Author:  "Ulrich; Joyce L."  at Internet
Date:    05/09/2000  2:50 PM

CC: "'Don Kittell'"  at Internet
CC: "'Scott Abbey'"  at Internet
CC: "'John Panchery'"  at Internet
CC: "Abbaei; Mike"  at Internet
CC: "Scheve; Timothy C."  at Internet
CC: "Brinkley; James W."  at Internet

Subject: Comments on File No. 4-430

------------------------------- Message Contents 

Legg Mason appreciates the opportunity to comment on File 4-430. Legg Mason has been pursuing the implementation of decimal pricing, both as a firm and as an active participant in three separate SIA-sponsored committees. We feel that this level of participation gives us a good insight into the issues surrounding both Dual Pricing and Decimals Pilot. Our comments are presented below.

In brief, we believe the Dual Pricing plan is not feasible, nor is it advisable. Legg Mason's preference would be to implement the original plan presented by the SIA, but on a delayed timetable - beginning not on July 3, 2000, as originally proposed, but on March 31, 2001, the date that has been published for full NASD readiness. The Decimals Pilot plan, as presented, represents a compromise position and contains some components that make it feasible but not necessarily advisable. With modified dates, the Decimals Pilot may become an acceptable alternative. The suggested modifications are discussed later in this comment letter.

It is important to understand the background of the original SIA plan in order to understand our concerns about the Dual Pricing plan and the Decimals Pilot plan. The original decimalization plan was devised with input from all interested industry participants, and was hammered out by the SIA Decimalization Steering Committee over the course of several months. It is of particular note that even this multi-phase plan had opponents who felt it was too complex to test and administer, and there was much discussion to determine the best course of action. After a great deal of healthy debate, the proposed phases were compressed to make them as short as reasonably possible. The actual number of phases (4) was chosen to permit the industry to cautiously introduce the additional volume that decimalization would generate, and to provide enough time to realistically measure capacity and performance and to react to same. The paramount goal, of course, is to efficiently implement decimalization in an orderly fashion without disruption to the markets and to maintain client confidence.

The proposed Dual Pricing plan calls for listed securities only to begin trading on September 4, 2000. Whether or not there would be two phases (first a nickel increment, followed by penny trading) is not yet decided. Non-listed securities would be added once NASD was fully "decimal ready" for non-listed securities. It is unclear if there would be an additional pilot phases (nickels, then pennies) for non-listed securities, but given the uncertainties about volumes, it would seem prudent to conduct a phase-in pilot of non-listed securities. As an alternate, the Decimals Pilot plan would begin on September 4, 2000, would include a small number of exchange-listed securities and their options, would go straight to penny pricing, and would continue until NASD was ready to implement decimals trading on non-listed securities, not later than March 31, 2000.

Our comments as to the feasibility and/or advisability of these two plans are as follows:

* Edits on order entry

It is important to understand the edits that have been planned, coded and tested in support of the "July 3rd" SIA implementation plan. In most standard cases, firms have opted to not pre-edit orders for the correct MPV ("minimum price variation"); rather, the order entry systems will send orders to the execution points and expect rejections for an incorrect MPV. The alternative was to create special edits for (slightly more than) 30 securities for the 5 week pilot period, and this is considered to be wasted effort. So, in lieu of special edits, many firms, (Legg Mason among them), intended to provide a "quick look up" of the 30 securities that were to be included in the nickel pilot. While this would have required manual maintenance of a small table that would have been made available to people entering orders, it was considered manageable and the most efficient method to manage the 5 week pilot period.

The proposed Dual Pricing plan would require a large and complex table, and it would not be possible to create and maintain these tables with the information needed to support the quantities of securities involved. Firms will be unwilling to accept an execution rejection problem with potentially high numbers of trades being rejected by the exchanges, or worse, customers could miss the market due to rejects. To get around these problems, firms will have to code, test and implement the special edits that could have been avoided otherwise. While the specs for this change may seem easy, the sheer number of places it would have to be installed, and the requirement to conduct regression testing make it almost certain this could not be ready by September. Also, note that this is truly "throw away" code that will not be useful once decimalization is complete.

