Information Technology in the Service Society: A Twenty-First Century Lever (1994)

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 Impacts of Information Technology at the Enterprise Level The differing structure, environment, and competitive situation in each service industry explain only some of the differences in performance rela- tive to investments in IT. Other differences reflect the experiences of indi- vidual companies within each industry. Both performance and the use of IT can vary greatly among companies in the same field (see Appendix A for references to relevant studies). Most of the variation reflects the fact that decisions concerning IT investment programs are made at different points in time by managers with different abilities and perspectives. Since most major decisions about IT investment are made by individual executives at the enterprise level, understanding how these decisions are made and evalu- ated is critical if national policy initiatives are to be effective. This chapter examines why firms invest in IT, how firms invest in IT and the categories of applications that are typical, and how firms assess the impacts of their use of IT. It also discusses cross-cutting observations on IT and examines the widespread belief among executives that IT has en- hanced performance overall, regardless of what macroeconomic measure- ments may indicate. The chapter is based on committee discussions in- formed by a set of semistructured interviews (Appendix D) held with about 80 senior managers in 46 major firms drawn from each of the major service industries transportation, communications, retailing, wholesale distribution, health care, financial services, and professional services. The interviews sought (1) to discover what types of enterprise-level data and experience might be available, (2) to validate further or raise 97

98 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY informed questions about the basis for industry- and macroeconomic-level data and analysis, (3) to understand what factors managers consider impor- tant in making decisions about investing in IT, (4) to obtain some useful insights into how companies manage their IT investments for greater effec- tiveness, and (5) to explore further the causes for and implications of the IT paradox. This exploratory investigation was designed to yield insights, not to substitute new statistical data for data obtained in other studies. The par- ticular firms whose managers were interviewed were chosen because they had verifiably strong reputations as leaders in the use of IT. As such they were more likely to have both positive and negative experiences to draw on, to have thought carefully about critical issues, and to be willing to discuss them. However, their response cannot be generalized beyond this sample, because these firms were clearly more likely than the average company to provide evidence of a positive relationship between use of IT and perfor- mance. Some conclusions follow. WHY FIRMS INVEST IN IT Executives generally invest in IT to increase expected profits, margins, or returns on capital. Some investments those made to reduce costs, to generate new products, to make measurable improvements in quality may be discernible as gains in productivity in standard published reports. How- ever, executives may also invest in IT for reasons that do not show up readily in such reports but may still be vital to the success of enterprises. For example, in some endeavors (e.g., research, engineering design, or medicine), the capacity to work at the intellectual frontier or to solve advanced prob- lems may depend on having the most current technologies. In others (e.g., accounting, consulting, legal work, credit clearance, or trust management), flexible access to the most powerful current databases and efficient data- manipulating capabilities may be essential to deliver a competitive level of quality. Managers generally regard IT as only one of many possible investments that may benefit their firms. When they choose to invest in IT, they do so in the belief that such an investment will provide better returns as compared to other alternatives, for example, not investing or investing in other pro- grams. But many of the factors they consider and many of the effects of their successful investments may not show up in aggregated economic mea- sures such as productivity, return on investment (ROI), sales, or margins. For example, since companies can neither calculate nor report their "alter- native cost" gains or losses where their companies would have been posi- tioned without an investment versus where they are with it it is impossible

IMPACTS OF INFORMATION TECHNOLOGYAT THE ENTERPRISE LEVEL 99 for more aggregated databases to reflect such gains or losses.2 Managers, especially in professional firms, may consider the cost of overinvestment in IT as insignificant against (the potentially much larger) losses the firm might incur if it fails to attract top talent, sustain its competitive edge, or hold market share because its IT systems are inadequate. Amona the more important rational reasons for investing in IT that are . not picked up in aggregate databases are the following: Expanding market share or avoiding catastrophic losses. While aggregate industry output is a key economic indicator for policymakers, managers focus more on the performance of their individual companies. A company may gain or lose share even though statistics on an industry's performance in the aggregate may indicate relative stability. For execu- tives, market share is a key parameter of successful performance as well as a basis for marketing, purchasing, and personal power for those in the com- pany. Investments in IT that only shift market share may indicate little or no measurable benefit at the industry or macroeconomic level. The same is . true of investments that successfully prevent very large losses-like air- plane crashes or environmental catastrophes. Yet both types of investment are economically rational. Creating greater flexibility and adaptability for future business en- vironments. Changes in the business environment (e.g., greater regulation, increased market or operational complexity) often demand significant changes in the way a company operates. In some cases, investments in IT may represent essential elements of the infrastructures that allow companies to survive despite rapid and unforeseeable changes in the external environ- ment. In other cases, companies may invest in IT to gain future flexibility for creating products that may not yet be planned or even conceived. In still other cases, IT may enable a company to maintain or reduce its costs while providing greater variety and flexibility in its services. · Improving the quality and stability of internal environments. Some firms invest in IT to obtain and analyze information that can provide a greater degree of predictability or stability in their operations and that can help them to avoid undue fluctuations in sales, profitability, or employment. Other firms may invest in IT to improve employee satisfaction or to en- hance worker safety. IT can help to eliminate burdensome tasks, make jobs more attractive, shorten training cycles before an employee becomes pro- ductive, and improve morale. IT is an especially powerful tool that compa- nies with life-critical operations can use to manage large amounts of data in real time to improve the safety of employees, customers, and the general public. · Improving the quality of products and of a firm's interactions with  customers. For many service firms, IT is an important element in creating

100 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY services that are better customized and tailored to meet the needs of indi- vidual customers. Companies increasingly compete in other aspects of cus- tomer service quality, often using IT to serve customers more rapidly, pleasantly, responsively, accurately, and completely. Such improvements help gener- ate long-term loyalty among customers. Similarly, IT can play an important role in improving reliability by ensuring more consistent levels of perfor mance and minimizing errors and can thus enhance customers' perceptions of a company and its products. As indicated above, individual firms have invested in IT to achieve a variety of results that are not generally captured in aggregate databases. A recent study by Brynjolfsson and Hitt3 provides quantitative support for these ideas. Specifically, Brynjolfsson and Hill analyzed output data and data on IT spending from 380 large firms that generate about $2 trillion in output annually, and found that increases in output at the firm level corre- late strongly with IT expenditures, even if the macroeconomic data do not demonstrate increases in productivity across the national economy or in individual service industries. These data were taken from the 1987 to 1991 time frame. HOW COMPANIES USE AND INVEST IN IT Types of Applications Senior managers interviewed by the committee tended to cluster their selected uses of IT into several broad categories that they evaluated using various techniques. The uses of IT included systems to support (1) basic infrastructure for communications and data handling, (2) mandated require- ments, (3) cost reduction, (4) specific new products, (5) desired improve- ments in quality, and (6) major strategic repositionings (perhaps including a complete reorientation of a company's business). ~ ' ' evaluate investments in IT on a program-by-program basis, rather than be- ing concerned with measures of overall productivity. In the committee's (companies tended to sample, the techniques used to estimate or evaluate the impacts of IT were comparable to those used for other major investments. Basic Infrastructure for Communications and Data Handling Information technology infrastructures consist of those basic informa- tion storage, retrieval, communications, processing, and distribution sys- tems that provide the backbone or framework interconnecting the multiple information-generating and information-using nodes in an organization. Although definitions may differ in detail from company to company, the IT part of a

