Securities Industry: Additional Testing Needed to Ensure Efficient Post-Trade Processing of Stocks

Published by the Government Accountability Office on 1990-09-26.

Below is a raw (and likely hideous) rendition of the original report. (PDF)

                                          Additional Testing
                                          Needed to Ensure
                                          Efficient Post-Trade
                                          Processing of Stocks

- ..--.---_-~~               -___   --.----
(;A( I,, I kI’1’15( ‘-!wcl

I       ’
~   4                                                                                                                                              I
            ,,. ,_ .,,, ..I   ..l,l _..._ . l-l.   .._ -.._-   -.....   . - ._-.-   .._.....---...   -.   - -.----   . -__--   ..--..______...__
GAO                General Accounting Office
                   Washington, D.C. 20548

                   Information Management and
                   Technology Division

                   The Honorable John D. Dingell
                   Chairman, Committee on Energy
                     and Commerce
                   Houseof Representatives
                   Dear Mr. Chairman:
                   During the October 1987 market crash, the U.S. stock markets experi-
                   encedbacklogs in correcting problem trades, where buyers and sellers
                   disagreeon the terms of their transactions. The backlogs during the
                   week of October 19, 1987, resulted in delays in determining market par-
                   ticipants’ financial obligations totalling over one billion dollars. This
                   report respondsto your June 27, 1989, request for information on (1)
                   automated systems developedto correct problem trades more expedi-
                   tiously; (2) the extent to which these systems have beentested to ensure
                   that they can processincreasedwork load volumes such as those experi-
                   encedduring the 1987 market crash; (3) the impact these systems will
                   have on reducing the 5-day period neededto clear and settle stock
                   trades; and (4) the role the Securities and ExchangeCommission(SIX)
                   played in reviewing and approving these systems.
                   As specified in your request, we reviewed the New York Stock
                   Exchange’sOvernight Comparison System,the American Stock
                   Exchange’sIntra-Day ComparisonSystem, and the National Association
                   of Securities Dealer’s (NASD) Automated Confirmation Transaction
                   System. As agreedwith your office, we also reviewed the Redesigned
                   Comparison System used by the National Securities Clearing Corpora-
                   tion (NSCC) becauseit plays a critical role in identifying problem trades
                   and transmitting them to the stock markets for correction processing.
                   These four systems are a collective effort by the exchanges,NASD, and
                   NSCCto streamline the processfor correcting problem trades. Details of
                   our objectives, scope,and methodology appear in appendix I.

                   The exchanges,’NASD, and NSCCare improving the post-trade processing
Results in Brief   of stocks. Specifically, they (1) implemented computer systems in the
                   last year that processthe results of each day’s trading activity earlier
                   than before, and (2) reduced reliance on inefficient manual correction

                   ‘For purposes of this report, the term exchanges refers to the New York Stock Exchange and Amer-
                   ican Stock Exchange.

                   Page 1                                                    GAO/IMTECXO-83Securities Industry



             processes.Consequently,these organizations have cut the time needed
             to correct problem trades from 3 days to 1 day. However, obstacles
             remain to reducing the overall time required to settle stock transactions.
             As a result, clearanceand settlement still takes 6 days, and the risks to
             market participants and clearing organizations associatedwith waiting
             6 days to receive payments remain.
             We also found that the new systemsused to improve clearanceand set-
             tlement were not stress-testedto assesstheir capability to processan-
             ticipated peak work loads. Specifically, while these four organizations,
             to varying degrees,performed systemsdevelopment tests prior to sys-
             tems implementation, they did not fully test these automated systemsin
             an operational environment that simulates extreme market conditions.
             For this reason,the securities industry and its participants cannot be
             completely assuredthat these new automated systemswill be able to
             correct trades promptly and accurately during peak processingperiods.
             However, since these systemswere implemented, these organizations
             have tested or plan to further test their systemsunder high-volume
             transaction work loads.
             With regard to SEC’s role in reviewing the development and implementa-
             tion of these systems,the commissionhas begun to strengthen its over-
             sight. It has (1) issued an automation policy that recommendsthat the
             stock exchangesand NASD assessand test their systems’ capacities and
             (2) established a group to assesscompliance with the new policy. How-
             ever, as we previously reported, the commission’soversight of auto-
             mated systems remains superficial.2 For example, SECdid not assess
             these systemsduring development and testing to ensure that the
             exchanges,NASD, and NY.32adhered to critical systems development prac-
             tices. As a result, SECcannot be certain that these systemscan consis-
             tently processproblem trades, especially during extreme market

