This action might not be possible to undo. Are you sure you want to continue?
1-19 Published by: Management Information Systems Research Center, University of Minnesota Stable URL: http://www.jstor.org/stable/249287 Accessed: 11/07/2010 02:14
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=misrc. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact email@example.com.
Management Information Systems Research Center, University of Minnesota is collaborating with JSTOR to digitize, preserve and extend access to MIS Quarterly.
Managers, ComputerSystems, and Productivity*
By: Franz Edelman
Productivity has become the number one concern of American industry. The largest and fastest growing part of the labor force is "the office," that most labor intensive of all American institutions. The vast bulk of office activity centers around managing information, in one way or another. The significant issue is, therefore, the identification and control of the organization's "total cost of information" and the most effective deployment of the resources dedicated to managing this information. The traditional approach has been to address the white collar productivity problem almost entirely through "cost displacement" systems; that is, and operated to systems explicitly designed mechanize and routine manually performed clerical tasks, and thereby to replace expensive labor with less expensive machines. One purpose of this article is the careful assessment of this approach.
The work reported in this article concerns the development and application of a radicallynew system architecture. This has been found, through controlled experimentation, to have substantial impact on managerial effectiveness and on the productivity of the firm's supporting white collar labor force. The article describes the motivation which first gave rise to these efforts and the solution developed to address the motivating problem. It describes a general model of the "cost of information" concept and its implications on one specific company. Finally, it reports an experiment performed to test the basic hypothesis, the results achieved, and the lessons learned from the total experience. The implications of these findings appear to be farreaching. They affect dramatically the concept of "system efficiency" currently in vogue in most places, the design process of computer based systems in general, and the emphasis on applications software, especially with regard to the use of non-procedural languages. In the final analysis, it affects the outlook and activities of the information professional of the future. Keywords: Decision support systems, cost/benefit analysis, productivity ACM Categories: 2.0, 2.2, 3.5, 3.7, 4.2, 4.33 *Thisworkwas sponsored and funded by RCACorporation while the author was its Staff Vice President, Business Systems &Analysis.
The study here is based on reported measurements of actual business operations over a period of several years. It shows quite conclusively that "value added" systems, designed and operated to support the managerial and professional activities of the firm, have the potential to favorably impact white collar productivity in terms of greater effectiveness and lower costs of actual operations, to an equal or greater extent than that associated with cost displacement systems specifically designed for this purpose. The implications of this conclusion are profound and far-reaching. They represent major opportunities for improved returns from the firm's investment in information resources in the 1 980's and beyond.
The Original Problem
the management of the RCA By mid-1972, that some of its Corporation had concluded benefit programs needed to be employee upgraded and decided to begin this process by evaluating the then existing retirement plan with the intention of making any required and/or desirable changes. The corporate Operations
turned out to be upon close scrutiny inaccurate. at a meeting on February 15." As a result. What was needed for data support of this model was real information about real people in the various segments of this population. The latter group saw this extended charter. and was therefore a sister activity of the design and programming group. this was installed at relatively few locations and even there a multiplicity of systems existed. the merits of the case aside. the support of business activities. corporate management made it clear that the prioritiesof the new system. First. were to be two-fold (see Figure 1). requested the appropriate data from all of RCA's operating units. some of which was. Even so. was inadequate for the accuracy and reliabilityrequirements of this task. usually one for each different major personnel application. both divisions and wholly owned subsidiaries. therefore. Thus. a mathematical model of the retirement plan was built as a means for testing the current plan under a variety of external economic conditions. for such major variables as population age. granted to the OR group. it had become apparent that this process of gathering vital data from operations to support a major company-wide management analysis and decision was not working and would not result in the desired outcome. was given the task of preparing the information required for this evaluation and subsequent action. care was taken to model this process in sufficient depth and detail to yield adequate and reliable information with which to support this large and important decision. The difficultyencountered in the productive use of the model . Accordingly. toward the end of 1973. usually again more than one. The OR group was at that time a part of the corporate MIS staff. the well known "bell shaped" type of curves. etc. which were being transmitted up the line during this period. as well as a number of alternative new plans which management might want to consider in the future.is related to data support of this straightforward model. Consequently. decided that "there had to be a better way. with the more or less routine and certainly essential EDP back office activities being viewed as a means to that end. This unusual situation was to have organizational and operational impact for years to come. whatever its eventual embodiment. It soon became apparent that the standard procedure of using stylized statistical distributions. Some twelve to fifteen months later.. or had none at all and depended on manual procedures. as a flagrant invasion of its territory. as perceived by the managers and professionals of the enterprise. became the first priority of the system activity. Since the retirement plan has large impact on current financial status and over the long term in the actuarialsense. It is importantto note that RCA had at that time a corporate Personnel Data System (PDS) in place. The Vehicle- The General Solution From the very outset. the model building task proved perfectly manageable. a reaction which is not entirely unjustified. which had existed since 1957 as a respected and credible analyticalresource to corporate and operating unit management. 1974.Managers. since the relationships involved were all well defined and relatively few in number. most salutory. However. and generally insufficiently reliable to support such a critical process. although some of them were quite complex. The remaining units either used their own system or systems. in the opinion of the author. which had responsibility for the existing corporate personnel system PDS. which has overall responsibility for all matters relating to Human Resources management. rather than relatively high level statistical aggregations.and thereby hangs the entire case study . Specifically. incompatible. Computer Systems. the corporate Operations Research group was invited to develop a more promising approach to such problems in general and to this one in particular. and Productivity Research group. Corporate management. the corporate IndustrialRelations (IR) staff activity. it served to redirect the thinking in many of RCA's system activities toward focus on managerial support through effective business analysis. and to submit an appropriate proposal to this effect. the system was to 2 MIS Quarterly/September 1981 . out-of-date. incomplete. Much of the data. length of service.
