April 26, 2000
Mr. Jonathan G. Katz, Secretary
Securities and Exchange Commission
450 Fifth Street NW
Washington, DC 20549-0609
Dear Mr. Katz:
I, Marshall E. Blume, Howard Butcher III Professor of Financial Management at the Wharton School of the University of Pennsylvania, strongly support the request of the NYSE to rescind Rule 390.
I do not support the enactment of the second proposal of the NYSE that only would allow trading as principal away from the primary market when there is price improvement. I believe that this proposal would reduce competition. This proposal would eliminate any trading off the primary market in a one-tick market, as price improvement is not possible. Only when the spread is two or more ticks could a broker-dealer potentially make a profit. Since the time intervals in which a broker-dealer could make a market would be random and outside its control, my judgment is that most broker-dealers would withdraw from such a market. The consequence would be a nullification of Rule 19(c)3 and a reduction in competition in the trading of NYSE-listed stocks.
Fragmentation and Competition
Electronic communications, computing power, and competition are rapidly changing the mechanisms for trading options and equities. Individuals today have access to information that was only available to major institutions in the past. The Internet has enabled individuals to trade in ways that were not even thought of a decade ago. Competition has brought with it major changes.
Of course, competition leads to some degree of fragmentation, and some of the desirable characteristics of the old ways of doing business vanish. For example, before the break up of AT&T, a single phone call was all that was needed to have a phone system fixed. Today, it is more difficult to obtain repair service. Yet, long-distance rates have dropped. New communication products are continually coming to the market. Are consumers better off today than before? Yes. Should we go back to the old way? No.
Similarly, as the market for the trading of securities becomes more competitive, there will be changes. Some services available under the old system, such as providing for price-time priority, may not function as well. But, as discussed below, modern communication systems may make price-time priority less important today than in the past. Initially, there may and probably will be more market fragmentation than before.
The question is whether on balance investors and the economy will be better off in a highly competitive market or in a government-imposed structure. Competition almost always leads to more socially desirable outcomes than a government-imposed structure unless the competitive solution involves some major negative externality. To illustrate, it could be that substantial fragmentation of the market would lead investors to lose confidence and, and as a consequence, raise the cost of capital, resulting in curtailed economic growth. Today's high level of economic growth suggests that the current level of fragmentation has had little effect.
Unless it can be shown convincingly that there are substantial externalities to fragmentation, public policy and the SEC should embrace competition. The SEC should ferret out any rule or process that hinders competition, and let the markets evolve on their own. Perhaps, a single venue is optimal, but let competition determine which venue that will be.
However, as discussed below, the SEC should take a proactive role in the dissemination of information, particularly trade information.
The Definition of a CLOB
The National Market System visualizes that all orders will interact. A CLOB (Centralized Limit Order Book) through which all trades must be executed allows full interaction of all orders. The key here is that all trades must be executed through the CLOB. Investors submit limit orders, and other investors hit these limit orders with market orders. No other trades are permitted.
The CLOB proposed by the SEC and others is quite different. To use the words of the SEC's request for comment, "the SEC could order the establishment of a national market linkage system that provides price/time priority for all displayed trading interest." The key word here is "displayed." Routinely, large institutional investors use not-held orders to be worked on the floor of the NYSE. The upstairs market matches buyers and sellers of large blocks of stock. These orders are not displayed and therefore would not be included in the CLOB.1
Today, individuals with their small orders usually use only limit or market orders. Their orders are just too small to justify other types of orders, like not-held orders. Thus, individuals for the most part would have to participate in a CLOB, while institutions have additional ways to execute their orders. Institutions would use a CLOB only when it was to their advantage, and possibly to the disadvantage of individuals.
Forcing individuals to use a CLOB and allowing institutions to use a CLOB only when it is to their advantage seems to me fundamentally unfair.
I shall assume in the remainder of this letter that a CLOB will be mandated only for displayed limit orders. Many institutional types of orders are thereby excluded.
Orders and Price-time Priority
In the following, I shall assume that there is strict price-time priority within any market arena and that market makers are not permitted to trade in front of their customer orders. It is my understanding that this assumption corresponds to the current practice for NASD broker-dealers under the Manning rules.
Trading on the NYSE does not always conform to strict price-time priority. This can occur when there is a clean cross, for instance. It can also occur when a floor trader with a newly arrived order trades in front of and at the same price as a previously submitted limit order. This possibility results from the NYSE rules for determining standing within the crowd.2 I know of no evidence of the impact of the violations of strict price-time priority on the probability of the ultimate execution of a limit order submitted to the NYSE. Since I have never heard of any complaints about the effect of such violations of strict price-time priority, I conclude that investors have found the execution of limit orders on the NYSE satisfactory. Investors do not require an absolute certainty of limit order execution but only a high probability. This point is important in evaluating the need for strict price-time priority over all market arenas as distinct from a single market arena.
The SEC has asked for comments on whether price-time priority should be extended to cover all public orders in any market. This is a complicated issue and will be addressed below.
