On 4 February, Market Structure Partners published a report on market data pricing, covering five European exchange groups, including Euronext. The report was commissioned by AFME, FIA EPTA, Plato Partnership, EFAMA and BVI.
The report contains a number of factual errors and inaccurate representations. Euronext regrets that although its name is continuously cited throughout the report, Euronext was not contacted in advance by Market Structure Partners or by the commissioning associations about the elements included.
Euronext covers below what it believes to be some of the most pertinent inaccuracies in the Market Structure Partners report. Given the number of errors in the report, the absence of comment from Euronext on other areas of the report does not constitute endorsement.
- The report’s pivotal claim that exchanges have offset declining equity trading revenues by increasing market data prices is factually incorrect. The report claims that the share of market data revenues over the total revenues of Euronext has increased from 11% to 19%, when in reality (and based on data publicly available on Euronext website) this ratio remained stable over the period at 11%.
- The report does not accurately represent market data pricing, notably omitting to account for Euronext’s acquisition of three new markets. For example, the report claims that non-professional user display market data pricing increased by +75% on Euronext, whilst in reality, accounting for the new markets integrated within Euronext, it has decreased by 40%.
- The report’s claim that Euronext underinvests in its equity markets is again false. This is demonstrated by the data given in the report for Euronext, which shows that Euronext has increased its spend on infrastructure by more than 300% between 2017 and 2023. This is also demonstrated by the numerous well-documented investments made by Euronext in recent years to upgrade its equity markets, including the upgrade of its Optiq trading platform, the migration of its data centre to a new green facility and its clearing migration in 2024. Euronext’s parallel diversification into new asset classes and services has had no impact on its continued investments in equity markets.
1. The report’s pivotal argument is based on incorrect data. The report claims that the share of market data revenue in total Euronext revenue has increased, whilst in fact it remained stable
The report’s cornerstone argument is that exchanges rely increasingly on market data revenues to offset falling equity trading revenues. It uses what it refers to as market data revenue ratio to evaluate the share of market data revenues over the total revenues of exchanges. According to the report, Euronext’s market data ratio increased from 11% to 19% between 2020 and 2023 (pages 2, 4, 67 and 69). This result is incorrect. In reality, and using accurate and public data, this ratio has remained stable at 11% over the period. Data can be found here: Real Time Market Data Contracts | Connect, Financial reports | euronext.com.
2. The report does not represent Euronext’s market data pricing accurately, notably omitting Euronext’s acquisition of three new markets over the period evaluated and focusing only on selected examples
The report contains numerous incorrect representations of Euronext market data pricing. These include the following non-exhaustive list.
Not factoring in Euronext’s acquisition of three new markets
The report states that non-professional fees have increased by +75% on Euronext between 2017 and 2024 (pages 7, 39, 60), omitting Euronext’s acquisition of new markets over this period. The report does account for the fees charged by Euronext across the markets it operates in 2024, including Dublin and Oslo, but omits to include the fees that were charged in Dublin and Oslo in 2017, before their acquisition by Euronext. In reality, when Dublin and Oslo fees are factored in throughout the period, non-professional display user fees have decreased by 40%.
Similar omissions are made in numerous instances in the report, including in reference to Index/Benchmark creation fees (page 56), Euronext’s market data revenue increase (page 69), and Euronext’s price list (pages 5, 41, 43 and 69).
Incorrect representation of Euronext’s customer base evolution
The report states that exchanges have a declining customer base (pages 2, 3, 62, 72). This claim is inaccurate and solely based on assumptions. In reality, over the period 2017-2024 used for the report and corrected for the acquisitions of Dublin, Oslo and Milan, Euronext has seen a substantial increase in its customer base. The loss of clients due to M&A activity in the market has been more than offset by new client entrants, including new execution venues, retail brokers, redistributors and other client segments.