through automatic restructuring and extension procedures. which will be reviewed here briefly to put the role of the system into proper context.effective managerial support. which is at the direct and personal disposal of the IR manager on an interactive. it was nonetheless the first time that it was so stated. timely. Number 2. in support of the managerial function above. at a plant site. Volume 8. explicitly and unamibiguously. CLERICAL/ADMINISTRATIVESUPPORT REGULAR TRANSACTION PROCESSING STANDARD REPORTS DATA INTEGRITY Figure 1. Winter 1977. the system was to be instrumental in providing all regular clerical functions and procedures required to assure continued integrity of all IR data in the system. This set of priorities led to four major design decisions. it was clearly understood that the back office was the means for attaining the top priority objective . with an absolute minimumof data redundancy. In other words. This was to include personnel managers at the local level. and all the way to senior IR executives in the corporate offices." Second. MANAGEMENTSUPPORT ANALYSIS POLICY DEVELOPMENT DIRECT INTERACTION 2. e. Objectives and Priorities be capable of supporting the managerial requirements of the IR function on a companywide basis." However. it was taken for granted by the managers that the informationfunctionwould create. on-line basis. This was felt to be essential. and operate a reliable. Specifically." Paul Berger and Franz Edelman.' Design decision #1 Data Base Management (DBM)technology would be employed. in order to * create and maintaina single repository for all employee information. and did not come as a surprise. pp 22-29 MIS Quarterly/September 1981 3 . and Productivity 1. by the senior management at the outset of a major system task.g.. Additional particulars about this system are available elsewhere. Data Base. the system was expected to "provide a facility for 'ad hoc' inquiryand analysis. While this priorityranking should not. at the division or subsidiary level. * guarantee effective data security and privacy to the individualemployee and the operating unit. Computer Systems. maintain. '"IRIS: A Transaction-Based DSS for Human Resource Management.Managers. and * provide a capability for system adaptation to changing business needs. and orderly "back office.
