May 21, 2000

Mr. Jonathan G. Katz
Securities and Exchange Commission
450 Fifth Street, NW
Stop 6 - 9
Washington, DC 20549-0609

Re: Commission Request for Comment on Market Fragmentation (File No. SR-NYSE-99-48)

Dear Mr. Katz:

American Century Investments1 appreciates the opportunity to comment on the Securities and Exchange Commission's concept release regarding market fragmentation. The structure of the securities markets has a significant impact on our two million investors who have invested more than $100 billion with American Century for their retirement, college savings and other long-term needs. Investor trust requires that we fight for highly competitive, transparent and efficient markets, and that we support regulatory structures that encourage, rather than impede, liquidity, transparency, and price discovery. 2 Consistent with these goals, we have strongly supported past regulatory efforts to improve the quality of the U.S. markets.3

American Century applauds the steady progress of the Commission as it has fostered markets with lower cost access and also illuminated practices that prey on investor inexperience. The Commission exercised courage as it called for decimal pricing of securities, limit order display rules across markets and the elimination of anti-competitive practices like New York Stock Exchange Rule 390. Such pro-competition, forward-looking initiatives advance the interests of investors who daily face the obscure and archaic rules of member-owned exchanges. Historically, much of the debate about market structure has been influenced and controlled disproportionately by the deep pockets and economic interests of entrenched financial intermediaries. Today's dialogue promises that the protection of investors' interests and growing objective data about hidden trading "charges" might supplant the nostalgia that has cradled and protected our nation's securities exchanges as they have grown increasingly inefficient.

The method and costs for securities trading have a direct bearing on investment performance for our investors. Both large and small investors suffer equally and proportionately. American Century consistently advocates policies that will encourage markets to provide investors with:

Investor Benefits from Emerging Trading Technologies

The 1992 American Century response to the Commission's Market 2000 Concept Release was one of only four filed by investor constituents. The issues in that release are eerily similar to those contained in this request for comment on market fragmentation. In the early 90's study, opponents to newly emerging and efficient technology platforms spawned language that asked whether such systems threatened "fragmentation," "segmentation," and "balkanization" of the nation's securities markets. Now, as then, such language seems only to reflect the howls from entrenched exchanges and brokers who have been insulated from competition by rules that masquerade as investor-friendly safeguards.

Our data indicate that American Century's overall trading costs, including commissions and market impact have fallen steadily and progressively throughout the 1990s.

American Century has relied on SEI Corporation over the years to help measure trading costs against industry metrics. The chart above demonstrates that:

It should be emphasized that investors bear the entire cost of securities trading in a mutual fund portfolio. The cost of commissions constitutes the most visible and easy to measure variable of trading costs. That said, numerous academic studies suggest that the largest component of trading costs are the "invisible" costs attached to market impact - or how much the market moves when intermediaries identify the presence of a large order in the market. The presence of such "market-moving" orders are regularly communicated to a host of competitive interests, including hedge funds, other mutual fund customers, day traders and proprietary trading interests (e.g. the specialist). 5

SEI analysis of American Century trading practices suggests that the use of electronic brokers wrings significant excess costs out of the trading process.

The SEI methodology for measuring market impact indicates that:

Some suggest that such trading cost data does not effectively account for differences in trading difficulty or other elusive, more difficult to measure best execution variables. Recent work by Domowitz and Steill suggests that electronic markets significantly reduce the cost of trading, when adjusted for trade difficulty, as compared to traditional exchanges both for listed and NASDAQ securities.7 The ten-year measure of American Century trading costs augments the conclusions from more recent studies.8

The above chart highlights the following:

The enactment of the Order Handling Rules in mid-1997 generated significant additional savings. During data periods from mid-1997, American Century's trading costs for NASDAQ securities have been steadily declining despite a period of relatively high stock price volatility.9 In three out of four periods since enactment of the Order Handling Rules, American Century experienced lower trading costs for NASDAQ securities than for NYSE-listed securities in marked contrast to our historical experience.

