Commission and the public of those updates. As noted in the baseline, the average exchange makes approximately seven changes to its fee schedules per year. While recognizing the possibility that as a result of the Pilot, exchanges may revise their fee schedules more or less often during the Pilot, the Commission has no basis to expect an increase in the number of Form 19b–4 fee filings other than at the beginning or end of the Pilot and has no basis to expect a decrease. The Commission also recognizes that as an outcome of the Pilot, the complexity of the Form 19b–4 fee filings could increase if exchanges seek to impose different fees within Test Groups 1, Test Group 2, and the Control Group, thereby increasing the overall costs for exchanges to revise their fee and rebate schedules.768 As discussed above, the Pilot may require exchanges to design new fee structures to comply with the Pilot’s Test Groups, which would then translate into additional information in each Form 19b–4 fee filing submitted during the Pilot. These costs are likely to increase because the exchanges could take more time to design and describe fee
structures in each filing than they do designing fee structures today. As discussed above in the baseline, the average fee schedules of exchanges are complex, with many different categories of fees or rebates assessed to NMS stocks. Assuming the frequency remains constant, then the Pilot could increase the incremental costs incurred by exchanges to file the expected Form 19b–4 fee filings during the Pilot.769 The additional costs would only be relevant for Form 19b–4 fee filings that occur during the Pilot Period, and would not apply to Form 19b–4 fee filings in the pre-Pilot or postPilot Periods, as the Commission does not believe that there will be any incremental costs associated with increased complexity of these filings during these periods. The Commission estimates that each exchange would bear an incremental cost of $10,600 per Form 19b–4 fee filing to account for the increased complexity associated with the requirements of the Pilot, or approximately $1,929,000 for the anticipated 182 Form 19b–4 fee filings for fee and rebate revisions across the 13 U.S. equities exchanges during the twoyear pilot duration.770 If the Pilot were to automatically sunset at the end of the first year, the Commission estimates that exchanges would bear costs of approximately $964,000 for the anticipated 91 Form 19b–4 filings for fee and rebate revisions across the 13 U.S. equities exchanges during the first year of the Pilot duration. In sum, the Commission expects the pilot to impose on each exchange a onetime cost of $48,400 at the beginning and end of the pilot for the additional 19b–4 filings required by the pilot, as well as an ongoing cost of approximately $10,600 per additional 19b–4 filing to account for increased complexity in 19b–4 filings caused by the Pilot. If we assume that exchanges continue to file 19b–4 filings at an average rate of 7 per year and if the pilot lasts for 2 years, these incremental costs sum to approximately $148,400 per exchange—which declines to $74,200 if the pilot ends after the first year. Combining the cost of the two additional 19b–4 filings with the cost of potential increased complexity provides an estimated cost of the pilot associated with 19b–4 filings of approximately $245,200 per exchange—or $3,187,000 in aggregate—if the pilot lasts two years, or $171,000 per exchange—or $2,223,000 in aggregate—if the pilot expires after the first year.

771On a given trade, an exchange earns the margin between fees and rebates. For example, if an exchange charges a take fee of $.0030 per share and offers a make rebate of $.0025 per share then the margin captured by the exchange is $.0005 per share traded. 772A number of commenters expressed concern that the Pilot would lead to decreased exchange revenue largely through decreased trading volume. See, e.g., Cboe Letter I, at 7; NYSE Letter I, at 3 and 15–16. 773See Proposing Release, supra note 2, at 13063. 774See Proposing Release supra 1 at 13067. 775This was the case in Canada when in January 2017 the Canadian Securities Administrators (CSA) approved the lowering of the fee cap for noninterlisted Canadian stocks from $0.0030 to $0.0017. In response to this regulation, the Toronto Stock Exchange retained its $0.0004 margin per trade for continuous trading—high priced securities ($1.00 and over) by lowering both fees and rebates by $0.0008. Fees for taking liquidity on noninterlisted securities reduced from $0.0023 to $0.0015 whereas rebates provided to liquidity providers declined from $0.0019 to $0.0011. Fees and rebates for inter-listed securities remained unchanged. See https://www.tsx.com/resource/en/ 1501. 776See Section IV.D.2.a. See also Section IV.C.2.b.iv 777See, e.g., Leaf Letter, at 1–2; Ennis Letter, at 2. 778T. Rowe Price Letter, at 4–5. 779See, e.g., TD Ameritrade Letter, at 7. Other commenters also either expressed their belief that the Pilot would not reduce trading volumes (see, e.g., IEX Letter II, at 8; T. Rowe Price Letter, at 5) or expressed uncertainty about the outcome of the Pilot on trading volume (see, e.g. Credit Suisse Commentary, at 5).