would utilize "interactive" computer technology and a strong "human engineered" interface with managerial/ professional end users. standard reports and all other routine. once the became Thus. they were to feel that the system would be a distinct help to them in the execution of their day-to-day activities and responsibilities. instead of depending on "data extracts" as by-products of other operating systems. where proper validation routines could be exercised. COBOL for the most part. It is based entirely on an ultra-high level language. is supported by reliable. the OR group was commissioned to construct and install the entire system. nonprocedural based front end ." i. Computer Systems.Managers. maintenance. The design and construction of an initial back office facility was accomplished by the end of 1974. The system's front end and particularlythe nonprocedural language have more than lived up to this expectation. Following a successful demonstration of the front end pilot in mid-1974.with a transaction processing system . such as payroll. This decision was the most controversial at the time. also referred to as a fourth generation or "non-procedural" language.e. In addition. relevant. This result has indeed been achieved and has materiallyreduced maintenance costs. results are always delivered on an overnight basis. it was decided that all updating of the database be accomplished through the back office. independence of all application programs from data structure. and in which they could begin to develop a perception of the system's direct and personal utility. that portion which provides regular database updating and 4 MIS Quarterly/September 1981 . virtually for the first time. An interactive front end provides data and the procedures for getting at and manipulatingthem on an "online"basis. Thus.. The non-procedural language was to facilitate an environment in which managerial users would be entirely comfortable with the system at their command. Thus. and installation at the pilot operating unit began in 1975. because of its critical operational responsibilities. if the particulartask at hand is judged urgent enough by the manager to warrant the additional expense of "immediate response.the traditionalback office in the opinion of the writer. the system operational. this decision also turned out to be the one with the most overall impact and warrants the largest share of the credit for the system's successful implementation and operation. In short. which is one aspect of DBM. would be equipped to receive and process directly all transactions of a personnel nature. On the other hand. design and documentation are also greatly assisted by the DBM approach and there is today very little argument about the overall desirability and beneficial impact of the DBM concept. It guaranteed. but these are merely buffered there and are later added to the normal transaction stream for validation and entry through the back office. Otherwise. and Productivity The last point was considered especially criticalin order to reduce "maintenance" effort. and IRIS has become an indispensible day-to-day working tool of the managerial and professional communities. recompiling. The architectural unification of a management support system . and timely data. that portion which is visible to managerial end users.. Design decision #4 This management support system. For the sake of both safety and end user comfort. Design decision #2 The system's "frontend." then this option is available.the human engineered. has turned out to be the most significant architectural development of this entire effort. periodic operations such as payroll. employing standard. since it affected the existence and operations of a number of other majorsystems then in existence. would eliminate constant patching. middle level languages. and retesting of these applications programs as real world changes required modifications of the data structures. a management support system which. This procedure has added greatly to the reliabilityof the data and to the feeling on the part of managerial users that they cannot inadvertently damage their databases. It required almost five years to install this system at all eighteen of RCA's Design decision #3 The system's "back office.e. would utilize traditional programming methods. the user of the front end may enter data changes at the terminal." i.
a much broader outlook on the entire issue of business information. By 1979. Itshould be noted that the post implementation audit measures. orders. The remaining sections will now relate the impact of this system on the productivity of the user function. be viewed as encompassing three sectors: 1.approximately 25% of the total. fascimile reprographics. It is not the absolute figures which seem important here. and $350 billion. and especially the costs associated with its care and feeding. and a host of others dedicated to handling informationin one way or another. and diverse organizations. word processing. and thereby the cost effectiveness of this system effort. usually two or three units being in varying stages of implementation at one time. yet it accounts for the largest and most frequently used set of tools and facilities for handlinginformationtransactions. The Office. The payoff from all sorts of office devices and facilities. Traditional EDPIMIS. in its broadest sense. and Productivity operating units. which represents a large cost element all by itself. Since the third sector is composed entirely of people. lies in increasing the productivity of office labor . or at least systematized. 2. This point is particularly critical to making the correct interpretationof the productivity results presented later on in this article. Computer Systems. He points out correctly that "this sector is rarely aggregated under a single expense heading. accounts for the single largest occupational category in the U. nor even normally associated with the tasks of informationmanagement. and inquiries of all sorts .that host of office employees classified as nonmanagerial and nonprofessional." Just how far beyond did not become clear until some operational measurements were taken in subsequent years.Managers." This third sector is neither aggregated under a single expense heading. Strassman. as shown on Figure 2. respectively. The numerical results presented in the Strassman paper are summarized in Figure 3. The three sectors were estimated in 1973 to have accounted for a national expenditure of $26 billion." Paul A. Thus." These include communications facilities. This makes the measurement of its nature and size an extremely difficulttask. switchboard operators and clerks. In order to begin to cope with the issue thus identified. September-October 1976. The results were. the most labor intensive institution in American life as it has been called. the entire corporation was on the system. MIS Quarterly/September 1981 5 . from computers to mail rooms. It is important to note that this classification scheme specifically excludes managerialand professional personnel. $42 billion. to say the very least.S. neither data processing nor administrative processing is an end in itself. Administrative Processing. devices. a capable Human Resources Management (HRM)system is an absolutely vital prerequisite for such measurements in all large. labor force . Strassman one more time: "After all.secretaries and typists. the large and growing "set of tools and facilities for handling information transaction. To quote Mr. Strassman proposes that information processing. were programmed into the system and are produced for performance evaluation on a monthly basis. but rather the sizes of these sectors relative to one another. micrographics. claims. complex. Recent estimates for the 1979 economy put the totals at more than double those numbers. sion makes measurement of its size a more difficult matter than in the case of the first sector. startling and disquieting. it came as a shock to even the best informed experts in the field that 84% of the cost of managing information is tied up in labor The "Cost of Information" Concept An important paper2 published in 1976 introduced. 3. the well defined and clearly identified activity relating to the use of computers and computer services. Strassman suggests that "managing information systems now goes far beyond just managing computers. HarvardBusiness Review. administrativepersonnel and people who process applications." This disper2"Managing the Costs of Information.