IMPACTS OF INFORMATION TECHNOLOGY AT THE ENTERPRISE LEVEL 101 firm's basic infrastructure generally includes the core voice-transmission, data-communication, record-keeping, and transaction-processing systems necessary to conduct business. What a firm regards as infrastructure may change over time as new technologies become available and customers come to expect certain basic facilities from their service providers. Interactive data communications, for example, may be essential today, although this was rarely the case a decade ago. Basic IT infrastructure systems are often approached as a "cost of being in the business." The majority (67 percent) of companies interviewed by the committee had done preproject analyses of IT investments in this area. But they often found that it was not worthwhile to separate out and justify individually the IT component of infrastructure any more than it would be to use ROI techniques to justify the purchase of hammers for a carpenter, the acquisition of a telephone for a lawyer, or the installation of a roof on a new building for a real-estate developer. Some items are integral to being in a certain business. Firms tend not to give serious thought to doing without such elements of infrastructure in toto, and they often do not make a formal evaluation of their overall payback. But they frequently do evalu- ate incremental investments in infrastructure on a cost-benefit basis relative to other means of accomplishing a particular goal. On this basis, a majority (73 percent) of interviewed companies said that they had attained accept- able to high measured returns on such investments. But 27 percent reported negative or indeterminate returns. Measurement complexities represented serious problems. For example, a study by the Aetna Insurance Company once analyzed the impact of IT on expense-revenue ratios over a 15-year period. Although the study concluded that IT had had little impact on these ratios, it also noted that the company could never have increased its handling of health care claims from 250,000 per week to 1.3 million per week without using IT and acknowledged the fact that this business was very profitable. Back- office automation in financial services, electronic point-of-sale systems in retail and wholesale trade, computer reservation systems in travel services, communications systems in all companies, and automatic flight control sys- tems in aircraft for air transport are examples of large IT infrastructure systems without which individual companies could not compete today but for which specific financial impacts are difficult to calculate. Observed Richard Liebhaber, chief strategy and technology officer at MCI, There is an extensive infrastructure of holes in the ground, structures on the ground, and apparatus in the sky. That is our fundamental infrastruc- ture. We regard it as a commodity asset: the transportation mechanism of our company. We view it from the standpoint of a cost factor, and we either build it, buy it, or lease each component depending on comparative cost and performance characteristics. The real problems in evaluating

102 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY productivity for these IT applications lie in those situations where the outputs are difficult to quantify in financial terms: "If I do this I will save 2 minutes here; if I do that I will save 8 minutes there," versus another alternative, "If I do this, I will save $9 million." As noted above, managers are not indifferent to infrastructure costs and impacts. When IT is regarded as basic infrastructure, the managerial deci- sion is not whether to invest in IT, but rather how to obtain needed IT capabilities at the lowest cost. If a firm's IT infrastructure requires some new capability, managers may consider alternative sources of that capabil- ity (e.g., performed in-house vs. purchased from an outside source) and seek to obtain it at the lowest long-term cost. When providing new levels of service to existing customers requires that incremental power or features be added to the base structure, companies may compare the net incremental costs and benefits of each of these features, in financial terms if possible. If features cannot be evaluated directly in financial terms, many companies use other metrics such as improvements in response times, levels of cus- tomer satisfaction, market penetration, or company image. Several of the firms interviewed by the committee suggested that the only truly rigorous way of evaluating the payoff from investments in IT infrastructure would be to calculate the opportunity costs of not being in the business, that is, the losses that would have been incurred if the investment had not been made. However, such a calculation could be highly tenuous because of difficulties in estimating what would have happened if the in- vestment had not been made. As Roger Ballou, president of the Travel Services Group (USA), American Express, noted: A good example of an infrastructure investment would be our automated authorization system. To determine the benefit of that system even today, I'd have to recreate mathematically what would exist if we didn't have it and only had our old personal telephone authorization system. For that you'd have to guess how a 20-year-old system would operate in today's world. That would be a highly theoretical exercise. The complications of evaluating the impacts of infrastructure and the possibilities of basing some decisions about investing in IT on cost and others on factors besides cost-were illustrated by J. Raymond Caron, president of CIGNA Systems, who observed that there are two types of infrastructure investments. For some, specific pay- back and evaluation mechanisms are used; for others, they are not. For example, for our data or voice communications we have put together a program based on AT&T's Tariff 12 and network costs. We know what our costs are, and we have goals to reduce those costs per minute, per connect, to the point where there are clear measures of benefits. In this case we can show dramatic evidence of our communications facilities cost

IMPACTS OF INFORMATION TECHNOLOGYAT THE ENTERPRISE LEVEL 103 ing far less today than they did 5 years ago.... When it comes to PC platforms (our CIGNA-Link platforms, which include PCs, LANs, the soft- ware to go along with them, plus e-mail), we find it very difficult to develop a useful cost-benefit measure. We've taken a position that we shouldn't waste time trying to do it. Caron's comments about personal computers (PCs) illustrate a special problem in measuring enterprise-level impacts of infrastructure. The falling costs of computer hardware and applications software, the greater user friendliness of many computer systems, and the increased capability to interconnect local and remote systems have started to change the very nature of business infrastructure. IT systems are used to increase the level of information and flexibility within an entire enterprise. Such generalized impacts are exceed- ingly difficult to measure in the short run. However, similar PC-based hardware and software can be specifically targeted to improve the performance of a particular operation. When used in this way, PC applications (e.g., to improve order handling, purchasing, customer inquiries or complaints, truck dispatching, field repair, editor-to- printed-page cycles, and broker-to-customer interfaces) have had consider- able payoff. When specific goals are set, processes are properly reengineered, organizational interfaces are redesigned, and new incentives are provided, performance gains from specific applications can be impressive. Cycle times can be cut from days or weeks to a few hours or minutes. Layers of management or entire departments can be eliminated. For example: · At the New York Times Company and Time Inc., PCs enable edit- ing to continue right up to the moment when a story is electronically typeset for printing. Entire intermediate hierarchies and the attendant complexity of coordinating different functional centers have been eliminated. · At McKesson Corporation, retail customers can use PCs to interact directly with the company's computers to eliminate many delays in order processing and much of the bureaucracy that used to exist to handle this function. The number of order takers at McKesson has been reduced from more than 700 to almost zero. The size of the sales force has been cut in half, even while sales have risen by a factor of 8. The number of buyers has been reduced from 150 to 200 located at over 100 distribution centers to just 12 sitting in front of desktop computers at a central site. Levels of management have been reduced in keeping with these changes. In other settings (e.g., universities, research laboratories, or publishing houses), professionals often prefer to write and edit directly at their own computers, thus decreasing the number of secretaries needed for these tasks and flattening and downsizing the organization. It is not clear at this point whether measured productivity improves when (often) higher-priced em- ployees like professors or researchers undertake such tasks.4 In other situa

104 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY lions, built-in desktop computer controls have eliminated much of the need for personal oversight in insurance, brokerage, and accounting applications- eliminating layers of bureaucracy. Despite the undeniable utility of PC applications in certain settings, many executives interviewed by the committee were ambivalent about the impact of across-the-board access to desktop computing in their firms. Measures of performance are unclear, and personal intuitions about the potential and real payoffs from this type of IT vary widely. For instance, a majority of managers interviewed believed that the printed appearance and visual clar- ity of correspondence, graphics, reports, and desktop printed matter are better because of PC use. Communication between knowledge workers (i.e., the professionals responsible for interpreting and analyzing informa- tion made available to a company), as well as the information available to such workers, may also have been improved. But the measurable value of these improvements remains unclear when compared to the significant costs of having desktop computing available throughout an organization. · At CIGNA Corporation, James Stewart, executive vice president and chief financial officer, said, "I'm not yet convinced that dispersal and utilization of PC-based technology have proven to be efficient. The busi- ness people in our organization drive me crazy on the subject of cost and benefit. I see increasing expenditures for what I perceive occasionally as 'toys in the business world,' which don't add up to measurable output or improve our results. I think better discipline in the effectiveness of PC technology is one of our great challenges." · At BankAmerica, Martin Stein, vice chairman, noted: "The area where I would be the first to agree is that the banking industry has not done a good job in utilizing the PC as an office automation tool. We are faced with the typical problem. Everybody's secretary must have a 486 chip in his or her PC because it's much faster. And the question becomes, So what? The metrics for measuring this kind of productivity are not very good." Interviewees reported that, to better manage the costs of their basic IT infrastructures, some companies regarded them as a "utility" and charged divisions for their use, just as divisions would be charged for externally purchased services. To further control costs, a few companies (10 to 20 percent) had begun to benchmark the utility's costs and performance against the prices and performance available for similar services bought externally. (Benchmarking is discussed in greater detail in Chapter 5.) Skepticism concerning the degree of payoff from general office use of desktop computers seemed widespread among interviewed executives. This may reflect the fact that PC software in its first decade was not well suited to the needs of nontechnical users and that many senior executives used comput

