Speech by SEC Commissioner:
Remarks before the Council of Institutional Investors

by

Cynthia A. Glassman

Commissioner
U.S. Securities and Exchange Commission

Boston, MA
September 28, 2004

Market Structure: Separating Myth from Reality

Thank you, and good morning. I'm very happy to be here this morning to talk about the Commission's recent market structure proposal - Regulation NMS (short for National Market System). Before I go any further, let me make clear that the views I express here this morning are my own and not reflective of those of the Commission or the staff.

Given how busy you all are, you may not have focused very much on the issues surrounding securities market structure, figuring that they involve far too many technical, arcane issues to deal with. And if you thought that, you would be right - this is a complicated area. It has been the most difficult area for me since I've been at the Commission. I guarantee you that, if we were starting from scratch, nobody would have invented the system we have today. But we're not starting from scratch. We have a market structure that began under a buttonwood tree on Wall Street in 1792 and that has evolved over time in response to market conditions, legislative and regulatory changes, new technology and, in some cases, simply from happenstance.

There is no question that increases in market efficiency and transparency over the last 25 years have benefited investors -- executions are quicker, spreads have narrowed, and transaction costs have been lowered. Nonetheless, there is general consensus that there is still room for improvement. This spring, after much study and discussion, the Commission proposed Regulation NMS, which makes several substantive recommendations for change. The decisions the Commission makes about market structure will ultimately affect investors, both retail and institutional, which is why I've been asked to discuss market structure reform this morning.

Before we take a look at the Commission's market structure proposal and the reaction to it so far, I would like to define some terms. What is market structure anyway? The markets are where buyers and sellers meet with their supply of, and demand for, shares of stock, and where the price of the stock is determined. Market structure has to do with how the markets are organized, how the various types of market participants relate to, and compete with, one another, and what the rules are that govern their behavior.

For example, we have the "listed" market for the trading of securities listed on exchanges, and we have a separate market -- the Nasdaq market - for the trading of securities that are not exchange-listed, except for the securities that trade on the OTC Bulletin Board or in the pink sheets. Nasdaq securities are not exchange-listed because Nasdaq is not an exchange. It has applied to become an exchange, but the Commission has not yet taken action on the application.

There has always been some overlap between the listed and Nasdaq markets. Nasdaq market makers trade exchange-listed stocks in the so-called "third market," and some regional exchanges trade Nasdaq stocks under special rules called "unlisted trading privileges." Nasdaq recently obtained SEC approval for a pilot program under which a handful of large cap companies listed on the NYSE have also listed on Nasdaq. There is a great deal of competition between the listed and Nasdaq markets for issuer listings, but on the whole, the markets operate independently of one another.

On exchanges such as the New York and American Stock Exchanges, business has traditionally been conducted through an auction process conducted on a physical trading floor. Floor brokers holding customer orders in a security await the right time to enter the continuous mini-auctions conducted by the specialist assigned to the stock. They literally stand "in the crowd" in front of the specialist's post. Specialists also get orders through electronic routing, and they may use the electronically entered orders to fill the floor broker's orders. Specialists are responsible for maintaining fair and orderly markets in the securities designated to them by the exchange, and they are permitted to, and in fact are expected to, use their own capital to correct imbalances of buy and sell orders. In that role, specialists make money off the spread between the bid and the ask. In addition to floor trading, most exchanges also offer some degree of automated trading.

We see a very different business model in the Nasdaq market. Business on Nasdaq was traditionally conducted over the telephone by multiple competing market makers. Today, Nasdaq is a fully electronic, screen-based system. As dealers, Nasdaq market makers have traditionally operated on a principal basis, buying from and selling to customers from their own inventories. Beginning in the late 1990's, Nasdaq market makers found themselves up against a new breed of high-tech competitor - the electronic communications networks or ECNs. ECNs are computerized matching systems that allow customer orders to interact with each other without an intermediary. Offering quick, cheap, anonymous trading, ECNs operate on an agency basis, charging their subscribers a fee to access their liquidity. ECNs became so popular with investors, market makers and other broker-dealers that Nasdaq itself built an ECN. As a result of competition from the ECNs, market share in the Nasdaq market has been shifting from Nasdaq to the ECNs.

