Beyond the Speed Bump: The New IEX Stock Exchange, and What Happens Next?

June 23, 2016


After a lengthy and rancorous public comment period, the Securities and Exchange Commission (SEC) on Friday approved the application of Investors Exchange (IEX) to become a full-fledged stock exchange.  As such, its displayed prices will have “protected” status under the SEC’s Regulation NMS (National Market System), obligating the other twelve U.S. stock exchanges to treat these prices as immediately available bids or offers.

The greatest controversy has centered on IEX’s use of a 350 microsecond delay, intended to disrupt predatory High Frequency Trading (HFT), as described in Michael Lewis’s Flash Boys.  Rival exchanges complained that, because of the programmed delay, IEX’s prices will not really be available immediately, leaving them stuck with an obligation to honor prices that are actually stale when displayed.  The SEC staff did not agree:

The Staff believes that . . . delays of less than a millisecond are at a de minimis level that would not impair fair and efficient access to a quotation, . . .  Today, a one millisecond intentional access delay is well within the current geographic and technological latencies already experienced by market participants when routing orders between trading centers.  Accordingly, the Staff believes that such a delay would be de minimis and consistent with the Commission’s interpretation of “immediate” as used in Rule 600(b)(3) of Regulation NMS.

The Commission did the right thing in allowing IEX to offer, as part of the National Market System, a “damped” exchange that operates a little more slowly than the fastest speeds that technology will allow.  As I indicated in a 2012 comment to the SEC, unproductive HFT latency races can consume real resources.  The use of brief delays, such as IEX’s deterministic delay, promises to improve the economic efficiency of financial trading.   

The Commission also did the right thing in committing to study developments over the next two years.  While the internal logic of the IEX exchange is straightforward, the interactions among the now 13 stock exchanges, each with a different microstructure and timing, will inevitably be complex.  Some thoughts on the problems the SEC will need to confront:

  1. As the Staff guidance (quoted above) recognizes, the word “immediate” in Regulation NMS cannot possibly be read to mean instantaneous, in the literal sense.  Even at the speed of light, orders traveling between spatially separated trading venues must take some time to travel.  In that de minimis one millisecond, light can travel 186 miles.  Thus there is always a risk that any “protected” quote displayed by an off-site exchange will be slightly stale.  But the problem is even more complicated, in that the nature of space-time, as described in Einstein’s special theory of relativity, does not allow for an absolute measure of time.  Not only is it impossible to be sure that a quote is not stale in real time; even after the fact, it will not be possible to construct an unambiguous sequence of events, when those events take place at high speed on spatially separated exchanges.  See “Space-Time Trading:  Special Relativity and Financial Market Microstructure.”
  2. The one millisecond de minimis exception to Regulation NMS will create space in which needed innovation can occur, but it is not a permanent solution to the problem of timing.  Regulators will need to develop a more systematic solution that accommodates a complex and variegated market microstructure.  One guiding principle should be that market relevant information should generally be transmitted unimpeded, even if marketable orders are subjected to delays.  The goal is to improve market efficiency by reducing the volume of trades taking place during the ephemeral information asymmetries that are ubiquitous at high speed.  This principle is easier to state than to apply, particularly in the face of complex order types with numerous contingencies.
  3. It should be possible for trading platforms operating at different speeds to co-exist and to compete – including those with and without speed bumps, using continuous or batched trading, or using innovations not yet contemplated.  The SEC will need to be alert to the possibility of harmful forms of cross-platform arbitrage; in general, however, arbitrage trading will serve to keep prices on the different platforms in sync.  As long as traders comply with the rules that apply on each exchange, arbitrage should not be discouraged.
  4. In seeking to achieve efficiency, stability, and fairness in the National Market System, the SEC need not seek to impose uniformity.  The real challenge will be to come up with a set of rules that can accommodate variety.  Innovation holds the most promise for resolving residual problems associated with predatory latency arbitrage, and competition is the best means of identifying the most useful innovations.  Convergence on a single solution is unlikely, however; exchanges with different internal microstructures will likely appeal to different segments of the market.  Difficult as it is to manage, diversity in microstructure will turn out to be a virtue.
  5. The SEC, and financial regulators around the world, will need to bring additional expertise to bear on the problem of market microstructure.  With the development of high-speed computing and communication technologies, global financial markets are evolving into a single massively parallel computing system – one that makes highly consequential decisions that humans, including regulators as well as market participants, find out about only after the fact.  In order to ensure stability and efficiency, parallel computing systems generally incorporate various features (master clocks, wait states, buffers, hypervisors, etc.) to ensure the various components are cooperating with each other.  This is not true of our financial markets, however.  Regulation NMS quite properly uses economic competition as its organizing principle, but a parallel computing system that is governed only by competition will be unstable and vulnerable to latency racing.  The SEC should seek outside advice on the design principles for a stable, fair, and efficient automated financial marketplace.  The National Academy of Engineering might be one mechanism through which such expertise might be assembled.  A suitable advisory panel should include, not only economists and financial experts, but also experts in parallel computer architecture, in signal processing, and in the stability of complex systems.  Competition should remain the touchstone of Regulation NMS, but algorithmic competition will need additional organizing principles if it is to operate successfully.