NetCoalition v. Securities & Exchange Commission

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued February 16, 2010 Decided August 6, 2010

No. 09-1042

NETCOALITION,
PETITIONER

v.

SECURITIES AND EXCHANGE COMMISSION,
RESPONDENT

NYSE ARCA, INC. AND
THE NASDAQ STOCK MARKET, LLC,
INTERVENORS

Consolidated with 09-1045

On Petitions for Review of an Order
of the Securities & Exchange Commission

Carter G. Phillips argued the cause for the petitioners.
Dennis C. Hensley, Kevin J. Campion, Richard D. Bernstein,
and Roger D. Blanc were on brief.
Mark R. Pennington, Assistant General Counsel, Securities
and Exchange Commission, argued the cause for the respondent.
Michael A. Conley, Deputy Solicitor, and Luis de la Torre,
Senior Litigation Counsel, were on brief.
 2

Douglas W. Henkin argued the cause for intervenors NYSE
Arca, Inc. et al. in support of the respondent. David S. Cohen,
Eugene Scalia and Amir C. Tayrani were on brief.
Before: HENDERSON and GARLAND, Circuit Judges, and
EDWARDS, Senior Circuit Judge.
Opinion for the Court filed by Circuit Judge HENDERSON.
KAREN LECRAFT HENDERSON, Circuit Judge: In 2006,
NYSE Arca, one of the largest securities exchanges in the
United States, proposed to begin charging a fee to investors for
access to its proprietary “depth-of-book” product, ArcaBook.
ArcaBook lists pending orders placed on NYSE
Arca—specifically, bids to buy at prices lower than, and offers
to sell at prices higher than, the prevailing market price. The
Securities and Exchange Commission (SEC or Commission)
approved NYSE Arca’s proposal, finding the proposed fees for
ArcaBook were “fair and reasonable” and “not unreasonably
discriminatory.” Petitioners NetCoalition, a public policy
corporation representing approximately 20 internet companies
(including Google and Yahoo!), and the Securities Industry and
Financial Markets Association (SIFMA), a trade association
representing more than 600 securities firms and banks, challenge
the SEC order, arguing that it violates the Securities Exchange
Act of 1934, 15 U.S.C. §§ 78a et seq. (Exchange Act), and the
Administrative Procedure Act, 5 U.S.C. §§ 551 et seq. (APA).
For the reasons set forth below, we conclude that the SEC did
not adequately explain the basis of its approval nor, on this
record, support its conclusion with substantial evidence and,
accordingly, we remand to the SEC.
I. Background
The Exchange Act provides the framework by which the
SEC regulates the transactions and the classes of participants in
the major securities markets in the United States. See Bradford
Nat’l Clearing Corp. v. SEC, 590 F.2d 1085, 1090 (D.C. Cir.
 3

1978). NYSE Arca is registered with the SEC as a national
securities exchange, see 15 U.S.C. § 78f, and as such is a
securities industry “self-regulatory organization,” or SRO, see
id. § 78c(a)(26) (defining “self-regulatory organization”).
Although self-regulatory, NYSE Arca remains subject to
comprehensive SEC oversight and control. See Karsner v.
Lothian, 532 F.3d 876, 880 (D.C. Cir. 2008) (citing 15 U.S.C.
§ 78s(b)). For example, an exchange is required to file its
governing rules with the SEC and comply with them. See 15
U.S.C. § 78o-3(a)-(b). An exchange’s rules must, among other
things, “provide for the equitable allocation of reasonable dues,
fees, and other charges among its members and issuers and other
persons using its facilities,” id. § 78f(b)(4); “promote just and
equitable principles of trade” and not “permit unfair
discrimination between customers, issuers, brokers, or dealers,”1
id. § 78f(b)(5); and “not impose any burden on competition not
necessary or appropriate in furtherance of the purposes of” the
Exchange Act, id. § 78f(b)(8). As an SRO, an exchange must
also file any proposed rule change (including a fee change) with
the SEC for approval. See id. § 78s(b)(1)-(2). The SEC notices
a rule change proposal for public comment and either approves
it if it is consistent with the requirements of the Exchange Act
or disapproves it. See id. The SEC has delegated the initial
consideration of a rule change to its Division of Trading and
Markets (which until 2007 was known as the Division of Market
Regulation). See 17 C.F.R. § 200.30-3(a)(12).
In 1975, the Congress expanded the authority of the SEC
through a major overhaul of the Exchange Act. See Bradford,
590 F.2d at 1091. Among other things, the 1975 Amendment
directed the SEC to facilitate the establishment of a “national

1
A broker trades on behalf of investors while a dealer trades on
his own account. 15 U.S.C. § 78c(a)(4)-(5). Most securities firms do
both and are referred to as “broker-dealers.”
 4

market system for securities,” or NMS, to link securities markets
nation-wide in order to distribute market data economically and
equally and to promote fair competition among all market
participants. Id. at 1094 (citing 15 U.S.C. § 78k-1); see
Regulation NMS, Release No. 51808, 70 Fed. Reg. 37,496 (June
29, 2005). The Congress specified five factors to guide the SEC
in establishing the NMS: the Commission is to assure (1)
efficiency, (2) fair competition, (3) availability of market data,
(4) practicability of order execution in the best market and (5)
the opportunity for an investor’s order to be executed directly,
that is, without the participation of a dealer. 15 U.S.C. § 78k-
1(a)(1)(C). Under SEC oversight during the next three decades,
the NMS grew to encompass the securities of over 5,000
companies with a collective U.S. market capitalization of more
than $14 trillion. Today NMS stocks are traded simultaneously
on one or more of the nine national exchanges that participate in
the system, including the New York Stock Exchange (NYSE),
the NASDAQ Stock Market (NASDAQ) and NYSE Arca, as
well as at non-exchange trading sites like (1) alternative trading
systems; (2) electronic communication networks (ECNs),
including the BATS Exchange (BATS ECN); and (3)
market-making securities dealers.
Because NMS stocks are traded so many places at once, one
of the important innovations of the NMS system is to make
available to investors a stream of “core” market data
consolidated from all of the exchanges. Regulation NMS, 70
Fed. Reg. at 37,503 (consolidated data stream “form[s] the heart
of the national market system” (quoting H.R. Rep. No. 94–229,
at 93 (1975), as reprinted in 1975 U.S.C.C.A.N. 321, 324
(Conference Report))). Core data for each NMS security
consists of three things: (1) last sale reports, which include the
price at which the latest sale of the security occurred, the size of
the sale and the exchange where it took place; (2) the current
highest bid and lowest offer for the security, along with the
number of shares available at those prices, at each exchange;
 5

and (3) the “national best bid and offer,” or NBBO, which are
the highest bid and lowest offer currently available in the
country and the exchange(s) where those prices are available.
To ensure that an investor can obtain core data for each
NMS security, the SEC requires each exchange, including
NYSE Arca, to report last sales (i.e., the price and size of the
most recent trade on NYSE Arca) and the current best bid and
offer to one of several central data processors for consolidation.
See 17 C.F.R. §§ 242.601, 242.602. Each central data processor
then disseminates the consolidated information, including the
NBBO, to broker-dealers and data vendors. See id.
§ 242.603(a)-(b). The SEC must approve the fees charged for
core data, see 17 C.F.R. § 242.608(b)(1), and a broker-dealer is
required to purchase the consolidated core data and make it
available to an investor seeking to place a trade,2 see id.
§ 242.603(c). In this way, today’s core market data system
provides investors in the U.S. equity markets with real-
time access to the best quotations and most recent
trades in the thousands of NMS stocks throughout the
trading day. For each stock, quotations and trades are
continuously collected from many different trading
centers and then disseminated to the public in a
consolidated stream of data. As a result, investors of all