The proposed Decimals Pilot plan lends itself to the use of a small, manually maintained table.

* Special Considerations for Internet trading

Almost without exception, firms that provide their clients with Internet trading facilities had planned to install an order entry edit to validate the correct MPV. With an Internet client, one must perform all of the requisite edits while the client is still logged on, because even with proper disclosures, a firm would not want to risk returning a (preventable) rejection to an account where the client is no longer logged on. The extra coding (and testing) to properly perform edits during the duration of the Dual Pricing plan is a daunting amount of work; so much so, that it is not feasible to accomplish before the proposed September start. The Decimals Pilot plan does not have the same complications.

* Testing

Good, controlled testing is the cornerstone of successful implementation. Good, controlled testing requires preparation of thorough test plans and scripts, (including expected results); careful execution of testing; a detailed review of results; corrective efforts and re-testing, as needed. Neither of these proposed alternatives include time to plan and execute testing, in fact, there are not enough known details to even permit test planning, and certainly neither can be thoroughly tested by the proposed start dates. The Dual Pricing plan is complex, so the testing is harder -- more scripts, and more iterations of point-to-point and extended point-to-point testing. To be assured of a smooth implementation, each phase, including conversion and fall back steps, would have to be tested independently.

The financial services industry has an excellent track record for good testing (T+3, Y2K), but those successes were only achieved by dint of hard work, detailed planning and excruciating attention to detail. There simply is not enough time to do a thorough job on testing either of these plans. Poor testing preparation leads to questionable testing and potentially unreliable results.

* Duration

The length of the implementation of the Dual Pricing plan increases the complexity to manage and administer the transition to decimals. This will require dedicated support resources for more than double the amount of time we had originally planned. This will affect the plans for other development projects for the remainder of 2000 and well into 2001, and we (at Legg Mason) use a service provider for our core back office functions. The support effect is magnified for firms that use their own in-house systems. One can only surmise that there may be an effect on the proposed date(s) for the industry's migration to T+1.

* Complexity

The Dual Pricing plan is too complex. There are too many steps, and too many conversions. Each additional phase introduces risk to the participants, and confusion to the investing public. The Decimal Pilot plan, while less complex, is potentially confusing, too.

It has been suggested that decimalization is the right thing to do to save the "investing public" money. While one can argue the monetary value of the purported savings to the investor, it is clear that the investing public will be confused by these plans. Take the case of two stocks that are perceived to be in the same industry - what is the different between Microsoft and IBM? How about AOL and Does the average investor know (or care) that MSFT is not listed and IBM is, so during this plan, MSFT trades in fractions and IBM in decimals? AOL will trade in decimals and won't? Even more confusing...AOL used to be traded via NASDAQ, but is now NYSE. Or, what of the case where an investor has securities that are both executed on the NASDAQ (one listed, one not), and one confirmation shows decimals and the other fractions? How many times will our brokers be required to explain this to their clients? It may even be supposed that client confusion could have an adverse affect on trading activities.

The uncertainties surrounding the method and timing of implementation of decimal trading have made it impossible to communicate any facts to our clients. Other than vague announcements that "securities will trade in decimals", no firm has been able to conduct a thorough educational campaign for clients that included concrete information such as the phases of decimalization, the timing, etc. It is already May, and we do not know enough facts to start the consumer education process. Even if a decision were made tomorrow, there simply is not enough time to conduct the proper consumer education.

* Volume Uncertainty:

The industry continues to experience and process record volumes. In 1999, the SIA commissioned SRI, a consulting firm, to conduct a study and predict the volume increases that could be expected with the implementation of decimalization. The industry is already handling volumes that exceed the original projections of volumes following the implementation of penny pricing (also known as "market price variations", or MPV). There is concern that the infrastructure is straining under the weight of the current volume, and to add more traffic due to decimalization is imprudent. Many firms already experience delays at normal market opens due to the volume of traffic and are working to implement increased capacity - what will happen when nickels, much less pennies, cause the volume to increase?