In the listed markets, customers compete for shares by placing limit orders or they accept the market price - a process that results in "price discovery." Large orders may be personally handled by a floor broker, who works the order to try to get the best price. The specialist displays its best quote in the consolidated quotation system, but if you're in the crowd at the specialist's post, you may be able to get a better price than the displayed quote - we call that "price improvement." In other words, the best price quoted isn't necessarily the best price. Am I sounding like Alice in Wonderland here?

In the Nasdaq market, price competition is driven by competition among market makers and ECNs. The best quote among all Nasdaq market centers is displayed on the consolidated quotation system as the National Best Bid or Offer ("NBBO"). When a dealer receives an order from a customer, it may "internalize" the order, meaning that it crosses customer orders internally or fills the order from its own inventory, rather than exposing it to the market as could occur on an exchange. Normally, retail customers will get at least the NBBO, and may get an improved price. Internalization is controversial because it creates a conflict between the dealer and the customer, both of whom want to buy low and sell high. According to Nasdaq, however, the level of internalization on Nasdaq has decreased over the past few years.

Whether you trade in the listed markets or the Nasdaq market, your broker owes you the duty of best execution. In releases over the years, the Commission has analyzed the important concept of best execution in terms of price, speed of execution, trading characteristics of the specific security, availability of accurate market data, technology and cost and difficulty of executing in a specific venue. While the Commission has stated that price is not the sole factor in determining best execution, price historically has been the most important factor. As technology has advanced, however, other factors have taken on greater importance. I'll explain this in a minute.

Now that we've got the basic concepts and terms down, let's see how this all fits together. In 1975, Congress revisited the nation's securities markets for the first time since creating the SEC and enacting the federal securities laws in the 1930's. Congress directed the SEC to establish a national market system that would provide for

• the efficient execution of securities transactions,
   
• fair competition among broker-dealers, among exchanges, and between exchanges and other markets,
   
• ready availability of quotation and transaction information to broker-dealers and investors,
   
• the ability of broker-dealers to execute orders in the best market, and
   
• an opportunity, consistent with the other goals, for investors to execute orders without the participation of a dealer.
   

Congress did not dictate a particular market structure to accomplish these goals, and this led to controversy. One concept was to send all orders to a consolidated limit order book (referred to as a "CLOB") where they would be executed on the basis of price and time priority. In other words, the first one in with the best price - across all markets - would get the execution. This concept was never implemented, however, because of concerns not only about the feasibility of such a structure, but also about the advisability of creating a centralized structure that might be less conducive to innovation and competition than markets competing with each other for orders. Instead, the exchanges developed an electronic linkage known as the Intermarket Trading System ("ITS") whose central tenet was and remains today the "trade-through" rule.

Trade-Through Rule

Under the ITS trade-through rule, markets are not supposed to trade at prices worse than another market's quote, and if they do, they are supposed to satisfy that market's quote, if requested. When adopted in 1978, ITS, with its trade-through rule, was intended to serve the dual and complementary purposes of facilitating the broker's duty of best execution and promoting competition among the exchange markets. With advances in marketplace technology, however, some market participants and institutional investors have come to believe that the rule is no longer necessary. In fact, they argue that it is impeding competition in the listed markets.

Here's how it works. If a market receives an order and is unwilling to match the price quoted by another ITS participant market, it must route the order (referred to as a "commitment") through ITS to the market displaying the better quote. Many orders are routed to the specialist on the floor of the NYSE because that's where the best "displayed" price can usually be found, but the price "displayed" in a manual market such as the NYSE may not be available. The quote may be stale or in the process of changing. In the time it takes the specialist to decide whether to honor the commitment, the market may have moved, an outcome even more likely under decimalization.

Therefore, it's not a sure thing that the customer will get the displayed price -- there's only the possibility of getting that price. And if the market has moved away, there is no guarantee that the order can be filled elsewhere. Given these realities, many customers have come to look at best execution in a different way. While price has historically been the primary factor, certainty of execution has become as important, if not more important, in achieving best execution. Some traders are willing to accept a slightly worse price than the best displayed quote because of the uncertainty that they will actually obtain the quoted price. I liken this to the "bird in the hand, two in the bush" concept.

Some commenters argue that the absence of a trade-through rule would discourage limit orders because limit orders that were traded-through would go unexecuted. But other commenters point out that limit orders can be left behind for another reason. The trade-through rule does not require that you trade with the better quote - you just have to match it. So limit orders at the NBBO may not be executed even with a trade-through rule.