2
The SEC has determined that because of the mandatory nature
of this regime, core data fees should bear some relationship to cost.
See Regulation of Market Information Fees and Revenues, Release
No. 34-42208, 64 Fed. Reg. 70,613, 70,627 (Dec. 17, 1999). To date,
however, the SEC has instead approved fees based on agreement
among market participants. See Concept Release Concerning
Self-Regulation, Release No. 34-50700, 69 Fed. Reg. 71256, 71272
(Dec. 8, 2004).
 6

types have access to a reliable source of information
for the best prices in NMS stocks.
Regulation NMS, 70 Fed. Reg. at 37,503.
All other market data is “non-core” data.3 This case
involves non-core data referred to as “depth-of-book” data.
Depth-of-book data consists of outstanding limit orders4 to buy
stock at prices lower than, or to sell stocks at prices higher than,
the best prices on each exchange.5 Because depth-of-book data
is non-core data, the SEC does not require that it be included in
the consolidated data stream or made available to an investor at
the time of trade execution.6

3
The terms “core” and “non-core” are SEC creations; the
Exchange Act makes no such distinction. See Order Setting Aside
Action by Delegated Authority and Approving Proposed Rule Change
Relating to NYSE Arca Data, Release No. 34-59039, 73 Fed. Reg.
74,770, 74,779 (Dec. 9, 2008) (Exchange Act and regulations “do not
differentiate between types of data and therefore apply to exchange
proposals to distribute both core data and non-core data”).
4
Generally speaking, a national securities exchange handles two
types of orders: a “market” order, which is an offer to immediately
trade at the prevailing market price on that exchange, and a “limit”
order, which is an offer to trade a certain amount of a security at a
specific price. A limit order that is not at a price that can be executed
is added to the exchange’s “order book” at the quoted price level.
5
If the current best bid or offer for a security in an exchange’s
order book is better than or equal to the best price available nationally,
it is necessarily reflected in the core data as the NBBO.
6
In fact, the SEC specifically rejected such a proposal in 2005 in
order to “allow market forces, rather than regulatory requirements, to
determine what, if any, additional quotations outside the NBBO are
displayed to investors.” Regulation NMS, 70 Fed. Reg. at 37,567
(“Investors who need . . . more comprehensive depth-of-book
 7

Depth-of-book data matters because of a concept called
“depth,” which refers to the number of shares of a security
available to trade at any given price point. If a trader wants to
buy a certain number of shares that exceeds the depth (volume
of shares available) at the best price, depth-of-book data will tell
him the number of shares available at prices inferior to the best
price.7 In this way, depth-of-book data allows a trader to gain
background information about the “liquidity” of a security on a
particular exchange, i.e., the degree to which his total sale or
purchase price will differ from what he would receive if the
entire trade were made at the prevailing best prices. For
instance, even a very large buy order for a security with high
liquidity on a certain exchange will trade at or close to the best
price while a similarly large order for a security with lower
liquidity on that exchange will cost more in toto due to the fewer
number of shares available at or near the best price.
A simplified example may help illustrate the concept of
liquidity and the utility of depth-of-book data. Assume an
investor wants to make an offer to sell 3,000 shares of company

information[] will be able to obtain such data from [securities] markets
or third party vendors.”). Nor does a broker-dealer need to purchase
depth-of-book data in order to meet its duty of best execution (which
requires it to exercise reasonable diligence to obtain favorable order
execution terms for customers) or any other statutory or regulatory
requirement.
7
In 2001, stock prices began to be measured in one-cent
increments instead of “sixteenths,” i.e., 1/16 of a dollar. Before 2001,
significant depth accumulated at the NBBO price because the
difference between the best price and the next inferior price was
1/16th, or 6.25 cents. With the initiation of decimalized trading,
however, the difference between those prices was reduced to only one
cent, which substantially decreased the depth at the best prices and
substantially increased the depth at the various one-cent price points
inferior to the best prices.
 8

XYZ. The best bid price reflected in the core data at NYSE
Arca for XYZ is 1,000 shares at $10. The investor knows he can
sell up to 1,000 shares at $10 but he does not know at what price
his remaining 2,000 shares will sell until after his order is
placed. This is where depth-of-book data comes in. Assume
further that NYSE Arca’s depth-of-book data tells the investor
that, apart from the best bid price from the core data, an
additional 1,000 shares of XYZ are available at the next price
level of $9.99 and yet another 1,000 shares at $9.98. Before he
places his order, then, he knows that his 3,000-share sale will
fetch $29,970 (the sum of 3,000 shares sold in three 1,000-share
blocks at $10, $9.99, or $9.98 each). Lower liquidity—i.e.,
fewer shares of XYZ available at or near $10—will result in a
lower total sale price. It is in this sense that depth-of-book data
“is useful primarily as background information on liquidity
outside the best prices.” See Order Setting Aside Action by
Delegated Authority and Approving Proposed Rule Change
Relating to NYSE Arca Data, Release No. 34-59039, 73 Fed.
Reg. 74,770, 74,792 (Dec. 9, 2008) (Order).8

8
This example is simplified because depth-of-book data reflects
only displayed orders; some brokers, however, request that a portion
of their orders be kept in reserve (i.e., not be displayed) in the
exchange’s order book. In addition, and due in part to the reserve
liquidity, most trades are made at the NBBO price or better, including
orders approaching 10,000 shares. See Order, 73 Fed. Reg. at 74,792
(at least 90% of NYSE Arca’s, NYSE’s and NASDAQ’s “executed
share volume of marketable orders were at prices equal to or better
than the NBBO in May 2008”). Non-professional (or retail) investors’
trades tend to be smaller and thus are especially likely to execute at the
NBBO price or better. See id. at 74,793 (“This undisplayed liquidity
enables retail investors to receive executions for most of their orders
at prices equal to or better than the NBBO, regardless of the displayed
size at the NBBO.”).
 9

In May 2006 NYSE Arca filed a proposed rule change with
the SEC pursuant to SEC Rule 19b-4. NYSE Arca proposed to
charge a fee for its depth-of-book data (which data it had
formerly made available at no cost)—in particular, ArcaBook,
“a compilation of all limit orders resident in the NYSE Arca
limit order book.” NYSE Arca Application at 7. It proposed to
charge a broker-dealer a $750 monthly fee for access to the
ArcaBook data feed. It also proposed to charge an additional
user fee of up to $30 for a professional subscriber and up to $10
for a non-professional subscriber per device displaying the
ArcaBook data.9 Finally, it proposed to cap the monthly fee
charged to a broker-dealer at $20,000 for 2006 with an
escalation provision for future years.
As noted earlier, NYSE Arca is required to show that a rule
change is consistent with the Exchange Act—that it provides
for, among other things, an equitable allocation of reasonable
fees among users and does not unnecessarily burden
competition. See 15 U.S.C. § 78f(b). NYSE Arca is also an
“exclusive processor” of securities information under the
Exchange Act because it distributes on its own behalf
information regarding its quotations and transactions. See
Order, 73 Fed. Reg. at 74,779 n.175; 15 U.S.C. § 78c(a)(22)(B)
(defining “exclusive processor”). This means that its
distribution of non-core market data also has to be done on “fair

