Streetwise Professor

Several of my academic articles are about competition between exchanges.  The basic conclusion of these articles is that the network effects of liquidity make price discovery a natural monopoly.  This means that a competition between two exchanges that do not attempt to “cream skim” uninformed order flow is likely to result in a winner-takes-all outcome, where all trading “tips” to a single exchange.  If a new entrant challenges an existing market, given the stickiness of order flow, the winning exchange is likely to be the incumbent.

Jeremy Grant in today’s FT provides a text-book example of the phenomenon.  About a year ago a group of banks  (UBS, Morgan Stanley, Goldman Sachs, Credit Suisse, BNP Paribas, Societe General, Deutsche Bank, Merrill Lynch and Citi)  that control a large fraction of order flow set up the Turquoise exchange to compete with the London Stock Exchange.  The banks hoped that the competition would force LSE to cut fees.

Turquoise operated an open, limit order book market like the LSE, as well as a dark pool.  Thus, it went head-to-head with LSE in competition in price discovery.  Not surprisingly, the venture never succeeded in attracting a large volume, despite the fact that the banks that started it dominate the equity markets.  The exchange gained at most 8 percent of FTSE 100 volume, and was unprofitable.

As a result, the banks have given Turquoise–yes, given it–to LSE.  LSE has agreed to invest to upgrade Turquoise’s technology, and will use the platform to trade a broader slate of European stocks.

Which all goes to show that, as the theory predicts, competing with an incumbent exchange in “lit” markets is extremely tough, though not impossible (as Eurex showed a decade ago in its competition with LIFFE).  It also demonstrates that control of large amounts of order flow is not a sufficient condition for an entrant to succeed, though it is arguably a necessary one.  It also shows that a for-profit, investor-owned exchange can compete successfully against a mutualized one, and that the return of the mutual model is not predestined.  Finally, this represents a cautionary tale for other nascent exchanges, such as ELX in the US.