The efficiency of such systems appears to extend only to NASDAQ-registered securities. The NYSE somehow escapes obligations under the Order Handling Rules and continues to refuse direct electronic access by investors to the specialist order book. Such behavior contravenes the spirit of the 1975 Amendments to the Securities Exchange Act that calls for markets to provide buyers and sellers an opportunity to discover prices "without the intermediation of a dealer."

Regulatory Requirements

The SEC has long sought to respond to the language of the 1975 Amendments to the Securities Exchange Act and its responsibility to foster markets that:

A recent survey of more than 40 traders of major institutions showed strong consensus that "ideal market" attributes would include:

American Century supports regulatory attempts to compel transparency and "quote" competition. The current state of affairs in the NASDAQ market with increasing fragmentation attributable to internalization practices of wholesale market makers, and at the NYSE where order priority rules are not clear and transparent, means that intermediary self-interest trumps minimum standards that the buyside requires to obtain best execution for investors. 10

The explosive growth of ECNs in the U.S. marketplace and across Europe validates such surveys of investor preference. If ECNs were allowed to compete directly with the NYSE, sizeable economic benefits would accrue to investors.

For years, American Century has advocated a strict price and time priority intermarket linkage. The ideal market for investors would foster maximum order interaction and transparency. At the same time, recent experience with the ITS trading system, the rules of trading at the NYSE and the structural problems related to ECNs and mandated SelectNet linkages suggests that the SEC should focus first on order disclosure, priority and interaction rules within individual markets before attempting to impose such rules across markets.

Markets are Fragmented by Internalization and Payment for Order Flow

While we do not perceive a "fragmentation" threat from the emergence of new automated transaction systems, we are nonetheless troubled by the "fragmentation" foisted upon the market by dealer intermediaries who either internalize or pay for order flow. Many wholesale broker dealers and exchange markets rely on the national market's disclosure of visible, limit orders to control, price and trade market orders generated by captive constituencies. Investors who display the desire to trade by using limit orders subsidize internalization practices. Ultimately, if the Commission allows internalization business models to proliferate then limit orders will be used with less and less frequency by investors, with a concomitant increase in stock price volatility.

Internalization and payment for order flow threaten transparency of trading processes. For instance, brokers now regularly receive so-called "VWAP"11 orders for specific securities - sometimes hundreds of thousands of shares - on both sides of the market. Potentially millions of shares are now internalized at the VWAP price, without participating in the market's price discovery processes and without order interaction with other investor orders. Many of these orders are now "book entered" overseas by U.S. brokers at the end of the trading day.

Major wholesale market-making firms, whose business models rely on payment for order flow, were engaged in about one-third of all trades in NASDAQ securities last year and have seen increasing market share gains as more and more retail investors trade individual stocks. At the same time, market makers often are represented in the market as the "best" buying or selling price only 25% of the time in many high profile, actively traded stocks while ECNs typically represent the market's best price more than 50% of the time.12

The chart above shows that:

Internalization practices prove anti-competitive in a host of trading venues - and the traditional exchange markets retain "members-only" benefits, marketing agreements and other "practices" that erode our confidence in trading on even the most well branded exchanges.

The NYSE Relies on Layers of Internalizing Rules

The NYSE suggests that fragmentation caused by internalization might be mitigated by a market-wide price improvement rule. In other words, orders could only be internalized by dealers who pay a price "better" than the national market's best prices for both buys and sells. In principal, we believe that this simple change in intermarket rules would be favorable for investors. At the same time, we can't understand how the NYSE might impose that rule on other exchanges or dealers when much of the physical, floor-based model of that exchange relies fundamentally on layers of such internalizing rules:

Excess Rents Charged by Non-Competing Specialists at the NYSE

At the New York Stock Exchange, orders cannot be traded without the intervention of a dealer - the specialist. Not much data exists about the profitability of specialist operations but the recent prospectus offering by LaBranche, a large NYSE specialist firm, provided a glimpse of the following:

How do they earn such economic rents? The designation of the NYSE as the "primary" market on which all other pricing should be based has established the NYSE as the principal "operating system" for the market. That status has been conferred upon member-owned exchanges by the Congress and by SEC regulatory interpretations over time.

