This is part 2/3 of a series of blogs on the history of the Market Data Wars. Part 1 is here. Feedback and suggestions more than welcome at benjamin.connault@iextrading.com.
Part 1 of this blog series took us from December 1999 to January 2011 and described the beginnings of the Market Data Wars.
Today’s Part 2 covers January 2011 to March 2015, focusing on the “ArcaBook II” series of market data court cases. This series of cases extends to the June 5, 2020, decision by the U.S. Court of Appeals for the D.C. Circuit (“D.C. Circuit”), for which I provide some brief background (full disclosure: IEX wrote an amicus brief in support of the SEC in this case).
Part 3 will cover March 2015 to January 2020 and focuses on the rise of “platform theory” arguments in the Market Data Wars. Each Part contains multiple sections; each section revolves around one expert report. Details and link to each expert report are provided at the beginning of each section.
5) 2010–2015: NYSE files ArcaBook II, SIFMA challenges the filing on “limitation of access” grounds
Ordover (with Compass Lexecon) for Nasdaq, January 2015. Available here (p.38).
We pick up the ArcaBook thread again. A lot happened between 2008, where we left off in Part 1, and January 2015. The next few paragraphs provide the procedural history for that period, which sets the background for the Ordover report in in January 2015.
In January 2009, shortly after the SEC reaffirmed its initial decision to approve SR-NYSEArca-2006–21 (“ArcaBook I”), NetCoalition and SIFMA challenged that decision by filing a petition in the D.C. Circuit. In August 2010, the D.C. Circuit found that the SEC’s market-based framework was valid, but that its application to ArcaBook I was incorrect, meaning that there was not enough evidence that the ArcaBook I fee was constrained by competition.
But there was another factor at play — one month prior to the D.C. Circuit decision, in July 2010, Dodd-Frank had been passed.
A subtle but very impactful change was made to the exchange rule filing process as part of a few lines in Dodd-Frank: exchange fee filings became effective upon filing and no longer required affirmative SEC approval. The SEC would now have 60 days to affirmatively suspend such filings. By default, in the absence of SEC actions, the filings would be “approved.”
Immediately after Dodd-Frank was passed, on November 1, 2010, NYSE proceeded to refile the ArcaBook I fee under the immediately effective SR-NYSEArca-2010–97 (“ArcaBook II”). The SEC took no action and the fee was “approved.” In sum, the effective-upon-filing Dodd-Frank amendment allowed NYSE to sidestep the D.C. Circuit’s finding on ArcaBook I.
In December 2010, SIFMA petitioned the D.C. Circuit for a review of ArcaBook II, but because of the “effective upon filing” change, the D.C. Circuit found in April 2013 that it had no SEC action to review and thus lacked jurisdiction to rule on the challenge. In a fateful sentence, however, the D.C. Circuit left open the door for SIFMA to bring a claim using a certain procedure known as a “limitation of access” or “section 19(d)” application: “we take the Commission at its word, to wit, that it will make the section 19(d) process available to parties seeking review of unreasonable fees charged for market data.” On May 31, 2013, SIFMA brought such a claim in front of the SEC.[1] The January 2015 Ordover report is Attachment 1 to Nasdaq’s Prehearing Brief in that case.
(19(d) was an unprecedented avenue to challenge market data fees. It is beyond the scope of this blog series to describe these procedural aspects in detail, but they became a highly critical part of the ArcaBook II cases. On June 5, 2020, the same D.C. Circuit Court found that 19(d) was not a proper ground to challenge market data fees after all.)
To summarize, the January 2015 Ordover report marks the next step in the series of ArcaBook court cases — already nine years after introduction of the original fee. The report itself is unremarkable. It is mostly a reiteration of the previous Nasdaq arguments. Prices are said to be constrained by “competition from alternative depth-of-book products” and “competition for order flow from other trading platforms.” Nasdaq is described as producing “joint products” and incurring “joint costs.”
6) January 2015: NYSE defends ArcaBook II in a report that was not made public
Hendershott and Nevo for NYSE, January 2015. Not publicly available.
Like Nasdaq, NYSE commissioned an economic expert report to support its brief in the ArcaBook II case in front of the SEC. As opposed to all the other reports covered in this blog post and for reasons unknown to me, the Hendershott-Nevo report does not appear to be publicly available.