vi. Other Costs to Exchanges
The Pilot may result in more complicated fee structures that could also increase an exchange’s processing costs of tracking and calculating monthly invoices for its members during the Pilot; however, the Commission does not have any information on the costs to exchanges for tracking and calculating monthly member invoices and therefore cannot provide estimates of quantified costs and no commenters provided such information. b. Other Costs Associated With the Pilot This section considers additional costs that may occur as a result of the Pilot. Specifically, this section discusses how the Pilot may impact exchanges’ fee revenue, broker-dealer compliance costs, brokerage commissions, liquidity, and issuers.
i. Loss of Exchanges’ Fee Revenue
The Commission analyzed whether exchanges could experience a change to their fee revenues associated with transaction-based fees and rebates for either of two reasons: A decline in the margin between fees and rebates,771 or a decline in overall trading volume on an exchange as a result of the Pilot.772 In the Proposing Release the Commission stated its belief that only stocks in the test group with a cap of $0.0005 (former Test Group 2) would experience narrower margins and estimated that these narrower margins could result in exchanges incurring revenue losses of up to $7,650,000 per month.773 With the removal of the former Test Group 2 the Commission now believes that the Pilot may not have a significant effect on Exchange revenue, at least not because of narrower margins between fees and rebates. For stocks in Test Group 1 (the Test Group with a fee cap of $0.0010) the Commission does not believe that the Pilot will result in narrower margins earned by the exchanges because the fee cap in this group is double the current typical average net capture (i.e., margin) that exchanges earn which is approximately $0.0005.774 Consequently, the exchanges can maintain current margins by reducing the rebate offered at the same level as fees charged are reduced.775 For stocks in Test Group 2, the Commission also does not believe that the Pilot is likely to shrink margins. For these stocks the fee cap remains at $0.0030, while exchanges are prohibited from paying rebates and offering Linked Pricing. Consequently, the prohibition of rebates for securities in this Test Group would conceivably allow the exchanges to reduce fees to as low as $0.00025—if charged to both parties in a transaction—without reducing the exchange’s average net capture per trade. While less likely given competitive dynamics, if the exchanges wanted to increase their net capture, it is possible under the pilot terms for total net capture in group one to be as high as $0.002 (if both side were charged the maximum of $0.001), and in group two to be $0.003, both of which are far in excess of the average net capture that exchanges receive today. For these reasons, the Commission now expects the effects of the Pilot on exchange revenue through impacting per-trade margins to be minimal. In addition, the Commission believes that the impact of the Pilot on exchange revenues through changes in trading volume are difficult to determine in advance, but recognizes that the magnitude of such changes could be significant and some potential lost revenue could be in a transfer to investors or among exchanges. The Commission considered whether individual exchanges could experience a decline in trading volume for four reasons, as multiple commenters suggested: If exchanges lose volume to off-exchange venues, if volume declines because of increased transaction costs, if the Pilot reduces excessive intermediation, or if volume shifts among exchanges. The Commission recognizes that the Pilot presents a risk
that the Pilot could result in less fee revenue for exchanges due to lower trading volumes. However, the Commission believes that decreased trading volume, while one possible outcome of the Pilot, is not the only reasonable outcome, and that the ex ante effect of the Pilot on trading volume is difficult to determine. First, several commenters stated that reducing or eliminating the ability for exchanges to pay rebates may cause exchanges to become less competitive relative to off-exchange venues like ATSs, which would not be so constrained. The analysis in Section IV.D.2.a identifies significant uncertainty in the potential for exchanges to be less competitive relative to off-exchange venues such as ATSs, and identifies conditions in which they could actually be more competitive.776 Consequently, the Commission cannot determine in advance of the Pilot whether exchanges will lose volume to off-exchange venues. Second, total trading volume, and consequently exchange revenue, could decline if the Pilot increases transaction costs. A number of issuers expressed the concern via comment letters that the Pilot would lead to lower levels of trading volume because of their experience with the Tick Size Pilot.777 However, not all issuers felt that the Pilot would result in lower trading volumes. One issuer ‘‘welcome[d] the opportunity for [its] stock to be included in the Pilot’’ and did not ‘‘expect that a reduction or outright removal of rebates will have any significant or harmful effects on . . . [its] stock’s trading volume.’’ 778 On the other hand, if the Pilot decreases the cost of trading on the exchanges, then the Pilot could increase trading volumes on the exchanges. This view was expressed by one commenter who stated their belief ‘‘the Pilot will reduce the costs of trading on exchanges, which may increase trading volumes on the exchanges.’’ 779 Lower costs of trading, caused by the reduction in fees, might increase trading volume on exchanges for at least two reasons. First, lower trading costs may induce trades that would otherwise not have occurred by allowing investors the ability to trade on smaller increments of information. Lower trading costs may also induce the participation of new traders, such as short-term traders for whom transaction costs are of greatest concern, to transact in a given stock

780NYSE Letter I, at 17.