             Clearanceand settlement takes place after trades have been executed
Background   and are the processesthat traders use to complete their transactions.
             Clearanceinvolves collecting and matching data on the terms of a
             trade-such as price- from traders who buy and sell stocks, and typi-
             cally takes 1 day. Settlement is the processwhereby buyers and sellers
             2SeeFinancial Markets: Oversight of Automation Used to Clear and Settle Trades Is Uneven (GAO/
             IMTE-90-47 July 12 1990), Financial Markets: Tighter Computer Security Needed (GAc,IMTEG
             00-16, Jan. 6, iSSO>,&d Financial Markets: Preliminary Observations On The October 1087 Crash
             (GAO/GGD88-38,Jan. 26,1988).

             Page 2                                                   GAO/lMTEG9O-83Securities Industry

exchangefunds for stocks traded, and generally takes 6 days. Because
the two processesoccur concurrently, completing them only takes 6 bus-
iness days after the trade date. Prior to the implementation of the new
automated comparison and correction systems,trades used to clear by
the third day and settle on the fifth.
Clearanceand settlement is performed by exchanges,NASD, and clearing
organizations that work together to gather and compare trades and
arrange for them to be settled by the traders. Most trades are success-
fully matched and settled without complications. Trades are matched
using key elementsof each transaction, such as stock price and quan-
tity, to ensure that buyers and sellers each have accurate data and as
such, agreeto the terms of the trade. Disagreementsbetween buyers and
sellers on trade terms result in problem trades that must be resolved
before the transactions can be cleared and settled.
The extraordinarily high trading volume during the 1987 market crash
created backlogs and increasedthe number of problem trades. For
example, before the crash, the normal percentageof unmatched trades
on the New York Stock Exchange,the American Stock Exchange,and
the NASD was 1.6,2.4, and 6.7 percent, respectively. However, during the
market crash, these percentagesfor the New York, American, and NASD
more than doubled to 3.4,6.6, and 12.8 percent, respectively. To resolve
the October 1987 backlogs,the stock markets closedearly for several
days, and trading firms and exchangepersonnel worked evenings and
weekendsuntil the trade correction problems were eliminated. NMD also
extended the operating hours of its automated correction system-the
Trade Acceptanceand Reconciliation System-to provide traders with
more time to resolve their transactions, and NSCCprovided trading firms
with additional time to report corrected problem trades to the clearing
After the crash, the primary federal regulator of the stock markets
(SEC);the President’s Working Group on Financial Markets;3and an
international private-sector group-called the Group of Thirty4 -
studied the manual trade correction processesand the S-day period

“The Working Group on Financial Markets was created by the President in March 1088 to identify
issues, make recommendations, and seek resolution of the complex problems raised by the crash. The
Working Group is chaired by the Secretary of the Treasury, and its members include the Chairmen of
the Commodity Futures Trading Commission, the Federal Reserve System, and the SEC.
4The Group of Thirty studied the clearance and settlement systems in the world securities markets
and made recommendations to standardize these systems across international borders. The group’s
members are from the United States and other countries’ financial services industries.

Page 3                                                     GAO/IMTJSG90-83Securities Industry

                                            neededto clear and settle stocks and made recommendationsfor
                                            improving the clearanceand settlement processin these two areas.SEC
                                            and the Working Group generally recommendedthat (1) exchanges
                                            automate their trade correction processesand (2) the securities industry
                                            shorten the 6-day period for stock clearanceand settlement. The Group
                                            of Thirty recommendedthat, by 1992,trades be settled by the third bus-
                                            iness day after the trade.

