Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
3 THE CHALLENGE OF CHANGE IN THE PRESENT By the nature of its mission, the Social Security Aciministration (SSA) serves the general population of the United States--its 50 states and territories. From time to time Congress modifies the details of the programs that the SSA administers ancI/or adds new programs to the agency's repertoire. Also, from time to time the SSA will decide that it must upgrade its facilities or level of service as a general action unrelated to any program. Thus, many factors combine to influence the workload that the SSA's computer systems must support. . . . Because the population and work force of the country are constantly growing, even if all other things were to remain constant, the number of enumerations and the volume of taxes collected will increase correspondingly. The age distribution of the population is independently changing; the number of inclivicluals of retirement age is increasing. Hence, the volume of retirement-related payments will increase correspondingly as will collateral decisions and actions related to retirements. As Congress changes the details of the SSA's programs, changes in the software are usually required. Inevitably, the software grows more complex and in turn imposes an increasing workload on the computer systems. As Congress creates new programs to be acIministered by the SSA, such as Supplemental Security Income (SSI), there is an impact on the software and hardware anal, in general, a growth in computer workload. The complexity of the new program, the segment of the population covered, integration with one or more other extant programs, anti basic characteristics of a new program will all combine to increase the SSA's computer needs. As Congress modifies tax laws, this too can have consequences for the SSA (e.g., social security numbers [SSNs] for dependents). The nation's life styles are more complex and are changing in~fundamental ways. Joblessness, single parentage, mobility of individuals--all such things in their own way increase the complexity and volume of the SSA's workload and its diversity. 25
26 The health history of the country is also changing as medical science progresses. This contributes to aging of the population and increases the scope and complexity of the health care programs that may involve the SSA in some way. The job environment is also changing, in many ways becoming more complex but also introducing new health and accident threats as higher technology and new substances appear in the workplace. These phenomena also impact aspects of health care programs, the volume of disputes over disability claims, and possibly the complexity of the disputes. Everything points to ever-increasing complexity in many dimensions for the SSA. The SSA, as the administrator and overseer of vital social programs, can do nothing to slow or redirect the forces of change. The forces are all external. The one factor that is essentially certain is the consequent growth in demand for computer services and the computer-based systems that provide the services. Quite beyond the natural events in the country that drive the magnitude of the SSA's total workload, there are pragmatic events that derive from growth and the simple passage of time. · Computer equipment becomes obsolete and must be replaced when spare parts and maintenance expertise become unavailable or excessively costly. Computer technology continues to advance in major ways, and replacement is sometimes the only way to improve cost to performance factors or accommodate growth. At some point the concentration of resources in one location becomes too risky for the country to accept, and steps to disperse computing facilities are not only prudent management actions but also valid technical options for responding to workload growth. As the number of clients grows and as the diversity of entitlements available to them increases, the simple problem of handling the flow of people, relating them to the data in the computer system, and making appropriate decisions outgrows the state of art of the field office equipment. Upgrading the environment in the field offices becomes a necessity, not a luxury to make life more pleasant for the SSA's representatives. So far, the various points have discussed here dealt only with the consequences of natural forces at work in the country and the pragmatics of keeping pace. Such forces drive the SSA and its computer needs but cannot be controlled by the SSA or, for that matter by the federal government. There are, in addition, things that the SSA wishes to do on its own initiative. For example, it is expedient to upgrade its level of service by instituting entirely new kinds of service or entirely new dimensions of old services. Depending on the details of such an improvement effort, the consequences for computer workload can vary from a major impact to one that can be swallowed as a part of the normal growth progression. However, even in this connection the SSA does not always have control over the level of