781See Proposing Release, supra note 2, at 13042. 782See NYSE Letter I, at 15–16. 783See State Street Letter, at 2. 784See Section IV.C.1.a.iii, supra. 785See Section III.A. See also, e.g., Fidelity Letter, at 8. who would not otherwise participate. Consequently, if the Pilot leads to decreased trading costs, then trading volumes in those stocks may increase— increasing exchange revenue. In Section IV.C.2.b.iv below, the Commission discusses its belief that the effect of the Pilot on liquidity and transaction costs is not clear and could either increase or decrease. Given this uncertainty in the impact on liquidity, the Commission cannot determine in advance whether the Pilot will result in a reduction in liquidity that would reduce trading volume. Third, the Pilot may decrease trading volume, and thus exchange revenues, if it results in a reduction in intermediation by market makers for two reasons, but this would be a transfer to investors. The first reason is that market makers from time to time will use marketable orders to balance inventory. If these market makers decline to participate due to reduced rebate incentives, then their marketable orders will not arrive—diminishing trading volume. The second reason decreased intermediation may lead to lower volumes is that non-market makers might begin to execute their trades via non-marketable orders. Nonmarket makers may submit marketable orders because of an inability to achieve high fill rates with non-marketable limit orders due to significant competition from market makers. The reduction in intermediation may result in situations where two traders with offsetting trades, who would have generated two separate trades with market makers as the counterparty instead execute their trade with each other resulting in one trade. This could occur, for example, if the Pilot reduces queue lengths for investors as discussed in Section IV.C.2.b.iv.(2) below. While this effect would result in a loss in revenue to exchanges that would collect margin on a smaller number of trades, it would be a net gain for investors because executions on a smaller proportion of marketable orders would mean that the investors pay less in transaction fees and would more often capture, or earn, the spread where previously they would have paid the spread to transact. Fourth, even if overall trading volume does not decline or shift to off-exchange venues during the Pilot, individual exchanges may experience a decline in trading volume if the Pilot leads to a change in market share among the lit exchanges. This mechanism is discussed in Section IV.D.2.a. This
section also highlights the difficulties in
determining the expected redistribution of market share among the existing exchanges due to potentially countervailing economic effects. To this point, one commenter noted that the loss in revenue estimated above relied on ‘‘exchanges’ existing market share percentages’’ which ‘‘assumes that exchanges would remain equally competitive for order flow.’’ 780 The Commission agrees that the Pilot may impact the level and distribution of trading volume on lit exchanges but notes that such a redistribution would be a transfer among exchanges rather than an economic cost. The Commission acknowledges that the reduction or elimination of rebates may particularly affect smaller exchanges due to the liquidity externality, especially if their primary competitive differentiation is based upon a modified fee model. As discussed in the Proposing Release liquidity tends to consolidate.781 Thus, the restrictions on rebates resulting from the Pilot could harm smaller exchanges that may be competing by paying large rebates rather than by producing better prices or execution quality. In the short run, this could lead to lost revenue for these exchanges. It could also have longer-term effects if smaller exchanges consolidate or exit as a result of the Pilot. However, the Commission does not believe that consolidation or exit is likely during the pilot because only about a quarter of NMS stocks will be included in test groups. The Pilot could also impact exchanges’ fee revenue after the conclusion of the Pilot if as a result of the Pilot broker-dealers permanently alter their order routing decisions after the Pilot is completed. One commenter argued that this may be the case and suggested that the Commission’s claim that the Pilot’s effect on broker-dealers’ routing decisions would be temporary ‘‘[was] contradicted by the Commission’s own finding that brokerdealers would not change their behavior unless the Transaction Fee Pilot lasts for at least one year.’’ 782 To this point, given the competitive nature of financial markets, the Commission does not expect that it would take most brokerdealers up to one year to alter their behavior. Indeed, this commenter supports this belief by stating that exchange and non-exchange trading centers