                                            In responseto the trade correction problems experiencedin October
Securities Industry                         1987, the stock exchanges,NASD, and NSCC developedand implemented
Has Implemented                             computer systemsthat speedup the gathering, comparison, and correc-
Computer Systems to                         tion of trade data. The four systems,their functions, and implementa-
                                            tion dates are summarized in table 1 below.
Address Trade
Correction Delays

Table 1: Trade Comparison and Correction Sybtemr, Primary System Functions, and Implementation Dates
                                                                           Primary System Functions
                                                                     W;parer trade unmatched trade                        Date of
Orasniratlon                       System name                                          data                              imolementation
American Stock Exchange              Intra-Day Comparison System                                    X                     November 1989
;t;p,:;l Association of Securities   Automated Confirmation Transaction       X                     X                     February 1990
National Securities Clearing         Redesigned Comparison System             X                                           August 1989
New York Stock Exchanae              Overniaht Comoarison Svstem                                    X                     Julv 1989

                                            These systemsenable these stock markets to compare and correct all
                                            trades, including problem ones,within 1 day after the trade date. NASD’S
                                            system can perform these tasks even soonerin that trades can be com-
                                            pared and corrected on the sameday they are executedm6   By speedingup
                                            trade comparison and correction, these four systemsoffer the stock
                                            markets the capability to avoid the delays associatedwith correcting
                                            trades that were encounteredduring the 1987 crash. While these sys-
                                            tems have yet to perform in an operational environment under extreme
                                            work load conditions, to date they have successfully processedproblem
                                            trades within established time frames. A detailed description of how
                                            each system operates is included in appendix II.

                                            “The American Stock Exchange designed its system so it could be modified in the future to compare
                                            and correct trades on their execution day.

                                            Page4                                                       GAO/IMTJ3G9O-t33!SecurkiesInduetry



                      Testing systemsto assesstheir ability to processdata during periods of
Tests to Assess       peak work load, commonly referred to as stresstesting, helps identify
Performance at Peak   and correct system weaknessesbefore they causedata processingdis-
Work Volumes Have     ruptions in a live operating environment. For this reason, in November
                      1989, SEC issued a policy statement recommendingthat the stock
Not Been Completed    exchangesand NASD, amongothers, conduct such tests.6We evaluated
                      the extent to which these organizations had conducted stress tests to
                      validate that each system and its market participants could process,
                      within established time frames, peak work loads anticipated by these
                      organizations during extreme trading periods. In each case,anticipated
                      extreme work loads were in excessof those encounteredduring October
                      We found that none of these organizations conducted stresstests to sim-
                      ulate the behavior of systems and market participants in an operational
                      environment with anticipated peak work load levels. While each organi-
                      zation, to varying degrees,conductedtests that provided them indica-
                      tions that their systems could processvolumes roughly equal to or in
                      excessof those experiencedduring the October 1987 crash, this testing
                      failed to provide complete assurancethat these systemscould process
                      such work loads. Specifically, the tests either did not validate that (1)
                      system performance (including, for example, its responsetime) was
                      acceptable;(2) the system could handle the number of corrections antici-
                      pated under extreme market conditions; (3) the system could support all
                      users expected to be on the system; or (4) the system could process
                      trade data simultaneously with other applications expectedto coexist on
                      the system. For example, before its system becameoperational, NSCC
                      successfully demonstrated that it could processwork loads in excessof
                      those experiencedduring the October 1987 market crash. However, in
                      this test, the system was dedicated solely to comparing trades and did
                      not run the other major systems applications it normally executes
                      Officials from the exchanges,NASD, and NSCCsaid they did not conduct
                      such tests becauseit is extremely difficult to accurately predict and sim-
                      ulate the post-trade processingenvironment during stressful market
                      conditions. Further, it is expensive and time-consumingto coordinate
                      the participation of all securities industry organizations and personnel
                      involved in the post-trade processingof transactions. We agreethat rig-
                      orous stress-testingof automated systems is complicated and not
                      without costs. However, until these organizations conduct thorough
                      “See Securities and Exchange Commission Release No. 34-27446, 64 Fed. Reg. 48703 (1989).

                      Page5                                                     GAO/IMTEG90-33 Securities Industry

                    stress tests, they cannot give market participants complete assurance
                    that the new automated systemswill be able to efficiently compare and
                    correct trades during periods of extreme market activity.