Twenty or thirty years ago, limit orders with prices away from the current market played a useful role. A retail investor could decide on the correct price of a security and submit a limit order at that price. It might be weeks or months, if ever, before it was executed. An investor who placed such an order had the right to expect that it be executed at the appropriate moment, and price-time priority was very important.
Although I do not have exact data, specialists and others have told me that the limit order book contains less depth than formerly. If so, limit orders with prices far away from the market are probably less common than they were twenty or thirty years ago. This is as it should be. Today, the Internet and television provide real-time market information. It is easy to place an order and then withdraw it.
It is my impression that retail customers today use limit orders for two purposes: first, to protect themselves against price movements during the interval between the submission of an order and its execution and second, to improve upon the best bid or offer to become essentially market makers.
As an illustration of the first purpose, assume that the NBBO (National market system Best Bid and Offer) is a bid of 20 and an offer of 20 1/8. To protect against price movement between submission and execution, the investor could submit a limit buy at 20 1/8. Since this limit order is at the market, an investor would expect that it would be executed if there were subsequent prints at 20 1/8. If there were subsequent prints and the order were not executed, the investor would have a valid reason to complain. If this happened frequently, the market arena would lose its customer base.
As an illustration of the second purpose, assume again that the NBBO is a bid of 20 and an offer of 20 1/8 and an investor then narrows the NBBO by placing a limit sell at 20 1/16. If the market arena in which this order is submitted receives a market buy and if the marker arena follows strict price-time priority, this limit order will be executed. If however another market receives a market buy order, there is no guarantee that this limit order will be executed without market-wide price-time priority.
The SEC raises the question of how strong the incentive would be for an investor to submit a NBBO-improving limited order when there is no guarantee that it will be executed, particularly in the presence of preferencing and payment for order flow. The answer depends upon the concentration of order flow among market arenas. To maximize the probability of the limit order being executed, an investor would want to send the order to a market arena with a large amount of order flow. In that way, the investor will insure a greater probability of the limit order being exercised.
Contrary to some observers' initial reaction, a market arena that can increase order flow through preferencing may actually encourage an increase in the submission of limit orders that better the NBBO. The argument is the following: The increased order flow makes it more likely that a limit order will be executed. As investors perceive that there is a greater probability that their limit orders will be executed, there is a greater incentive to submit NBBO-improving limit orders.
Competition and market forces themselves will eliminate extreme market fragmentation. It is likely that order flow will tend to concentrate in a limited number of market arenas. Order flow begets more order flow.
In my judgment, price-time priority within a market arena will give sufficient incentive for investors to submit limit orders that narrow the NBBO. Remember that it was not until the last few years that investors could be confident that their NBBO-improving limit orders would be displayed or executed. The 1997 order handling rules have had a substantial effect on narrowing NASDAQ spreads. The NYSE now insists that limit orders that narrow the spread be executed or displayed.
The SEC should let the markets evolve competitively. Only if there is clear evidence of a market failure should the SEC impose a specific structure, like a mandated CLOB.
Internalization and Payment for Order Flow
The auction process used by the NYSE relies heavily upon trading interest in the crowd on the floor. This trading interest cannot easily be displayed on CQS (Consolidated Quote System), such as a large not-held order. The result is that the displayed spread on the NYSE, when it exceeds one tick, is often greater than the effective spread at which market buys and sells are actually executed. Academic studies have confirmed that when the quoted spread is two or more ticks, the effective spread is much narrower and frequently averages just over one tick. The effect is that many market orders are executed within the NBBO when the spread is two or more ticks. This possibility of executing at a better price than the NBBO is what the NYSE terms "price improvement."
The basic problem is that the trading interest on the NYSE floor cannot be incorporated into CQS. The result is that investors who trade through a market arena that merely matches the NBBO for NYSE-listed stocks will face greater effective spreads than if the orders were sent to the floor of the NYSE. That the trading interest on the NYSE floor cannot be incorporated into CQS is not a criticism of the NYSE; it is just a statement of fact. The NYSE and CQS are two different systems, and they just do not meld well.
Some of the regionals, as well as Madhof, have addressed this incompatibility between the trading process of the NYSE and the CQS in two ways: First, they have developed algorithms and other order-exposure methods to mimic the possibility that a market order submitted to them will be executed within the NBBO as it might be if sent to the NYSE. Second, they have paid for order flow, either directly or through reciprocal arrangements.
This payment for order flow represents a reduction in the spread if it is passed onto the investor through reduced commissions. The unresolved empirical questions are: Do these algorithms and procedures indeed mimic the trading process on the NYSE? Are the payments for order flow set at competitive levels and are they passed onto investors through reduced commissions?3
Today, I would oppose any prohibition on payment for order flow as it may reduce competition. This payment for order flow may be only a competitive response to the incompatibility of CQS and the way in which trades are conducted on the NYSE. It allows the trade prices of NYSE competitors to be competitive with NYSE trade prices.
