[ICE, NDAQ, CME, CBOE] Exchange Operators: Fragmentation and New Points of Differentiation

Intercontinental Exchange (ICE)

Intercontinental’s multifaceted data and exchange platform began as an electricity marketplace, founded in 2000 by current CEO Jeffrey Sprecher with the backing of two investment banks and seeded by volume from energy majors Shell, Total, and British Petroleum (the collapse of Enron, the dominant electricity broker at the time, fortuitously diverted a flood of trading volume to ICE as well).  Through its acquisition of International Petroleum Exchange the following year, ICE picked up the Brent contract (a component of the Brent complex against which 2/3 of crude futures volumes are benchmarked) and expanded into oil futures trading in direct competition with the New York Mercantile Exchange (which would be acquired by CME in 2008).  

Since going public in November 2005, the company has acquired its way into various non-energy domains, including soft commodities (NY Board of Trade, $1.1bn; 2007)¹; credit default swaps (Creditex, $625mn; 2008)²; exchange-traded emissions (Climate Exchange, $606mn; 2010); and, most notably, equities.  Through its industry-quaking acquisition of NYSE Euronext in November 2013³, an $11bn cross-border deal ($2.7bn in cash and 42.4mn shares representing ~37% of the company; 13x/8x pre/post-synergy EBIT), ICE more than doubled its operating profits, shifted its revenue mix towards data and listing fees, and brought cash equities trading into its stable of exchange operations.  Here is a before (2012) and after (2014) decomposition of revenue:

The ICE/NYSE merger was part of a frenzy of deal activity in the exchange space during the mid/late-2000s⁴, most of them motivated by the same industrial logic:

1/ Eliminating redundant technology, trading platforms, and corporate overhead.  The high fixed costs of exchange operations make them highly scalable.  In a typical merger, 25%-40% of the target’s cost structure might get hacked within a few years.  The cost cutting opportunity is especially rich for a technology-focused operator like ICE, who did not mature in the opulent, stained oak halls corseting the pens of the flesh and blood exchanges.  ICE has, without compunction and to the great consternation of floor traders, replaced open-outcry with electronic market making systems.

2/ Bolstering scale, especially in light of intensifying fragmentation.  Exchanges benefit from network effects…more buyers and sellers --> deeper liquidity --> lower transaction costs and better price discoverability —> more buyers and sellers. 

Just under half of ICE’s total revenue, net of transaction expenses, comes from exchange transactions and clearing, broken down across asset types like this: 

While only 17% of net transaction revenue, “cash equities and equity options” punches above its weight, driving listing revenue – NYSE remains a marquee listing venue for companies going public and NYSE Arca is the flagship listing venue for ETF listings⁵ – and data fees. 

Equity trading venues have multiplied over the last 20 years.  In the late ’1990s/early 2000s, advances in technology coupled with accommodative regulation gave rise to a host of non-exchange Alternative Trading Systems (ATS) and dark pools who didn’t provide the listing functions of a regular exchange and so weren’t required to publicly disclose their trading operations.  They competed fiercely for order flow, offering fees and rebates to liquidity providers as part of a “maker-taker” payment model that was later adopted by exchanges and became ubiquitous by the mid-2000s.  Non-exchange share of trading in NYSE listed stocks has gone from 17% in 1990 to 31% in 2015, per the OECD⁶.  Major US exchange operators have been losing share since the mid-2000s, as this SEC memo elaborates: 

“…lit venues [as opposed to “dark pools”] have been losing market share for a variety of reasons to non-exchange dark venues that do not display quotes or orders. From February 2005 to February 2014, the share of volume executed by displayed venues declined from 70.6% to 61.4% for NASDAQ stocks and from 87.0% to 65.4% for NYSE stocks. During the same period, the collective share of dark venue trading in NASDAQ stocks increased from 29% to 39%, and the collective share of dark venue trading in NYSE stocks increased from 13% to 35%”.

Source: Bloomberg

Source: Wall Street Journal

In response to the proliferating number of trading venues, stock exchanges consolidated.  There are 13 approved stock exchanges in the US today. All but one, IEX, are owned by one of three publicly traded companies: Intercontinental, Nasdaq, and CBOE.  But there are many other trading venues outside these exchanges, testament to lower entry barriers brought on by electronic trading.  Of course, an unintended consequences of pro-competition regulation, which was supposed to reduce exchange fees, is that exchange operators began charging connectivity and market data fees – fees that are the subject of so much controversy today – to traders seeking price discovery in an increasingly fragmented landscape.  But the point remains that equity exchanges, on their own, are a less good business than they were a decade ago.