Erroneous comparison of professional display market data and inaccurate representation of the Natural User concept
The report states that professional display market data fees for level 2 data have increased by 42.53% on Euronext between 2017 and 2024 (pages 7, 39, 41, 43 and 59). In reality, over this period, professional equity display data fees have increased by 23% on Euronext (less than the accrued inflation in the relevant EU countries). The reports calculation of 42.53% reflects a lack of understanding of the Natural User concept. The report states that the Natural User concept results in a fee increase, whereas in fact it is an opportunity for clients to reduce fees for display use by netting users who take data from multiple sources to a single user.[1]
Unrealistic comparison of display and non-display market data
In Figure 14 on page 45, the report draws a conclusion by comparing the annual costs in 2017 of level 2 display data used by a single person, against the costs in 2024 of having machines using Level 2 data for non-display usage such as for principal trading, broker activity and to run a trading platform. Level 2 display data and non-display data are two fundamentally distinct types of products. Moreover, Euronext considers that it is not possible for a single human to perform the same activity as multiple machines.
Unrepresentative examples
The avatar example for a retail broker as provided in the report in paragraph 6.3 (pages 58-62) makes use of an unusual combination of parameters that only applies to 1% of Euronext client population. In its avatar section, the report exclusively focuses on the most expensive pricing options and omits to present more cost-effective pricing options which significantly decreased fees for small and mid-size clients between 2017-2024.
3. The report incorrectly represents market data costs and the investments made by Euronext in its markets, claiming that Euronext underinvests in its equity markets when data and facts demonstrate the opposite
The report states that there is no cost to exchanges for producing data and that there was no significant expenditure by exchanges on equity markets in recent years, including Euronext (except for the acquisition of Borsa Italiana). It thereby concludes that exchanges’ market data fees bear no relation to the costs of production of market data (pages 2, 8, 9, 10, 21, 22, 34, 41, 76, 77, 81, 82, 83). This claim is factually inaccurate and unfounded.
Market data is a core product of exchanges’ operations, IT infrastructure and investments. Market data is not a marginal product derived from trading services. The costs incurred in producing market data are therefore shared with trading services.
In addition, as demonstrated by the data presented in the report and yet ignored in the report’s narrative, Euronext continuously invests in its equity markets. Page 87 shows a 300% increase in Euronext’s spend on infrastructure between 2017 and 2023. This is illustrated by the numerous well-documented investments made by Euronext in recent years to upgrade its equity markets, including the upgrade of its Optiq trading platform, the migration of its data centre to a new green facility, and its clearing migration in 2024. This is also reflected by the absence of severity 1 outages on Euronext markets in the last four years; the list of outages presented on page 82 of the report as “some examples” of exchanges’ outages is in fact an exhaustive list as far as Euronext is concerned. Euronext’s parallel diversification into new asset classes and services has had no impact on its continued investments in equity markets. A full list of the main upgrades and innovations delivered by Euronext over the last ten+ years can be found here: Our Journey | euronext.com.
Further, Euronext is required by MiFIR to charge market data based on costs. In contrast to the report’s claim, Euronext complies extensively with this requirement and provides extensive disclosures on the topic. Data can be found here: Real Time Market Data Contracts | Connect
Conclusion
The Market Structure Partners report on market data is flawed, with many factual errors and inaccurate representations. It should be approached with caution. The question of market data merits rigorous research: we hope that the above elements can be viewed as insightful observations.
Euronext is a core market infrastructure operator in European equity markets and is a core supporter of market transparency. Euronext operates in full compliance with the relevant regulations regarding both market data production and publication, and market data pricing, and is committed to the highest standard of integrity and transparency on the topic. As continuously illustrated throughout its history, Euronext is committed to nurturing European equity markets through technological excellence and continuous innovation, including in the area of market data.
[1] The Natural User unit of count was introduced by ESMA as part of MiFID II in 2018. Clients may choose to adopt Natural User as a unit of count for display data usage, but there is no requirement for clients to opt for Natural User. A client wanting to adopt Natural User has to ensure they can manage their market data inventory to an individual user level. They are then able to net a user with multiple sources of the same data, i.e. they can take data from one or more data vendors, but report and pay for the user just once. This is an alternative to other units of count e.g. user per source where a user pays once per source of data.