9
A “non-professional subscriber” is an end user who is a natural
person and who is neither (1) registered or qualified with the SEC, the
Commodities Futures Trading Commission, a state securities agency
or a securities exchange or association, (2) engaged as an “investment
adviser” under the Investment Advisers Act of 1940, 15 U.S.C.
§§ 80b-1 et seq. nor (3) exempt from a securities association
registration requirement by virtue of being employed by a bank or
other exempt organization. See NYSE Arca Application at 22. All
others, including end users that are not natural persons, are
“professional subscriber[s].” See id.
 10

and reasonable” and “not unreasonably discriminatory” terms.
Id. § 78k-1(c)(1)(C)-(D); see 17 C.F.R. § 242.603(a) (same).
NYSE Arca sought to establish that its ArcaBook fees
conform to Exchange Act and regulatory requirements in two
ways. First, it “believe[d] that the proposed market data fees
would reflect an equitable allocation of its overall costs to users
of its facilities.” See NYSE Arca Application at 6-7. Second,
it suggested that its fees “are fair and reasonable because they
compare favorably to fees that other markets charge for similar
products.” Id. at 7. On the latter point, NYSE Arca noted that
at that time the NYSE charged professional subscribers $60 for
monthly access to its depth-of-book product, OpenBook, while
NASDAQ charged up to $76 for combined monthly access to its
two depth-of-book products, TotalView and OpenView (which
provide real-time depth-of-book data related to (1) NASDAQ-
and (2) NYSE- and American Stock Exchange-listed securities,
respectively). See id.
The SEC published the proposal for public comment in the
Federal Register on June 9, 2006. The SEC, through delegated
authority (at that time, its Division of Market Regulation),
approved the proposal on October 12, 2006. NetCoalition
petitioned the SEC for review of that order one month later. The
SEC granted review on December 27, 2006 and invited parties
or persons to file statements supporting or opposing the October
12 Order. After receiving twenty-five comments, the SEC
published a notice of a proposed “Draft Order” on June 4, 2008,
after which it received sixteen additional comments.
 11

Although the SEC set aside the October 12 order,10 it
approved the proposal in its own 114-page order dated
December 2, 2008 and published in the Federal Register seven
days later. See Order, 73 Fed. Reg. 74,770. Briefly, the Order
evaluates the NYSE Arca’s proposal under a “market-based
approach.” That approach has two parts. The SEC first asks
“whether the exchange was subject to significant competitive
forces in setting the terms of its proposal for non-core data,
including the level of any fees.” Order, 73 Fed. Reg. at 74,781.
If so, the SEC approves the proposal “unless it determines that
there is a substantial countervailing basis to find that the terms”
violate the Exchange Act or SEC rules. Id. If not, the exchange
must “provide a substantial basis, other than competitive forces,
in its proposed rule change demonstrating that the terms of the
proposal are equitable, fair, reasonable, and not unreasonably
discriminatory.” Id. The SEC approved NYSE Arca’s fees for
ArcaBook data on the grounds that (1) NYSE Arca is subject to
“[a]t least two broad types of significant competitive forces,” id.
at 74,782, and (2) there is no countervailing basis under the
Exchange Act to disapprove the proposal, see id. at 74,794.
Petitioners NetCoalition and SIFMA both timely petitioned for
review on January 30, 2009 and we subsequently consolidated
the petitions for review. See Order Granting Resp’t’s Mot. to
Consolidate, No. 09-1042 (D.C. Cir. Mar. 30, 2009). NYSE
Arca and NASDAQ intervened in support of the SEC.
II. Analysis
Our jurisdiction arises under 15 U.S.C. § 78y(a)(1) (“A
person aggrieved by a final order of the Commission . . . may

10
The Order does not specify why the SEC set aside its Division
of Market Regulation’s October 12, 2006 order other than to say that
“[t]he Petitioner and commenters raised a number of important issues
that the Commission believes it should address directly at this time.”
Order, 73 Fed. Reg. at 74,771.
 12

obtain review of the order in the United States Court of Appeals
. . . for the District of Columbia Circuit . . . .”). We review the
Order under the APA’s arbitrary and capricious standard, i.e.,
we “hold unlawful and set aside agency action” that is
“arbitrary, capricious, an abuse of discretion, or otherwise not in
accordance with law” or “unsupported by substantial evidence.”
5 U.S.C. § 706(2)(A), (E). Under the APA, the SEC is required
to “examine the relevant data and articulate a satisfactory
explanation for its action including a ‘rational connection
between the facts found and the choice made.’” Motor Vehicle
Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463
U.S. 29, 43 (1983) (quoting Burlington Truck Lines v. United
States, 371 U.S. 156, 168 (1962)). The SEC’s factual findings
are conclusive if supported by substantial evidence. See 15
U.S.C. § 78y(a)(4).
The petitioners challenge the Order on three grounds: first,
the SEC violated the Exchange Act by failing to engage in cost-
based ratemaking; second, the SEC arbitrarily rejected its earlier
cost-based approach to determine whether market data fees are
“fair and reasonable”; and third, the SEC’s conclusion that
NYSE Arca was subject to significant competitive forces that
constrained its ability to set fees for market data is not supported
by substantial evidence.
A. SEC’s “Market-Based” Approach Does Not
Contravene Exchange Act
The petitioners assert, first, that the SEC’s “market-based”
approach to evaluating the fairness and reasonableness of NYSE
Arca’s ArcaBook fees conflicts with the Exchange Act. As
noted earlier, the Congress has expressly delegated to the
Commission the power to determine, after notice and comment,
whether a proposed rule change is consistent with the Exchange
Act. 15 U.S.C. § 78s(b)(2); see supra at 3. Accordingly, we
review the agency’s interpretation of the Exchange Act under the
familiar two-step analysis set forth in Chevron, U.S.A. Inc. v.
 13

Natural Res. Def. Council, Inc., 467 U.S. 837, 842-43 (1984).
See Fin. Planning Ass’n v. SEC, 482 F.3d 481, 487 (D.C. Cir.
2007); cf. United States v. Mead Corp., 533 U.S. 218, 226-27
(2001) (“Chevron deference [appropriate] when it appears that
Congress delegated authority to the agency generally to make
rules carrying the force of law, and that the agency interpretation
claiming deference was promulgated in the exercise of that
authority.”).
Chevron step one applies if the “Congress has directly
spoken to the precise question at issue . . . [because] that is the
end of the matter; for the court, as well as the agency, must give
effect to the unambiguously expressed intent of Congress.”
Chevron, 467 U.S. at 842-43. If the statute is “silent or
ambiguous with respect to the specific issue,” we proceed to step
two, at which step we accept the agency’s interpretation of the
statute as long as it is reasonable. Id. at 843. We bear in mind
that “[i]t is irrelevant that this court might have reached a
different—or better—conclusion than the SEC.” Am. Equity Inv.
Life Ins. Co. v. SEC, — F.3d —, 2010 WL 2757499, at *7 (D.C.
Cir. July 12, 2010) (citing Nat’l Cable & Telecomm. Ass'n v.
Brand X Internet Servs., 545 U.S. 967, 980 (2005)). In
considering the permissibility of an agency’s construction of a
statute, we determine not only whether the agency’s
interpretation is reasonable but also whether the agency has acted
within its authority under the statute. See Goldstein v. SEC, 451
F.3d 873 (D.C. Cir. 2006) (“As always, the ‘words of the statute
should be read in context, the statute’s place in the overall
statutory scheme should be considered, and the problem
Congress sought to solve should be taken into account’ to
determine whether Congress has foreclosed the agency’s
interpretation.” (quoting PDK Labs. Inc. v. DEA, 362 F.3d 786,
796 (D.C. Cir. 2004)); see also Am. Bar Ass’n v. FTC, 430 F.3d
457, 468 (D.C. Cir. 2005) (rejecting agency suggestion “that
Chevron step two is implicated any time a statute does not
expressly negate the existence of a claimed administrative
 14