IMPACTS OF INFORMATION TECHNOLOGY AT THE ENTERPRISE LEVEL 105 BOX 3.' thong Lotus Notes at (:hase Manhattan Bank Chase Manhattan is beginning ~ project to deliver financial informa- t~on to its customers using Lotus Nlotes, a '~groupware" application de- signed to facilitate transfers of information contained in large databases. Senior Vice President Craig Goldman argues that Flutes creates a new attitude for the information services department, allowing managers to focus on the Font office where employees interact with customers rather than on the back offices Goldman believes that Notes provides an envi- ranment that will let employees react quickly to new business ooDortuni ties. . . SOURCE: Vizard, Michael. 1993. "Chase Banking on Notes to Reach Customer Desktops," Computerwor~d, April 12, p. 47. ers sparingly because of unfamiliarity or discomfort with them. However, this is clearly changing; software is improving rapidly, and upper-level managers increasingly are individuals who grew up with personal computing. Recently, a number of companies have reported observing enhanced performance when new, specialized "groupware" communication tools (like TeamFocus, Lotus Notes, or VisionQuest) are used to capture system knowledge, make meetings more efficient, and realize some of the communications benefits sought for PCs (Box 3.1~. The emergence of simpler network software, multimedia capabilities made possible by autosynchronous trans- fer technologies, and greater capacities for electronically connecting fixed and mobile workstations may increase the benefits derived from desktop computing. Mandated Systems Regulators such as the Internal Revenue Service, the Food and Drug Administration, the Securities and Exchange Commission, state and local regulatory agencies, and other constituents like the investment community now require companies to provide specific reports or to have in-line moni- toring systems for operations. Most respondents to the committee's ques- tions considered the necessary investments as costs of being in their par- ticular business and used IT as an element in reducing the cost of complying with increasing demands from regulators and other constituents. These investments in IT are made with no possibility of showing a specific profit or gain to the company except as compared to performing the task by more manual means. On that basis, most interviewed executives believed that the returns on investments in mandated systems were satisfactory. For

106 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY many companies, mandated requirements for reporting and monitoring have escalated in recent years, perhaps because computers have made more com- plex reporting easier or even possible. Almost by definition, mandated investments in IT do not contribute to a company's measurable overall productivity or gains in performance. But interviewed company executives did not manage mandated invest- ments in IT casually. Many estimated the relative cost of achieving the desired result using system A versus system B. While no evaluations were made of direct payoffs or returns for such investments, they were carefully monitored and evaluated using benchmarks, performance metrics established on installation, and operations costs versus other alternatives. Cost Reduction Cost reduction is often a major goal of investing in IT. This was especially true in first-generation applications, in which IT often substituted directly for manual or clerical labor in repetitive transactions such as mate- rials-handling activities or in large record-keeping departments such as ac- counting, purchasing, or personnel. Other first-generation uses allowed firms to achieve substantial gains by avoiding costs associated with exces- sive capital float, inventories, and billing or payment errors and by saving on the costs of paper processing, check handling, production set-up, engi- neering drawings, or shipping. While the benefits of these systems were said to be substantial, independent studies and the committee's interviews suggest that large companies have begun to reach limits in automating re- petitive functions and achieving what might be called the industrialization of services.5 They are now focusing on using IT for other, more complex- and often strategic purposes. Second-generation cost-reducing systems are often focused on consoli- dating or eliminating facilities such as data centers, production-planning departments, or order-processing bureaucracies. An interesting application in this area was illustrated by Craig Goldman, senior vice president and chief information officer of Chase Manhattan Bank: There is a significant cost-reducing phenomenon that is not well docu- mented in the industry. That is picking a single site [from which] to deliver an information-intensive product on a global basis. It means you have common software in one location that delivers outputs to multiple sales locations or distribution centers. Anytime you have software in mul- tiple locations, it is never common and this lack of commonality is usually very expensive. That is where we see large potential efficiencies. A large majority (85 percent) of the companies interviewed made for- mal evaluations of their cost-reducing IT projects. These firms generally found that direct financial savings and returns on investment could be cal

IMPACTS OF INFORMATION TECHNOLOGY AT THE ENTERPRISE LEVEL 107 culated with relative ease. Standard formal capital planning and payback analysis techniques widely accepted in industry are applicable in this con- text. Although often not implemented consistently or well in the early days of IT, ROI calculations for these projects were said to be fairly straightfor- ward now. In most cases, interviewed companies now claimed to achieve targeted results or to exceed them through secondary uses of the same IT technology. At the same time, the payoffs from cost-reducing systems often were not as high as they could have been because of inadequate reengineering of processes and inadequate reorganization of personnel. (This point is addressed further in Chapter 4.) Specific New Products or Opportunities IT can enable the creation and delivery of specific new information- based service products that are extensions of a company's basic repertoire. Sometimes, such extensions are relatively modest, for example, a new "shopping channel" offered by a cable TV company. In other cases, such extensions have a major impact; for example, a telecommunications company may install new software that allows differential billing rates for calls placed to special parties or at a given time of day (MCI's Friends and Family pro- gram or AT&T's Reach Out America program). If a new product is distinctive enough, a company can measure its impact on profit and sales quite explicitly. However, many new products merely apply new applications software to flexible IT infrastructures already devel- oped, as when a new insurance product is offered using existing systems. In such cases, it is difficult to assess the total costs of the new product, although its incremental costs may be easy to evaluate. Such joint cost situations may introduce other complexities. By further utilizing existing IT infrastructure, some particularly innovative new products may simultaneously lower average costs in the system, thus allowing greater pricing and servicing flexibility on existing products. A prime example is MCI's Friends and Family program, which lowers costs for residential customers yet uses the network's otherwise excess capacity in nonbusiness hours. In the words of Richard Liebhaber at MCI: The decision to do Friends and Family was a combination of retention and market share: 62.8 percent of MCI's traffic was during the day; 33 percent of the traffic (in 1990) was off shift. The percentages in residen- tial traffic were approximately reversed. Even today much of our infra- structure is idle during the night. So when we introduce a new product for nighttime residential traffic, such as Friends and Family, it may not require much capital. Many of the key investments will be for marketing rather than technological infrastructure. Returns are going to be largely the dif- ference between increased market volume (share) and marketing costs.

108 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY The creation and delivery of specific new service products have a more modest and often a much more predictable impact than do strategic innova- tions (see below), because such products can be offered within the context of a well-understood business concept. A majority (64 percent) of the companies interviewed by the committee said that IT investments used to support new service products could usually be analyzed explicitly for profitability and re- turns. The vast majority (90 percent) of responding companies noted that they received satisfactory to high returns on these investments. It is important to note that the corporate financial statements that are the basis for macroeconomic statistics reflect total returns from specific products, rather than the prorated returns that might be more appropriately assigned to IT as just one component of total investment. Any bias in reported data would therefore overstate positive returns on IT at more ag- gregated levels. Even for investments in IT that seemed to have quite measurable results, companies whose representatives were interviewed only had an overall correlation between investments and total project returns. This is a special problem in interpreting results when as has been noted- many new information-based products capitalize on installed infrastructure and marginal investments for the new products themselves are minimal. Conversely, if new products do not achieve desired returns, one cannot conclude that the fault lies in the IT system. Poor planning and implemen- tation may be the principal culprits. Further, since potential payoffs from new products are always probabilistic, multiple projects and investments in IT are often necessary as in any other innovative field to obtain a single payoff. Hence, average payoffs on new products will be considerably less than those on established ones. And actual returns may not appear until long after investments have been made. Improvements in Quality Improved quality is often a most important output of IT systems. Yet most service companies encounter major problems in appraising the impacts of IT on service quality, and especially in measuring that impact in finan- cial terms. This is true when two services are compared simultaneously or when a service output at one time is compared to an output at an earlier date. Service quality is often an intangible whose real value exists only in the mind of a customer. Respondents to the committee's interviews noted that it was impossible to quantify how much more valuable it is to a cus- tomer to receive a monthly bank statement within 48 hours versus a week later; how much more a cleaner, neater, more professional letter is worth; how much better served a customer may be in choosing from 100 financial service products rather than from 50; or how much it benefits a customer to have a shorter wait, a more accurate sales slip, or a more relaxed clerk in a