IRIS Cost Performance .34 4.0 16.03 DOWN 4 14.05 + PERCEN CHANG UP DOWN UP o0 OC).45 4. OF TRANS / EMPL / YR I) ) 3. 'AD HOC' COST / EMPL / YR ( COST PER TRANSACTION ( USAGE DATA NO.0 DOWN UP DOWN TOTAL NO.7 120.24 0.03 12/31/79 7.09 3.1 136. / YR (000) Figure 2.' COST / EMPL / YR ( ) ) 12/31/80 7.0) (D CD (c q) oCD I12 12 MONTHS ENDING UNIT COSTS * TOTAL COST / EMPL / YR ( 'REG.7 2007.6 1929.42 0. OF RECORDS (000) TRANSACTION VOL.
PROC.(United States Expenditures for 1973) EDP/M IS $26 BILL ADMIN. The National Information Iceberg (D -o CD CO CC) . / OFFICE INFORMATION PROC. 0 q) 840/ Figure 3.
and geographically dispersed throughout the entire enterprise. non-supervisory employees holding "non-policy making"jobs.Fortunately. and summarizing these costs. organizationally. This result is doubly disconcerting when one realizes that actual computer hardware. This definition eventually boiled down to three labor cost elements for a specific subset of the employee population: base pay. an informationsystem about people with at least the minimumof occupation coding to allow classification by job category relevant to the task at hand. and the accounting system is therefore usually helpful in identifying. What do these numbers look like for RCA Corporation. temporary help. any errors or approximations were to be on the side of omission. The fact that this piece accounts for the bulk of the problem complicates the matter and makes it all the more urgent. direct labor." which was clearly needed to assess the size of the labor content segment of the information cost iceberg.provided a company-wide utility for experimentally developing a model leading to a consistent definition of the "information worker. collecting. however. and a modest overhead factor. it is startling to realize that traditionalEDP accounts for only 6%. in particular? 2. exceedingly conservative. and when there was any debate or reasonable doubt as to whether or not a certain job category was to be included among the "informationworker" class. especially the first of these. but also. Conversely. the RCA IRISsystem for Human Resources Management not only contained such occupation coding satisfying the minimum requirements. thus facilitatingreasonably accurate estimates of the sizes of each of these cost components. if anything. The term "knowledge worker" is also beginning to be used in the literature. In addition. This was the intent because the problem. they are closely and explicitly associated with information handling. and all forms of supervision. Are there reliable ways of measuring the size of the third sector and how might this be accomplished? 3. the employee benefit package. and equally importantly. as gleaned from the nationalstatistics.particularly. Computer Systems. The accounting system is of no help at all in identifying and collecting these cost elements which are functionally. is normallyvery difficult to get reliable readings on. an effort began in early 1977 to address this vital issue by first tackling the measurement problem. What is needed to begin addressing this measurement task is what could be called a "Human Resources Accounting System. The last qualification. This general definition was tested extensively and appears to serve the purpose adequately and consistently. and Productivity content. according to this breakdown represents only 2? to 3( of the total informationcost dollar. Therefore. most if not all large companies require some form of staff approval of device acquisitions in these areas. three large questions immediately emerged: 1. The thirdsegment. in this case white collar labor involvement in handling business information. The "informationworker" subset was defined to consist of all indirect. since it excludes overtime. The reason for the measurement difficulty is the pervasiveness of white collar labor in every 8 MIS Quarterly/September 1981 .required a good deal of custom tailoring to some operating units because of their unique activities. what can systems technology do to help alleviate the problem at RCA? Consequently.This means something Implications on One Specific Company The top two sectors of the information cost iceberg are relatively easy to measure. it was left out. In the first place." in short. was so large that it was judged critical to avoid any form of overstatement.Managers. Was there ever a more telling signal that it may be time to adjust our priorities and broaden our horizons? When these results began to be digested at RCA. The last two items are computed as a percent of base pay. salaried. It should be noted that this definition is. If the results of such measurements indicate a relationship even approximatingthat in the nationalpicture. aspect of business activity. which is estimated today to account for 30% to 40% of total EDP cost.