An analogous situation might be the establishment by Congress of Microsoft as the official operating standard for the computer. Obviously Microsoft was the first to discover the true power of a standard operating system for the desktop computer. Recent events would suggest that Microsoft also discovered that bundling increasing numbers and kinds of software applications into the operating system pleased consumers - but displeased those who believe that competition spurs innovation.

The NYSE sits as the sole arbiter on a number of shared exchange operating committees like the Intermarket Trading System - the effective operating system for the nation's exchanges. The use of veto power in a number of these venues successfully stymies efforts to stimulate interconnectivity of markets. And the NYSE has, over the years, incorporated innovations begun at regional markets only after a concept is proven and considered a real threat to the established hierarchy of exchanges.

Internalization Subverts Competition Between Dominant and non-Dominant Market Centers

The recent launch of the OptiMark trading utility of the Pacific Stock Exchange (PSE) provides needed insight into the anti-competitive practices at both the primary and regional exchange markets.14 OptiMark's system relied on effective linkages between markets - as promised by the Intermarket Trading System (ITS). The historical record reflects that the SEC was forced to broker a compromise between competing parties on the ITS operating committee after principals exhausted more than a year in fruitless, back room debate on how OptiMark could or should be linked to the NYSE market. That process helped draw down a new competitor's intellectual and economic resources required to compete effectively. The final solution reflected the economic staying power of NYSE monopoly position rather than a rational attempt to further the goals of a National Market System.

Shortly after effective launch of the new system, the NYSE effectively shut down ITS access to the OptiMark utility - based on arbitrary volume limitations. As an early user of the utility, American Century documented numerous failures by NYSE specialists to abide by conventions of the ITS agreement and filed that report with both the Commission and exchange officials.15 In several documented cases, American Century was unable to execute orders sent from the Pacific Exchange to New York that were subsequently and immediately executed when sent through the NYSE-proprietary SuperDOT, order delivery system.

American Century experience with the OptiMark utility also produced an eye-opening understanding of the marketing arrangements between regional exchanges and wholesale market makers that subvert competitive quote-making among exchanges. American Century recognized early that the Pacific Stock Exchange (PSE), in combination with the OptiMark utility, might provide an effective mechanism to generate competing quotes with the NYSE. Archipelago, LLC constructed an order display link to the PSE similar to that already in existence at the NYSE. American Century traders subsequently sent orders to the PSE that "improved" the NYSE best bids and offers. Those orders were intended to serve as an advertisement on the Consolidated Quote System of potentially larger block trading opportunities available through the OptiMark utility. Instead, American Century found that PSE specialists almost immediately sent these "price improving" limit orders across ITS from the West Coast to the NYSE specialists' books.

In trying to provide price competition, American Century discovered that PSE specialists maintain marketing agreements with a number of broker-dealers who internalize order flow. To protect those firms' ability to internalize customer orders, the specialist offers "primary market protection" on the regional floors. That protection essentially promises "internalizing" firms that if stocks trade at the retail order's price on the "primary" exchange, the PSE specialist will use his capital to execute the order at that price on the regional exchange.

There were three problems with the new competing orders that American Century sent for display to the PSE.

When American Century offered to forego "primary market protection," the specialists and staff asked that American Century allow small retail orders to be traded in front of the large order, even if they arrived at a later time. The PSE could not excuse one firm from that "marketing arrangement" without compromising the entire business model of the exchange.

This is prima facie evidence that regional exchanges - under current rules - provide neither competitive quoting nor innovation. If markets find it difficult to attract order flow on the basis of quote competition, and instead rely on internalization and payment for order flow, they should not exist. Regulatory subsidies of such empty pro-competition promises must stop. The recent announcement that the PSE and Archipelago will attempt to create a truly competitive and automated stock exchange might well change this existing state of the business. Promises of price and time priority and electronic, non-intermediated access to the PSE should not be perpetually stalled by entrenched and dominant exchanges, operating committees and member firms - all cloaked in concerns about investor protection.