IMPACTS OF INFORMATION TECHNOLOGY AT THE ENTERPRISE LEVEL 109 checkout line. Companies reported that it was also very difficult to define the relative worth to customers of the more precise analytical and diagnos tic capabilities now available in, for example, medicine or design, versus the less powerful but sometimes cheaper techniques of the past. Often IT allows the handling of enormous increases in complexity. Although large benefits accrue to customers, competition may prevent the service company from increasing its sales or profits. For example, travel agents must complete in minutes arrangements that they could not have handled in days or in some cases weeks-only a decade ago. To optimize their load factors, airlines make an average of 175,000 fare adjustments a day and up to 600,000 on some days. While these may pay off in higher profits at the airline level, travel agents must often re-search and re-issue tickets multiple times for a given customer's flight. Complexity, com- pounded by tasks such as worldwide arrangements, has increased by orders of magnitude; however, agents' average commissions have held steady at between 7.5 and 9.5 percent on airline prices that have dropped in real terms since deregulation. Other factors further complicate determinations of quality. First, the quality of outcomes is heavily dependent on the actions of the buyer. Su- perb educational, health care, consulting, or financial services can be ren- dered useless if the buyer does not follow up properly. Alternatively, ex- cellent customer implementation can convert poor-quality services into astounding successes as successful graduates from poor schooling systems so often prove. Second, final outcomes may not be known until long after the deliv- ery of a service. The true costs and benefits of education, surgery, den- tistry, or insurance, for example, may not be known to either the customer or the provider for years or decades. Even then exogenous events may largely determine outcomes. But the customer will perceive-and receive- the service's value in terms of actual outcomes, not only those over which the producer might have had reasonable control. Measures of IT's impact on performance are essentially useless unless they adequately reflect the quality of output. While a significant majority (over 80 percent) of the responding companies had established formal metrics to measure the quality of their service, these were almost always engineer- ing metrics (such as system response times, percentages of time computers were available, and cycle times for operations) or data from customer sur- veys (expressing customers' views of the services of the company) that could not be converted into financial measures of the quality of output. Even when they had elaborate engineering and survey mechanisms to moni- tor quality, companies could not isolate the relative contributions of IT, versus those of other factors such as management, in creating these outputs. Nor could they relate such measures directly to financial results. American Express's $6.8 billion Travel Related Services (TRS) Group

110 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY is a case in point. The TRS Group samples client satisfaction monthly on such matters as timeliness, accuracy, and problem resolution for all its transaction types. It tracks some 46 measurements defined as important by client research broken down into six main categories. American Express asks its clients to weigh the importance of each issue, leading to a weighted customer satisfaction index that summarizes the key metrics and aggregates the number of "dissatisfying incidents" recorded in interviews and in cus- tomer complaints. American Express uses these indexes as direct measures of the quality of its performance. These metrics are tied into a total quality management approach throughout the TRS Group division for improving the overall effectiveness (efficiency and quality) of its operations. Yet American Express cannot place a direct dollar value on improvements in quality. Unless sales increase or capital or labor requirements decrease as a direct result of investing in IT to improve the quality of output, companies are generally unable to place a financial value on the benefits they receive from using IT. Surrogate measures of quality-engineering metrics or data from customer surveys-were used by a vast majority (90 percent) of the companies interviewed by the committee. Yet no company stated that it could reliably convert such measures into metrics of financial payoff. In- tangibles such as a firm's better responsiveness to customers and increased coordination with suppliers do not always increase the total volume or the margins and prices on products a company sells or services.6 Competitors may quickly match the new capability, and each company may merely maintain its share of a total market that has not grown in size. Although companies often try hard to measure the impact of IT on quality, capturing such impacts and presenting them in ways that can be aggregated at higher levels of data are real problems. Major Strategic I,z~ovat~ons Strategic investments in IT are those that significantly change the posi- tioning of a company in its markets, redefine the basic nature of the com- pany, or are necessary to maintain the viability of the company despite unforeseen competitive incursions. Strategic systems tend to simultaneously change many different elements of customer, cost, and competitive relation- ships not just revenues or costs. They affect such things as quality of customer service, flexibility, break-even points, response times, and market image, which cannot be measured in simple financial terms and hence be picked up in macroeconomic data. Strategic changes almost always induce a competitive response whose timing, scope, and force are unpredictable. So many things happen at once in the marketplace that respondents reported that it was difficult, even post

IMPACTS OF INFORMATION TECHNOLOGY AT THE ENTERPRISE LEVEL 1l l hoc, to sort out impacts precisely. As in the introduction of new products, major effects may not be felt for a long time. For example, both the retail trade and telecommunications industries are in the midst of huge transfor- mations-involving optical-scanning, computing, and digital communica- tion technologies that were largely unforeseen when these innovations were first introduced. In some cases, incremental applications of IT may ultimately restruc- ture a company so much that they redefine the very nature of the company's business in ways that could not have been foreseen at the time of installa- tion. For example, Bankers Trust automated its back-office functions so successfully in the early 1980s that it was able to handle many wholesale banking operations with a sophistication much greater than that of most of its competitors. As its wholesale volume grew, it sold its credit card opera- tions and its New York retail and upstate commercial banking operations. The core of its business became proprietary trading in bulk capital market activities, and Bankers Trust reoriented itself toward wholesale banking and trust operations. Within the wholesale markets it understood well, Bankers Trust could take higher risks with newer, more complex instruments, mak- ing possible much higher margins. In terms of efficiency, automation en- abled Bankers Trust to cut its staff by two-thirds, while tripling its volume. But more importantly, Bankers Trust had redefined its total business con- cept and its options for future growth. When they are successful, strategic systems offer the highest potential payoffs from IT. But, by definition, strategic investments in IT systems are infrequent and the results are difficult to forecast precisely because no one has developed or exploited such systems before. Classics among systems representing strategic investments are ECONOMOST and its successor sys- tems at McKesson, Cosmos II at Federal Express, automated teller systems at Citicorp, and AANET at Arthur Andersen & Co. Other examples include Bank of America's COIN, Levi Strauss's Levi-Link, CHIPS (Clearinghouse for Interbank Payments system), and the Depository Trust. All of these systems enabled their sponsoring companies to offer new services to clients, but it was their capacity to restructure the business in its industry that qualified the systems as strategic. In some cases, investments in IT that the sponsoring companies hoped would have strategic impact failed or yielded disappointing returns: these include videotext, the New York Times Refer- ence Service, Federal Express's Zap-mail, and AMR-AMRIS's CONFIRM system. Systems in the latter category show up both in higher cumulative IT investment figures (until written off) and in decreased profits at the macroeconomic level. Perhaps the best known example of an IT application with strategic im- pact is the SABRE system of American Airlines. In addition to providing efficient and accurate flight reservations, SABRE enables the company to

