agency or Commission has been formed, presumably the Boston area could begin implementing a system within 12 to 18 months. Extensions of existing dividend reinvestment systems to new securities issues could be done on a similar 12 to 18 month timetable. Development of settlement systems for NASDAQ would take longer, perhaps two to three years. Pressure of competition, by prohibiting extension of the NASDAQ testing phase until the settlement system is implemented, will encourage adherence to a minimum time schedule. The major point we would like to convey is that any systems change is major and requires time— the objective is to be sure the expenditure is justified in terms of the benefits to be realized.

The cost to implement these systems can only be guessed at. The principal though, of how these costs should be met, can be established. The transaction charge previously mentioned, perhaps an extension of the present SEC charge on security sales, should cover the cost of central planning and development group. Development of individual systems should be paid for directly by the members of the industry involved. These costs, especially to banks, will be recouped in transfer fees from more profitable operations. It is not necessary for the Federal Government to finance any of these systems changes.

Implementation of certificateless systems such as those proposed would provide for a diversity of industry processes, compatible with our concepts of free competition between capital markets. What would be standardized is the coding systems and formats for information interchange. In this area, legal changes could be helpful to improve the flexibility allowed in format of legal documents. For example, substituting reels of magnetic tape for individual transfer instructions, or delivery instructions.

Elimination of stock certificates, and the related improvement of communications in the settlement process between brokers and transfer agents could conceivably provide for next day settlement of all security trades. Without the need for manual intervention after the trading process there will be fewer problems of manpower shortages and clerical errors as volume increases. Without the need to physically present certificates, problems of mail delivery and physical movement and security thefts will be eliminated.

We estimate that a certificateless system of settlement and its related benefits could reduce payroll costs for clearing brokers by as much as 40-50%. The work of the cage, dividend, proxy, and stock record departments would be virtually eliminated. Margin and customer accounting operations would be simplified. One medium sized firm has estimated their annual expenses would be reduced by 2 million dollars. This amount is equal to half of their pre-tax net income.

Equally important would be the savings to out-of-town brokers if certificates were eliminated. It is conceivable that they no longer would be required to have a New York correspondent, and could thereby reduce clearing costs which now amount to as much as 20-25% of gross commissions. This would be an important competitive advantage for local brokers, outside of New York in competing with the large wire houses and New York based firms.

We believe the elimination of the certificate is vital and inevitable. The question is whether it can be done quickly enough to help the financial community before the next crisis. With prompt action, and strong support from this Committee we believe this goal can be achieved within the very near future.

At this point, we would like to respond to the specific questions outlined in the Committee letter of September 29, 1971.

(a) We believe it is entirely feasible to develop a fully automated stock transaction system. There are a number of examples of systems in other industries which have characteristics similar to that of the proposed automated stock transaction system. In addition, there are important elements of the proposed system which are in fact already implemented within the securities industry. Lybrand, for example, has conducted a study for several of its clients to compare the relative performance of automated security clearance systems. There are no less than a dozen different such systems available or proposed for use by individual brokerage firms. These security clearance systems control all movement of stock and transactions from the point of agreement of the trade with the other broker to the ultimate settlement with the customer. Examples of these systems include those in use by brokerage firms such as CBWL-Hayden Stone, Goldman Sachs & Co., Smith Barney, and several others. The study, which the Lybrand Data Processing consultants prepared, indicated such a system

would range in cost from $2,000,000 to $2,500,000 and could be installed in six months to one year.

The security clearance operation represents only one part, perhaps in terms of effort expended, from one third to one half of a brokerage firm's activity in processing a trade. The remaining brokerage effort, which would all be part of the proposed automated stock transaction system, includes the Trade Entry Procedure, or sending of the customer's order to the trading floor or order room, and then the Trade Execution Procedure, which is the matching of buy and sell orders of different brokers.