power”). Finally, a statute’s “general declaration of policy” does
not protect agency action that is otherwise inconsistent with the
congressional delegation of authority for “[a]gencies are . . .
‘bound, not only by the ultimate purposes Congress has selected,
but by the means it has deemed appropriate, and prescribed, for
the pursuit of those purposes.’” Colo. River Indian Tribes v.
Nat’l Indian Gaming Comm’n, 466 F.3d 134, 139 (D.C. Cir.
2006) (quoting MCI Telecomms. Corp. v. AT&T, 512 U.S. 218,
231 n. 4 (1994)).
Under the Exchange Act, the SEC has a duty to ensure that
NYSE Arca’s proposed fees for market data are, among other
things, “fair and reasonable.” 15 U.S.C. § 78k-1(c)(1)(C) (fees
must be “fair and reasonable” and not “unreasonably
discriminatory”); see 17 C.F.R. § 242.603(a) (same); 15 U.S.C.
§ 78(f)(b)(4) (exchange must also “provide for the equitable
allocation of reasonable dues, fees, and other charges among
. . . persons using its facilities”). The petitioners believe that the
SEC’s market-based approach is prohibited under the Exchange
Act because the Congress intended “fair and reasonable” to be
determined using a cost-based approach. The SEC counters that,
because it has statutorily-granted flexibility in evaluating market
data fees, its market-based approach is fully consistent with the
Exchange Act.
We agree with the SEC. None of the statutory language
answers the “precise question” whether the SEC is required to
evaluate non-core data fees under a cost-based approach.11

11
In contrast, another Exchange Act provision also enacted as part
of the 1975 Amendment requires that the fees charged as commissions
be “reasonable in relation to the costs of providing the service for
which such fees are charged.” See 15 U.S.C. § 78f(e)(1)(B) (emphasis
added). In other words, the Congress knew how to tie a fee’s
reasonableness to its underlying cost but declined to do so for non-
core data fees. See Catawba County, N.C. v. EPA, 571 F.3d 20, 36
 15

Chevron, 467 U.S. at 842-43. Because the Congress delegated
authority to the Commission to determine whether an SRO’s rule
change is consistent with the Exchange Act and because the
statute is silent as to whether a market-based approach is
consistent with the same, under Chevron step two we uphold the
SEC’s action if it is based on a “permissible construction of the
statute.” Id. at 843; see Sea-Land Serv., Inc. v. Dep’t of Transp.,
137 F.3d 640, 645 (D.C. Cir. 1998) (“[Chevron] deference comes
into play . . . only as a consequence of statutory ambiguity, and
then only if the reviewing court finds an implicit delegation of
authority to the agency”); cf. Michigan v. EPA, 268 F.3d 1075,
1082 (D.C. Cir. 2002) (“Mere ambiguity in a statute is not
evidence of congressional delegation of authority.”).
The petitioners rely on portions of the legislative history
suggesting the Commission was supposed to “assume a special
oversight and regulatory role” over exclusive processors by
treating them as public utilities, a role inconsistent with allowing
market forces to determine market data prices. S. Rep. No.
94–75, at 12 (1975), as reprinted in 1975 U.S.C.C.A.N. 179, 190
(Senate Report); see id. at 11, 1975 U.S.C.C.A.N. at 189 (“Any
exclusive processor is, in effect, a public utility, and thus it must
function in a manner which is absolutely neutral . . . .”);
Conference Report at 93, 1975 U.S.C.C.A.N. at 324 (“[W]here
a self-regulatory organization or organizations utilize an
exclusive processor, that processor takes on certain of the
characteristics of a public utility and should be regulated
accordingly.”). These statements, however, refer to an
“exclusive central processor for the composite [i.e., consolidated

(D.C. Cir. 2009) (“When interpreting statutes that govern agency
action, we have consistently recognized that a congressional mandate
in one section and silence in another often ‘suggests not a prohibition
but simply a decision not to mandate any solution in the second
context, i.e., to leave the question to agency discretion.’” (quoting
Cheney R.R. Co. v. ICC, 902 F.2d 66, 69 (D.C. Cir. 1990))).
 16

core data] tape or any other element of the national market
system,” not to an exchange acting as the processor of its
proprietary non-core data. Senate Report at 11, 1975
U.S.C.C.A.N. at 189 (emphases added); see also Conference
Report at 93, 1975 U.S.C.C.A.N. at 324. In fact, the legislative
history indicates that the Congress intended that the market
system “evolve through the interplay of competitive forces as
unnecessary regulatory restrictions are removed” and that the
SEC wield its regulatory power “in those situations where
competition may not be sufficient,” such as in the creation of a
“consolidated transactional reporting system.” Conference
Report at 92, 1975 U.S.C.C.A.N. at 323; see Senate Report at 12,
1975 U.S.C.C.A.N. at 190 (“[I]n situations in which natural
competitive forces cannot, for whatever reason, be relied upon,
the SEC must assume a special oversight and regulatory role.”).
This reading is consistent with the notion that the Congress did
not intend to take away the SEC’s authority to rely, at least in
some circumstances, on the market. See id. (“[A] fundamental
premise of the bill is that the initiative for the development of the
facilities of a national market system must come from private
interests and will depend upon the vigor of competition within
the securities industry as broadly defined.”).
Moreover, while “competition” is only one of five objectives
included in the 1975 Amendment, 15 U.S.C. § 78k-1(a)(1)(C),
the SEC’s market-based approach does not impermissibly
elevate competition above the other objectives. In its Order, the
SEC responded to the congressional desire that it rely “on
competition, whenever possible, in meeting its regulatory
responsibilities for overseeing the SROs and the national market
system.” 73 Fed. Reg. at 74,781; see id. at 74,771 (Congress
intended that “competitive forces should dictate the services and
practices that constitute the U.S. national market system for
trading equity securities”). Accordingly, the SEC took a more
traditional “regulatory approach,” id. at 74,781, to core data
(“where competition may not be sufficient,” Conference Report
 17

at 92, 1975 U.S.C.C.A.N. at 323) and a “market-based
approach,” with a greater reliance on competition, to non-core
data, see Order, 73 Fed. Reg. at 74,780-81 (SEC prefers
consolidation for core data but deconsolidation for non-core data
to “allow[] greater flexibility for market forces to determine data
products and fees”). We conclude the SEC’s interpretation—that
a market-based approach to evaluating whether NYSE Arca’s
non-core data fees are “fair and reasonable”—is a permissible
one.12
B. SEC Did Not Arbitrarily Reject Prior Cost-Based
Approach to Calculating Market Data Fees
The petitioners next contend that, in approving NYSE
Arca’s proposed fees, the SEC arbitrarily “rejected” its prior
cost-based approach. Under APA review, an agency may not
“depart from a prior policy sub silentio or simply disregard rules
that are still on the books.” FCC v. Fox Television Stations, Inc.,
129 S. Ct. 1800, 1811 (2009). An agency acts arbitrarily by
“fail[ing] adequately to justify departing from its own prior
interpretation” of a statute. Goldstein v. SEC, 451 F.3d 873, 883
(D.C. Cir. 2006).
The petitioners cite two examples of the cost-based
approach from which the SEC allegedly departed without
explanation but neither example—each of which is distinguished