Relative to equities, futures trading in asset classes like energy, commodities, and rates have proven more immune from competition, grounded as it is in extensively referenced benchmarks – including well-known ones like Brent, US Dollar Index (USDX), ICE swap rate, and LIBOR – to which ICE owns the underlying IP and against which many $trillions of assets are managed and hedged⁷. 

Energy trading is dominated by CME and ICE.  A few competitors have tried to challenge the duopoly.  In early 2015, NASDAQ OMX predicted that by dramatically undercutting transaction prices on a small sliver of CME’s and ICE’s energy complex, its futures and options exchange (“NFX”) would grab ~10% of market within a few years.  But after coming out the gate to great fanfare in July 2015, interest in NFX seems to have petered out.  An oil price benchmark launched by Shanghai International Energy Exchange (INE) has seen more promising uptake, grabbing an impressive ~6% spot market share within its first year of launch.  Still, unlike Brent and WTI, which are traded by market participants with real commercial interests, INE’s futures are traded for speculative purposes by Chinese retail traders and suffer from jerky trading volumes.  It doesn’t appear that this new benchmark functions as an efficient hedging tool for global oil producers and shippers, at least not yet.

Banks, brokers, and energy merchants regularly consult ICE’s commodity, rates, and FX benchmarks and trade the futures underlying them, in turn further reinforcing their benchmark status and attracting still more transaction activity on ICE’s exchanges. A benchmark can be monetized multiple times.  Moody’s makes money not only by charging issuers for ratings, but also by charging investors for the research backing those ratings⁸.  Likewise, Intercontinental makes money not just from transactions on its exchanges, but also on the data spilling off them.  

At first, ICE’s Data Services segment (~45% of revenue and EBITA) mostly disseminated the data generated on ICE’s exchanges to third party data vendors and directly to end users who, rather than wait for delayed aggregated data from wholesalers, paid for direct data feeds that powered their custom applications and risk models.  But through acquisition, ICE has expanded into Pricing and Analytics [bond and futures valuation, risk analytics, regulatory compliance solutions, index design – i.e. the benchmarks I discussed earlier], Desktops [ICE’s version of Bloomberg, basically, where “10s of thousands” of users trade, chat, and manage risk with analytic tools], and Connectivity [secure and low latency connectivity to exchanges and clearinghouses].

2015 was a pivotal year in this journey.  Around that time, European regulators were crafting new regulations under MiFID II, aimed at expanding the reach and tightening the constraints imposed by MiFID I with nebulous goals related to pricing transparency and consumer protection.  This ambitious, unwieldy piece of legislation went into effect early last year and has far reaching implications across the financial landscape, but there are two that seem relevant here.

First, the regulation imposes onerous record keeping and reporting requirements on their trades, including price and volume information, that demonstrate best execution.  Second, it expands the number of regulated platforms authorized to host commodity derivatives trading, potentially fragmenting the markets and obscuring price discovery⁹.  ICE seized on the prospect of heightened demand for data and analytics catalyzed by this regulation and in December 2015, bought two companies that more than doubled its data services revenue, transforming a data business that was captive to its exchange operations to one that stands firmly on its own, carving out unique avenues of value creation.

ICE’s major foray into this space was its $5.2bn (5.5x revenue / 10x post-synergy EBITDA) purchase of Interactive Data Corp (IDC), which it “won” by outbidding rival suitors Nasdaq and Markit. Whether by through broad distribution or better execution, ICE has double IDC’s standalone ~3ish% growth rate.  At the time of acquisition, around 70% of IDC’s revenue came from valuing thinly traded fixed-income securities and selling this data to portfolio managers and risk managers at banks and asset managers (with 98% annual retention)¹⁰.  Fixed income securities are far more complicated and varied than equities.  There is only one AIG common stock, but a thicket of AIG fixed income SKUs, some rather arcane, with different maturities, covenants, coupons, and liquidity profiles. 

During the mid/late 2000s, I’d get bond pricing data from broker “runs” delivered to my Bloomberg inbox, and if I wanted to know the bid/ask on a particular security, I’d have to ping various dealers.  TRACE quotes were too sparse to be meaningful, especially for off-the-run securities.  Through a manual, ad hoc, and somewhat dubious process, you could triangulate historical bond yields from third party data sources and dealer runs or impute them by averaging different points along the curve.  It sucked and even back then there was bubbling acknowledgment that reliable, systematic pricing of debt securities was long overdue.  Little progress has been made in the intervening dozen years; only 20% of the bond market trades electronically, mostly through RFQ (“request for quotes”, where an electronic expression of interest for a certain security is sent to market participants).  IDC, with continuous pricing on 2.7mn fixed income securities, sourced either directly from trading data or indirectly from market data-fed models, is a step in that direction.