Many brokerage firms have already installed sophisticated Trade Entry Systems. The Control Data Corp. has designed such a system and installed it at Hornblower Weeks (that's the Horace System). Several other brokerage houses have similarly automated the procedure for routing orders from customers directly to the exchange floor or to the point of trade, and then sending the report of execution back to the customer together with a confirmation.

The third part of a broker's stock transaction system, the Trade Execution phase, has been the subject of extensive research by the American Stock Exchange, the N.Y. Stock Exchange and the NASD. The American Stock Exchange has made the greatest progress towards installing automated trade execution systems. The AMEX Floor Derived Clearance is an automated trade reporting system. It is already in use for processing Odd lots and is being implemented for 100 share round lot orders to be executed at the market price. This system is designed to tie directly into the computer processing systems of individual firms.

These examples illustrate the state of the art for those brokers with the more advanced data processing systems. The picture we have drawn demonstrates, we believe, that the essential elements from a broker's point of view, of an automated stock transaction system have been defined, and in some cases, solutions are being implemented.

Another important area, perhaps equal in size to that of the broker's processing problems, is the work of the bank transfer agents and registrars. A stock transaction system must also include the functions performed by the banks in their role as agent for issuing corporations. A number of banks have recently installed, or are in the process of installing improved transfer systems. The First National Bank of Boston has designed and installed a new stock transfer system which permits share transfer without the use of stock certificates. There are other examples of sophisticated computer systems now installed in banks which are equal in complexity, in terms of the equipment required, to that probably required by an automated stock transaction system.

In addition there are several examples, outside of the securities industry, of sophisticated data processing systems which are somewhat comparable to those required to solve the problems we are addressing here. Examples of such systems include the Social Security Administration's Computer Complex, and the Internal Revenue Service System.

(b) As stated earlier in this presentation, we believe it is premature to define a stock transaction system in any degree of detail. What is needed is the establishment of an appropriate Agency or Commission to guide the design and implementation of required systems. It is important to appreciate though, the large number of operating entities that will be affected. These include the customers (30 million shareholders), the stockbrokers (5,000 broker/dealers), stock exchanges (about 10), clearing houses (also about 10), transfer agents (2,000 to 3,000 banks) and registrars (also 2,000 to 3.000 banks).

As a result, the problems of installing the types of systems suggested is not one of hardware, or even of system design, but rather a huge problem of organization and coordination of the many independent entities. We are not aware of any precedent for a system which would match the proposed stock transaction system when compared on the basis of the number of diverse businesses involved. (c) The cost to design and implement a system of the magnitude being discussed would probably cost at least to $15 million, exclusive of equipment costs. the NASD has stated publicly that their new clearing system will cost $3.5 million to develop and install. The Lybrand study of automated security movement systems referred to earlier estimated systems costs of $2 million to $2.5 million. The potential benefits are difficult to assess, but we believe the savings to the brokerage industry would exceed $2 billion annually. We expect that most of these savings would be passed on to customers via lower commission rates. We

noted earlier in our presentation that certain brokerage operating costs would be reduced by 40% to 50%.

The benefits to bank transfer agents is less certain. They would clearly benefit by better, more controlled and more automated operations. A seminar for bankers which Lybrand sponsored in March, 1970 examined this subject in some detail. The economic benefits to bankers are not as significant as those to brokers, although indications are that transfer costs would be reduced, and transfer fees would also go down.

The benefits to shareholders, in terms of lower costs and more reliable service are the principal objectives to be realized. Without an improved stock transaction system, our trading markets would probably collapse when volume reaches the high levels projected for 1975 and 1980. Corporations obviously benefit from a reliable and effective capital market. They would also probably have lower transfer fees.

(d) The time to implement complicated processing systems is directly related to the effectiveness of the organization charged with that responsibility. The new SIPC organization was in operation, and collected over $13 million from 5,000 members in less than six months. The majority of banks installed magnetic ink encoding of checks in less than 5 years. It took American Airlines 10 years to install the Sabre Reservation System. We estimate such a system could be designed within two years. Effective implementation, assuming complete support from all parties and strong, effective direction with appropriate legal and regulatory support could be accomplished in another three to five years.