2 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY handle hundreds of thousands of pricing and routing changes each day to optimize customer value and its own costs. The airline can hold Friday after- noon seats open for business travelers who are willing to pay a higher price than leisure passengers will pay. While offering discount fares early in the year for the Christmas season, the airline can ensure that seats are available for passengers who wait until later in the year, thus balancing and increasing profits over the long haul. It can also control oversales on flights and the customer unhappiness that results. By using a post-departure checkout proce- dure, American can also sell at later stops the seats opened up by "no-shows." And it can better control subsidiary services, such as special meals and handi- capped support, as well as flight operations. American believes its greater capability in these respects and its capacity to finely segment its customer base because of the detailed information it has available have enabled it to grow steadily during the post-deregulation shakeout when many other airlines have lost control of their costs. Many credit SABRE for converting American from one of the weaker airlines in the late 1970s to one of the strongest airlines today (although, like other airlines, it faces financial challenges). The strategic use of IT to change a company's positioning in the mar- ketplace and the difficulty of measuring that impact is illustrated in tele- communications by Richard Liebhaber of MCI, who said, We deal with 20 million customers every month. It's the only business in the world where companies invoice every minute, every second, every line item. And people send us letters about details in their bills. Wouldn't it be powerful, as a service provision, if we could convert all our worksta- tions in customer service to image capability and provide our representa- tives with the invoice and any letters about it in image form? When the customer call comes, we could answer questions immediately and better, but there would be no measurable productivity increase there.... It is an opportunity to improve the image, the feel, the touch of MCI. Intuitively, I would tell you it's going to provide market share, customer satisfaction, bottom line. I can't prove it. I'm not going to waste my time trying. The strategic use of IT may also enable industries to develop additional revenue streams of considerable value. One example is the sale of mailing lists created and maintained by IT systems among mail-order retail houses; in some instances, the revenue from such sales is an important fraction of the revenue generated by actual sales. A second example is the revenue stream that goes to owners of major computerized reservation systems (CRSs); airlines unable to afford their own CRSs must buy services provided by existing CRS owners. The revenue from such sales is substantial. IT may also help to create whole new lines of business for an entire industry, as mortgage-backed securities, swaps, and derivative securities have in investment banking. As Daniel Schutzer, vice president at Citibank, noted:

IMPACTS OF INFORMATION TECHNOLOGYAT THE ENTERPRISE LEVEL 113 The quality of the total service we deliver and the fact that these are whole businesses now handling trillions of dollars and making hundreds of mil- lions in profit leave no doubt in my mind that there has been productivity. But if we're talking about productivity in terms of the number of people that run branches and how much profit we're making on our old lines, we probably haven't really decreased the number of people; for years, we were building many more branches. The use of IT has facilitated the formation of many new industries or markets, as the examples in Table 3.1 indicate. The list there is incomplete, and the data indicate only approximate scales. The full impact of creating these new industries often will not be reflected as benefits for the originat- ing firmest or to the original service industries they might supplant. There can often be substantial delays before macroeconomic data are restructured to reflect such new industries. Although other forces helped to create these new markets and industries, most would have been essentially impossible to operate at their existing scales and efficiencies without IT. How much they substituted for other businesses is unknown. Even the incomplete list in Table 3.1 suggests that new service benefits to customers substantially out- weigh aggregate investments in IT. One of the most powerful strategic applications of IT is the large-scale capture, segmentation, and integration of data about both markets and op- erations (Box 3.21. Use of the resulting data and information has helped revolutionize the ways in which business (and even government) is con- ducted.7 For example, · Retail chains, by keeping track of product features and sales trends in detail, can help innovators design" products most suited to the market. By using these data to constantly update operations, producers can access mar- kets more quickly, shorten cycle times, and decrease their risks from inno- vation. Customers are presented with the products they most want. And shortening production-to-market times (as Levi Strauss's Levi-Link does) reduces inventory risks for both producers and distributors. · Researchers in companies routinely use electronic networking to communicate internationally and with more centers of expertise than in the past, bringing together the best minds in an entire company (and the outside world) to bear on a problem. Professional service firms can quickly access their expertise anywhere in the world. The potentials for synergy and value are well recognized in research and consulting networks such as Internet or AANET. Arthur Andersen & Co. has developed a network, AANET, that helps link over 60,000 people in some 300 offices in 70 countries. The company's cumulative experience is growing so rapidly that executives say, "Even those in the know may not have the best answer to the totality of a complex

14 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY TABLE 3.1 Scalar Estimates of New Markets and Industries Largely Facilitated by IT Sales Volume or Revenues Industry Typea (billions of dollars)b Year Source Catalog retailing Computer software and services Fast-food restaurants Wholesale clubs 47.2 79.8 75.6 33.0 Discount merchandise retailers 151.5 Computerized reservation 1.8 services (worldwide) Public information resources 11.0 (North America) Cellular telephones Home video sales and rental Basic cable television (subscription revenue) Satellite services ,.2 12.2 12.6 1.5 Video games (worldwide) 7.0 Computer animation (worldwide) > 5.0 1992 Catalog Age, March 1993 1989 The 5th Annual Computer  Industry Almanac, 1992 1992 Frozen Food Digest, March 1993 1992 Financial Times, May 21, 1993 1989 Discount Store News, July 2, 1990 1989 PC Week, August 7, 1989 1991 Information Week, August 26, 1991 1992 U.S. Industrial Outlook, 1993 1992 U.S. Industrial Outlook, 1993 1992 U.S. Industrial Outlook, 1993 1992 U.S. Industrial Outlook, 1993 1992 Forbes, January 18, 1993 1994 Computer Graphics World, March 9, 1991 Assets Managed Credit card receivables 206.0 1990 Faulkner & Gray, 1992 Swaps (worldwide 300.0 1991 The Bankers Magazine, outstanding balance) May/June 1993 Mutual funds (worldwide) 161.0 1992 Business Week, January 18, 1993 Credit card charge volume 464.0 1991 Faulkner & Gray, 1993 "Securitized" home mortgages 100.0 1991 Business Week, July 20, 1992 Asset-backed securities 50.9 public 1992 The Bankers Magazine, 8.0 private May/June 1993 Transaction Volume New York Federal 150,000 over Fedwire transactions/minute 1990 Computerworld, April 22, 1991 Automated teller 600 million machines transactions/month 1991 American Banker, December 7, 1992

IMPACTS OF INFORMATION TECHNOLOGYAT THE ENTERPRISE LEVEL 115 TABLE 3.1 Continued Industry Typea Transaction Volume (continued) CHIPSC Foreign exchange market Treasury bond market Depository trust (outstanding commercial paper issues) aU.S. only unless otherwise indicated. ball figures are approximations only. CClearinghouse for Interbank Payments system. Sales Volume or Revenues (billions of dollars)b Year $250 trillion 1990 > $600 billion/day 1990 $100 billion/day 1990 Source Interviews with Clearinghouse for Interbank Payments The Economist, July 21, 1990 The Economist, July 21, 1990 $555 billion 1992 American Banker, September 28, 1992 question." AANET attempts to capture and leverage knowledge wherever it exists at Arthur Andersen. Partners say that the company's distinctive com- petency has become "empowering people to deliver better-quality technol- ogy-based solutions to clients in a shorter time." Auditors and consultants who find unique solutions to problems introduce the solutions to the system through specially indexed subject files maintained in the Chicago headquar- ters. Field professionals who encounter a new problem can query everyone on the system for solutions through an electronic bulletin board. On its  audit reference and resource disk (a CD-ROM system connected to desktop microcomputers), Arthur Andersen & Co. collects and distributes as much up-to-date data as it can on taxes, customers, solutions to special problems, Financial Accounting Standards Board pronouncements, court rulings, and professional standards to guide its auditors and consultants in the field. Field personnel can both tap into the headquarters database directly through AANET and carry onto a customer's premises a CD-ROM with the power of many firms' central libraries. The capacity of many companies to develop a knowledge-based com- petitive advantage depends largely on the level of detail to which they can break down their contact point data and then analyze, mix, and match these data in different ways (see Box 3.21. The company with the largest and most detailed information base can segment and target its marketing and customer service activities with a precision and unit cost that smaller com