The markets are currently in a state of transition. Decimalization is on the horizon, but the real change will be the reduction in the tick size. Some ECNs may become exchanges themselves. NASDAQ has just voted to become a profit-making institution. After these changes and others have finally worked themselves out, the SEC might want to reexamine the effect of payment for order flow upon competition. It may be that with narrower spreads, the magnitudes associated with payment for order flow will become smaller, making payment for order flow a less important aspect of market structure than it is today.
Comments on Policy Options
I support the SEC's first option of requiring greater disclosure by market centers and brokers concerning trade executions and order routing.
A retail customer who is well informed will make better decisions than one who is not. To this end, the SEC should require that portals, defined as entities where orders are submitted, provide meaningful information as to their routing policies.
Additionally, the SEC should require each portal to provide detailed information of each trade.4 Such information should include: the type of order, the number of shares, the trade price, the commission, taxes, the name of the portal, the market arena in which the order was executed, the time the order was given to the portal, the time of execution, and the time of the report of the trade to the client. With this information, third parties would be able to publish rankings of the portals.
The SEC should be cautious in specifying how the information should be summarized. If the SEC focused on the wrong summary measures, portals could change their business practices to look good on the summary measures to the detriment of the retail investor. It is probably best to allow the portals themselves to summarize the information in their advertisements and to give the data to third parties for their analysis. The only requirement would be that the advertisements are not misleading.
I have become increasingly impressed with the impact of the real-time reporting of transactions on national television. Based upon my experience with both undergraduates and MBAs, I have concluded that the televised real-time tape has had a substantial impact on their understanding of execution quality. Students place trades on their cell phones, and then watch one of the national television stations to assess the quality of the execution. If they perceive that they received a bad execution, they are not hesitant to complain to the portal and they often prevail.
If the SEC causes the collection of this detailed trade information from the portals, it would not be long until some entity made these data available over the Internet, and the SEC should craft any regulations to encourage such display. With this type of data, individual investors would be able to evaluate the quality of their executions. Since there would be a substantial amount of data, the site might make available only data for a user-specified day for a user-specified security for a 10 to 15 minute interval, or it might even present some analysis of the quality of execution. In any case, competition would determine the best way to present and analyze these data. While I would expect that only a limited number of investors would avail themselves of these data, even a small number of investors need analyze these data to have a major impact on the business practices of most portals.
Before imposing any of the other proposed options to deal with market fragmentation, I would suggest that the SEC consider causing the portals to report the above detailed information. The effect of the Order Handling Rules and the Alternative Trading System initiative have already had substantial impact, but their ultimate effect is still to be determined. The SEC should let competition pursue its course.
At this time, the SEC should not prohibit payment for order flow. As I argued above, payment for order flow might be viewed as a competitive response to the inability of CQS to display trading interest on the floor of the NYSE. After decimalization and other changes in the market place take place, the SEC might wish to revisit the issue of payment for order flow.
The last four options would impose specific structures on the market. The most extreme option would mandate a CLOB. It should be noted that the seemingly less intrusive mandate of linking the individual trading arenas through the exchange of time and order information is fully equivalent to a mandated CLOB. The different approaches undoubtedly involve different technical issues and costs, but both represent government-mandated market structures.
Such a mandated CLOB, or its equivalent, would stifle future competition among competing market mechanisms, particularly for retail customers. It would further establish a technical standard that could in the future limit new innovations in trading.
In sum, there is a tension between price-time priority and competition. As argued above, having strict price-time priority within a market arena may provide adequate protection to an investor, even to an investor who is submitting a limit order that improves the NBBO. Extending price-time priority across market arenas could reduce competition, and thus is not warranted at this time.
In closing, the SEC should encourage competition among market arenas. If a CLOB represents a market equilibrium, the SEC should let competition make this determination.
Marshall E. Blume
1 The NYSE report dated March 23, 2000, "Market Structure Report of the New York Stock Exchange Special Committee on Market Structure, Governance and Ownership" acknowledges the difficulties of defining the exemptions from the CLOB. This report concludes that CLOB proposals "seem narrowly tailored to accommodate the specific business models of their proponents."
2 Nonetheless, the NYSE does preserve strict price-time priority within the limit-order queue itself.
3 Yannis Bakos, Henry C. Lucas Jr., Wonseok Oh, Sivakumar Viswanathan, Gary Simon, and Bruce Weber (Electronic Commerce in the Retail Brokerage Industry: Trading Costs of Internet versus Full Service Firms, Working Paper S-00-1, New York University Salomon Center) find a negative relation between price improvement and commissions. On average, the increase in commissions swamps the gains from price improvement. Specifically, they find that the total savings of using an Internet broker rather a full-service broker is 22.96 cents per share for a 100-share trade and 6.64 cents per share for a 2500-share trade. Of course, the full-service broker could be providing other services for this greater cost.
4 Today, the exchanges and NASDAQ collect substantial revenues from the sale of market quotes and prices. If the SEC wishes to maintain this source of resources, the SEC might only require that this suggested information be made available after the close of trading each day or with some other time delay.