vigorously compete for trading volume, and that market participants are sensitive to revisions in transactionbased pricing models.783 Given this competition, the Commission believes market participants will likely adjust their behavior quickly both upon the implementation and conclusion of the Pilot, and that the Pilot duration will incentivize broker-dealers not to ‘‘wait out’’ the Pilot who could be otherwise inclined to do so if the duration were not sufficiently long.784 If the Pilot results in a decline in fee revenue for exchanges, then this could lead to other costs borne by investors as a result. Exchanges could promote additional order types and may even initiate new types of markets as a result of the Pilot, which would only
serve to further fragment markets and add to their complexity, the costs of which could be borne by investors.785 In particular, the Commission recognizes the remote possibility that an exchange holding company could attempt to optimize its overall performance during the Pilot by further diversifying with other exchange models. The Commission believes, however, that a new equity exchange registered in direct response to the Pilot would be unlikely to become operational before the conclusion of the Pilot. In addition, the Commission believes that it is unlikely that exchanges will promote additional order types as a result of the Pilot. In sum, the Commission believes that the costs to the exchanges due to narrower margins earned per trade are likely to be minimal—if any—due to the removal of the proposed Test Group with the fee cap of $0.0005. However, the Commission does expect that there could be a change in trading volume or a redistribution of market share among exchanges as market participants reoptimize their order routing systems as a result of the Pilot. However, due to the reasons discussed in this section, the Commission cannot determine in advance of the Pilot whether these market share/trading volume changes will increase or decrease exchange revenue. Consequently, the Commission acknowledges that the Pilot may lead to lower trading volume/market share for exchanges, which would impose a cost in terms of lost transaction fee revenue, but is unable to quantify the expected magnitude of this potential cost and no commenter provided an estimate of the amount of the lost transaction fee revenue. While the Commission cannot determine in advance of the Pilot its impact both in terms of direction and magnitude, the Commission has attempted to estimate the costs should volume decline. Using data from Table 3 in section IV.B.2.e the Commission

786See, e.g., Larry Harris Letter, at 10–11; STANY Letter, at 2; FIA Letter, at 3; Nasdaq Letter I, at 10; Nasdaq Letter III, at 9. 787See, e.g., Vanguard Letter, at 3; Healthy Markets Letter I, at 34. 788One commenter stated that the Commission did not consider implementation and coding costs. See STANY Letter at 2. However, the commenter does not elaborate on why the Proposing Release estimates of $3,741,000 implementation costs for broker-dealers to adjust their order routing systems at the beginning and end of the Pilot and the $20,726,000 costs for broker-dealers to update their order routing systems for fee changes during the Pilot failed to consider implementation and coding. See Proposing Release, supra note 2, at 13063– 13064. 789See Bacidore, Jeff, Hernan Otero, and Alak Vasa, 2011, Does smart routing matter ?, Journal of Trading 6, 32- 37. (available at http:// jot.iijournals.com/content/6/1/32), which found that smart-order routers designed to maximize rebates delivered worse execution quality to their clients. 790See FIF Letter at 8. See also FIA Letter at 3; Nasdaq Letter I at 10. 791Even in the absence of smart-order routers, broker-dealers could still adjust their execution determinations to take advantage of the changes implemented during the Pilot and these adjustments would incur costs. While the Commission does not estimate these particular costs, the assumption that all broker-dealers have automated order routing systems is reasonable and is necessary to enable the estimation of cost estimates. The Commission believes, however, that the costs to adjusting manual systems could be lower than the costs to adjust automated routing systems. If any broker-dealers still route orders manually, they likely do so because setting up and maintaining manual systems is not economical for them. It is likely that such firms utilize exchange routing services. 792See sections IV.C.1.a.iii and IV.E.5.a for additional discussion. 793See Table 2 in Section IV.B.2.f supra. 794 In addition, the Commission recognizes the potential costs to exchanges of this complexity above in Section IV.C.2.a. 795See Larry Harris Letter, at 10–11. 796This estimate is based on the following, which reflects the Commission’s experiences with and burden estimates for broker-dealer systems changes:
[(Attorney (5 hours) × $401) + (Compliance Manager (10 hours) × $298) + (Programmer Analyst (10 hours) × $232) + (Senior Business Analyst (5 hours) × $265)] ≈ $9,000 per broker-dealer that is a member of at least one exchange. As of December 31, 2016, estimates an annualized upper bound on the profit earned from transaction fees for each exchange. Aggregated across all exchanges this number equals approximately $3 billion per year. However, only about 1/4th of NMS securities will be in a test group subject to the Pilot, so the Commission estimates the Pilot could affect the approximately $750 million per year across all exchanges, depending on how much volume changes. Consequently, if the Pilot were to cause a 10% reduction in trading volume on exchanges then this change could reduce fee revenue on the exchanges by approximately $150 million over two years or $75 million if the Commission were to allow the Pilot to automatically sunset at the end of the first year. However, the Commission does not believe that the 10% reduction in trading volume is a reasonable assumption. While the Swan Study shows that Nasdaq lost 10% of its volume using one volume measure, it lost only 200,000 shares in another. In addition, the Swan Study finds no change in overall volume or in offexchange volume, just a migration from Nasdaq to other exchanges. Therefore, the Commission does not believe that either the 10% volume reduction or, consequently, the estimate of $150 million revenue reduction is reasonable.
ii. Broker-Dealer Systems Costs
Although the costs of compliance with the Pilot will primarily affect the exchanges, broker-dealers and other market participants are also likely to incur costs as a result of the Pilot. Commenters provided mixed information on the magnitude of these costs. While some commenters stated that the costs to broker-dealers of the Pilot would be substantial,786 other commenters stated that the costs for broker-dealers associated with the Pilot would not be significant or expensive because exchange fee schedules change regularly and broker-dealers are used to adapting their order routing algorithms to new and changed fee schedules.787 This section provides the Commission’s estimates for broker-dealer compliance costs associated with the Pilot.788 In response to the Pilot, market participants might have a one-time cost at the onset and the conclusion of the Pilot to adjust their systems to reflect the shocks and potential additional complexity of transaction-based fees. In addition, there may be additional modifications to routing strategies that are made in subsequent months to adjust to changing liquidity dynamics as behavior changes associated with the pilot settle in. Many broker-dealers have smart-order routing systems that use algorithms to route orders based on certain criteria, such as fill rates, time to execution, lowest fees, or highest rebates.789 One commenter agreed, stating that such systems changes ‘‘should primarily consist of modifications to the routing tables and other associated operational activities.’’ 790 The Commission understands that some of the associated changes and modifications may already be coded into the smart order router (SOR) algorithms such that changes to associated liquidity shifts may be dynamic and automated, i.e., in need of little additional modification. To estimate these costs, the Commission assumes (1) that all brokerdealer members of exchanges will adjust their systems for the pilot and (2) that all broker-dealer members of exchanges have automated order routing systems.791 While the Pilot does not directly require broker-dealers to adjust their systems, the Commission expects broker-dealers who do not update their systems may incur significant costs relative to those who do in terms of potential impacts on execution quality and in their ability to manage fees and rebates. Broker-dealers might choose to adjust their systems for the Pilot for many reasons, including to recognize that the Pilot could affect execution quality for investors and/or to better manage fees and rebates.792 Therefore, the cost estimates assume that brokerdealers will adjust to their existing systems to capture changes in fees and rebates associated with each Test Group of securities, rather than bearing start-up costs associated with implementing new order routing systems. In its estimates, the Commission recognizes that the costs associated with adjusting the execution algorithms by broker-dealers for the Pilot are likely to be