12
We have similarly held that a “just and reasonable” standard did
not “preclude[] the FERC from relying upon market-based pricing.”
Elizabethtown Gas Co. v. FERC, 10 F.3d 866, 870 (D.C. Cir. 1993)
As long as “there is a competitive market the FERC may rely upon
market-based prices in lieu of cost-of-service regulation to assure a
‘just and reasonable’ result.” Id. (quoting Tejas Power Corp. v.
FERC, 908 F.2d 998, 1004 (D.C. Cir. 1990)); see also EarthLink, Inc.
v. FCC, 462 F.3d 1, 13 (D.C. Cir. 2006) (contrasting “cost-based”
pricing required under 47 U.S.C. § 271 with “the more relaxed ‘just
and reasonable’ standard of 47 U.S.C. §§ 201 and 202”).
 18

in the Order—fits the petitioners’ characterization. They first
cite several statements contained in a 1999 SEC “Market
Information Concept Release” proposing changes to, among
other things, the SEC’s then-regulation of market data fees. See
Regulation of Market Information Fees and Revenues, Release
No. 34-42208, 64 Fed. Reg. 70,613 (Dec. 17, 1999). The
Concept Release noted that “the fees charged by a monopolistic
provider of a service (such as the exclusive processors of market
information) need to be tied to some type of cost-based standard
in order to preclude excessive profits if fees are too high or
underfunding or subsidization if fees are too low” and therefore
“the total amount of market information revenues should remain
reasonably related to the cost of market information.” Id. at
70,627. But this statement (and the Concept Release in general)
addressed market data fees charged by a central exclusive
processor of consolidated core data. See 64 Fed. Reg. at 70,615
(discussing dissemination of “consolidated market information”
pursuant to plans that “govern all aspects of the arrangements for
disseminating market information,” including “fees that can be
charged”) (emphasis in original). The SEC did not intend by this
statement to require that fees for an exchange’s proprietary non-
core data must also be cost-based. While costs are “[o]ne
standard commonly used to evaluate the fairness and
reasonableness of fees, particularly those of a monopolistic
provider of a service,” the Congress “did not require the
Commission to undertake a . . . strictly cost-of-service (or
‘ratemaking’) approach to its review of market information fees
in every case.” 64 Fed. Reg. at 70,619; see id. (“Such an
inflexible standard, although unavoidable in some contexts, can
entail severe practical difficulties.”) (footnote omitted). Instead,
as the SEC recognized, the Congress gave it flexibility in
evaluating “fair and reasonable” market data fees.13 Id.

13
Moreover, the Concept Release merely stated the SEC’s
“belie[f] . . . that it may be possible to develop a more flexible, cost-
 19

The petitioners’ second example is no more persuasive. In
1984, the SEC stated that an “exclusive processor’s fees should
be based strictly on the expenses it incurs in providing the
information to vendors.” Order Announcing Commission
Findings, Modifying Interim Relief, and Instituting Proceedings,
Release No. 20874, 49 Fed. Reg. 17,640, 17,647 (Apr. 24, 1984)
(Instinet Order). The issue arose when the SEC rejected a
proposed fee to access market information because the SRO that
proposed the fee was not only charging market data vendors (like
Instinet) but was also directly competing with the vendors by
“providing enhanced information products to its own direct
subscribers.” Concept Release, 64 Fed. Reg. at 70,623. The
SEC emphasized that the cost-based approach was used due to
“the particular facts of the present case.” Instinet Order, 49 Fed.
Reg. at 17,646; see Order, 73 Fed. Reg. 74,787 n.254 (citing
Concept Release, 64 Fed. Reg. at 70,622 (“The Commission
repeatedly emphasized . . . that the scope of its decision was
limited to the particular competitive situation presented in the
proceedings.”)). Nor is the SEC’s interpretation of the Instinet
order a novel one, given that it had already distinguished the
order on its “peculiar competitive context” in the 1999 Market
Information Concept Release. 64 Fed. Reg. at 70,623.
In any event, more recent examples illustrate that the SEC’s
“market-based” approach is not an unexplained shift from its

based approach” that avoided cost-calculating difficulties and at the
same time maintained “a reasonable connection between the cost of
market information and the total amount of revenues derived from
market information fees.” 64 Fed. Reg. at 70,628. The Release, which
identified itself as a “request for comments,” did not purport to
suggest the “belief” as a rule. Id. at 70,613-14 (“[T]he Commission
has decided to invite public comment before taking further action.
This release is intended to assist the public . . . by identifying a variety
of specific issues . . . on which the Commission is particularly
interested in receiving comments.”).
 20

earlier cost-based pronouncements. In 2001, the SEC’s Advisory
Committee on Market Information made clear that “the extent
and manner of dissemination of deeper market information can
be determined by market forces—i.e., by the business decisions
of individual market participants—rather than by regulation.”
Report of the Advisory Committee on Market Information: A
Blueprint for Responsible Change, § B.2,
http://www.sec.gov/divisions/marketreg/marketinfo/finalreport
.htm (Sept. 14, 2001) (last visited July 22, 2010) (Advisory
Committee Report). Likewise, the SEC itself intended in
Regulation NMS that “market forces, rather than regulatory
requirements” play a role in determining the market data (other
than the NBBO) to be made available to investors and at what
cost. 70 Fed. Reg. at 37,567. In fact, in 2002 the SEC approved
NASDAQ’s proposed fees for its own depth-of-book product
without analyzing costs. See Order Approving Proposed Rule
Change, Release No. 34-46843, 2002 WL 31554080 (Nov. 18,
2002). We conclude that the SEC’s “market-based” approach
does not arbitrarily depart from prior practice.
C. SEC Failed to Show that NYSE Arca is Subject to
Significant Competitive Forces in Pricing ArcaBook
Although we uphold the SEC’s market-based approach
against the petitioners’ cost-based challenges, we do not mean to
say that a cost analysis is irrelevant. On the contrary, in a
competitive market, the price of a product is supposed to
approach its marginal cost, i.e., the seller’s cost of producing one
additional unit. See Tejas Power Corp., 908 F.2d at 1004 (“In a
competitive market, where neither buyer nor seller has
significant market power, it is rational . . . to infer that price is
close to marginal cost, such that the seller makes only a normal
return on its investment.”). Supracompetitive pricing may be
evidence of “monopoly,” or “market,” power. See United States
v. Microsoft Corp., 253 F.3d 34, 51 (D.C. Cir. 2001) (“Where
evidence indicates that a firm has in fact profitably [raised prices
 21