Retaining a belief in the continued benefits of a trade-through rule, yet acknowledging the difficulties for investors caught in the gap between electronic and manual markets, proposed Reg NMS retains the trade-through rule, but offers two exceptions. The first would permit quotes that are not electronically accessible to be traded-through. In other words, manual quotes could be traded-through, but electronic quotes couldn't. The second would permit informed investors to "opt-out" of the trade-through rule on a transaction by transaction basis. The first exception seems to be well accepted at this point, but the "opt-out" provision has proved extremely controversial.

Proponents of the trade-through rule and opponents of the "opt-out" - principally the New York Stock Exchange and some institutional investors -- argue that all investors want the best price and that the trade-through rule provides a necessary structural support for the broker's duty of best execution. They assert that the rule protects customer limit orders, which contributes to liquidity. Without the rule, they argue, customers whose limit orders are traded-through - that, is, customers whose limit orders don't get executed even though trades have occurred at less advantageous prices -- will have less of an incentive to place limit orders in the future, which will reduce liquidity.

Proponents also make the argument that if electronic markets are permitted to trade-through manual markets, there is no need for an opt-out. The New York Stock Exchange has announced its intent to become a predominantly automated market. The NYSE plans to expand its automated trading system and to allow automated executions against its best quote and its limit order book most of the time. According to the NYSE, the new system would limit the degree to which human beings, that is, specialists, could interact with incoming automated orders. One reason it is important for the Exchange to expand investors' ability to trade automatically has been demonstrated in the Commission's recent settlements with the seven current NYSE specialist firms. By virtue of their intermediary position between buyers and sellers, and their control of the NYSE display book, specialists gain an informational advantage, giving them an opportunity to trade for their own accounts ahead of customers. In the settlements, without admitting or denying the allegations, the firms agreed to findings that they "froze the book" and traded ahead of customer orders. Therefore, an important test of whether the NYSE's new system will be an acceptable solution will be its success in limiting the specialists' discretion to shut down the electronic system.

Opponents of the trade-through rule, as well as those supportive of the "opt-out" provision, include many broker-dealers, ECNs and institutional investors. Despite their success in the Nasdaq market, the ECNs have struggled to make inroads into the exchange-listed markets, largely, they believe, as a result of the trade-through rule. As I already described, an ECN trying to execute a large customer order in a listed security has to route the order to the Exchange floor if there's a better price displayed there and the ECN has no matching price in its book. Taking valuable time to reach what may turn out to be an inaccessible quote is contrary to the ECNs' business model of quick execution speed and certainty of execution.

Another controversial aspect of the Reg NMS proposal would make the trade-through rule applicable to all markets. Until now, the trade-through rule has been unique to the listed markets, but under the proposal, the rule would be expanded to Nasdaq. In evaluating whether this step is advisable, it will be important to me to see an objective analysis of the data on the extent of trade-through activity in the Nasdaq market and to learn how the application of the rule would affect that market. It is important to remember that, while Nasdaq has not operated under a trade-through rule, all brokers, regardless of the market on which they trade, are subject to the obligation to provide their customers the best possible execution.

Access and Access Fees

Another component of proposed Reg NMS is to set standards for access among market participants on a fair and non-discriminatory basis. This has been viewed favorably by commenters. However, the companion proposal to limit so-called "ECN access fees" has generated controversy. When the Commission adopted rules to promote the growth of the ECNs in the late 1990's as a way of increasing competition in the Nasdaq market, it permitted the ECNs to charge their subscribers a fee to access their liquidity, provided the fees were minimal. Access fees permitted the ECNs to flourish and even to pay rebates for limit orders, but since the markets' move to decimalization in 2001, they have become increasingly controversial. One concern is that only ECNs are permitted to charge these fees to access their best quote, while all market participants have to pay them. Another concern is that, in some cases, the fees may have either been set beyond minimal levels or assessed unfairly. Moreover, they are not displayed in the ECNs' quotes, which could confuse investors. An obvious solution would appear to be requiring access fees to be included in the quote, but that would require quoting in sub-pennies, which nearly everyone seems to think would create more problems than it would solve, and which another part of the market structure proposal would prohibit. The compromise proposed in Reg NMS is to place a cap on access fees.