6 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY BOX 3.2 Comprehensive and Company-wide Databases Properly designed systems far comprehensive dam capture coffer companies one of the few and most important economies of scale in the current marketplace. just as the utility of a telephone network increases dramat'- cally with the number of telephone subscribers connected to the net- work, so also do comprehensive data sets enable highly sophisticated business operations. Computerized reservation systems provided import tent competitive advantages for individual companies, but they had major structural impacts on the entire industry when they began to carry flight and Are information far carriers ether than their owners. These more comprehensive data sets enabled much tighter integration of services provided by different companies (e.g., feeder airlines and national carri- ers, auxiliary services such as car rentals and hotels) and facilitated "one- stop" shopping, thereby making trip planning much less onerous for indi- viduals or travel agents. In the telecommunications industry, detailed per-call records of telephone traffic have enabled long-distanre compa- nies to develop a wide variety of call pricing plans to suit different calling patterns. In the retail industry, detailed profiles of customer preferences have increased the efficiency of marketing by enabling focused and tar- '~geted sales campaigns. The databases of Dow Jones' Merrill Lynch, Read- ers Digest, The New York Times, and many other service companies are unique competitive resources individually worth billions of dollars. A firm that has taken pa~icuiar advantage of comprehensive data- bases is Toys `'R" {Js, which enjoys $5.5 billion in annual sales and three times the revenues of the world's largest toy suppliers. With its com- mand of information about the marketplace and capacity to control the positioning, display, retail advertising, and pricing of toys, Toys "R" Us can sell products virtually year around while its competitors can afford to carry significant inventories only doffing the short pre-Christmas season, when over 60 percent of all toys are sold at retail. Toys `'R" Us, with its high volumes and soph~sbeated electronic irlYentory and point-of-sale sync tems, has a powerful mechanism for pretesting which toys are likely to sell-- and in what for~when the big Christmas sale season comes In addition, with its soph~st'tated models of toy sales, toy production, and changing Ashes, it can invest in its inventories with considerably less risk than other companies. By insisting on '~just-;n-time" deliveries, it can further decrease ~nark~dr~wn losses, which are as much as its percent of the toy industry~s cost for other retailers. To create such comprehensive databases' it is virtually essential to capture ~nforn~ation at the point of its creation. For example, point~c, sale systems the capture price and item information in real time are a critical element of the foundation for the use of IT in the retailing indus- try. Modern telecommur~ications stems generate large amounts of transaction information with each call (e.g., calling number, called number, time and

IMPACTS OF INFORMATION TECHNOLOGYAT THE ENTERPRISE LEVEL 117 length of calI), thus enabling a host of new services berg., Caller ID and Friends and Family, as well as very detailed billing information). Airline companies capture transactions at the source through their computerized reservations systems, enabling downstream improvements in scheduling, food preparation, maintenance, and so on. Securities companies must capture each element of a transaction instantaneously and be able to display both that transaction and market averages continuously. The most valuable resources of a brokerage firm are often its electronic database, data-tracking, and computer-tracking programs without which it could not compete. Customer expectations about the professional cap- ture and use of this Information - and legal liab~lides for its nor`-use have grown apace. Point-of-creation data capture, coupled with a higher degree of auto- mat~on in the actual mechanics of capturing data, reduces both the time required and the error rate for data entry; bar-code scanners make many fewer errors than human cashiers. In addition, capturing data at the source means that data need not be repeatedly entered (further el~minat- ~ng opportunities for error), and the avaitabR~ty of data at the beginning of the data-pracessing cycle means that the benefits of automation can be captured at any point downstream. Transactions with customers are the originating point for all work processes w~tl,~n an industry and the key to providing new and better customer services. petitors cannot match. Data systems have thus become one of the most important economies of scale for service enterprises. This fact has strategic significance in systems design. Companies that do not consider the future strategic implications of comprehensive data capture at the time they develop their systems-or that attempt to save up- front costs during early systems design and installation often lose impor- tant data and associated opportunities, as well as incur higher costs when they try later to exploit more refined breakdowns of data. Many companies have discovered that it is much easier to aggregate overly refined data later rather than to try to introduce disaggregations to a structure not built for them. A good example is provided by MCI, whose Friends and Family program is based on flexibilities designed into its IT systems from the beginning. As Richard Liebhaber of MCI said, How we do this is probably the most important trade secret in our compa- ny. We didn't think about Friends and Family specifically at the time we designed our IT system. What we did was focus on what we had to be able to control from an information base viewpoint because we are an infor- mation service business and designed in extra flexibilities at that time. Making quantitative judgments about the impact of investments in IT for strategic purposes, as for infrastructure, presents unique problems. Stra

8 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY tegic IT systems are not necessarily less intensive in either labor or capital than existing operations; indeed, they may well require more of both. Pre- cise preproject calculations of expected return are difficult because ultimate consequences are often unknown. And postproject calculations require massive assumptions about what a company's cash flows would have been with and without the chain of investments involved. For example, as noted by Liebhaber, To analyze payoffs of Friends and Family we would have to introduce assumptions both about what our share would have been without [the prod- uct] and about how much longer you [the customer] will be a member of the MCI family as a result of this product. In some cases, potential payoffs from success may be so obviously high as not to warrant detailed evaluation. As Robert Elmore, partner and worldwide director of the Business Systems Consulting Group at Arthur Andersen, said, It is often not necessary to take a detailed look at paybacks on certain strategic investments. Strategic use of information is often dealing with a 100-to-1 return or at least a 10-to-1 return, and in these cases returns should be obvious. For example, our electronic bulletin board was installed about 4 years ago, and there is no question among our practitioners that it gives us a competitive advantage and that it has made a large direct contribution to sales growth. We expect our line of business to triple in size in the next few years. We couldn't even think about being that large if we did not have this capacity to leverage our intellectual capital across the entire division. In other instances, broad gauges such as changes in market share, total profitability, growth rates, overall sales, and returns on investment may be relevant to judging strategic impact. When a major strategic change results from a series of incremental decisions (as in the case of Bankers Trust), it may be possible to justify each incremental step by using traditional measures. However, when the aggregate impact is considered post hoc, these detailed numerical analyses may have little meaning because of failures in some projects and unexpected synergies among others. In some cases, the risk posed by a strategic project may be unassessable by strict financial analyses. The only reason some projects succeed is that an individual entrepreneur evaluates risks differently and actually enters the field to test the assumptions underlying a project. Turner Broadcasting Company and Federal Express's overnight pack- age service are success stories in this category. Decision Making About Investing in IT and the Rigor of Program Evaluation Given the multitude of reasons for investing in IT and the different ways in which firms use IT, processes for decision making about IT of course vary from firm to firm. Nevertheless, two common characteristics

IMPACTS OF INFORMATION TECHNOLOGYAT THE ENTERPRISE LEVEL 119 stood out in the committee's sample group: (1) the similarity between the practices these companies use for decision making about IT and those used for other types of corporate technology investments and (2) the difficulty of separating IT's impacts from those of other factors affecting the success or failure of a project. The committee's sample of firms tended to use decision-making proce- dures for IT investments that were (1) comparable to those they use for other projects and (2) similar to those other companies use when investing in high-technology projects. A large majority of the companies interviewed by the committee (see Question Box 2 in Appendix D) reported that they routinely undertook preproject evaluations of investments in IT program by program. Depending on the nature and purpose of a given project (IT or not), preproject assessments varied greatly in the degree of quantification used. Few companies felt they had effective systems for evaluating the aggregate paybacks from all IT programs. A vice president of a major airline reflected the position that many respondents expressed. He said, The process that we've followed for the last 10 years for all of the projects that have been large investments is to put them through a very rigid justification process. The focus is on ROI justification. There are some strategic implementations or implementations which you can't justify, no matter how hard you might try, in terms of dollars and cents. But there must be the implication that implementation will be an advantage for the company from a strategic viewpoint. It's basically a system-by-system process that is very similar to our capital allocation process. As a matter of fact, all the resources that are used at the IT end of the business are allocated through a process that very similarly matches our capital alloca- tion process within the company. In certain cases in which the desired result is relatively easy to specify (e.g., for cost-reducing systems and certain new-product projects), compa- nies tend to use the same rigorous ROI and financial evaluation techniques for IT investments that they would use for non-IT investments devoted to the same ends. But for many systems, data forecasts are less precise, and calculations of ROI are often not as useful as other metrics. For example, because there are no positive returns from mandated systems, companies often merely compare various IT systems to see which will give the desired effect at the lowest long-term cost. Because it is frequently impossible to measure the impacts of improvements in quality precisely in terms of sales or profit, firms often use engineering metrics like response times, error rates, or service availability times when evaluating investments in systems intended to improve quality. Similarly, the long-term effects of strategic flexibility, more refined data-handling capabilities, or more detail in data- bases may defy financial quantification. Hence strategic investments may be evaluated in terms of their probable effects on market share, customer