(e) There are a number of computer systems which are relied upon to process and control functions of a critical nature at least equal to that of the proposed stock transaction system. The command and control systems of our Defense Department is a notable example. The Apollo moon program has demonstrated the high degree of reliability and backup which can be built into sophisticated systems.

Within the business sector, the one-line savings systems of many banks are illustrative of critical operations which have been performed with complete reliability. The Howard Savings bank system has complete duplication of equipment including even its own auxiliary power supply. The degree of reliability which is needed can be designed into a system.

(ƒ) Duplicate equipment, and system design features referred to in the answer to question (e) above, can provide continuous operation under any foreseeable conditions.

(g) Conversion to any new system which involves many elements of a company's operation is a significant undertaking. With careful planning, and adequate training, changes can be implemented without destroying the ability of the industry to service its customers.

(h) Implementation of an integrated system cannot be achieved without the full cooperation of all related associations and businesses. We believe this is the most critical element of the entire problem. For this reason, as we stated earlier, we recommend the creation of a special Commission, or Task Force, similar in some respects to the recent Postal Commission. This Commission would have the responsibility, and the regulatory controls, to analyze the operations problems of the securities industry, and to direct the installation of an appropriate system or systems to resolve those problems.

(i) The design of new transaction systems, making extensive use of automated data processing equipment, will provide opportunities for more effective monitoring of security transactions. The key questions of uniform accounting procedures, however, usually concern the classification of related revenue and expense items. This will not be affected by a new stock transaction system.

(j) We have proposed the creation of a new Commission, specifically charged with the responsibility for developing a program to resolve the near term and long term problems of the securities industry. This Commission should also be given the necessary regulatory powers to properly direct the implementation of their program, and then to monitor its effectiveness.

(k) Legal changes are critical to insure the necessary uniformity and design flexibility to effectively change the processing of security transactions.

Professor John Steadman of the University of Pennsylvania Law School has done extensive research to support the use of Federal legislation in the area of stock transfer and the operation and functioning of the national securities market. His letter to former SEC Commissioner Richard B. Smith of April 5, 1971 (which he has made public) carefully documents this position.

Our firm, Lybrand, Ross Bros. & Montgomery, sponsored a seminar in May, 1970, on the legal problems involved in the operations problems of the securities market. The transcript of this seminar is available from Lybrand. It was also published in the January, 1971 issue of "The Business Lawyer". These works all describe in some detail the required changes in the legal structure.

(1) The method and extent of public information which is required will depend on the nature of the systems installed. It is very probable that the major part of any systems changes will not affect the general public or even most shareholders. For example, we don't envision a system which requires 30 million shareholders to send in all stock certificates for redemption. Any change involving the shareholders will almost certainly have to be one of evolution, for example by not issuing certificates on new purchases. Experience in the mutual fund area, and with no-passbook savings systems indicate the public has no trouble adopting to new techniques, as long as it is not detrimental to their interests, and certainly if it is an obvious advantage.

(m) We believe the establishment of a National Commission, as described earlier in this presentation, provide the best approach to identifying all problems concerned with the development and implementation of a stock transaction system. Such a Commission is also the best vehicle for resolving these problems. The concern we have is that in the absence of such a Commission, solutions will be adopted, and enacted into law, which appear attractive, but are not based on adequate investigation, and will not truly solve the operations problems of the security industry.

We very much appreciate the opportunity to present our views on this important subject. If we can be of further assistance, please do not hesitate to call on us.

Mr. Moss. The next panelist is Mr. Clifford Noyes, regional vice president of North American Rockwell Information Systems Company.

STATEMENT OF H. CLIFFORD NOYES, REGIONAL VICE PRESIDENT OF THE NORTH AMERICAN ROCKWELL INFORMATION SYSTEMS COMPANY

Mr. NOYES. Thank you, Mr. Chairman.

I appreciate being invited to be here today. I think what we are talking about primarily is the development of a major new system. Even more major, if we are talking about the trading process as well as the settlement process, as your questions seemed to imply.