The idea of capping access fees, which are a form of commission, is opposed by the ECNs, which depend on them as a source of revenue, but is favored by Nasdaq market makers, who, as dealers, have not been permitted to charge them. (Note: Brokers charge commissions; dealers charge mark-ups.) In an effort to level the playing field, proposed Reg NMS would permit all market participants, including dealers, to charge access fees, at least when accessed through an SRO's system, but cap the fees at 2 mils ($0.002) per share. I'm keeping an open mind about all of these issues, but I'm generally uncomfortable with the Commission's acting as a rate-maker, and am considering whether relying on the market centers to address this problem would be more appropriate. At least one market center has taken action in this area. Nasdaq has put into effect a 3 mil cap for participants in Nasdaq's systems.

Market Data Revenue

The final piece of the rule proposal is a reworking of the formula by which revenue from the sale of market data is apportioned among the securities industry self-regulatory organizations ("SROs") such as the NASD, the Amex and the NYSE. How the process works now is that markets send their quote and trade information to an entity known as a securities information processor or SIP, which consolidates the data, sells the data back to the market participants and third party vendors, and divides the proceeds among the SROs. The SROs use the money to fund their regulatory programs, and some rebate part of their market data revenue to their members to attract order flow.

Generally speaking, the SROs' share of market data revenue is calculated on the basis of a formula that rewards quantity -- the number of trades or share volume reported by the SROs. The Reg NMS proposal would change the formula to reward quality. In addition to quantity, SROs would be rewarded for posting the best quote.

The proposal has been criticized by the affected SROs as overly complex and by other market participants as not going far enough to resolve the market data question. Most commenters called for a much more comprehensive review of market data fees, given the current importance of market data revenue to SRO operations and its relation to the broader issue of the Commission's oversight of the SROs. In light of recent reports of breakdowns of SRO governance and regulatory systems, the Commission is going to examine whether business and regulatory functions can effectively co-exist within a single institution or holding company structure or whether we need to consider other alternatives.

Now that the comment period on proposed Reg NMS has expired, and we begin the process of reviewing all of the comment letters we've received, the Commission's most difficult - but most important - task will be to separate myth from reality. As I have tried to point out, there are significant stakeholders on every side of nearly every issue in the market structure arena. Sifting through the rhetoric and the dueling charts, graphs and statistics is no small undertaking, even for an economist like myself. We need to make sure that our rules are not impeding efficient markets or subsidizing an inappropriate status quo, so we must look at the issues in terms of what is best for investors - rather than for existing institutions trying to protect their turf.

As an economist, my general philosophical preference is for market, rather than regulatory, solutions. In my view, the Commission shouldn't decide who should compete in the market or how they should compete. We have far too much on our plate to get involved in micromanaging the markets. So long as there are no natural or artificial barriers to competition, competition among market participants and among markets drives innovation and progress. But because our rules do affect competition to the extent that they can have the effect of entrenching certain business models, it is imperative that our focus be on getting rid of rules that impede competition.

Therefore, I am looking at the market structure proposal in terms of our goals for market structure reform: efficiency, transparency, competition and fairness. To determine whether our proposals further these objectives, we need to know the answers to some basic questions, and this is where we have to separate myth from reality. For example,

• Do customer limit orders need protection?
   
• If so, does the current trade-through rule protect customer limit orders?
   
• Would eliminating the trade-through rule or adopting an opt-out provision to the trade-through rule discourage the posting of limit orders?
   
• Is how the size of the market data revenue pie is determined and how the pieces of the pie are allocated a good way to fund SRO regulatory functions?
   

These are just a few of the questions we need to be able to answer. The questions are important to investors - both retail and institutional -- because the rules the Commission adopts will have an impact on investors' trading strategies and profitability. We need to make sure that our proposals do not have the effect of widening spreads, increasing transaction costs or impeding investors' ability to choose how and where to trade. To make sure this doesn't happen, we must look for solutions that increase efficiency and liquidity and remove impediments to competition. If we can do this, the Commission will have served investors well, and I will certainly be working to make that happen.

Thank you for your patience in following me through the market structure maze. I'd be happy to take some questions.

http://www.sec.gov/news/speech/spch092804cag.htm