The IEX Application: Some Further Comments

We think IEX is a useful innovation and serves the needs of institutional investors well in its current form. These benefits exist from the combination of a dark liquidity pool and a speed bump, allowing for larger fills that traditional exchanges typically can’t cater for. However, IEX as an exchange — with displayed, automated quotations — poses certain challenges to the National Market System and likely erodes some of these benefits. The Securities and Exchange Commission’s (SEC) latest consultation is a welcome development but stops short of the more wholesale market structure reform examination that is due.

Our recent article for TabbFORUM on the IEX exchange application has generated a lot of feedback. A two-sentence summary of our position in that article could be as follows:

• We do not think the SEC will approve the current IEX application given current rules, but since the SEC is now consulting on reinterpreting these rules, this point may soon be moot.
• We are skeptical that encouraging time-delay competition among exchanges will result in sunlit uplands for investors, and think that a more holistic reconsideration of the ecosystem of payment-for-order-flow, maker-taker, Rule 611, market fragmentation, and so on would be a better approach.

Perhaps we would say that given our fondness for encyclopaedia-length regulatory consultations.

The incredibly technical nature of the IEX debate, coupled with the outsized passions involved as a result of the huge public awareness raised by Flash Boys makes it challenging to reach consensus. Amongst the very diverse CFA Institute membership, comprising over 135,000 members in over 150 countries, the responses to the article have ranged from support to outrage.

There have been some constructive and specific criticisms of the article. First, there is a passage in Reg NMS that implies that the interests of long-term investors should supersede other interests (and some critics imply this should supersede any other passage) when these are in conflict. Many critics of our article took it to be anti-IEX and contrary to long-term investor interests. As a wealth of CFA Institute content will attest, we believe in long-termism, but principles-based arguments such as these are always difficult to enforce when discussions devolve into highly prescriptive issues.

The argument in our article was that it appeared to us that the IEX application did not comply with existing rules. It seems to us that the fact that the SEC is now consulting on reinterpreting some of these rules, specifically to decide whether to regard sub-1 millisecond latencies de minimis, effectively concedes the point we made — that the issue is not black and white.

This new consultation can also be interpreted as the SEC indeed wanting to put long-term investors’ interests first, having heard the views of the undoubtedly large number of institutional investors supporting the application, and thus itself being implicitly supportive of the IEX application. However, the SEC may have found it necessary to formally consult on reinterpreting some of its previous rules to be confident of granting the exchange application without offering opponents any ground for legal recourse.

A second criticism is that our reference to the National Best Bid and Offer (NBBO) becoming unreliable as a result of an ecosystem of time-delayed exchanges inspired by IEX is not sufficiently precise. The NBBO as defined by Reg NMS is generated by one of two defined securities information processors (SIPs). At points in the article, we also refer to the NBBO in the context of a ‘virtual’ NBBO generated by individual exchanges from their direct feeds.

This distinction is important, and we have received criticism that the argument that IEX may disturb the virtual NBBOs of individual exchanges is not a valid critique of the IEX application. This criticism is based on the fact that exchanges do not send their quotes and trades to the SIP on their best and lowest-latency infrastructure, access to which is offered for sale to high-frequency trading (HFT) firms and other trading venues. Although exchanges must release the information to the SIP at the same time as they release the information via their proprietary data feeds, the step of consolidating the data from all the exchanges, performed by the SIP, adds further latency that a trader with a direct feed is able to bypass.

As a result, there is an argument that the distinction between the IEX-programmed delay and the ‘built-in by-design’ delay of antiquated network infrastructure is academic to a non co-located trader. Allowing IEX to delay by 350 microseconds may have no material impact on a non co-located trader because of these existing network latencies. A trader relying on the SIP feeds would probably notice an even smaller difference or none at all.

As an aside, the argument that some market participants are taking advantage of antiquated SIP infrastructure to alternately sell valuable direct feeds and also trade on faster virtual NBBOs seem to us to be a good reason to encourage ongoing efforts to improve the SIP. There is some evidence that this wish may be naïve. If attempts to improve the SIP are unrealistic then this is another reason to welcome the SEC’s consultation on potentially ignoring anything sub-1 millisecond.

However, for better or worse, the bulk of market activity and liquidity is being provided by algorithmic traders that use co-location and direct data feeds to construct their own view of the market. For these market-making activities, much lower latencies are critical as they reduce information uncertainty and thereby lower the cost of liquidity provision, which is ultimately reflected in spreads.

In an ecosystem full of copy-cat, speed-bumped exchanges, will the increased uncertainty caused by time delays cause these electronic market-makers to raise spreads? Will price discovery be impeded? Will spreads increase materially? These are important questions because the reduction in trading costs brought about by increased speed has brought well-documented benefits. However, we do not argue that trading costs and price discovery should be the sole criteria for evaluating the efficacy of the current system — we agree that capital formation should not be overlooked.

It is possible any impacts from these speed bumps will be immaterial, and some argue that the whole concept of selling co-location with a speed bump does not make economic sense so that copy-cat exchanges will not happen. Equally there is an extensive track record of unintended consequences following market structure changes, and reports of incumbents developing post-IEX speed-bump plans have recently surfaced.

However one views the aforementioned arguments, it seems clear to us that fragmenting the market along the time-delay dimension will likely yield more complexity and even more scope for questionable ‘innovation’ in the HFT space. It is possible that supporters of IEX are fine with this and expect IEX, particularly its hidden peg order pool, to at least be one trading venue acting as an oasis for the average punter. However, this is true today with IEX operating as an ATS, and several commentators have pointed out that it is not immediately obvious why IEX needs the quote protection afforded by exchange status for its displayed pool as well. In the UK and Europe, there is no trade through quote protection and some have argued that it may be time to reconsider Rule 611 in the US.

Finally, our article argues that there are several aspects of modern market structure that could likely be improved. Perhaps this should have been foremost in the article as it seems to have been completely missed among all the controversy. It seems to us a positive that the SEC is commencing a new consultation to come up with a coherent policy position on these issues, rather than bundling everything up in a decision on a single exchange application.

The tone and conclusion of our article was meant to convey the impression that if one wants to ameliorate some of the negative effects of modern market structure, it seems odd to do this by focusing all efforts on allowing a single participant to pursue a time-delay strategy with the justification that the incumbents are all pursuing time-delay strategies by default.

It seems more sensible to have a debate about whether we want these time delays, and if not how to get rid of them? If the answer to this question is that it is too hard and there are too many vested interests, again it seems to us that the SEC’s new consultation may provide a way out, for better or worse.

If you liked this post, consider subscribing to Market Integrity Insights.

Image Credit: iStockphoto.com: a-wrangler