27 service that its clients expect. The citizens of the United States derive their expectations for service from what they receive from systems in the private sector, state governments, and other federal agencies. The dominant force is undoubtedly the private sector because it is always ready to be innovative in its computer-based offerings in the expectation of achieving a competitive advantage. Therefore, the private sector and its measures of service quality are yet another force outside the SSA's control that drives SSA's clients, expectations for level and quality of service and therefore the agency's workload as well. As a consequence of the observations made in the foregoing paragraphs, we conclude that: Because of forces beyond and outside its control, the Social Security Administration will require periodic infusion of capital to expand and improve its computer base and scope of its information systems. Beyond what is required just to keep up, we also conclude that: The Social Security Administration will require occasional infusion of capital to institute and implement new computer- and/or communication-based services to keep itself in line with societal expectations for its interface with the public. The Feast to famine,. Wood condition to impending disasters circumstances that have characterized the SSA's cyclical computer posture over time derive in part from failing to identify the different reasons for which the SSA computer workload has grown in a given period. Hence, a proper case for capital funds for new or upgraded equipment has not been made. A simple argument that We need more computer capacity to avert a disaster" is not a satisfactory basis for defending a budget request. The Congress and the country deserve to know what is driving computer needs and how funds will be required to meet the demand. We recommend that: The Social Security Administration adopt management processes of strict analysis and control to avoid the recurrence of an information systems crisis. For example, it may prove to be that on-line computer support for the recently installed 25,500 computer terminals is inadequate, perhaps as a result of the SSA's having only early experience functioning in a terminal environment. Problems can be caused by a release of unperceived latent user demand. To deal with changes in demand the agency must (1) thoroughly forecast and justify expansion and upgrade of the existing information infrastructure on a continuing basis, (2) quantify and define in advance the specific performance goals to be provided its clients, (3) measure and monitor actual performance, and (4) plan for the orderly introduction of new services consistent with available personnel and budgetary resources.
28 In the following sections of this chapter we focus on several of the near-term challenges--ones that we believe are the most important but that have not been addressed in the plans we reviewed. LEVEL OF SERVICE The specific levels of service to be offered to the agency clients need to be defined more clearly. Use of such qualitative terms in the Agency Strategic Plan (ASP) goals as Improve or Tests can lead to endlessly chasing ephemera that can only be subjectively analyzed for achievement and that will always be subject to disagreement between the SSA's management and its oversight authorities. We recommend that: The Social Security Administration quantify each aspect of service quality (e.g., elapsed time to complete), monitor its overall performance, and manage against such a priori service goals. By quantify, we mean, for example, completing an enrollment in 3 days or issuing a social security number in I week. Unless this is done, the tendency will lead to excessive and unnecessary expenditures. Based on our observations, we found that the service levels provided by the SSA are good to adequate but that disability processing experiences significant delays in validation by state authorities and appeals of rejected claims. However, automation of disability functions has not received the same level of attention as the retirement and survivors insurance programs.; Service Levels Affect Workload The agency has moved toward on-line telephonic services for its clients through introduction of the area code 800 service. Such "on-demand" services create a workload that cannot be controlled or even accurately anticipated by the agency. Furthermore, providing such services will require a greater degree of computer system and budgeting flexibility than the agency, as currently controlled, is able to exercise on its own. We urge the SSA to recognize that on-demand service is likely to result in an increased workload that is not necessarily tied to increased productivity. Providing on-demand services will also increase complexity and cost. Therefore the SSA must specify in detail the services to be offered its clients while carefully assessing their cost and planing for their impact on operations. The agency must also must be constantly aware of the fact that on-demand processing results in uncontrollable volumes with significant daily, monthly, and annual peaks. Operating efficiency will need to be measured and improved over time through the creation of procedures, and organizational actions and the application of technology. As efficiency improves, improvements in service performance specifications can be realized or, alternatively, service levels can be held as they are but provided at lower cost.