Block Trading Deserves No Exemptions from Order Interaction Rules

The concept release asks for specific guidance about the need for block trading exemptions from any price and time priority and/or inter-market price improvement requirements. Such exemptions, if granted, confer separate benefits and establish institutions as a separate class from other investors. Institutions should not be allowed to use the visibility of limit orders submitted by other institutions and retail investors to "size" the market without protecting those orders during the consummation of a block trade.

Such an exception creates enormous economic subsidies that artificially lower the negotiated cost of capital available from brokers for certain clients. Brokers provide a valuable service when they "rent" liquidity services to investors. The pricing of that service should reflect the needs and desires of the institution demanding such a service without extracting high economic cross subsidies from all market participants.

Many market commentators defend the provision of broker capital to facilitate block trades as an important contributor to market liquidity. They make a corollary argument that block trading cannot be performed effectively on ECNs and that block trades should be exempted from order interaction obligations. Brokers traditionally "rent" capital when trading a block and natural counter-parties occur only a small percentage of the time.16 Historically, brokers negotiated block transactions on the phone and acted as "small order aggregators" using networks of small retail market makers, so-called wholesale market makers, to "work" large blocks into the market's flow by telephone and more recently on SelectNet or on ECNs. Today, ECNs help eliminate the risk premium that large institutions pay to trade as technology replaces capital in that small order aggregation process.17

American Century data help illustrate how ECNs facilitate the trading of extremely large blocks of stock - effectively replacing broker capital as a more cost effective way to effectively piece large blocks of stock into the marketplace.

The above chart illustrates the following:

To further examine the efficacy of ECNs in aggregating small order volumes, the ten largest single day trades on ECNs for the same three-month period are listed above. The illustration of the single day trading volumes combined with the objective measurement of trading costs suggests that technology provides a better alternative than the high costs of broker capital required to achieve an immediate execution in the market.

The SEC Should Regulate Within Markets Before Reaching Across Markets

The explosive growth of the Internet in part provides the solutions to these vexing issues. Electronic, non-intermediated auctions on-line are drawing huge resources and attention from buyers and sellers of Beanie Babies, airline tickets and auto parts. Ford, Chrysler and GM seek such a venue to reduce supplier costs. Why do member firms of the major exchanges resist the major virtues that their own securities analysts extol as "beneficial" to the economy in enterprises outside of securities trading?

The Commission should scrutinize and regulate order interaction rules within individual markets and refrain from regulating the technology that might be used to integrate markets. Already, eight ECNs are discussing the building of a virtual private network to link order books and dispersed pools of liquidity. These robust networks tie together systems that already incorporate the four major tenets of the buyside investor's "ideal market" without creating a single point of technology failure - a true virtual limit order book. The systems recognize that the "market" no longer is limited by access to the physical trading floor on Wall Street.19

The Commission's time is misspent in trying to regulate the technology capacity of markets. If competitively-limiting linkages like ITS were simply eliminated, a market unimpeded by outdated rules would quickly establish network linkages as a direct response to customer demand. These linkages would recognize and penalize inefficient systems - like those currently operated by the NASDAQ stock market - where legacy systems impede the delivery and reliability of trading information.

Insight into the regulator's role over tariff setting within the securities markets can also be gained from the implementation of the Order Handling Rules. The sudden mixing of dealer systems (where all customer charges are implicit) and of agency auction systems (ECNs who explicitly charge for access) squarely placed the Commission in the role as a rate-setter. Dealers argued they should not have to pay for access to ECNs because dealer-to-dealer trade in NASDAQ is "free." However, dealer trading is free only if dealers have zero profitability in the business model. Rather, dealer trading imputes hidden tariffs. Most investors would prefer to see charges assessed explicitly.