substantially above the competitive level], the existence of
monopoly power is clear.”). Thus, the costs of collecting and
distributing market data can indicate whether an exchange is
taking “excessive profits” or subsidizing its service with another
source of revenue, as the SEC has recognized. Concept Release,
64 Fed. Reg. at 70,627 (cost-based standard “preclude[s]
excessive profits if fees are too high or underfunding or
subsidization if fees are too low”); cf. MCI Telecomms. Corp. v.
FCC, 675 F.2d 408, 410 (D.C. Cir. 1982) (“A basic principle
used to ensure that rates are ‘just and reasonable’ is that rates are
determined on the basis of cost.”) (footnote omitted). Even
NYSE Arca’s proposal acknowledges that costs are relevant in
assessing the reasonableness of its fees. See NYSE Arca
Application at 6-7 (“NYSE Arca believes that the proposed
market data fees would reflect an equitable allocation of its
overall costs to users of its facilities.”); see also Comment Letter
of Mary Yeager, Corporate Secretary, NYSE Arca, Inc. at 13, 16
(Feb. 6, 2007) (in setting fee levels NYSE Arca considered “the
contribution that revenues accruing from Arca Book Fees would
make toward meeting the overall costs of NYSE Arca’s
operations” and noted “market data revenues compare favorably
to the markets’ cost of producing the data”).
Moreover, the risk that NYSE Arca could exercise market
power appears to be elevated in the pricing of its proprietary
non-core data. There is no dispute that NYSE Arca is the
“exclusive” provider of this data. While many exchanges sell
Google stock, only NYSE Arca offers access to the Google limit
orders included in its depth-of-book product, ArcaBook. As of
June 2008, NYSE Arca was one of the largest exchanges in the
nation in terms of shares traded. See Order, 73 Fed. Reg. at
74,783 tbl. 1 (NYSE Arca had third-highest share of volume in
U.S.-listed equities among national exchanges). Although the
SEC is correct that “[m]any businesses . . . are the exclusive
sources of their own products” without also having market
power, id. at 74,786, as recently as 2005 the SEC was
 22

“concerned that an SRO with a significant share of trading in
NMS stocks could exercise market power in setting fees for its
data,” including data “beyond the NBBO.” Regulation NMS, 70
Fed. Reg. at 37,559. The Order thus recognized that if an
“exchange could, in fact, exert monopoly power over its pricing
of non-core data, it obviously would be inappropriate for the
Commission to rely on non-existent competitive forces as a
basis for approving an exchange proposal.” 73 Fed. Reg. at
74,787.
The SEC concluded, however, that NYSE Arca could not
exercise market pricing power as to ArcaBook. But the
Commission did not require NYSE Arca to substantiate its
market data costs—not at the time of NYSE Arca’s fee
application, not before the SEC published the draft proposal and
not before the Order issued. Instead, the Commission noted
more than once the difficulty of cost calculation in determining
whether a fee is fair and reasonable, often by referring to earlier
orders. See, e.g., Order, 73 Fed. Reg. 74,794 (cost-based
approach entails “whole host of difficulties in calculating the
direct costs and common costs of market data—an endeavor that
the Commission discussed at length in 1999 and will not repeat
here”); id. (cost-based regulatory approach “would be
extraordinarily intrusive on competitive forces, as well as quite
costly and difficult to apply in practice”). At one point the SEC
seemed to suggest that it might allow NYSE Arca’s fees to be set
by competition simply because of the difficulty of cost-
calculating. See id. (“faced with the pragmatic challenge of
determining whether non-core market data fees are fair and
reasonable,” SEC “strongly believes that the current level of
competition in the U.S. equity markets provides a much more
useful basis to make this determination than a regulatory attempt
to measure market data costs”).
The petitioners maintain that the SEC’s failure to consider
NYSE Arca’s costs make its determination that NYSE Arca is
 23

subject to significant competitive forces arbitrary and capricious
because cost-collecting is in fact not difficult. The APA
“establishes a scheme of ‘reasoned decisionmaking.’ . . . Not
only must an agency’s decreed result be within the scope of its
lawful authority, but the process by which it reaches that result
must be logical and rational.” Allentown Mack Sales & Service,
Inc. v. NLRB, 522 U.S. 359, 374 (1998) (quoting State Farm, 463
U.S. at 52); see also United Techs. Corp. v. U.S. Dep't of Def.,
601 F.3d 557, 565 (D.C. Cir. 2010) (agency failed to provide
reasoned basis for conclusion because “[a] naked conclusion
. . . is not enough”). While “we have long held that agency
determinations based upon highly complex and technical matters
are entitled to great deference,” Domestic Sec., Inc. v. SEC, 333
F.3d 239, 248 (D.C. Cir. 2003) (internal quotation omitted), “we
do not defer to the agency’s conclusory or unsupported
suppositions.” McDonnell Douglas Corp. v. U.S. Dep’t of the
Air Force, 375 F.3d 1182, 1187 (D.C. Cir. 2004). In addition, an
agency may not shirk a statutory responsibility simply because
it may be difficult. See Chamber of Commerce of U.S. v. SEC,
412 F.3d 133, 143 (D.C. Cir. 2005) (difficulty in formulating
cost estimate does not relieve SEC of “statutory obligation to
determine as best it can the economic implications of the rule it
has proposed” because even “in face of uncertainty, agency must
exercise its expertise to make tough choices . . . and to hazard a
guess as to which is correct, even if . . . the estimate will be
imprecise”) (internal quotation and citation omitted).
But the SEC did not approve NYSE Arca’s proposed fees
based primarily on “difficulties in calculating the direct costs and
common costs of market data,.” Order, 73 Fed. Reg. at 74,794.
It did so instead based on its determination that consideration of
costs was unnecessary because of an alternative indicator of
competitiveness: NYSE Arca is subject, the SEC said, to “[a]t
least two broad types of significant competitive forces” in
pricing ArcaBook—namely, “(1) [its] compelling need to attract
order flow from market participants[] and (2) the availability to
 24

market participants of alternatives to purchasing the ArcaBook
data.” Id. at 74,782. That determination fails the APA’s
reasoned decisionmaking standard for another reason advanced
by the petitioners, as we now explain.
1. Order Flow Competition
We begin with the points on which the SEC and the
petitioners agree. No one disputes that competition for order
flow is “fierce.” Order, 73 Fed. Reg. at 74,782 (“Attracting
order flow is the core competitive concern of any equity
exchange—it is the ‘without which, not’ of an exchange’s
competitive success.”). As the SEC explained, “[i]n the U.S.
national market system, buyers and sellers of securities, and the
broker-dealers that act as their order-routing agents, have a wide
range of choices of where to route orders for execution”; “no
exchange can afford to take its market share percentages for
granted” because “no exchange possesses a monopoly,
regulatory or otherwise, in the execution of order flow from
broker dealers”; and therefore “NYSE Arca must compete
vigorously for order flow to maintain its share of trading
volume.” Id. at 74,782-83. The parties also agree that there is a
connection between order flow and market data, i.e., that “[a]n
exchange’s ability to attract order flow determines whether it has
market data to distribute.” Id. at 74,783.
The dispute centers on whether the connection works both
ways: not only that increased order flow makes market data more
valuable but that more modestly priced market data drives
increased order flow. The SEC concluded that it does: an
“exchange’s distribution of market data significantly affects its
ability to attract order flow.” Id. (“The wide distribution of . . .
depth-of-book order data[] to many market participants is an
important factor in attracting . . . orders.”); id. at 74,791 (“wide
dissemination” of market data is “important tool” to attract order
flow). The SEC contends that order flow competition constrains
NYSE Arca’s pricing of its depth-of-book data for two reasons.
 25