29 Planning for Quality It is important to distinguish between services to be provided (e.g., by telephone) and the quality of services of fered. A quality assurance mechanism should be established with independent reporting relationships to monitor service delivery quality. In this regard the quality of service should be limited to delivering services to specifications rather than maximizing them. Such a mechanism will form the basis for complete planning. A complete planning process considers such factors as workload volume, service levels, and efficiency as parameters that can be traded against one another in determining resources and cost. Each factor becomes a Control levers, so to speak, that can be adjusted to arrive at an appropriate overall mix. For example, given fixed budget constraints and an uncontrollable volume, either efficiency must be improved or service specifications reduced in order to achieve balance. If a complete planning process is followed, it should begin by establishing the basis for objectively evaluating and presenting agency performance. Without such a process, performance will always be in the Eye of the beholder"--something over which the SSA can have little control or influence. In the on-line environment the general public is the Driver in the sense that the public's behavior governs the demand for services and therefore the workload. While it is true that by throttling service, notably response time, the SSA can discourage the public from demanding attention, it is likely that the SSA would not do so because it could and probably would provoke public outcry and criticism. Previously, the demand for computer services was driven by the SSA employees, over which the agency has tighter control and who generally are tolerant of less responsive service delivery. Thus, the conversion to an on-line environment, as desirable as it may be for many reasons, opens the agency to a vulnerability that it has never faced. . _ ,< . ~ , . . The Potential for Overload The control factors described above must be balanced. Otherwise, a self-induced crisis can be precipitated. For example, the nationwide rollout of area code 800 service implemented in October 1989 and the annual workload peak that will follow in January 1990 have the potential for overloading the agency's work force and information infrastructure, consisting of its computers and communications. We recommend that: The Social Security Administration thoroughly analyze the impact of new client-driven services or the quality upgrade of services that entail additional identified or latent on-line computer support to assure the adequacy of supporting automation. CONTINUITY OF SERVICE The SSA has been a long-time user of automatic data processing (ADP), and, as a result of the System Modernization Plan (SMP), its ADP capabilities have been dramatically
30 improved. During the period of modernization the National Computer Center (NCC) has become a pivotal element in SSA operation, and this dependence should grow in the future. The agency has progressed in its transition from paper-based operations to the installation and use of over 25,000 computer terminals located throughout the agency. From each terminal, an employee can access central data files and enter data for processing. With a downsized work force and increasing workload, the agency has appropriately chosen to increase the productivity and efficiency of its work force through automation. We endorse these improvements and support the continued transition to on-line processing. Increasing Dependency and Current Plans As the agency's reliance on information technology increases, the adverse consequences of failure become more severe. As it progresses toward paperless processes, its ability to fall back to a paper-based system will diminish. At present, the SSA's plan for backup, should the NCC become inoperative, is to revert to batch processing and move to a commercial hot site.1 Under such conditions this would permit restoration of only limited batch processing capability within 72 hours. Only critical work consisting primarily of claims and payment adjustments will be processed and the performance of even these tasks will degrade because less computer capacity will be available. Other important workloads such as data matching, wage posting, and cost-of-living adjustments will have to be suspended along with software development and management information support. There would be no support for on-line access to the SSA's databases which is currently needed at the teleservice centers and district offices nationwide. The SSA's backup and recovery plans need to be reconsidered in light of the agency's growing dependence on its information systems-and their on-line use throughout the agency. Two Basic Approaches The risks and consequences of loss of the NCC warrant immediate attention to assure that adequate backup capabilities, fullback plans and procedures, and system safeguards are in place to guarantee continuity of operation across a full spectrum of disaster scenarios. Short of a major and long-term redesign of the SSA's applications and systems architecture to a distributed or hierarchical configuration, there are two basic variations that can provide adequate backup for the NCC and be rapidly implemented: 2. 1. An on-line, remotely located, highly automated dedicated system routinely processing a portion of the workload. In this variation either the NCC or the dedicated backup would be able to support all critical functions. A standby system available to run critical elements of the operation in the event of a catastrophic failure at the NCC. {A hot site is a fully operational data center--equipped with host computer systems, front- end processors,and network necessities--where an organization may move its operations in the event of a disaster.