                    Sincethese systemshave been made operational, the organizations have
                    taken or plan to take steps they believe provide additional assurance
                    that these systemscan handle increasedwork loads. Specifically, NSCC
                    conducted another test that demonstrated its system could successfully
                    processthree times the trade comparison volume experiencedduring the
                    October 1989 market break. This test included the operation of other
                    major applications, that run simultaneously on the computer, at normal
                    operating levels rather than those anticipated during increasedmarket
                    activity. The American Stock Exchangeprocessed2 normal days of
                    unmatched trade data during a Saturday work session.Although this
                    exercise did not fully validate the system’s ability to handle anticipated
                    high volume work loads, exchangeofficials believe it showed their
                    system should be able to processincreasedwork loads, and added that
                    the exchangealso recently supplementedthe hardware in its system to
                    further speedup the processingof such trades. NASD has also conducted
                    stresstests to determine whether its system can operate at required
                    work load levels. To date, these tests have shown that vendor-supplied
                    software cannot consistently processanticipated high volume work
                    loads. NASD is working with the vendor to correct the problem and plans
                    to retest the system in the near future. New York Stock Exchangeoffi-
                    cials also plan to stresstest their system in mid-September1990 at work
                    load levels anticipated by the exchangeduring extreme market

                    Becauseof the importance of following a structured systemsdevelop-
Additional System   ment approach, in gathering information on system testing, we also doc-
Development         umented the systemsdevelopment practices followed by these
Weakness            organizations. Failure to follow a structured systems development
                    approach can contribute to the development of systemsthat do not per-
                    form neededfunctions or cannot operate at required work load levels.7
                    In this regard, we found that one organization overlooked certain critical
                    systemsdevelopment practices. Specifically, the organization did not

                    ‘Generally accepted systems development practices involve a structured, step-by-step process for
                    developing, operating, and maintaining automated systems over their whole life. By using a struc-
                    tured approach, organizations (1) help assure automated systems are developed, evaluated, and oper-
                    ated in an effective manner at the lowest overall cost; (2) assure management accountability for the
                    successor failure of the system’s development; and (3) reduce the risk that the system will not
                    operate as intended.

                    Page 6                                                     GAO/IMTECBO4l3Securities Industry


                       fully (1) assesssystem requirements and (2) project future system work
                       load needs.Thesesteps were omitted to more quickly develop and
                       deploy the system. However, the system’s implementation was post-
                       poned several times becausethe system did not perform at levels consid-
                       ered acceptable.Organization officials recognizedthat these delays were
                       attributable to lax compliance with their systems development proce-
                       dures and are now requiring systemspersonnel to follow structured sys-
                       tems development practices in developing all future automated systems.

                       We also evaluated the role the automated comparison and correction
New Systems Will       systemswill play in reducing the 5-day period for clearing and settling
Help Shorten Stock     stock transactions. The potential benefits of reducing the S-day cycle
Clearance and          include (1) reducing payment risks and uncertainty by having the pay-
                       ment for and exchangeof stock occur closer to the trade date; (2) coordi-
Settlement but         nating the stock clearanceand settlement processmore closely with the
ObstaclesRemain        l-day clearanceand settlement processin the options, futures, and gov-
                       ernment bond markets; and (3) increasing the efficiency and competi-
                       tiveness of the U.S. securities industry in global financial markets. In
                       this regard, the automated comparison and correction systems imple-
                       mented by the exchanges,NASD, and NSCCrepresent progresstoward this
                       goal. Specifically, these systems have reduced the stock comparison and
                       correction processfrom 3 days to 1. The 2-day reduction, fully imple-
                       mented in February 1990, is attributable to (1) NSCC'Sfaster processing
                       system for matching trade data and (2) efficiencies gained by the
                       exchanges’automating their trade correction processesand NASD'S
                       enhancing its automated trade correction capability.
                       However, while trade comparison has been reduced to 1 day, stock set-
                       tlement still requires 5 days. Settlement remains at 5 days due to com-
                       plex legal, financial, and other issuesassociatedwith shortening this
                       processand reducing settlement risks. Theseissuesinclude
                     . reducing the time required to bring physical stock certificates, especially
                       those of small investors, to settlement by, for example, (1) accelerating
                       the delivery of such certificates; (2) storing them in a central location
                       and recording their movement amongtraders in book-entry form, or (3)
                       eliminating the use of certificates entirely;
                     . cutting the time available to trading firms, large institutions (e.g.,
                       banks), and retail customersto arrange for (1) trade confirmations and
                       (2) payments and bank financing of certain security purchasesor
                       delivery obligations;