With the development of any new system, we must consider using the methods and techniques which have been built up over the years and tested in the development of new systems.

Step 1 is normally the development of a concept based on defined and stated objectives. For instance, some of these objectives might be to minimize processing cost to an individual investor, perhaps to retain the control and stability that the specialist on the exchanges provides, perhaps to protect the interests of registrars, perhaps to centralize the trading of each security issue, perhaps not to deprive investors of the benefits of certificate possession. I think this hypothetical list could probably go on into the hundreds.

Some of the objectives or tentative objectives are important. Some perhaps trivial, some perhaps mutually exclusive. Some are achieveable and some are not. But we need the goals and we need the objectives. Then, as step 2, we can hypothesize a system and system alternatives, and we can evaluate these alternatives against our defined goals. Once having these, we can proceed with the development of a system by a

team with the knowledge and experience and resources and, perhaps most importantly, authority to implement the system.

It may include certificates and it may not. It seems to me, for the short term, a balance is necessary-with part of the system operating with certificates, but primarily automated.

Thank you, sir.

(The prepared statement submitted by Mr. Noyes follows:)

RESPONSES TO QUESTIONS RELATIVE TO THE DEVELOPMENT OF AN AUTOMATED STOCK TRANSACTION SYSTEM

Q. Can a fully automated system for effecting stock transactions be developed and implemented with equipment currently available?

A. Adequate communications equipment and computer processing equipment to accomplish the basic functions required of a fully automated system are currently available and currently operational. Similar applications-in general scope and function-have been in operation for several years. Certainly, in toto, the communications systems now exist which allow entry of orders at any spot in the nation, which transmit the order electronically through the computers of any one of a dozen firms and into the Exchanges. Similarly, reports of trades are passed via the same networks back to the branch offices of the member firms. Market data reporting systems are in operation—are operating in an online mode. They rapidly disseminate the information about each trade on the Exchanges out to a multitude of random access files which can be queried by a variety of cathode ray tube terminals.

The computing capacity required by an automated stock transaction system is not particularly large. While a total system for automated trading and settlement has never been built-never to my knowledge even been designed in detail-it seems that the implementation of any of the various concepts proposed would not require as much innovation or expense as systems like the American Airlines SABRE system, the Air Force's SAGE system, or the computer and ground communication networks required to support the Apollo missions.

The hardware exists, but the system to use it has not been designed. And, its design may well reveal both significant problem areas and opportunities for increased efficiency as the designers try to tie the hardware together.

So while the answer to the question proposed can only be "yes," the answer must be contingent to some extent upon the detailed design of the system, the development of detailed hardware specifications, and the matching of currently available equipment against those specifications.

Q. Describe the essential features of such a system.

A. The system already exists. What we are considering in essence is how much further to go with automation. Should we continue the development of an automated settlement system along the lines of CCS? Should we replace all or some of the certificate processing now being done with electronic bookkeeping? Should we replace some or all of order execution by manual means with order execution by computer? I view two basic functions involved; one, the trading function and two, the settlement function. If the trade system is automated, at least from the point of execution, the processes of matching and confirmation and clearance become either trivial or not required at all. At the time the trade is actually made, the complete data about that trade is available. All parties to the trade are known. The price and quantity are known. The issue is known. It is only when this data becomes separated through the present system of manual processing that the matching and confirmation (clearance) process becomes required. Similarly, if the settlement process is automated-such as within CCSnot subject to intervention by the parties involved, the clearing steps of netting and allocation are no longer necessary.

While we can talk about the advantages of automation in concept, we soon come to the realization that it may not be practical to completely automate the processes involved. One of the first questions relating to the automation of trading is, "What types of trades should be automated?" The Exchanges are already developing programs to automate the trading of odd lots and single round lots. The price of the trade is determined by the price of a previous non-automated trade. One side of the trade is the specialist. Now, if we are contemplating auto

67-228 O 72 pt. 3 - 16