31 For the first option we suggest dividing the processing load approximately equally between the NCC and the on-line Backup National Computer Center (BNCC). Thus, the latest versions of the software and databases and required communication access to support district, regional, and local offices will be in place and operational for both the NCC and the BNCC, and either could take over for the other in an emergency. For the second option, which would be only employed in the event of an emergency, it is critical that effective methods be in place to assure that the latest version of the software, databases, and operational staff can be rapidly relocated to the standby system. The SSA's hot-site plans reflect its selection of the second possibility. However, this approach currently precludes access to the community of on-line users unless an extensive investment is made to establish network connectivity to the standby system and possibly to enhance its hardware and software capability as well. Having processing capability is necessary but not sufficient in itself to provide the current level of information system support. The data communications network is now a vital part of the SSA's information systems that link computer and data resources to the agency's employees. As time progresses, the SSA will become less able to fall back to manual methods. In another version of the second option, the standby system might ordinarily serve to support administrative or non-critical functions such as decision support or software development. In the event of a disaster at the NCC, these functions could be deferred and their processing assets reassigned to support the critical programmatic functions. The Risk Is Real Despite the excellent precautions and facility design measures taken by the SSA to protect the NCC, the possibility of an outage is real. For example, on May 12, 1989, the SABRE reservation system of American Airlines failed for about 13 hours due to the side effects of a routine installation of disk storage (PC Week, 1989~. In addition to causing a direct loss of revenue during the outage, discounted tickets were sold by other airlines against a seemingly unlimited inventory in the airline's noncurrent but operational systems. The impact of the loss of this system rippled through travel agencies, car rental companies, and hotels. Attempted falibacks to along abandoned methods of telephone calls and paper logs resulted in lost or double bookings. Regressing to Paper Unless all of the manual data capture forms are changed to precisely reflect their counterpart computer terminal screens, fallback to manual data entry and retrieval will become more and more difficult, even during short periods when the on-line automated system is unavailable. But even if this is done, at some point in the future as reliance on the automated system increases, regardless of the resources applied to the recovery process, it will be impossible for the SSA to recover from an extensive system outage without an on-line backup system and up-to-date databases.
32 Reevaluation Needed Since the NCC has become such a critical element in the SSA's ability to serve its clients, we recommend that: The Social Security Administration immediately develop ~ workable strategy for surviving ~ partial or major loss at the National Computer Center. One of the alternatives that must be considered is a second geographically separate computer center. A second site can, with minimal staffing, be quickly implemented, provide an effective backup, and also assist in meeting increasing capacity needs that will accompany the transition to on-line operations. DECISION MAKING In order to improve our understanding of the rationale behind the SSA,s selection of projects to be implemented, we reviewed its decision-making process. Our purpose was to arrive at an appreciation for the basic objectives that drive the agency and thereby influence its decision making. In addition to the discussions we had with the agency's executives on this subject, we also reviewed the internal administrative instruction that deals with the commissioner's decision-making process (Social Security Administration, 1988). Instead of revealing the agency's established and well-defined objectives, we found a reactive environment driven by external and largely unpredictable forces. Thus, while the ASP talks in long-range terms, the reality is that short-term exigencies and crises still dominate the decision-making process. The major and obvious steps have been taken to proceduralize decision making, but the agency must recognize the need to move from making reactive to proactive decisions by more comprehensive planning and consistent management. The sources of projects can come Top downs from SSA executives or higher governmental authorities or Bottom ups from the component organizations. The issues and actions that require the commissioner's decision are described in the administrative instruction referenced above. Once a proposed project or initiative is so identified, the following five-part executive review process takes effect. This process applies to all major projects and generally occurs in the sequence noted: 1. ComDonent Develooment--Each organizational component proposing a project is responsible for ensuring that its proposals state the general requirements, identify available options, assess costs and benefits, consider relative priorities, and are coordinated with other interested components. ADP Plan ImDact--All proposed projects requiring systems development resources must be considered in light of other ADP projects competing for those resources. Budget ImDact--Any proposed project requiring a procurement must be considered in light of other Information Technology Systems (ITS) expenses, including telephones, telecommunications, and ADP systems. Project budget proposals are
33 reviewed by an internal Systems Review Board (SRB), the commissioner, and executive staff. Proposals are prioritized. 4. SRB Action--The SRB was created to provide for an independent internal review of major ADP projects. It formulates recommendations to the commissioner and is composed of an executive board chaired by the chief financial officer with dedicated technical staff support. Commissioner's Decision--The commissioner is the final decision maker on major ADP projects at the SSA. At any of these stages the SSA can and does make use of outside consultants to review its strategies and provide expert advice. in terms of specific ADP responsibilities, duties are divided along three principal lines: 1. The deputy commissioner of operations has responsibility for all programmatic systems as well as for operation of the NCC. 2. The deputy commissioner of management has responsibility for office automation and management information systems. 3. The chief financial officer has authority over financial management systems and chairs the SRB. While such a three-way division of responsibility is rational in light of historical evolution and indeed is a workable arrangement now, in the long run it will impede the progress the agency seeks to make toward its goal of a fully integrated systems environment for the following reasons: . Each component of the tripartite is automatically and implicitly encouraged to develop systems for itself. In general, such systems would reflect the views and needs as seen internally to the organization, not the broader needs and views of the SSA overall. Each would stand alone, so to speak, with regard to other parts of the agency. There are bound to be organizational and jurisdictional barriers despite the best intentions everywhere. More to the point, there might well be satisfactory cross-component discussions and agreements for full integration with the individuals presently in place, but the basic structure that divides management responsibility implicitly encourages an intraorganization point of view. With different individuals in place in the future, the essential cooperation might well decay and progress toward integration might thus be deterred. We thus conclude that more centralized authority for information management and systems may improve decision making and the level of integration achievable and maintainable by the agency. 2 Programmatic systems are those computer systems that are used to support the SSA programs such as old-age retirement and survivors insurance.