120 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY satisfaction, capacity to attract or empower personnel, or potential for fu- ture flexibility, as well as their capacity to accomplish a specific current . . . mission or mayor goa .. Major effects of strategic investments may not be felt for a long time, and the timing and impact of specific events, if they occur at all, are also unpredictable. The entire cost structure and cost competitiveness of a firm within its industry may be affected, also unpredictably. In some cases, so many intangibles are involved that management's judgment may be a better indicator than detailed metrics. Although many interviewed companies noted that it is often not possible to develop accurate quantitative or financial metrics to measure the potential or actual benefits of some strategic invest- ments in IT, they tended to treat such decisions within the same framework used for other complex advanced-technology ventures. A good analogy can be found in research and development (R&D). Both R&D and IT are often intended to provide new options for the invest- ing company. In many cases, investments in R&D and IT are uncertain with respect to payoff. Payoff is usually not immediate, and the expected value is often not quantifiable or even estimable, let alone predictable. Executives try their best to forecast expected benefits, but success is often driven more by vision and superb implementation than by meeting accurate financial analyses. As with R&D, some IT projects succeed, and some do not. It is not surprising that the experience of firms in implementing new IT systems varies from extraordinary success to utter failure, a fact illus- trated by many studies at the enterprise level. Sometimes, the analogy goes much deeper implementing a new IT application may in fact be an R&D venture. If IT has never been used in a particular application before, managers may not understand whether, how, or how well the system will function in practice. There can be large uncer- tainties in the amount of time it takes to develop the application, apply IT to it, and develop specific software for it. Such efforts can be very expensive. Often a functionally complete IT infrastructure must be put into place be- fore the real pros and cons of possible applications can be reasonably tested. If the system is installed in an ongoing operation, there are additional risks of interrupting the business itself. For example, the New York Stock Ex- change had to install its automated trading system successfully without any interruption of operations; otherwise losses could have been catastrophic. Similarly, in the mid-1980s, Kmart invested in an expensive relational data- base system to store scanner data (hundreds of gigabytes) from retailers with only the belief (not the assurance) that once such data were made available, retailers would be able to exploit that data to their advantage. A second key characteristic of decision making for investing in IT in any complex situation is that it is virtually impossible to separate IT's contributions to success or failure from the effects of other factors such as

IMPACTS OF INFORMATION TECHNOLOGY AT THE ENTERPRISE LEVEL 121 planning, training, management, or other supporting investments or from other important changes that may be happening in the marketplace and business environment at that time and that may make it difficult to sort out net effects, even post hoc. Against a backdrop of extensive external change, it is often impossible to discern the financial impact of entire programs, let alone the contribution of IT to those programs. Like a milling machine or a grinder, IT is only one component of a system that enables a company to produce and deliver a good or service effectively; actual production usually requires many other components, in- cluding management, facilities, training, support, organizational factors, systems, equipment, and even materials. For example, when incremental increases in sales or levels of service require an incremental buildup of a firm's physical plant or capacity, the cost of IT may be just another component in the aggregate investment needed to bring the new business opportunity into being (e.g., the IT accompanying an extension of operations into a new geographical area). At best, it may be possible for a company to impute to the IT investment a prorated share of a project's total return based on its relative cost as one factor. In many cases, executive intuition and judgment may be as satisfactory as more complex numerical allocations of cost and benefits. In a strict sense, the proper way to evaluate the impact of IT is to measure the effects of a total system and then to analyze IT's direct contri- bution to them on a factor contribution basis.8 However, due to the inabil- ity to separate the influence of IT from that of other important factors at the enterprise level, such an approach to evaluation is rarely feasible, and no company interviewed by the committee had used it. However, this is also true for evaluating investments in fixed equipment in manufacturing indus- tries and other, non-IT business investments in most enterprises. Although a majority of those interviewed by the committee had tried to evaluate their firm's overall payoff from IT, none felt that they had satisfactory financial metrics for this purpose.9 To supplement or take the place of unsatisfactory financial metrics, many companies reported using engineering metrics to determine important impacts of IT. Engineering metrics often reflect more detailed aspects of an individual firm's work process (e.g., the number of individual transactions processed per employee in the banking industry, the number of claims processed per em- ployee in the insurance industry, the time taken per customer service request in the telecommunications industry, the time before the arrival of desired merchandise in the retail trade industry; Box 3.3 gives additional examples).~° Engineering metrics are very useful for diagnostic purposes and to compare the performance of one company to that of another (benchmarking). For those types of investments in IT for which financial measures are ineffective (par- ticularly for improvements in quality), managers rely heavily on engineering

22 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY BOX 3~3 Examples of Engineering Metrics AirI;nes Fuel cost per passenger-mile Load factors (average percentage of occupied seating on scheduled flights) · Response time for customer inquiries on reservations lines Telephone Companies Average repair time far service calls Number of calls processed per operator-Jay Cost per mile to install cable Sources and destinations Connection delays Number and duration of interruptions Frequency of disconnects Signal variation over different intervals, signal strength and other measures of signal quality · System `'down" times metrics to plan and oversee performance. Often such metrics cannot be re- lated directly to bottom-line measures of revenue or profitability. The fact that no company interviewed by the committee had been able to do so sup- ports the contention that macroeconomic data based on financial reports can- not accurately reflect changes in quality. Data from engineering metrics on the true impacts of IT on many aspects of quality are available only at the enterprise level and are not captured at any higher level of aggregation. CROSS-CUTTING OBSERVATIONS REGARDING ALL USES OF INFORMATION TECHNOLOGY Despite the wide variety of IT uses described above, several observa- tions seem to apply to all of them. Controlling the Costs of IT IT has become a significant element in the cost structure of many com- panies. From 1981 to 1989, absolute annual expenditures on IT in constant dollars grew by 86 percent nationally across the entire service sectored More importantly, this figure understates by a substantial amount the total cost of IT, since it reflects only expenditures on hardware mainframe computers, PCs, terminals, data communications networks, and the like. At least as