First, the wide dissemination of ArcaBook will attract interest in
the prices listed therein of those traders who wish to buy or sell
more shares than the number of shares shown in the core data.
See id. at 74,783 (“The extent to which a displayed [limit] order
attracts . . . interest will depend greatly on the wide distribution
of the displayed order to many market participants.”). Second,
because many traders who buy depth-of-book data from NYSE
Arca also select on which exchanges to trade—in particular, “the
large broker-dealer firms that control the handling of a large
volume of customer and proprietary order flow”—NYSE Arca
can risk its order flow share by overpricing market data. Id.
(“Given the portability of order flow from one trading venue to
another, any exchange that sought to charge unreasonably high
data fees would risk alienating many of the same customers on
whose orders it depends for competitive survival.”). For these
reasons, the SEC concluded, NYSE Arca’s proposal “cannot
reasonably be interpreted as that of a monopolist able to take
advantage of its market power over a small group of
professionals who value the data highly, but rather that of an
exchange facing significant competitive pressures in attempting
to sell its data to a large number of professionals.” Id. at 74,792.
The conclusion is not objectionable in theory. The problem
is that is it at odds with the SEC’s repeated statements
throughout the Order and in its briefs that depth-of-book data is
simply not very important to most traders, even professionals.
See, e.g., 73 Fed. Reg. at 74,791 n.295 (rejecting the notion that
“an exchange’s depth-of-book data is so critically important that
the exchange is not significantly constrained by competitive
forces in pricing that data”). This data, it appears, “is useful
primarily as background information on liquidity outside the best
prices” because “more than 90% of the time, traders do not
access the liquidity displayed in an exchange’s depth-of-book
order data, even for large orders.” Id. at 74,792 (92% of NYSE
Arca’s May 2008 “executed share volume of marketable orders
were at prices equal to or better than the NBBO” for orders
 26

ranging from 100 shares to 9,999). According to the
Commission, depth-of-book data is “most accurately
characterized as useful, but not essential, for professional
traders.” Id.; see Resp’t’s Br. 48 (“The availability of core data
makes resort to depth-of-book unnecessary for most investors
. . .”); id. at 50 (“The limited appeal of depth-of-book data is
reflected by the fact that of the 420,000 professional users who
purchase core data in NASDAQ-listed stocks, only 19,000
professional users purchase NASDAQ’s depth-of-book data
product.”); see also Intervenors’ Br. 7 (“Use of depth-of-book
data is uncommon even among professional investors . . . .”).
But if the SEC is correct that depth-of-book data is
“unnecessary” and of “limited appeal” to most investors, it is not
clear why “wide distribution” of the data “to many market
participants” is critical in attracting order flow. Order, 73 Fed.
Reg. at 74,783. If anything, the SEC’s order seems to suggest
that the depth-of-book tail is wagging the order flow dog—but
without a consistent explanation of how or why this is so.
This is not to say that wide dissemination of market data
cannot increase order flow but rather that it is not necessarily
so.14 More problematic is the lack of support in the record for

14
Perhaps one way to resolve the apparent contradiction between
asserting both the critical role of market data and at the same time its
limited appeal to most investors is to say that the data is important but
only to those “[i]nstitutional investors that need to trade in large size
[and] typically seek to assess market depth beyond the best prices,”
Order, 73 Fed. Reg. at 74,784, or to “algorithmic order routing
services” that use computer algorithms to determine the most
profitable trading sites, id. at 74,792. In other words, perhaps only a
minority of professional traders is interested in NYSE Arca’s depth-
of-book data but those few execute an outsized share of the total
trading volume so that unreasonable fees would cause them to place
their orders elsewhere and ultimately affect order flow. But if this is
the SEC’s theory, it is not supported by evidence of the traders using
 27

the SEC’s conclusion that order flow competition constrains
market data prices. Granted, the record includes statements from
NYSE Arca and other exchanges to support the conclusion. E.g.,
Order, 73 Fed. Reg. at 74,784 (“NYSE Arca, in setting the fee,
acknowledged that it needed to balance its desire for market data
revenues with the potential damage that a high fee would do to
its ability to attract order flow.”). The self-serving views of the
regulated entities, however, provide little support to establish
that significant competitive forces affect their pricing decisions.
Nor does the remaining evidence provide substantial support.
For example, the SEC quoted the 2001 Advisory Committee
Report’s statement that “a market’s inability to widely
disseminate its prices undoubtedly will adversely impact its
ability to attract limit orders and, ultimately, all order flow.” Id.
at 74,783 n.216 (quoting Advisory Committee Report
§ B.1). But this was a conclusion, not evidence. Moreover, it
was made in the context of “whether basic market information
should continue to be required to be provided in a consolidated
format to market participants,” see Advisory Committee Report
§ B—thus supporting the unremarkable proposition that failure
to disseminate core data could adversely affect order flow.15

NYSE Arca’s depth-of-book data or of the percentage of total trading
volume those traders may generate.
15
In fact, the Advisory Committee disagreed over whether
competitive forces would ensure that market data fees remain “fair and
reasonable.” While some Committee members were “concern[ed] that
the rise of publicly-held for-profit exchanges, and their obligation to
maximize shareholder value, will put upward pressure on market data
fees,” others argued that the exchanges’ inclination to increase data
fees “will be checked by the need to make data available to generate
order flow and attract listings.” Advisory Committee Report § D.3.d.
The Report did not resolve the disagreement but rather cautioned that
“the Commission may want to be more vigilant in assuring that a
 28

The SEC also cited two anecdotes to support its
determination that competition for order flow constrains non-
core market data fees. The SEC first described how Island ECN
“ceased displaying its order book to the public in three very
active exchange-traded funds” in order to avoid certain
regulatory strictures—and promptly lost fifty per cent of its
substantial market share in the three funds. Order, 73 Fed. Reg.
at 74,784. The SEC also cited the fact that BATS ECN and
International Stock Exchange (ISE) provided their depth-of-book
data to customers free of charge, ostensibly as a way to generate
order flow. Together, these examples show that depth-of-book
market data is apparently important enough to at least some
traders that it must be made available; they say nothing about
whether an exchange like NYSE Arca is constrained to price its
depth-of-book data competitively.16
2. Alternatives to Depth-of-Book Data
The SEC also asserts that an exchange must consider the
extent to which institutional and other “sophisticated traders
would choose one or more alternatives instead of purchasing the

for-profit SRO’s market data fees meet the statutory ‘fair and
reasonable’ standard.” Id.
16
The SEC points to a NASDAQ-commissioned economic study
described in the Order which employs the “total platform” theory
whereby market data and trade executions are “joint products” with
“joint costs” at each trading “platform,” or exchange. Resp’t’s Br. 43-
44; see Order, 73 Fed. Reg. at 74,779. Although an exchange may
price its trade execution fees higher and its market data fees lower (or
vice versa), because of “platform” competition the exchange
nonetheless receives the same return from the two “joint products” in
the aggregate. But this is not the theory of competition the SEC relied
on below and it may not press it for the first time on appeal. See
BNSF Ry. Co. v. Surface Transp. Bd., 604 F.3d 602, 613 (D.C. Cir.
2010) (citing SEC v. Chenery Corp., 332 U.S. 194, 196 (1947)).
 29