34 Divided Resources The SSA has chosen to divide its systems development and support resources to be more closely aligned with the agency's component organizations. Prior to this decision the agency had centralized its systems resources in a single organization. This change at the SSA resulted in specific ADP responsibilities being allocated as described in the Decision Makings section above and the reassignment of resources from one organization to essentially two organizations: the Office of Operations and the Office of Management. Such decentralization can result in overlapping responsibilities, duplication of effort, uncoordinated projects, confusing priorities, and systems that interoperate poorly. If these problems can be managed, then a separate organizational structure can serve to dedicate and focus resources for specific needs. However, in the long run the agency will be better served if its systems resources are centrally managed rather than separated. The SRB has not demonstrated that it has been an adequate mechanism for coordinating SSA components. Cost-Benefit Analysis We did not find that an analysis of costs versus benefits was being applied with consistent criteria by all organizational components. For example, the Office of Operations uses undiscounted costs and benefits, while the Office of Management used a discounted cost-benefit analysis for its office automation project. This does not allow projects to be effectively compared among sectors of the agency, especially return-on-investment comparisons. Consistent application of discounting at the discount rate specified from time to time by the Office of Management and Budget (OMB) will improve decision making. Furthermore, costs should be computed on a life-cycle basis that reflects the total costs to the SSA for acquisition and ownership of a system over its useful life. Life-cycle costs include the cost of development, acquisition, and maintenance. On the benefit side, the following three categories must be addressed: 1. Costs actually saved from real budget reductions. 2. Costs that will be avoided, such as those that will not be incurred because higher levels of output are achieved at the same level of budget. 3. Value of service improvements to the SSA's clients for which cost savings or avoidance cannot be readily calculated. We suggest that the SSA also consider including social benefits in its analysis. Other agencies, such as the U.S. Army Corps of Engineers and the U.S. Department of the Interior's Bureau of Reclamation, have used this as a means for communicating to oversight authorities what they are trying to do and justifying proposed actions. Monitoring Total Costs The agency's practice in deciding on a project includes an initial rough estimate of total costs. At the very beginning of a major project details are not known, so cost estimates are necessarily imprecise. Later as the project begins, precise estimates of the immediate tasks are -
35 made and adherence to them is monitored, but we found that total project costs and schedules are not updated as a project proceeds. The agency decision makers should have ongoing visibility of the projected total costs to complete a project in addition to the incremental costs they now review. Such a procedure is warranted for agency-initiated projects as well as those that result from mandate because it gives decision makers an opportunity to anticipate problems and look for better alternatives.3 The SSA also needs to ensure that feedback of results from all projects, not only major ones, is routinely provided to the project manager and superiors. This will provide a management accounting too! to assess performance and improve future estimates and projections. Systems Review Board In operation since November 1987, the SRB was created to be an independent adviser to the commissioner. It acts as an oversight authority for the SSA,s systems development operations and procurement activities. This organization provides some worthwhile checks and balances but it also consumes resources directly and indirectly. However, of even greater concern is its potential to stifle effective development by interjecting unnecessary delay. The SSA's chief financial officer, who chairs this board, should certainly be concerned with the level of operating expenses, the effective procurement of volume-driven processing capacity increases, and the benefits of a proposed development. We also agree that the chief financial officer should monitor development schedules and expenditures against the achievement of promised benefits. However, we do not believe that the SRB should review technological and systems design aspects. This function is more effectively performed as part of the design process by specialists within the organizations having such responsibility and by quality assurance and verification and validation functions applied to the design and development work products. MANUAL PROCESSING The SSA processes about 6 million new retirement claims each year. The effect of the baby boomers will be felt around the year 2010. At the present time about 20 percent of the retirement claims received are too complicated to be handled routinely and must be processed by the program service centers (PSCs). This fraction of complex cases referred to the PSCs creates a significant burden. The major reasons for claims requiring manual processing are as follows: Data are not available to complete the action and must be obtained from the client or another source. Some of the data are not available on computer files and must be retrieved from paper records. 3 Refer to Chapter 2, Project Management, for additional comments on this subject.