                       Page 7                                      GAO/IMTEG90-83 Securities Industry

                    . speedingup the availability of settlement payments by increasing reli-
                      anceon same-dayfunds transfers, along with eliminating payments by
                      checksthat do not allow funds to be available until the next day; and
                    . determining whether related federal and state laws and regulations
                      requiring the use of physical stock certificates and allowing a S-day set-
                      tlement period, for example, would need to be modified,
                        As of August 1990, a working committee of U.S. securities and banking
                        representatives is studying these areasin detail, with the goal of pro-
                        posing a strategic plan that the securities industry can follow to reduce
                        the settlement processto 3 days.” Specifically, the working committee
                        plans to survey stock clearanceand settlement organizations to deter-
                        mine the feasibility and acceptability of modifying the stock settlement
                        process,including analyzing the capability of certain existing automated
                        settlement systems used in the clearanceand settlement processto
                        operate under an accelerated3-day settlement schedule.The working
                        committee expects to complete its study and strategic plan by late fall,
                        at which time it hopesto involve the SEC,the Federal ReserveSystem,
                        and the stock clearanceand settlement organizations in implementing
                        the necessarychangesto reduce the entire processto 3 days. In this
                        connection,the working committee briefed SECon its efforts, and the
                        commissionis considering holding a conferencein November 1990 with
                        stock clearanceand settlement organizations to discussthe issuesassoci-
                        ated with making the changeto a 3-day settlement process.

                        Active federal oversight of the use of computers is critical to ensuring
SEC Has Not Fully       that these systems have sufficient capabilities and controls in place to
Ensured That Trade      compare and correct trades promptly and accurately. In this regard, SEC
Comparison and          has taken steps to help ensurethat certain securities industry organiza-
                        tions establish capacity planning and vulnerability assessmentprograms
Correction Systems      for automated systems and networks. However, SEX’Sdirect oversight in
Are Technically Sound   assessingautomated systems remains superficial.
                        SECofficials informed us that when the comparison and correction sys-
                        tems were being developed,they (1) visited the involved organizations
                        to gain a better understanding of how the systemswould work and
                        where the processingwould occur and (2) requestedfrom the exchanges
                        and NSCCwritten certifications that their systemswere designedto pro-
                        cesswork loads such as those experiencedduring the October 1987

                        ‘This committee is the U.S. contingent of the Group of Thirty and is working to implement the Group
                        of Thirty’s recommendations in the United States.

                        Page 8                                                     GAO/IMTEGQ          8ecuritiee Industry

market crash. However, as we have previously found in related finan-
cial market reviews, SEC oversight excludes direct technical assessments
of systems during critical development and deployment phases.QFor the
four systems we examined, for example, SEC’S review was not designed
to validate that these organizations had taken adequate steps to ensure
that these systems(1) have the capacity to support timely operations
under normal and high-volume conditions; (2) have controls in place to
prevent unauthorized accessand misuse;(3) are able to provide contin-
uous service in the event of equipment and software failures, natural
disasters, and intentional malicious acts; and (4) have adequatecontrols
to ensure that the systems’ hardware, software, and communications
will perform as intended.
In discussingthe need for more in-depth assessments,SECofficials
explained that the commissiondoesnot have sufficient technical exper-
tise and relies on the exchanges,NASD, and the clearing organizations to
ensure the soundnessand integrity of their own systems.This superfi-
cial systemsoversight by SECreducesthe public’s assurancethat these
systems have sufficient capabilities and controls in place to process
trades in a prompt and accurate manner. In addition, it also raises ques-
tions about the strength of the federal regulator’s oversight in this area
and highlights the need for the commissionto do more in the automated
systems area.
Sincethese four systemswere developed,SEC has begun to increaseits
ability to overseethe automated systems and telecommunications net-
works used by certain organizations in the securities industry. Specifi-
cally, the Commissionissued an automation review policy in November
1989 that asks certain securities industry organizations to establish
comprehensivecapacity planning and vulnerability assessmentpro-
grams for their automated systems and networks. In addition, SEC estab-
lished an automation review group in April 1990 and plans to provide it
with technical staff in the third quarter of 1990 with the expertise to (1)
assesscompliance with its automation policy and (2) assist commission
staff, as needed,in addressingtechnical issuesduring oversight

“See GAO/IMTEC-90-47, July 12, 1990; GAO/IMTEC-90-16, Jan. 6, 1990; and GAO/GGD-88-38,
Jan. 26,1988.