part of the same population. These figures provided the first hard evidence that white collar intensity and productivity were indeed large and important issues. at that time. the RCA definition of the knowledge worker includes all supervisory personnel. in effect. and five units during the first half of 1 977 would result in growing aggregate impact during these three years. They raised the all important question with RCA's Informationfunction. as measured on the basis of actual 1978 operations. but it also provided the basis for the test of the system's capability to impact the productivity of its user community. and Productivity quite different. with 42.Managers. showing a 72% labor content rather than the 84% reported by Strassman. and thereby help to improve the productivityof the white collar labor force. The "Experiment" In 1978 it became apparent that RCA's IRIS situation had become the basis for a large "controlled experiment. and three service companies. both groups were statistically balanced to assure with high probability that they were originally. high volume manufacturing. whether and to what extent computer based systems could replace some of these labor costs. two separate businesses: one high technology and one service. three high volume manufacturing companies. companies fell into two groups. not only did the system serve as the means of taking the measurements. It takes three to six months for the full impact of the system to be felt. Cost reduction at the clerical level had never been an issue. The remaining four units were ineligible because they did not at that time support RCA's standard occupation code structure. Itwas most fortuitous that the IRISsystem was.000 employees. and therefore selectively tractable. the differential in count. in fact. and service business. that the basic purpose of IRIS had always been the support of RCA's managerial processes. and was at no time even mentioned as a system objective. plus all those others who hold "policy making" positions. as was stressed earlier. The first or pilot unit had become fully operational by May of 1 975 and the last of the nine by June of 1 977. It is most importantto recall. two units in the first half of 1 976. measurable. it can be thought of as the flipside of the information worker. It contained two each of high technology. in the preIRIS period. and therefore could not contribute to the required productivity The fourteen participating measurements. In other terms. It can therefore be hypothesized that these nine units in the aggregate during the 1976-78 period. in the middle stages of company-wide implementation. This test is described in the next section of this article. Computer Systems. This statistical balance is important because these three different types of business have very different Human Resource Management requirements. The IRIS implementation schedule for these nine units was as follows.000 employees. in the sense that the combined populations of knowledge workers and information workers make up the totality of the white collar labor force. is shown in Figure 4. Hence. with later implementations drifting more toward the three months end of this range. with 21." supported by the information worker staffs in the various functions. The former are specifically excluded from the measurements reported here. This circumstance endowed the system with a dual role. will reflect whatever impact IRIS has had on their operations. The resulting breakdown at RCA Corporation of the cost of managing business information. the knowledge workers are the "principals. Thus. the overall relationships are not too dissimilar. The control group consisted of five operating companies.) Thus. While it is true that RCA's ratios are slightly less lopsided than the national picture. favorably impacting white collar productivityof its user community. (One large unit is. The target group consisted of nine companies. which were identifiable. It contained three high technology companies. It could be expected that the of two units in implementation sequence mid-1975." This could be used to test the question of whether or not a system of this architecture does in fact have the potential of MIS Quarterly/September 1981 9 . Fourteen of RCA's eighteen operating units participated in this experiment.
/ Figure 4.0 (1978 OPERATIONS) a (D "1I C) I) CO Cr :' ADMIN. The RCA Information Iceberg . OD PROC. OFFICE INFORMATION PROC.
and since the human resources area is very dynamic. This might have been anticipated since now all nine group A companies had been fully implemented. Thus. whenever necessary. By the end of 1 977. the "with and without" basis should be reliable. group A achieved a per capita white collar labor cost level more than 1 5% below that of group B."using the definition of "information worker" developed earlier. These unprecedented levels of activities clearly illuminated the existence of a responsive facility for supporting Human Resources Management functions. and to calculate that on the basis of annual cost per employee serviced by the department. namely. far more so than could have been expected during relatively tranquiltimes. Computer Systems. which were attempting. through overtime and temporary help rather than additions to force. and Productivity The earliest IRIS implementation in the control group took place in November of 1 978. the cost avoidance differential had risen to 30%. It must be pointed out that the 1976-78 experimental period. It must be understood that this is not a cost reduction in the traditionalsense. It was agreed to use the "cost of information. This situation appeared to satisfy the requirement for a large scale. represented an extremely fortuitous circumstance. The solution finally arrived at was rather simple and served the present purpose as an adequate first approximation. namely the personnel departments. Thus. Never before had personnel departments in large companies been subjected to such pressures. The Results The "information worker" definition was now applied with one additional qualification. During that time. which clearly was not a matter of choice. MIS Quarterly/September 1981 11 . This was the establishment of the "productivity measure" of the user activities. This cost of personnel informationper employee per year is plotted in Figure 5 as the vertical axis against time. One more significant issue needed to be resolved prior to taking the measurements. As a result. the term "cost avoidance" is more descriptive of the financial impact. provided that both groups are indeed from the same population. The level portion of the "B" curve resulted from tight headcount and cost controls. since no clerks and/or administrators were separated as a result of the system's installation. with only four of the nine group A companies fully operational