36 Data exist in computer files but are not available to the computer program that needs the cIata. The function is regarded as a low-volume special case that justifies a low priority on cleveloping automatic processing of it. The processing rules are so complex that they are very difficult to execute with a computer program and so their automation is ~ieferreci. The processing rules are so complicated that it is not feasible to incorporate them into a computer program. Manual processing for these reasons has a major impact on staffing and total system processing costs. The following is a list of specific instances where manual processing is necessary; the annual volume of such actions is also shown. ACTIONS REQUIRING MANUAL PROCESSING Miscellaneous benefit record corrections Supplemental Security Income (SSI) claims Direct entries from PSCs Returned-check processing Failures in benefit computations Disability follow-ups Subsequent claims on active accounts Split overpayment recovery or beyond 1999 Spouses each with their own benefit accumulations Corrections due to cost of living adjustment recomputation errors Unusual terminations such as foreign beneficiaries or returned checks Manual corrections of payment histories Death lump sums Foreign addresses Disability claims Primary insurance over- and underpayment amounts Status changes due to income added Errors in Medicare billing Corrections for duplicate payments Retirement benefits for more than 10 dependents Retirement benefits for widows Total THOUSANDS PER YEAR 8,380 4,400 1,920 1,500 1,440 1,200 720 336 260 240 192 192 170 150 70 48 36 24 24 12 12 21,326
37 All of these predominantly manual processing tasks performed each year have a greater chance of inclucing errors than those that are automated. Many cause corrections to be applied outside of the original process, accounting for much of the load in the Miscellaneous category listed above. The cost of these manual processing tasks is considerably greater than the relative volume counts would suggest. 1. Each is intrinsically more complex than those handled by automation. 2. The cost of doing them manually can easily be 10 times greater than it would be if automation were applied. The combination of these factors suggests that even though the aggregate of the transactions is less than 10 percent of the total transaction load of the SSA, the cost of manually processed actions is likely to be over half of SSA's processing costs. The SSA must assess the costs for manually processing transactions. The activities and costs of manual processing provide an important input to resource allocation and planning and are also the basis for cost-benefit assessments. Based on the foregoing argument, there are significant opportunities for automating more of the processing that is now done manually. Increases Expected Uniess automation is extended to handle more of the cases that are now considered special, the percentage of manual actions will increase. Two major reasons for this are due to changing demographics where (1) the number of dual income families is increasing and (2) the number of dependents from multiple marriages is increasing. Even with a stabilizing divorce rate, the effects of social change in the last 30 years have not yet affected the retirement system; when they do the SSA will need to have extended automation to an even greater percentage of the workload cases it now handles manually. We recommend that: The SSA aggressively carry forward to completion automation of the basic functions of the agency, including the many for which automation is currently under way or partially implemented, as well as some for which automation is not yet started. Disability Claims The process and decision making for handling disability claims are more complex than those for retirement benefits. The SSA receives about 1.2 million disability claims each year. After receiving a claim at a district office, the office requests from state authorities medical verification of the claim. About 40 percent of the claims are approved. Of the claims rejected, about 25 percent are appealed, with about 12 percent of the rejections being reversed in the appeals process. Because of continuous developments in information technology, automation may now be applied to processes where it was not economically or technically feasible in the past. Processing of disability actions as now done is a labor-intensive and
38 mostly manual process that is prone to inconsistent results and errors. Therefore, it is an area where the SSA must be on the lookout for technology that can economically and reliably automate these processes. In Chapter 4 we discuss several technologies that show promise for disability processing. DEGREE OF FUNCTIONAL INTEGRATION Complete functional integration means a user environment in which any computer-based action that is performed anywhere in the agency can be accomplished from the same terminal. For example, . An employee servicing a client for enrollment or filing a claim could perform all programmatic actions, have access to all relevant data, and have available all relevant support computer-based processes through a single terminal in the same . · on-line session. An employee adjudicating a disability claim could have similar full access to all relevant