more costly than the periodic updates that broker-dealers may make to incorporate changes to fee schedules implemented by exchanges or to fine tune their strategies. The additional expected costs may occur because changes for the Pilot are likely to require more complex programming that segments stocks into different fee regimes (assuming exchanges implement fees customized to each Test Group), rather than just altering codes or inputs. As of July 2017, exchanges have 18 fee categories and 21 rebate categories, on average.793 If exchanges maintain the same level of complexity in their fee schedules during the Pilot, up to a two-fold increase in the number of fee and rebate categories could occur, which would increase complexity for broker-dealers who incorporate fees into their order routing decisions.794 Additionally, the Commission agrees with the commenter who stated that, to the extent that broker-dealers’ order routing algorithms are programmed to the exchange, and not the individual security, the Pilot will increase complexity by requiring an adjustment to this methodology.795 The Commission estimates that the costs to broker-dealers that are members of exchanges to make the initial adjustment to their order routing systems at the outset of the Pilot would be approximately $9,000 per brokerdealer, or $3,573,000 in aggregate across the 397 broker-dealers that are currently members of equities exchanges.796 The

397 unique broker-dealers were members of exchanges (Form X–17a–5). The aggregate costs of updating order routing systems to reflect the Transaction Fee Pilot requirements would cost $9,000 × 397 ≈ $3,573,000. Note that smaller broker dealers will often use the smart order router of larger broker dealers or those offered by exchanges, and will therefore benefit indirectly from the work done by the providers of their smart order routing services. 797Several commenters made similar statements. See, e.g., Vanguard Letter, at 3; Healthy Markets Letter I, at 34. 798The Commission cannot estimate and commenters provided no insight into the degree to which the number of fee and rebate revisions by exchanges will increase or decrease during the Pilot. 799This estimate is based on the following, which reflects the Commission’s experiences with and burden estimates for broker-dealer systems changes:
[(Compliance Manager (0.5 hours) × $298) + (Programmer Analyst (0.5 hours) × $232)] = $265 per broker-dealer that is a member of at least one exchange. The aggregate costs updating order routing systems to reflect the periodic fee and rebate revisions would cost $265 × 397 ≈ $105,000. 800 If 182 total fee and rebate changes were to occur over the duration of the Pilot (13 equities exchanges × 7 revisions per year × 2 years = 182), each broker-dealer would bear costs of updating its order routing systems of $265 × 182 ≈ $48,000, or $19,056,000 ($48,000 × 397) in aggregate across all broker-dealers over the first year of the Pilot. The Commission estimates that costs would be $9,528,000 ($265 × 13 exchanges × 7 updates × 397 broker-dealers) if the Commission determined that Pilot automatically sunset at the end of the first year. 801These costs reflect the estimated cost of $9,000 at the outset of the Pilot to update the order routing system to reflect the changes to the fee structure for securities in the test groups, $48,000 to reflect the incremental costs of the estimated 182 revisions to fee schedules during the Pilot ($265 per revisions × 7 revisions per year × 2 years × 13 exchanges), and $9,000 at the conclusion of the Pilot to unwind changes to the order routing systems, for a total of $66,000 per broker-dealer. If the Pilot were to automatically sunset at the end of one year, then these costs would be approximately $42,000 ($265 × 7 revisions × 13 exchanges+2*$9,000) per brokerdealer. 802See, e.g., Angel, Harris, & Spatt, supra note 530. 803See Regulation NMS Subcommittee Recommendation for an Access Fee Pilot (June 10, 2016), available at https://www.sec.gov/spotlight/ emsac/emsac-regulation-nmsrecommendation61016.pdf (‘‘June Recommendation’’). See also Shawn O’Donoghue, 2015, ‘‘The Effect of Maker-Taker Fees on Investor Order Choice and Execution Quality in U.S. Stock Markets’’, Working Paper, available at https:// papers.ssrn.com/sol3/papers.cfm?abstract_ id=2607302. 804Several commenters made similar statements. See, e.g., NYSE Letter I, at 13; ASA Letter, at 2. 805The Commission acknowledges differing effects on brokerage commissions could occur as a result of the Pilot depending on whether the client is a retail customer versus an institutional customer. For instance, some brokerage accounts charge per-transaction commissions to retail clients. Institutional commissions, on the other hand, are highly negotiated and may be based on something other than a per trade or per share basis, such as a flat fee for use of a broker’s order routing algorithm; however, data on the structure or magnitude of institutional commissions is not publicly available. 806There are approximately 8,000 NMS securities and just under 800 will be included in Test Group 2. Commission further estimates that broker-dealers would bear a similar cost to adjust their order routing systems at the conclusion of the Pilot. Additionally, the Commission expects that