on IRISand all other factors remaining the same for groups A and B. Instead. Rather this difference represents the cost of the labor force which the group A companies did not have to hire in order to satisfy the large and rapidlygrowing requirements for people information. that it identify only those individuals who were engaged in personnel work. Too many other factors change during those three months to identify and isolate the effect of a single change. and the latest in early 1979. no impact of IRIS could possibly be felt during the 1 976-78 period. This cost of information per employee per year was then tracked for the three years for which the experiment can deliver statistically valid results. Since it does take at least three months for the impact of an IRIS implementation to be felt.Managers. This is not an easy matter and care was taken not to overcomplicate this issue for the sake of satisfying the special interests of some of the operating units. through traditionalmeans. The two graphs describe the behavior of the target group (A) and the control group (B). All three conditions were met in this case. By the end of 1976. controlled experiment in a rapidly changing environment. instituted in 1978 to cope with the headcount growth experienced by the group B companies. In addition. and possibly even misleading. it was clear that the typical "before and after" experimental setting would be unreliable. start the experiment under the same conditions. to keep pace with an evergrowing volume of information requirements. The IRISsystem was thus enabled to identify and calculate the annual "cost of personnel information"and to normalize this over the total number of employees serviced by each personnel function. and experience the same information demands and general circumstances for the duration of the experiment. the increasing demand for personnel informationwas now met in group B. or about twice the 1976 level. the long pent-up demand for internalchange at RCA caused massive internal programs to be initiated and executed during that period. These are discussed in the next section. governmental demands for information about people grew dramatically.
Cost of IR Information 1978 . I a ' '--I--- I i I 1975 1976 1977 Figure 5.03 150 140 130 - $per -per Employee year q) 0 CD - B: 5 Un 21.( est.000 on IR 6 I . ) $91 A: 9 Uni 42.00 NOT $130 (D - At $130 0X (D 120 OC I10 100 90 80 70 60 50 I a - $103 $103 $87 $82 .
from at least one point of view. and money on the part of the managers who use this facility. even a low priorityone.a Decision Support System commissioned to provide information support of RCA's personnel policy makers and managers. amounting to approximately $27 per active employee per year. Since the ad hoc percentage is still growing after several years. even greater than that in personnel functions. On the contrary. that the B curve is beginning to turn around. representing a current annual rate of about $2. The year 1979 is not one of the test years. hard measurements of actual operations indicate a majorcost impact on the support staffs in the user organizations .plus approximately $4 for IRISsupport staff. development to installation throughout the firm. It is interesting to note. four majordesign decisions were made at the outset. the cost avoidance of IRIS had leveled off at a little over 20%. The right hand bars show the system cost which replaced the labor cost. been reliably estimated that the relative cost avoidance potential in financial operations is. when white collar labor content is included.45. By the end of 1978. that reliable.42 out of a total of $7.15 per employee per year.5 million annual cost avoidance represents a change in labor content and does not reflect the system cost incurred to accomplish this. was spent on managerial usage (see Figure 2). These related to MIS Quarterly/September 1981 13 . Since the total cost of information in most companies amounts to about 10% of gross revenues. It has. The key word in the above claim is the word "effective. or 46%. effort. did the issue of "clerical cost saving" arise as a system objective. This $2. as would be expected or certainly hoped.no small potatoes in anyone's book. can benefit equally from this type of system architecture. To return to the Information Iceberg concept. the "novelty effect" can no longer explain this extraordinarily high level of managerial acceptance and participation. It is not at all inconceivable that the company-wide end result of an effective office informationprogram can be a displacement of 1 5% to 20% of the cost of office operations. the impact for a corporation having $8 billion in sales. if anything. Never once in the course of the system's from initial conception through existence. and Productivity Since neither overtime nor temporaries are accounted for as cost components in this measurement. The question must therefore be asked what it is about this particularsystem which might account for this phenomenon. This payback does not include the favorable impact which the system clearly has on the managerial processes. it must be inferred that the system performs its top priority mission adequately. The left hand bars show the cost avoidance level in each of the three years. since all but one of the group B companies were fully implemented by early 1979. Since this requires personal time. As was enumerated in an earlier section of this article. spent on the operation of IRIS. Even though we do not know how to measure that in dollars and cents. would be a pre-tax profit improvement of $1 20 to $160 million. Here was IRIS. Now it turns out. Thus.Managers. every business function. Interpretationand Lessons Learned The operating results reported in the previous section present a startling development. and that the A curve continues to hold its own quite satisfactorily. in fact. at all levels of the enterprise from plant location to the corporate executive offices. Computer Systems. all three of these being precious commodities in any managerial community. especially finance. if not with distinction. This amounts to a 60% return and is a profitable venture by any standard. this factor does not show up. after several years of use. making the comparison even more conservative. however. replaced $27 per employee per year of labor cost. Figure 7 depicts an overview of this productivity improvement experience and tradeoff.5 millioncompany wide. the record shows that in 1980 $3. in 1978. This relationship is shown in Figure 6. It claims that each dollar invested at the top of the pyramid in effective system technology can be expected to replace at least $3 in white collar labor cost." It is quite clear that the systems concepts of this particularcase are by no means restricted to personnel operations.a totally unexpected result. for example. $7.