data, perform all necessary programmatic actions, and be supported by other necessary processes (e.g., a capability to do arithmetic calculations, a capability to perform financial calculations such as those relevant to an annuity) through a single terminal in the same on-line session. An employee performing any of the actions required in the administrative and financial components of the SSA could have similar full access to everything needed for performing assigned tasks. While this sequence of examples might seem to suggest a simple merge of three independent computer-based systems in such a way that any one could flow through a single terminal at a time, it is much more subtle. There are bound to be instances in which the user will have to concurrently function in more than one part; for instance, if a letter is necessary during the processing of a disability claim, the user should be able to recess the review of information related to the claim, move into a word processing function, compose the letter, have copies automatically routed wherever necessary throughout the administrative system (e.g., to the general counsel's office), and do so without terminating the principal session on disability claim processing. Another example is a disability claim that requires concurrent access to retirement claim data for proper processing. Or every user action might automatically forward selected detail or summary information into the administrative system for preparation of management reports. A major advantage of integration is management of the work-force assignments. A common terminal and user interface throughout the SSA facilitates the movement of people from area to area as work demands change. Training might also be reduced to some extent. We have just described the ultimate in integration, namely, all systems within the SSA fully interconnected and serving all users. There are lesser degrees of integration that might be more cost effective and hence desirable. Another National Research Council committee (National Research Council, 1977) addressed the question of integrating telecommunications
39 services by asking: How completely can separate facilities and services be integrated before exceeding the optimum cost-benefit ratio? It was concluded that more integration does not naturally lead to more benefits at lower cost--particularly as the integration leads increasingly to increased complexity and counterproductive combinations. The integration of SSA information, systems, and processes might be encapsulated, so far as the client is concerned, in the construct of ~whole-person processing." An SSA employee should be able to handle any action pertinent to a client--from the simplest act of enrollment to the most complex--through a single terminal with ready access to all data, records, and processes required. The degree of integration needed to accomplish this is greater than that which the agency has achieved thus far but is much less than complete integration. Integrating Administrative and Programmatic Systems The second phase of an office automation project, which is discussed below, has the objective of integrating the administrative and programmatic systems. Such a level of integration can substantially increase system complexity, without an offsetting gain in productivity or avoided cost. Thus, even though the notion of a fully integrated environment may appear attractive, we suggest proceeding with caution in planning for the integration of administrative and programmatic systems unless it can be established that the cost of technical complexity can not only be offset financially but also that such a high degree of integration is needed to support the agency's mission. Integration of just the programmatic systems and functions would be an appropriate and preferable first step. Computer Security Not every SSA employee is authorized to perform every action or even to access every record or item of data. Thus, division of responsibility will be essential throughout the work force. Technical, managerial, procedural, and administrative safeguards must be in place to enforce, control, and monitor performance of responsibility as assigned. In short, what is usually called computer security4 will have to be fully in place and thoroughly implemented with contemporary approaches and technology. The SSA already attends to the security of its computer systems but not in the context of a highly integrated work environment. The safeguards now in place will be a valuable foundation on which to build, but security must be an ab initio design and implementation issue for all hardware and software that will become part of a more integrated system. The security issue must not be regarded as an after-the-fact add-on; it could then be too awkward, too expensive, and possibly even impossible to implement. Elsewhere~we have urged that workstation technology be incorporated into the overall system architecture; as such progress is made, security must extend to each workstation as well. In addition to technical safeguards 4 In addition to physically protecting computer rooms and encrypting data links, computer security also entails protecting information from unauthorized or inadvertent disclosure or modification while providing for needed accessibility.