Page 9                                               GAO/IMTEG99-83 Securities Industry


                      Progressis being made in the stock markets to improve the efficiency
Conclusions and       and timeliness of post-trade processingactivities. The exchanges,NASD,
Recommendations       and NSCChave streamlined the trade comparison and correction process,
                      which has enabled the securities industry to cut this processby 2 days.
                      We are also encouragedby the efforts of the Group of Thirty, the securi-
                      ties industry, and the regulators to shorten the overall settlement pro-
                      cess.However, until more action is taken, the payment risks associated
                      with stock traders and clearing organizations waiting 5 days to
                      exchangepayments for stocks are still present. In addition, SEC is begin-
                      ning to take positive steps to increaseits oversight of the use of automa-
                      tion in the stock markets. However, more needsto be done to help
                      assuremarket participants that trade comparison and correction sys-
                      tems are designedto operate smoothly during periods of market stress.
                      Although these systemshave operated well since they have been imple-
                      mented, stress-testinghas not been completed to ensure that they can
                      handle extreme market activity, and one organization overlooked crit-
                      ical systems development practices when developing its system.
                      Accordingly, we recommendthat the Chairman, SEC,ensure that (1) the
                      New York Stock Exchange,the American Stock Exchange,NASD, and
                      NSCCconduct complete stress tests to demonstrate that their trade com-
                      parison and correction systems can handle increasedwork loads antici-
                      pated during peak trading periods and (2) all organizations follow
                      structured systems development practices when developing automated

                      We discussedthe contents of this report with senior officials from the
Agency Comments and   New York Stock Exchange,the American Stock Exchange,NASD, and
Our Evaluation        NSCC.We also discussedit with responsibleSECofficials and a represen-
                      tative from the Group of Thirty’s US. working committee. We have
                      incorporated their commentswhere appropriate. Except as noted below,
                      these officials generally agreedwith the contents of our report.
                      SECstaff reiterated that the 1987 market break experiencerevealed the
                      need for further automation of the trade comparison and correction
                      processes.They added that acceleration of the trade comparison func-
                      tion to a next-day basis, coupled with automating the trade correction
                      processes,significantly addressesthe trade correction deficiencies iden-
                      tified during the 1987 break, Commissionstaff also believe that the
                      implementation of these systems,basedon reasonabledevelopment
                      testing along with complete stress-testingafter deployment, provides
                      considerableimmediate benefit to market participants at minimal

                      Page 10                                    GA0/IMTEG90-83 !SecuritiesIndustry


        market risk. We agreethat significant progress has been made in
        addressingthe trade correction problems that occurred during the 1987
        market crash. However, we believe that thorough stress-testingshould
        be performed on new automated systemsto help ensure that they are
        appropriately designedto handle work loads anticipated during extreme
        trading periods.
        Additionally, SECstaff questioned the need for the commissionto use its
        scarceresourcesto perform direct technical assessmentsdescribedin
        this report and others when independent assessmentsand internal eval-
        uations currently are performed by or on behalf of the exchanges,NASD,
        and the clearing corporations. We continue to believe that, given the
        important role automation plays in the securities markets, SEC needsto
        be more directly involved in reviewing and assessingthe markets’ use of

        We are providing copiesof this report to other interested membersof
        Congress,legislative and executive branch agencies,and the public. We
        will also make copiesavailable to others upon request.
        This work was performed under the direction of Howard G. Rhile,
        Director, General Government Information Systems,who can be reached
        at (202) 276-3466.Other major contributors are listed in appendix III.
        Sincerely yours,

        Ralph V. Carlone
        Assistant Comptroller General

        Page 11                                   GAO/IMTEC90-83 Securities Industry

Letter                                                                                          1