AV COST CLERICAL OPERAT $27 co CO 30 20 - $16 10 r-- L 1976 1977 Figure 6. IRISSystem Cost/Performance 197 .$_per C) cD Employee Year $39 per 40 CD a) (D.
OFFICE INFORMATION C) (b co co 0 Figure 7.INV ( EFF SYS TEC ADMIN. How Large is the System "ROI?" . PROC.
taken one or even two at a time. in the author's opinion. The economic benefits accruing from major reductions in cost and time of development cycles will be one of the incentives. professional effort. many years that cost displacement systems those systems intended only to produce clerical savings . though compound. in retrospect. as will equally majorreductions in the cost of maintainingsuch a system. appears to set this system apart from other DSS. Thus. raises an often quite innocuous question. Conclusions and Summary The IRIS system is now sufficiently institutionalized at RCA Corporation to have become an absolutely indispensible tool of the Industrial Relations communities throughout the firm. has resulted in perceived levels of managerial utility never before reached. It is now becoming quite clear that this is a faulty hypothesis. especially Finance. an interactive." with respect to relevant business information. 16 MIS Quarterly/September 1981 . the most powerful incentive will continue to be genuinely effective support of the firm's managerial and professional activities. and 4. to the local manager the system looks like a local system. able to call on ad hoc aggregates of real world data. combining all four of these attributes. All this is contained in a single. One need only visualize the often frantic activity which can begin at the clerical support level when a manager. extensive and complex modeling will require. for the time being. Of course. is to be learned and capitalized upon. integrated system architecture. But the "directlytransaction driven DSS" concept. while managerial users see only that portion which supports them." presenting radically different views of itself. online extract will deliver the same capability without impairing the system's architectural integrity in any major way. however. Not only can those two types of systems cohabit a single. the direct. architecture. but also a new and very different "system personality. as a clerical support facility for managing day-to-day matters of data handling. Conventional DP wisdom had had it for many. and this point leads to the major lesson which. 3. and in other companies. There is nothing new about DSS having the first three of these. integrated architecture most harmoniously. This facility has produced a systems intended only to beneficially impact the firm's managerial processes . not only avoids much of the clerical churning but also produces better and more timely information. but the synergy achieved thereby makes this combination very much greater than the sum of its parts. Thus. of course. company-wide system. a directly transaction driven decision support system. depending on who is doing the viewing. The cost avoidance potential at the clerical support level of this type of system architecture appears to be considerably larger than that of traditional systems specifically designed for that purpose (see Figure 8). use of database technology. with equal user friendliness. such direct interaction is economically inadvisable. again in retrospect. Computer Systems.and decision support systems . to have been the most far reaching in generating not only a new system architecture. available on an online basis. However. a conventional. the fourth design decision would have to be judged.are entirely different worlds and cannot successfully cohabit a single. It is easy to see and has now been reliably demonstrated that a higher degree of "managerial self-sufficiency. understood that there are applications where. while to the senior corporate executive it looks like a top level. It is expected that its architecture will become the standard in systems for other business functions. and Productivity 1.Managers. because of the high level of transaction volumes. The DSS concept.) The availability even a modest facilityfor ad hoc of modeling has added greatly to the managerial perception of high system utility. that the best road to more efficient office operations is through more effective support of the managers. since it is no longer necessary to have professionals program relatively trivialtasks. In those cases. (It is. especially a senior executive.those Perhaps it should not be too surprising. batch oriented back office. Clerical staffs see the system. 2. non-procedural front end.