broker-dealers would update their order routing systems with changes to fees or rebates submitted by exchanges through Form 19b–4 fee filings to the Commission during the Pilot. As discussed in the baseline, exchanges, on average, make changes to fees or rebates approximately seven times per year; therefore, broker-dealers are likely to have experience in updating the order routing systems to reflect these routine changes to fees and rebates.797 As in the estimates of the costs of the initial and final adjustments, broker-dealers are likely to face higher costs per update as a result of the Pilot because of the added complexity of having to update multiple modules within their order routing systems. The Commission’s estimates of these updates assume that exchanges update their fees schedules as often during the pilot as at present. Therefore, the costs to broker-dealers associated with the Pilot are the additional costs associated with the complexity of the updates and not the total cost of the updates.798 In other words, brokerdealers would have updated their systems (or routing tables) anyway in the absence of the Pilot to reflect the same number of exchange fee and rebate changes. The Commission estimates described below reflect the additional cost of the Pilot (‘‘additional costs’’), which is how much more an update might cost during the Pilot compared to a scenario without the Pilot. The Commission believes that the perupdate additional costs associated with these changes are likely to be a small fraction of the costs associated with the initial costs of adjusting the routing systems to reflect the required fee and rebate revisions at the outset of the Pilot. The Commission estimates that the additional costs to broker-dealers that are members of exchanges to make periodic adjustments to their order routing systems to reflect changes in fees and rebates would be $265 per adjustment, or approximately $105,000 in aggregate across the 397 brokerdealers that are members of U.S. equities exchanges.799 As shown above, the Commission expects that exchanges, if submitting changes to fees and rebates at the same rate as they have in the last five years, would submit 182 total revisions to fees and rebates over the two-year pilot duration. Therefore, the aggregate costs of updating order routing systems would be $48,000 per brokerdealer, or $19,056,000 in total across all broker-dealers.800 If the Pilot were to automatically sunset at the end of the first year, the aggregate costs of updating order routing systems would be $24,000 per broker-dealer, or $9,528,000 in total across all broker-dealers. In sum the Commission believes that the all in costs to broker-dealers of updating their order routing systems as a result of the Pilot will average approximately $66,000 per brokerdealer to update their systems over the entire Pilot Period. If the Pilot automatically sunsets at the end of the first year, the costs associated with these updates will be
approximately $42,000 per broker-dealer.801 The Commission notes that these estimates may be overstated. Not all broker-dealers are members of all exchanges, which would reduce the total number of changes to the order-routing systems that they would implement. Additionally, the exchanges could resort to more simplified fee schedules relative to the current baseline, which would reduce broker-dealers’ costs of updating their systems for the Pilot.
iii. Temporary Increase in Brokerage
Commissions
Beyond the implementation and compliance costs for exchanges and broker-dealers associated with the Pilot, the changes to the exchange transactionbased fee and rebate structure could lead to temporary increases in brokerage commissions charged to their customers. Several studies show, and several commenters concurred, that brokerage commissions today are at historically low levels.802 Brokerage clients may have a preference for low commissions with services provided by broker-dealers, and in turn, may allow broker-dealers to capture rebates (and bear the costs of access fees), either through explicit contracts or implicit agreements.803 As a result, the Pilot could lead to higher overall commissions as rebates obtained by broker-dealers fall,804 thereby temporarily reducing the overall welfare of retail brokerage clients as a result of increased commissions.805 For instance, the elimination of rebates and Linked Pricing in Test Group 2 could result in a transfer from broker-dealers to exchanges. Assuming, as discussed above, the margin between fees and rebates is approximately $0.0002 per share,806 with transaction fees of $0.0030 per share and rebates of $0.0028 per share, Test Group 2 could result in a transfer of $0.0028 from broker-dealers to the exchanges with

Transaction Fee Pilot for NMS Stocks

Similar documents from other regulators