Market Fragmentation

Linking of various competing markets should be implemented through improvements in upstairs order routing, not downstairs mechanishms. Market fragmentation is in the news. U.S. SEC Chairman Arthur Levitt has spoken against the harmful effects of market fragmentation, and the SEC has asked for comments on possible solutions to fragmentation in markets. The SEC has also requested plans for linking options markets and has approved a plan submitted by the Amex, Chicago Board Options Exchange, and International Securities Exchange. The heads of major brokerages have testified on the need for a national central limit-order book (CLOB) that would consolidate all order flow and assure price/time priority across all markets. The NYSE opposes both any new CLOB and the continuation of the old Intermarket Trading System (ITS), which links the NYSE and the regional exchanges.How serious is market fragmentation? Should the SEC impose a regulatory solution? Should the various competing markets be linked into one market? I discuss the importance and extent of market fragmentation, the difficulties of directly linking downstairs markets, and a possible solution to lack of links in the markets.Market fragmentation arises when investors send their orders to a market where the orders do not interact with orders from other markets. Price priority (by which the order receives the best price in any market) is maintained in fragmented markets when each market promises to match the national best bid or offer. Time priority (by which the order goes to the market or dealer that first displays the best price), however, is frequently violated. And even price priority can be difficult to maintain if, for example, a large order in one market trades through prices in another market. Certain new electronic markets do not promise to match the best price elsewhere.Most proposed market-linking mechanisms are structured to link downstairs trading floors and trading facilities after a customer order has reached a particular market. The idea is that a market receiving a customer order would take control of that order, either executing it or sending it via a direct linking mechanism to another market. Linking markets by a downstairs linking mechanism should, however, be questioned.First, requiring trading markets to link up directly is undesirable because it reduces competition among the markets. If markets must belong and conform to a linking mechanism, innovation and competition among markets are likely to be impaired. Second, a downstairs linking mechanism is unlikely to be successful because no one would own it. It would be the common property of the linked markets, but no market would have an incentive to improve and update the mechanism. Third, the need for a downstairs mechanism to link markets has been overstated. Markets are already linked by upstairs order-routing systems. If a problem exists with the links, we should improve the current upstairs linking mechanism by increasing transparency and improving upstairs routing systems. Two implications of this approach are that the ITS should be allowed to wither away and that a formal linking mechanism for options markets, such as that recently approved by the SEC, is not necessary.The challenge for regulators is not what new linking system to construct but whether to impose rules on brokers for routing their orders. Regulators may want to specify more clearly what constitutes best execution for upstairs brokers, including rules for upstairs order routing that forge the desired links. Order routing according to strict price/time priority across markets would be detrimental to markets because it would create artificial incentives to start new markets.Requiring orders to be sent to a market with the best posted price, however, would be a reasonable step toward reducing market fragmentation. Such a rule would eliminate the option that now exists for market makers to match the best price for orders they wish to execute and fail to match the best price for orders they do not wish to execute. To work, the “best posted price” rule would need to be limited to orders that can be guaranteed execution by the market. The rule would make preferencing more difficult because a market maker would have to be ready to trade at the posted price vis-à-vis all comers, not simply preferenced orders. To the extent that a decline in preferenced order flow occurred, more orders would interact in price determination.