.Managers. C) C) ILI z Cl) IP% 0 E E C'. LI 4: LLI C.)UJ _1 0 oc MIS Quarterly. Computer Systems.) 0.September 1981 17 . and Productivity U--o LJJ (C o C) LLJ Cl) ICo) cr ~ z I LLI I 4< am2i C0 an E a.<: LU Cl) LL -J c 4< (M =I CMC) - O 0. C-) o) 0 0c on pc LLU C. 0 0) I iz LL I II 0) O I.
there have been organizational changes in the direction of emphasizing the IR function's information resources and giving it more prominent organizational placement. which was originally commissioned to design. Inc. in general. In 1954. build.e.D. One major organizational change has resulted directly from this development. Edelman has wide experience over a broad spectrum of business functions and his current interests center around Information Resource Management. and the Society for Management Information Systems. with honors. of Princeton. a high performance state-of-the-art system for the management of informationin the Human Resources area. Dr. by the "throughput" of the computer installation. Since people now comprise 70% to 80% of the total cost of managing information. During his tenure at RCA.Managers. While this is perhaps an issue requiringlonger lead times. The corporate Operation Research group. New Jersey. He became its Director in 1962. he and his activity became part of the corporate planning function and corporate MIS. and is widely known for his work in the design and use of Decision Support Systems. in 19 77. more importantly. participating in the design of the RCA color television system.S. he helped set up RCA's first technical computing facility in support of research and engineering activities. Subsequently. Allthis has resulted in a change in outlook. He is a member of the Institute for Management Sciences. i. and rightly so. Acknowledgments The work described in this report consisted of many large and complex tasks. an independent consulting company. both system professionals and. and run this system." Whereas this used to be perceived as relating primarilyto how efficiently the hardware was utilized. Dr. About the Author Franz Edelman is President of Edelman Associates. toward computers. and M. It is said. it is now seen much more appropriately as relating to the productivity of people. 18 MIS Quarterly/September 1981 . It owes its success to two essential factors: a very small but exceedingly talented and dedicated staff of systems professionals and an equally dedicated and innovative functional user community. install. Dr. and for all aspects of computer applications in 1 977. especially as that relates to the support of managerial activities. from McGillUniversity. Dr. that one measure of a system's effectiveness is the degree of its organizational impact. the Operations Research Society of America. In 1975.. He is the originator and initial designer of the IRISsystem. whose total commitment to this effort was absolutely indispensible. Computer Systems. he was a member of the Planning Council on Information Technology of the American Management Association. He is under retainer to Index Systems.. under whose auspices he conducts a major portion of his business. Edelman was associated with manufacturing and engineering operations and spent several years at the central R & D facility. end users. where he rose to the position of corporate Staff Vice President for Business Systems & Analysis. Inc. he was given additional responsibility for systems development and. Prior to entering the consulting field and joining Index Systems. degrees in Applied Mathematics from Brown University. It has surely raised the credibilitylevel of the information function and has begun to yield more acceptable returns from the considerable investments in the resources of informationmanagement. and information management. dedicated to the analysis of RCA's business problems. computer people. He has lectured and published extensively. and in 195 7 he established the Corporate Operations Research group. was assigned the responsibility for all corporate system development in 1976. Until his retirement. he achieved the position he held at his retirement. on the part of the managerial community of the user function. and Productivity major reorientation toward the notion of "system efficiency. Edelman spent over thirty years in the employ of RCA Corporation. Edelman obtained a Bachelor's degree in Mathematics and Physics. and Ph. this reorientation is surely long overdue. with responsibilities for company-wide computer applications and business analysis.
who then represented the IR function of the pilot operating unit. and Productivity It is almost impossible to identify specific individuals without the risk of omitting equally meritorious performance. Mr. Their to this effort was dedication personal indispensible. the history of this entire project is based on a rare and exceedingly productive partnership between the systems and user communities and whatever success was achieved clearly belongs equally to both. Fuchs. responsible for the success of this project. this system effort has had two champions in the RCA Executive Offices. On the client side. played the key roles of the users on the system's design task force. Computer Systems. George H. has given the top management support needed to assure the system's acceptance with the managerial community. Robert F. RCA Executive Vice President for IndustrialRelations. especially during the early period before the system was able to survive on its own merit.Managers. MIS Quarterly/September 1981 19 . Arthur Bennett and Jack Noon. The project benefited greatly in its early stages from the participationof Woody Skillman and from the leadership of Paul Berger. Corporate Vice President for Organization and Management Resources. and his very capable staff. Dr. In short. who took over its management in 19 75. Messrs. as much as anyone. Maddocks. the group's present director. has been a constant and enthusiastic supporter of this work and is. Great contributions have been made by Denis Foley.