Competing venues should be able to charge whatever fee for access to a proprietary pool of liquidity would be economically competitive. That would imply that linkages could not be forced upon centers but that such centers must accept orders from other markets or exchanges. We would expect that technology integrators would quickly create algorithms allowing investors to choose a trading venue based on price, time, cost of access, and systemic dependability metrics.20 If one market's cost of access was too high or the system's response too slow, one would expect liquidity to migrate to the most dependable, lowest cost, and most secure venue.

Decimal Trading Should Put Economic Pressure on Payment for Order Flow

Recent concerns expressed by a number of trading constituencies reflect worries about proprietary trading interests "stepping ahead" of limit orders for no more than 1c and the effect that might have on limit order disclosure. A number of recent technology enhancements and other market technology changes "dead ahead" reduce greatly any worry that this might occur. Additionally, a rapid move to decimal prices should eliminate many of the economic incentives that promote internalization practices.

Island ECN already allows trading by wholesale market makers in increments as fine as 1/256. That equates to .333 cents. The move to decimals simply makes visible to the public what is seen by market makers and sophisticated traders, as was the case with the implementation of the Order Handling Rules. There are a number of venues in Europe already announcing plans to launch trading in SP 500 stocks there -- in decimals. Will they honor U.S. regulatory conventions or view "splitting of the spread" as a competitive advantage? Will U.S. alternative trading venues disallow smaller price increments despite U.S. trader preferences to use finer pricing increments?

The failure of the NYSE to establish clear rules for order priority has allowed stepping ahead by competing interests for years, as was explained earlier in this document. The move to decimals should curtail predatory activities of the NYSE floor "crowd" because the economics of trading and clearing costs make that a more unprofitable activity. Finer price increments will remove much of the floor information advantage that the speculative crowd gains whenever a large order arrives at the specialist post.

Reserve Books and Discretionary Pricing Features Encourage Limit Orders

Contrary to conventional wisdom, American Century does not expect that small orders will litter the marketplace when decimal pricing is introduced. Most firms have neither the technology nor the patience to move the supply and demand balance a penny at a time. Institutional firms will probably place orders in nickel, dime or quarter increments. Retail orders, handled by wholesalers, will either internalize against institutional orders or interact at the book's prices. Furthermore, modern electronic markets (e.g., most ECNs) contain "reserve" or "back book" features that allow participants to post a smaller order size to gain time priority at a price. Like an iceberg, the public quote may be only the tip of a big order hidden behind. With the elimination of NYSE Rule 390 and the possible inclusion of ECNs as participants in ITS and the public quote, this will create risk to participants who do not disclose orders.

Additionally, at least two ECNs (Bloomberg and Archipelago) offer a solution for possible "penny jumping" as they allow placement of orders "with discretion." For example, a buyer may place a buy order on the book to pay 20 with non-transparent instructions to pay as much as 20.15, if a counter-party reveals a selling interest at that price. The potential existence of discretionary orders, in conjunction with a "reserve" book, should dampen price volatility and compel posting of limit orders.

Summary and Conclusions

The Commission set forth a series of six possible regulatory frameworks for dealing with issues of fragmentation and internalization practices within the U.S. equity marketplace. None of the approaches currently envisioned can overcome the entrenched rules and practices of the U.S. Securities business. Rather than reach across markets to establish an effective National Market System, the Commission should first focus on the building blocks of such a system by requiring each U.S. national securities exchange and Alternative Trading Systems to adhere to ideas promoted in regulatory option "f" which would provide for:

An intermarket "price improvement" feature would be beneficial to investors. That rule would allow internalization of market orders only at prices better than those quoted among competing market centers - even a penny of price improvement changes the economics of the payment for order flow business.

As a second step, the Commission might consider "busting" monopolies that control the order interaction rules across markets (e.g. ITS) and the collection and pricing of transactions from primary exchanges to the investing public. Such important public utilities should be overseen by independent boards who work in the best interests of the investing public. A single "black ball" vote by a major exchange should not impede competition and efficiency. Such boards should represent exchanges, brokers, investors and listing companies.

