[The following excerpt is taken from David Weisberger’s comment letter to the SEC in regards to IEX’s potential impact on the stock market if it were granted permission to become a stock exchange. The views expressed here are not shared by the editors of Traders or its parent company Sourcemedia.]
In the debate over the IEX Application, the “retail is being hurt” mantra comes up repeatedly. Comment letters from retail investors and articles in the press position IEX as an innovation that helps individual investors. In addition, there was a form letter on the IEX website in 2014 available to retail investors so they could instruct their broker to route ALL orders to IEX.
To document the potential cost or benefit to retail investors of routing their market orders to IEX, we analyzed the Rule 605 public data for October and November 2015. The data demonstrates that, if retail market orders were sent only to IEX instead of to the top wholesalers (including Citadel, KCG, UBS, SIG, Citi and Two Sigma), those orders would have lost over $150 million dollars in execution quality; an annual rate of $900 million. A detailed description of our analysis follows.
The data for the analysis period contains over 132,000 market orders routed to IEX, which we deem to be statistically significant. For the period, the effective spread relative to the quoted spread (EQ) delivered by IEX was more than double that of the wholesale market makers (133 vs 60 EQ). During this period, IEX executed 21 million shares of market orders, while the wholesalers executed 17 billion shares. It is worth noting that the average order size sent to IEX was actually smaller than that sent to the wholesalers (464 vs 765 shares). Consequently, since there is a positive correlation between order size and higher execution costs, it logically follows that our estimate is more likely to underestimate the potential cost to retail.
We also analyzed the marketable limit flow executed by both IEX and a group of wholesalers in order to understand the context of the market order data. We found that market orders are a much higher percentage of marketable order flow for retail than for IEX’s current clients. Specifically, the data from the top seven wholesale market makers by volume (Citadel, KCG, UBS, Two Sigma, Citi, SIG and BNY) show that retail executed 56 percent of their liquidity seeking orders via market orders. Market orders sent to IEX, during the same period, represented less than one-half of 1 percent of the liquidity seeking orders that they executed. Thus, it is statistically necessary to use IEX’s execution quality on market orders to evaluate the execution quality of retail orders on IEX. Analysis that utilizes IEX’s “all marketable” execution quality, due to the fact that such a small percentage of their flow was market orders, would be statistically invalid, since there is such a large differential in execution metrics between limit and market orders.
We also studied the relative execution quality between IEX and the wholesalers for marketable limit orders, although we note that such comparison must be separately considered from the market orders analyzed above.
Before describing the analysis, it is important to understand that retail orders are deemed “held,” which means that the broker handling that order is obligated to either route or execute that order as soon as it can. Brokers receiving these orders have no option to cancel them unless the trading day ends and the limit price was never reached. As a result of this, during the period studied, the average execution rate of marketable limit orders handled by the wholesaler group studied was over 55 percent. IEX, however, executed less than 8.5 percent of the marketable limit orders they reported.
The data for the time period studied shows that IEX executed marketable limit orders with an EQ of roughly 85, compared with an average EQ of 97 for the seven top market makers. On the surface, this is a positive comparison for IEX, but it fails to take the execution rate into account. As noted, the wholesale market makers executed the retail marketable orders at a rate of roughly 6.5 times that of IEX, so we need to adjust the probable EQ that IEX would have delivered had they executed orders at a similar rate in order to produce a statistically valid comparison.
While estimating the EQ that would have been delivered by IEX had they been required to execute all orders that were executable, instead of cancelling them, we believe that a reasonable estimate can be calculated. The opportunity cost of unexecuted limit orders is not easily quantifiable without having the specific order data to analyze, but most models assume one-half of the bid offer spread. If we put that into EQ terms, it would suggest an EQ of 200 for the 48 percent of the unexecuted orders which might reasonably have been expected to be executed if treated as “held.” As large as that number is, based on experience, the average EQ received when fully routing aggressive retail orders does approach that number. That said, a more conservative proxy would be to use IEX’s own EQ from their handling of market orders. If we used that number for the 48.5 percent of the orders that IEX did not execute to predict a blended average EQ that aggressive marketable limits would have received, their EQ would rise from 86 to 127. However, utilizing the more conservative assumption that one-half of all the cancelled shares would be executed against displayed markets, then the predicted EQ would rise to 114. Whichever set of assumptions we use, it is clear that IEX would deliver an aggregate EQ on retail marketable limit orders of higher than the 97% shown by the wholesaler group in the analysis period.
This analysis is likely a consequence of IEX’s own design, which specifically rejects the concept of segmentation of different client groups. While this design has many positive aspects for a market that is applying for exchange status, providing retail investors with superior execution quality is the core concept of the wholesale business.
Retail investors deserve superior execution quality for the same reason that Wal-Mart gets the lowest prices from its suppliers. In both cases, the party that it is more economical to trade with receives a pricing benefit. For Wal-Mart, the reason is economies of scale, which allows producers to sell more predictable volume at a lower marginal cost. Orders from retail investors carry less risk than similarly sized orders from institutional investors. The reason retail orders carry less risk, however, has nothing to do with the intelligence or investment acumen of retail investors (as many media pundits claim). The reason, simply put, is that retail orders are WYSIWYG (what you see is what you get).
To explain why WYSIWYG is less risky, consider the following: When a retail order to buy 2000 shares of IBM is sent to a market maker, most of the time that order completes the trading interest of that investor. Institutions, on the other hand, on average continue to buy or sell. That “follow-through” buying or selling tends to move the price in the direction of the orders. Thus, market makers who take the other side of the initial orders have more risk. The term used by quants for this phenomenon is “correlation,” and retail orders are essentially un-correlated to future orders. Thus, it is not an accident that retail orders receive better treatment.
(Note: It is to the credit of the large retail brokers that so much of this benefit is now being passed directly to their clients. They do this by a commitment to analyzing execution quality and using that to select and oversee their market makers. As we have previously written, the price improvement provided to market orders amounts to over $600 million per year).
In conclusion, we believe that the SEC should approve the IEX Application with conditions. While we agree that approving IEX’s Application will allow the market to determine the value of their speed bump and pricing model, it is equally important to ensure that no regulatory advantage is provided to IEX or exchanges as a group. To accomplish this, we recommend that the SEC make the approval conditional upon IEX’s affiliated routing broker being subjected to the same delays as all other routing brokers. We further suggest that IEX be required to clearly disclose their execution quality statistics for individual client segments and types of orders as part of marketing materials.
We believe that such an approval order would accomplish the goal of supporting innovation, while mitigating the risks that approving IEX’s current model would create.
– – – – – – – – – –
David Weisberger is the managing director and head of market structure analysis of RegOne Solutions, by Markit. The views expressed do not reflect those of Traders.