Appendix I
Objectives, Scope,and
Appendix II                                                                                    16
Description of
Comparison and
Correction Systems                                                                              -
Appendix III                                                                                   18
Major Contributors to
This Report
Table                   Table 1: Trade Comparison and Correction Systems,
                            Primary System Functions, and Implementation


                        GAO       General Accounting Office
                        GGD       GeneralGovernment Division
                        1MTF.C    Information Managementand Technology Division
                        NASD      National Association of Securities Dealers
                        NS4X      National Securities Clearing Corporation
                        SEC       Securities and ExchangeCommission

                        Page12                                  GAO/IMTEG90433Securities Industry
   SecurItlea Industry
Appendix I

Objectives,Scope,and Methodology

              Our objectives were to provide information on computer systemsused to
              reconcile market participants’ trades executedon three major U.S. stock
              markets-the New York Stock Exchange,the American Stock Exchange,
              and the National Association of Securities Dealers(NASD). Specifically,
              our objectives were to (1) provide information on automated systems
              developedby these stock markets to more easily correct problem trades;
              (2) assessthe extent to which these systems have been tested to ensure
              that they can processincreasedwork load volumes such as those experi-
              encedduring the 1987 market crash; (3) determine the impact these sys-
              tems will have in reducing the 5-day period neededto clear and settle
              stock transactions; and (4) assessthe role that the Securities and
              ExchangeCommission(SEC) played in reviewing and approving the use
              of these systems.Becauseof the important role the National Securities
              Clearing Corporation (NSCC) plays in gathering trade data from market
              participants trading on the stock markets and identifying those problem
              trades needing correction, we also reviewed a redesign of the automated
              comparison system used by the clearing corporation.
              To determine the development status of these systems and to under-
              stand how they operate, we obtained available systemsdevelopment
              documentation and met with senior officials at the New York Stock
              Exchange,American Stock Exchange,NASD, and NSCC.We also toured
              these organizations’ computer and trade reconciliation facilities and
              observeddemonstrations of the systems’ capabilities. Further, we
              observedat the exchangesand NASD how traders use the systemsto cor-
              rect trades in a live working environment.
              In looking at testing of these systemsto ensurethat they can processthe
              increasedwork load volumes, we reviewed generally acceptedsystems
              development criteria and federal rules, regulations, and policies to deter-
              mine the systemsperformance tests that should be performed when
              developing and implementing computer systems.We met with and dis-
              cussedthis issue with senior exchange,NASD, and NSCCofficials. We also
              reviewed their systemsdevelopment documentson testing, including
              any test plans, to assessthe degreeof testing performed to demonstrate
              that these systems would be able to processincreasedwork loads in a
              prompt and accurate manner. Additionally, becauseof the importance
              of following a structured systemsdevelopment process,we reviewed
              other systemsdevelopment documentation to assesswhether these orga-
              nizations had adhered to generally acceptedpractices in developing
              their systems.In this connection,we reviewed and evaluated available
              performance statistics on these systems,such as responsetimes and

              Page 14                                     GAO/IMTEG90-83 Securities Industry
~-     ~~~        --~~
     Ame* 1
     ObJeaivea, &ope, and Methodology

     volumes processed,to determine whether the systemswere processing
     trade data at intended performance levels.
     To assessthe role of SECin reviewing and approving these systems,we
     met with responsiblecommissionofficials to discussand document how
     they processedthe exchanges’,NASD'S, and NSCC’S requeststo develop
     and implement these automated systems,including what technical
     assessmentsthey had performed before approving their use. We also
     reviewed the requests-called rules-submitted to SECby the
     exchanges,NASD, and NSCC to develop and implement these systems,
     along with any relevant industry comments.In addition, we reviewed
     and discussedwith appropriate SECofficials the commission’sautoma-
     tion review policy and documentedSEC plans to establish a group to
     overseecompliance with this policy.
     With regard to the impact these systemswill have on shortening the 6-
     day clearanceand settlement period for stocks, we obtained information
     from senior SECofficials on how the systemswill affect the S-day period.
     In addition, we discussedthis, issue with representatives from (1) the
     Group of Thirty committee responsiblefor implementing in the US. the
     group’s recommendation for shortening the clearanceand settlement
     processand (2) the Securities Industry Association. Further, we
     obtained and evaluated the views of senior exchange,NASD, and NSCC
     We conducted our review in accordancewith generally acceptedgovern-
     ment auditing standards, between August 1989 and July 1990.