A "hard CLOB" or central limit order book which consolidates all orders residing in disparate venues is undesirable and would be subject to a single point of technology failure.

Truly competing market centers and exchanges should be required to accept non-intermediated, electronic orders from other exchanges or markets; furthermore each market center should be free to establish the cost for access to that market.

True competition often creates fear and uncertainty. The Commission must not fear such competition as the structure of U.S. capital markets continues to evolve. American Century appreciates the opportunity to comment on these important matters relating to technology, fragmentation and the nation's securities markets.


Harold S. Bradley
Senior Vice President
Steven C. Klein
Head of Global Trading
John Wheeler
Manager, U.S. Trading
Gregory H. Bokach
Senior U.S. Equity Trader
NASDAQ Quality of Markets Committee

1 American Century Companies, a Delaware Corporation located in Kansas City, MO, was incorporated in 1958 and is the holding company of a registered investment advisor. American Century offers a variety of asset management and investment advisory services and acts as investment advisors to open-end investment companies registered under the Investment Company Act of 1940.
2 American Century actively invests in new companies and technologies that should enhance U.S. market quality by promoting competition, price discovery and lower costs for investors. Such investments include OptiMark Technologies, Archipelago, LLC, TradePoint Financial Networks, U.K. and W.R. Hambrecht and Co.
3 American Century traders have represented investor interests as members of the NYSE Institutional Trader Advisory Committee, the NASDAQ Quality of Markets Committee, the NASDAQ Subcommittee on Call Market Openings, the ICI Equity Markets Advisory Committee, Institutional Investors TraderForum and on various AIMR task forces studying market structure issues.
4 These concepts do not represent novel approaches to market structure. Rather they reflect the standard practices governing many established securities exchanges in Europe and most of the new technology trading platforms in the U.S., e.g. Instinet, Archipelago and Bloomberg TradeBook.
5 See "Automation, Trading Costs, and the Structure of the Securities Trading Industry" Domowitz and Steill, 1999, updated 2000.
6 Source Lipper Securities.
7 See "Automation, Trading Costs, and the Structure of the Securities Trading Industry," Domowitz and Steill, 1999, updated 2000.
8 Non-traditional data reflects life of trading at American Century; e.g. B-Trade since 1997.
9 The recent rapid increases in overall stock market capitalization and volatility are disproportionately represented in ECN data which comprise only about three years data on average.
10 Economides and Schwartz, "Assessing Asset Managers' Demand for Immediacy: Equity Trading Practices and Market Structure" substantiates the institutional traders' desire for such rules of engagement.
11 Many trading consultants now evaluate the efficacy of trading by measuring how closely trades approximate the day's Volume Weighted Average Price (VWAP) for all trades in a given stock.
12 Source: NASDAQ Stock Market data
13 American Century became familiar with this practice when documenting irregularities with the handling of orders transmitted on our behalf from the Pacific Coast Exchange to the NYSE through the Intermarket Trading System (ITS).
14 American Century, along with a number of major Wall Street firms, owns a small equity position in the OptiMark auction utility.
15 "Study of ITS Abuses at NYSE," American Century Investments, April 29, 1999, attached. See example of trading in BEL, 3-23-99 and MO, 3-25-99.
16 Anecdotal evidence suggests that less than 20% of block trades represent the agency matching of two customer orders for large broker dealers.
17 Brokers regularly avail generous capital for block facilitation to the largest commission payers. That commitment is reflected in the "loss ratios" of such accounts that reflect the percentage of gross annual commissions "lost" in the ongoing commitment of capital. Small institutions and retail investors help underwrite the commitment of block trading capital to the largest accounts.
18 The block equivalent was derived by dividing the principal amount traded in a specific security each day and dividing that amount by an average share price of $40, for comparison purposes.
19 See "Automation, Trading Costs, and the Structure of the Securities Trading Industry," Domowitz and Steill, 1999, updated 2000.
20 TradeScape regularly routes orders to the ECN with the quickest response times.