exchange’s data.” Order, 73 Fed. Reg. at 74,784. According to
the SEC, given the other sources (and potential sources) of
market data available to sophisticated investors—specifically, (1)
core data, (2) depth-of-book data from other exchanges, (3)
“pinging”orders and (4) the threat of independent distribution of
depth-of-book data by data vendors or securities firms acting in
concert—NYSE Arca must price ArcaBook reasonably or it will
lose these subscribers. Id. at 74,784-85; see id. at 74,791 n.295.
But the existence of a substitute does not necessarily
preclude market power. See 2B Phillip E. Areeda, Herbert
Hovenkamp & John L. Solow, Antitrust Law § 506(a) (3d ed.
2007) (Areeda); see also Ariz. Pub. Serv. Co. v. United States,
742 F.2d 644, 651 (D.C. Cir. 1984) (“[T]he mere existence of
some alternative does not in itself constrain the railroads from
charging rates far in excess of the just and reasonable rates that
Congress thought the existence of competitive pressures would
ensure.”). Rather, whether a market is competitive
notwithstanding potential alternatives depends on factors such as
the number of buyers who consider other products
interchangeable and at what prices. See Areeda § 506c. We
evaluate market competitiveness in antitrust cases, for example,
by asking whether there exists a “reasonably interchangeable”
substitute in the same market, which “depends not only on the
ease and speed with which customers can substitute it and the
desirability of doing so, . . . but also on the cost of substitution,
which depends most sensitively on the price of the products.”
FTC v. Whole Foods Mkt., Inc., 548 F.3d 1028, 1037 (D.C. Cir.
2008) (en banc) (citing United States v. Microsoft Corp., 253
F.3d 34, 53-54 (D.C. Cir. 2001) (per curiam)); see id. at 1037-38
(describing “small but significant non-transitory increase in
price,” or “SSNIP,” technique developed by the Department of
Justice and Federal Trade Commission to analyze reasonable
substitutability). The inquiry into whether a market for a product
is competitive, therefore, focuses on the customer and, in
particular, his price sensitivity—in economic terms, the
 30

product’s “elasticity of demand.” Areeda § 503 (“‘Elasticity’
here refers to the rate at which customers will turn away from the
firm’s product in response to a price increase or toward it in
response to a price decrease.”); see Order, 73 Fed. Reg. at 74,792
(“The availability of such alternatives increases the elasticity of
demand for an exchange’s depth-of-book data.”).
The SEC’s analysis of alternatives does not reveal the
number of potential users of the data or how they might react to
a change in price. We do not know, for example, how many
traders accessed NYSE Arca’s depth-of-book data during the
period it was offered without charge—and thus how many
traders might be interested in paying for ArcaBook. See Arg.
Tr. 32 (SEC: “[W]e don’t have numbers.”). Nor do we know
whether the traders who want depth-of-book data would decline
to purchase it if met with a supracompetitive price. The Order
does provide some idea of how many traders buy depth-of-book
data (albeit on a different exchange): of 420,000 professional
traders who subscribed to NASDAQ’s core data product as of
April 30, 2007, only 19,000 (less than 5%) purchased its depth-
of-book product.17 See Order, 73 Fed. Reg. at 74,785 (“[T]hat
95% of the professional users of core data choose not to purchase
the depth-of-book order data of a major exchange strongly

17
We also know that as of July 2008, about 15% of ISE
members—20 of 140— subscribe to its depth-of-book product even
though it is free. See Order, 73 Fed. Reg. at 74,779; Comment Letter
of Michael J. Simon, Secretary, International Stock Exchange, at 2
(July 10, 2008). Given that ISE’s share volume in U.S. listed stocks
is significantly smaller than that of NYSE Arca (.9% compared to
16.5% during June 2008), it is no surprise that its market data is less
in demand. See 73 Fed. Reg. at 74,784 (depth-of-book data provided
by exchanges and ECNs with greater trading volume “will be
proportionally more important in assessing market depth” and thus
“smaller exchanges may well be inclined to offer their data for no
charge or low fees as a means to attract order flow”).
 31

suggests that no exchange has monopoly pricing power for its
depth-of-book order data.”). But the fact that there are few
buyers does not by itself demonstrate a lack of market
power—which, after all, is “the ability to raise price profitably
by restricting output.” Areeda § 501(emphasis added).
Stated differently, evidence that few people buy the data
tells us little about whether the data is “critically important” to
those traders who do. And without additional evidence of trader
behavior, the SEC has not adequately supported its determination
that the alternatives it identifies in fact constrain NYSE Arca’s
depth-of-book fees. Order, 73 Fed. Reg. at 74,784-85. Core
data, for example, reveals only the best prices available; it is not
a “substitute” for depth-of-book data, which measures the
number of shares available at prices inferior to the best prices.
Depth-of-book data from other exchanges could be an alternative
for individual securities but that determination cannot be made
without knowing how actively the security is trade on those
exchanges. See id. at 74,785 (depth-of-book data provided by
exchanges and ECNs with greater trading volume “will be
proportionally more important in assessing market depth”).
Likewise, a “pinging” order, which involves placing a limit order
for a number of shares larger than that included in the core data,
reveals the number of shares available at the order price or better
and therefore does provide “background information on
liquidity.” Id. at 74,792. But a pinging order also executes a
trade for both displayed and undisplayed orders at (and superior
to) the “pinging” price, meaning that it is not an obvious
alternative, at least without evidence that traders in fact use it in
lieu of depth-of-book data. Finally, even if we assume that the
“threat of independent distribution of order data by securities
firms and data vendors” is not unduly speculative, id. at 74,785,
the SEC’s duty is to ensure that fees are “fair and
reasonable”—not to predict that, with the entry of a competitor,
they might someday get there. Cf. Microsoft Corp., 253 F.3d at
53–54 (“The test of reasonable interchangeability . . . consider[s]
 32

only substitutes that constrain pricing in the reasonably
foreseeable future, and only products that can enter the market
in a relatively short time can perform this function.”). In sum, the
SEC had insufficient evidence before it to conclude that a trader
interested in depth-of-book data would substitute any of the four
alternatives (or simply do without) instead of paying a
supracompetitive price.
***
For the foregoing reasons, the petitions for review are
granted—we are unable to perform our APA review on the
record before us in order to affirm the SEC’s determination that
NYSE Arca’s ArcaBook fees are “fair and reasonable” and
otherwise compliant with the Exchange Act. In addition, the
SEC has failed to “disclose a reasoned basis,” see Am. Equity
Inv. Life Ins. Co., 2010 WL 2757499, at *10, for concluding that
NYSE Arca is subject to significant competitive forces in pricing
ArcaBook. The Commission’s Order Setting Aside Action by
Delegated Authority and Approving Proposed Rule Change
Relating to NYSE Arca Data, Release No. 34-59039, 73 Fed.
Reg. 74,770 (Dec. 9, 2008), is therefore vacated and remanded
to the Commission for further proceedings consistent with this
opinion. See Am. Equity Inv. Life Ins. Co., 2010 WL 2757499,
at *12 (decision to vacate depends on seriousness of order’s
deficiencies and disruptiveness of vacatur (citing Allied-Signal,
Inc. v. U.S. Nuclear Regulatory Comm’n, 988 F.2d 146, 150–51
(1993))).
So ordered.