     Page 15                                   GAO/IMTEGfM)-S3 Securities Industry
Appendix II

Description of Comparison md
Correction Systems

               The following describesNSCC’Scomparison system and the exchanges’
               and NASD'S correction systems and includes a brief description of how
               the systemsoperate.

               NSCC'Sredesignedsystem comparestrades and identifies unmatched
               trades faster than its old system. Specifically, the redesignedsystem
               comparestraders’ transaction data, identifies unmatched trades, and
               transmits this information electronically to the exchanges’correction
               systems before the start of the next trading day.’ In the past, NSCCcould
               not complete these processingactivities until the end of the day after
               the trade execution date. The redesignedsystem involved extensive
               modifications to the computer system and data processingscheduleused
               by the clearing corporation to compare trades. The primary modifica-
               tions involved (1) the development of new software to processpost-
               trade data more quickly and efficiently and (2) new requirements placed
               upon traders to submit their data earlier than under the previous
               The New York Stock Exchange’sand the American Stock Exchange’s
               correction systems-called the Overnight ComparisonSystem and the
               Intra-Day Comparison System, respectively-are similar in that they
               consist of a network connecting a central computer to terminals. Specifi-
               cally, the New York Stock Exchangesystem consistsof a mainframe
               computer with 470 terminals, while the American Stock Exchange’s
               system usesa minicomputer with 70 terminals. Located at the
               exchangesand at traders’ offices, the terminals (1) receive unmatched
               trade data from NSCCbefore the start of the next trading day and (2)
               allow traders to more quickly correct unmatched trades. Oncesuch data
               is successfully matched-that is, when traders agreeto the terms of the
               transactions-the correction systems electronically transmit the trades
               to NSCCat the end of the day to be prepared for settlement.
               NASD'S system-the   Automated Confirmation Transaction System-also
               consistsof an extensive network of computer terminals and minicom-
               puters that are used by NASD traders to correct unmatched trades.2How-
               ever, unlike the exchanges,NASD'S system can also compare and match
               trades within minutes after execution. Specifically, NASD requires
               ‘NSCC used to compare trades executed on the exchanges and by NASD. However, since NASD’s new
               system compares trades, NSCC no longer performs such comparison processing for NASD.
               2NASD has over 3,000 terminals capable of accessing the Automated Confirmation Transaction
               system. These terminals can also be used to accessother NASD automated trading and support sys-
               tems, Conversely, the terminals of the exchange correction systems are devoted solely to trade correc-
               tion activities.

               Page 16                                                     GAO/lMTEG9O-S3 Securities Industry
Dcudptlon olCbmparlson       and
corrrsaion Sy*tenu

traders, after executing trades, to enter their trade data directly into the
new system for it to be compared and matched, rather than having NSCC
perform this step. NASD rules require traders to enter their transactions
into the system during the trading day, within 90 secondsof each
trade’s execution. The trade information is immediately transmitted to
the other party for verification or correction. When no differences exist,
or after traders resolve inconsistencies,the system automatically
matchesthe trades, and NASD transmits daily the matched trade infor-
mation to ~scc to be prepared for settlement3 As a result, NASD trades
can be matched and compared on the sameday they are executed.

3When differences exist, NASD requires traders to resolve them by 1:OOp.m. of the next trading day.
Trades not responded to by the opposite party within this time frame are automatically matched by
the system and transmitted to NSCC for settlement. Trades rejected by the other party are automati-
cally deleted from the system.

Page 17                                                    GAO/IMTEZo88        Securities Industry
Appendix III

Major Contributors~to This Report

                       Richard J. Hillman, Assistant Director
Information            William D. Hadesty, Technical Assistant Director
Management and         Gary N. Mountjoy, Project Manager
Technology Division,   Gregory P. Carroll, Staff Evaluator
Washington, DC.

                       Bernard D. Rashes,Senior Evaluator
New York Regional      Garry Roemer,Deputy Project Manager
Office                 Daniel P. Schultz, Staff Evaluator
                       Diana Gershon,Staff Evaluator

(810468)               Page 18                                   GAO/IMTEGBO-S3   Securities Industry