40 within the workstation, communication security is an adciitional requirement. There is much experience and a growing technology base that can be exploited for system and network security. An operational system that would largely have the right technical characteristics already exists. The SSA will not have to underwrite R&D projects or be the lead agency in an unknown new field, but it will have to become thoroughly conversant with current state of art, and it will have to handle a major engineering and development effort to achieve a properly secure integrated environment. OFFICE AUTOMATION The office automation (OA) project was presented to the committee by the SSA as a two-phase project with the objective of increasing the productivity and efficiency of the agency6 In its presentation, SSA assumed an 8 percent agencywide productivity increase, and projected that the overall program would produce a 245 percent return on investment. The SSA estimated that a breakeven point would occur in 19 months. A total investment of $258.2 million was estimated with $27.75 million planned for fiscal year 1989. The first phase of procurement for the OA project began with approximately 5,550 personal computers (PCs) and a second procurement of about 8,000 more is planned. The PCs are using a modest selection of purchased software. Phase I is to include PCs functioning as individual workstations and interconnected into an administrative network. The initial effort to link the SSA's field sites using local-area networks, data modems, and a commercial electronic mail package is in process. We agree with the modest scope of the initial phase and its objective to provide an integrated administrative and management information processing environment. However, the project must involve people experienced with the impact of office automation on the work environment, on organizational structure, and on ways of doing business in addition to those having computer expertise. The introduction of office automation usually changes the culture of an organization and requires clear goals and measurements to ensure that the productivity targets will be achieved. In the absence of such considerations the outcome in industry has often been that no productivity gain is achieved and that introduction of PCs ends up only raising the level of elegance by which tasks are 5The Integrated Computer Network (ICN), designed at Los Alamos National Laboratory under U.S. Department of Energy sponsorship, concurrently supports users operating with classified data, unclassified technical data, and unclassified but sensitive administrative data. The ICN incorporates extensive technical safeguards to control access to various categories of information and to monitor the actions of users. It has been successfully exported to and reimplemented in a commercial organization. 6 The term "office automation" is used loosely; and is not defined with precision. In general, it refers to the use of computer-based technology and techniques to automate the administrative processes of an organization or office. In the past its context was that of a central computer, commonly a minicomputer, connected to a number of simple terminals. More recently, it refers to the use of stand-alone, but often networked, microcomputers or other workstations for the same purpose. However, the functions performed by an MA system" (e.g., word processing, spread sheets, electronic mail, desktop publishing) vary with the specific needs of the organization it serves. Word processing is almost always present. -
41 accomplished rather than saving or avoiding costs7 We have reservations about the effectiveness of investment in the SSA,s office automation project as now structured in the absence of addressing the collateral issues of clear goals, measurements, verification of results, and prior office automation experience. We suggest there be a careful evaluation and appraisal of how the present 5,550 terminals are being used and how much of this is service enhancement versus productivity improvement. We suggest that it is important to answer such questions as: How well were the Phase I original project objectives met? How many of the expected benefits were achieved? As offices began to use the terminals, how many new applications were programmed, and how much service gain versus productivity enhancement was achieved from this experimentation? We also believe that continued procurement of hardware should be delayed until this evaluation has been carried out and that future funding requested for this project be examined in this light. The second phase of the office automation project is aimed at achieving a totally integrated processing environment in which users will have access to both programmatic and administrative functions from a single workstation. This objective raises a broader set of issues that were addressed in the preceding section on integration. As presented to us, the completion of Phase II is to be achieved by 1994. Given the considerable effort left to be done in completing the automation of all programmatic tasks (millions of transactions are still handled manually) and the priority that such work deserves, as well as the work yet to be done to complete Phase I and evaluate its results, we believe that Phase II is more than the SSA can or should strive to achieve in the time frame mentioned. Furthermore, the objective of a fully integrated computer-based information infrastructure for the SSA is unlikely to be achieved successfully given that the needed technical resources are split among several organizations and under different management responsibilities. REFERENCES National Research Council. 1977. Telecommunications for Metropolitan Areas: Near-Term Needs and, Opportunities. Washington, D.C.: National Academy of Sciences. PC Week. 1989. Rare Software Glitch Costs American Millions. Page 61. May 29. Social Security Administration. March 25, 1988. Commissioner's Decisonmaking Process. SSA Pub. No.: 24-033. SSA Administrative Instructions Manual System. 7 A study of 20 Fortune 500 companies conducted by the Massachusetts Institute of Technology, concluded that sophisticated information technology added nothing to overall productivity. The reason: whatever the potential of computers and allied technologies, it was undermined by cumbersome and inappropriate organizational structure.