The market data and vendor committee of the SIA has already expressed concern as to the ability of firms to establish a reasonable baseline of throughput, test for capacity issues and increase capacity in the time remaining before the proposed September start. One pressing question - what should the target volume be? Should firms just adjust the SRI results to begin at the current peaks and project growth from there? In the absence of better data, many firms (market data providers and others) must do just that.

The Dual Pricing proposal neither addresses volume-related testing, nor does it allow any time to evaluate the introduction of penny pricing. The industry could be overwhelmed with no lead time to take corrective measures. The Decimal Pilot cautiously introduces small numbers of penny traded listed securities, and these could serve as the baseline for modeling future volume increases.

* Rule Changes

Most exchanges will be required to file "rule changes" with the SEC. At the recent SIA Decimalization Steering Committee (May 1st), there was a discussion as to the issues that are unresolved and the work that remains to be done. If there are still unresolved issues for the current SIA plan, there are certain to be issues for both the Dual Pricing and Decimals Pilot plans. There should be careful discussion and deliberation about the questions, and it is not clear that they could be resolved for a September 4th start.

* Options mitigation

The projected volumes in options quotes is very high. Currently, the OPRA system will increase capacity to 8,000 messages per second (mps) in April, with an additional increase in capacity to 12,000 mps in the summer. One study predicts that options quote traffic (could) increase to 36,000 mps with penny pricing. Efforts by the Options Mitigation Committee to agree upon and impose a traffic standard have been unsuccessful; as of now, the participants are "self regulating" their message volume. No one can predict the effects of penny trading on this message volume, and it is unclear if the "self regulation" will continue to provide the necessary relief. The proposed Dual Pricing plan does not plan to gradually introduce and measure this volume, and is therefore deemed ill advised. The Decimal Pilot does provide a cautious introduction of penny option quotes and the length of the pilot gives time to measure, react and correct, as needed.

* Testing readiness

Across the board, the industry is re-using Y2K test facilities to conduct point-to-point (PTP) tests between participants (the ability to connect and send files) and to conduct extended point-to-point (EPTP) tests that prove the ability to process those files. This should have eliminated testing errors attributable to "set up" problems. In addition, firms must demonstrate their ability to convert open order(s) from fractions to decimals.

The original deadline to complete "mandated" point-to-point testing was March 31, 2000. This has already been postponed to May 31, 2000 for most exchanges, but the NASD systems have not yet been made available for testing. The SIA has announced additional extended point-to-point test dates of June 10th and August 12th. The NASD systems for listed security testing are scheduled to be available for general testing efforts during the June 10th EPTP test.

The results from the April 8th extended point-to-point test suggest that the industry is making progress, but there were problems to be resolved. It may turn out that these are "environmental set up" issues (despite re-use of Y2K environments) that are easily resolved, but it may also turn out that there are problems with the conversion programs and ability to process decimals that require fixing and retesting. The May 13th test will be crucial in determining the underlying state of readiness. Again, that test will be without NASD. If we learn from history, then we should all realize that we won't be ready to implement without additional testing dates, and enough time to plan and conduct same, using the same "best testing practices" outlined above. This suggests it is premature to predict a September 4th implementation of any sort.

* Summary

There are too many unknowns in the proposed plans, and we do not believe there is enough time to address them before the September 4th start. Given the number and nature of the unknown variables, Legg Mason believes the Dual Pricing plan is not feasible, nor advisable. The Decimals Pilot plan is feasible, but not advisable with the currently stated deadlines. The industry needs more time to flesh out the details, and we propose a mid to late October start.

Joyce L. Ulrich
First Vice President, Brokerage Applications