Other filings that cite the report provide an indirect view into its content. Unsurprisingly, the report argues that “(1) competition for order flow and (2) competition for depth-of-book data products both impose significant competitive constraints on NYSE Arca’s pricing of ArcaBook.” Some of the new arguments include the fact that “[a]fter ArcaBook became fee-liable in 2009, trade volume on NYSE Area decreased” and “all buyers do not need to purchase depth-of-book data products from all significant trading venues.”
Without access to the report, it is hard to evaluate the merit of the arguments. Ultimately, in an important 2018 opinion (which will be discussed in Part 3), the SEC found that analyses in the Hendershott-Nevo report “contain flaws that limit their persuasiveness with respect to Nasdaq’s assertion” and that their “failure to conduct apples-to-apples comparisons of trading on traditional exchanges also significantly undermines the persuasive value of their analysis.”
7) March 2015: SIFMA argues that the ArcaBook II fees are not constrained by competition and generally fall short of the Securities Exchange Act standards
Evans (with Global Economics Group) for SIFMA, March 2015. Available* here.
On the other side of NYSE and Nasdaq in the ArcaBook II case, SIFMA submitted their own expert report. As expected, the Evans report argues that “depth-of-book products are not substitutes for one another” and that “competition for order flow does not significantly constrain the pricing of depth-of-book data.” Because SIFMA’s brief came after Nasdaq’s and NYSE’s briefs, the Evans report also rebuts the January 2015 Nasdaq and NYSE expert reports. It argues for example that “Hendershott and Nevo’s data show that demand is highly inelastic and that NYSE Arca lost few customers following a massive price increase."
Those arguments are not surprising. However, the Evans report makes four important additional points. First, it argues that “cost and margin data are relevant to whether the exchanges’ depth-of-book data fees are significantly constrained by competition.” NYSE and Nasdaq have been and remain very reluctant to provide any data regarding the costs they incur in producing and distributing their market data products. This is despite the fact that both the SEC and the D.C. Circuit indicated that cost data would be useful, eg. “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” (from the D.C. Circuit’s 2010 decision in the first ArcaBook case). Second, the report makes the clear conceptual distinction between “multi-product platforms” and “multi-sided platforms.” By 2015, a novel theory known as (multi-sided) platform theory was getting a lot of traction in economic and legal circles. Multi-sided platforms are a different concept from the multi-product platforms of the 2008 Ordover-Bamberger report (see Part 1), which has created a lot of confusion in the exchange industry. 2 Unfortunately, the clarity of the Evans report was not enough to entirely dissipate the confusion, which continues to this day. Third, the Evans report correctly stresses that public policy goals around market data, as captured in the Securities Exchange Act, go above and beyond competitivity of the market for market data. “[T]he Exchange Act seeks to ensure that data are widely disseminated to increase market efficiency and transparency. Increasing depth-of-book data prices significantly above cost to cross-subsidize other exchange products is not consistent with this policy” and “[e]ven if competition fully dissipated the profits of the exchange overall […] this result would be inconsistent with public policy designed to ensure the wide availability of market data.”
In a 2019 Staff Guidance, the staff of the SEC’s Division of Trading and Markets expressed exactly the same principle: “If significant competitive forces constrain the fee at issue, fee levels will be presumed to be fair and reasonable, and the inquiry is whether there is a substantial countervailing basis to find that the fee terms nevertheless fail to meet an applicable requirement of the Exchange Act.”
Fourth, the report recognizes the importance of latency, as opposed to focusing on data content only, arguing that “users can obtain the data […] through the Exchanges’ direct feeds faster than through the consolidated data feed.” In the world of high frequency trading, the low latency of direct market data products is arguably the primary source of market power.
[1] In fact, SIFMA brought two claims: oneagainst SR-NYSEArca-2010–97, and anotheragainst twenty-three other market data fee filings, including a certain SR-NASDAQ-2010–110 from Nasdaq. SR-NASDAQ-2010–110 was consolidated with SR-NYSEArca-2010–97.
[2] Multi-product platforms are firms that sell multiple products and may incur common production costs across these products. Multi-sided platforms are firms that facilitate the interaction of two or more distinct groups of customers who need each other in some way. These two theoretical economic models of a firm are very different. They can lead to very different conclusions when used to analyze issues of costs, prices or market power. The academic literature typically reserves the term “platform” to refer to “multi-sided platforms.” The “multi-product platform” terminology of the Market Data Wars is unusual — although the concept of a multi-product firm is not. In a real-world analysis, the question becomes: which of the two concepts, if any, is the most appropriate to capture the economic structure under scrutiny?