IMPACTS OF INFORMATION TECHNOLOGY AT THE ENTERPRISE LEVEL 123 important for an organization that ' wishes to make effective use of IT are several other essential elements that seriously affect profits from IT's use: · Software, whether custom-made, off-the-shelf, or user-developed. Initial investments can vary from very costly domain-specific software to inexpensive "shrink wrapped" diskettes and user-developed spreadsheets, databases, and so on. · Training and support. Users often require extensive training and support if they are to use a technology effectively. Even then they are unlikely to achieve the ability to use a system's full capability before the next generation of technology and customer demands require further changes. · Maintenance and upgrading. Hardware and software in place must be constantly maintained and often upgraded or replaced long before they have reached the use life calculated at the time they were acquired. The latter two categories often represent over 60 percent of the ongoing cost for a sophisticated system. Despite the importance of the elements listed above, the accounting rules that determine a company's book value do not treat hardware, soft- ware, and IT service (e.g., maintenance, training, support) equally. IT hard- ware costs (and initial software costs, if demanded by IRS) are usually carried on a company's books as depreciable assets. The costs for equally important software support, training, and IT service are generally treated as expenses (see Appendix DJ. Since overall IT costs are typically divided about equally among hardware, software, and service, a company's true investments in IT are rarely reflected in statements of book value. Even for elements of IT captured on a balance sheet, the capabilities of IT improve so rapidly that any given IT investment may be obsolete before a company can obtain the full benefits for which the technology was purchased. Various studies have suggested the effects of over~nvestment, duplication, and dissipation of profits that may occur. Repeated replacement poses problems for all companies. But the committee's interviews confirmed studies indicat- ing that replacement costs are a special issue in professional services, such as accounting, consulting, medicine, scientific research, architecture, engineering design, advertising, entertainment, or the law. These concerns were well captured by Marshall Carter, currently chairman and chief executive officer of State Street Bank and Trust Company, who said: The investment side of this poses problems. We have traditionally invest ed in technology on a 3- to 7-year cycle. But now the shelf life of a lot of things is 18 to 36 months. If you're not putting in tens of millions per year, you'll find you're falling behind the competition. Yet you must constantly invest even more in your capital base before you can get a complete payoff. For example, we're now installing a global treasury system. Within 18 to 36 months, there'll be a new version or modules that we'll have to rebuy.

124 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY Once a significant IT infrastructure is acquired, costs for hardware, software support, and maintenance become more fixed than variable in na- ture. These fixed IT costs become especially burdensome when revenues are squeezed by recessions or competitive pressures. A special problem exists because IT systems are purchased in "lumps." Because of the power of individual systems and long learning times, companies tend to purchase more capacity than warranted by average levels of demand. As a result, entire industries tend to overcapacity. In such situations where costs are mainly fixed, each producer can (up to the constraints on its capacity) gain total margin by selectively lowering prices to increase volume. The result is strong downward pressures on average prices (and thus on the percentage of margins) in services using IT,~3 even though the services provided may be much more complex and of higher absolute value to the customer. Daniel Schutzer at Citibank expressed one strategic response to this problem: Today we would like to move to an architecture which is less fixed-cost based and more variable. This would let us make sure our businesses remain profitable as they're growing and as they're shrinking, without being stuck with big fixed costs just at the wrong times. We see the technology moving more toward totally network-based systems where our divisions provide services on top of an intelligent network. All of these things dictate a more open-systems approach, but we do not know now what their specific impact will be. Enhancing Technological Sophistication and Developing Standards Service companies have often driven the state of the art in IT software, computing algorithms, hardware, and standard setting. For example: (1) Expert system software is widely used to capture and disseminate "best practices" from one skill source to others (e.g., for paramedics, diagnosti- cians, maintenance personnel, or brokers). (2) Software programs for tele- phone switching systems (million-line programs controlling real-time op- erations and billing records) routinely push the software state of the art. (3) Airline scheduling problems have generated more efficient linear program- ming algorithms that can solve problems involving tens of thousands of variables. (4) Financial services firms' demands for analysis of huge quan- tities of data have driven the design and purchase of massively parallel processors. The need to exchange data for applications extending across different organizations has driven the development of standards for electronic data in- terchange (EDI). Standards for document formats, communications protocols, and inter-application data transfer have enabled disparate computer systems running different software packages to communicate with each other across a

IMPACTS OF INFORMATION TECHNOLOGY AT THE ENTERPRISE LEVEL 125 variety of applications. Standards have been crucial to growth and progress in a number of industries, including travel agencies and airlines (standards for CRSs), retail-manufacturing (universal product code and EDI standards), fi- nancial services and banking (magnetic ink recognition and message security standards), and standards for asynchronous multimedia (data, voice, and video) communications. Despite such progress, meaningful standards for inter-appli- cations data transfer have yet to be developed for important specific applica- tions, such as standards for recording medical information, for monitoring outcomes of various procedures, or even for billings There are, of course, important possible drawbacks to standards notably a premature "freezing" of technology. For standards to be accepted, members of the relevant community must be willing to forego some unforeseen future opportunities and potential technological advances in favor of those created by enhanced current interoperability and electronic data interchange. When the former outweigh the latter, interested partners will resist standards, causing higher current costs but perhaps enhancing future capabilities as did the opponents of analog standards for high-definition television. Problems in Assessing Enterprise Performance For companies in certain industries such as air transport and telecom- munications, there are some readily quantifiable and widely accepted mea- sures of output. Ton-miles or passenger-miles handled in transportation, packages or documents processed in overnight delivery, kilowatt-hours or cubic feet of gas delivered by utilities, and the number of calls connected, minutes connected, or access lines provided by the telecommunications in- dustry are widely accepted examples. In industries for which measures of physical output are available (e.g., telecommunications, transportation, electric utilities, large fast-food and merchandise chains), levels of productivity (as measured, e.g., by revenues or value added per unit of labor input) have traditionally compared favor- ably to those in manufacturing. However, the correlation between growth in these measures and higher profits is not always good. Fare wars in the airline industry have destroyed profits even when revenue passenger-miles have risen, and the profits of individual companies within the industry have shown even greater variations. Even a uniformly positive correlation be- tween measures of physical output and profitability or revenues may be the result of other business practices or investments, or the result of a regula- tory environment that assures operating companies a "reasonable" rate of return. For many firms (e.g., those in financial services, health care, medicine, law, accounting, design, software, or government services), it is almost impossible to define a consistent measure of physical output that will reflect

126 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY variety, complexity, quality, or value produced (see Chapter 1). In some cases, gross statistics such as the number of clients or customers served or the number of audits or medical procedures performed may be available.iS But the qualitative nature of the output of many service companies makes its precise numerical measurement much more difficult than in goods-pro- ducing enterprises. Yet it is often along these qualitative dimensions that individual service firms compete with one another. In such cases, financial metrics such as gross revenues or profitability may be the only alternative, although even financial metrics have their limitations. A company may be a nonprofit enterprise (e.g., a nonprofit hospital). Revenues or profits may not be valid measures of output in enterprises where prices have been regulated (as in property and casualty insurance or telecommunications), where price increases are due to increased risks (as in insurance or obstetrics), or where revenues are determined by budget allocations (as in museums, foundations, or educational institutions).~7 Assessing the performance of individual firms involves several critical difficulties, a number of which affect the usability of measures at the macroeconomic or industry levels and have been discussed from that stand- point in previous chapters. Pass-Through of Benefits A variety of respondents reported that the rapid advances in IT and the high levels of competition in both the IT-supplying and IT-using industries made it difficult for even innovative companies to capture benefits from their investments in IT. With rare exceptions, IT hardware and software- and supporting communication services are available from many compet- ing vendors. Hence it is hard to create technological barriers to entry in the service sector other than those attributable to software developed internally. Even then, since a wide variety of software approaches can (with almost equal efficiency) solve almost any given problem, there are few protectable intellectual property rights that prevent replication of results. Fast followers can move with relative assuredness toward known re- sults and often at costs lower than those of the innovators because of ongoing technological advances. Technologies diffuse so rapidly that even the most innovative respondents said that, with few exceptions, IT gave them only a temporary competitive edge. When asked how long this edge might last, the usual answer was "a few months," which on further inquiry resolved into 3 to 9 months' time. Once new products or improvements in quality are copied, companies are forced to compete largely in terms of price. The same is true for each competitor, with the result that overall industry revenues may not fully reflect the benefits of an innovation- which are largely passed through to the customer instead. The innovation

IMPACTS OF INFORMATION TECHNOLOGYAT THE ENTERPRISE LEVEL 127 may proliferate so quickly that it soon becomes a routine and "commoditized" expectation of all customers. Unless the total market is growing substan- tially, overall industry revenues will not increase significantly despite mul- tiple investments dedicated to innovation. The tendency for profits from IT-based innovations to be short-lived is exacerbated by IT hardware trends. The computing power of IT has been growing so rapidly relative to unit cost (recently 35 percent per year for processors and 45 to 50 percent per year for memory) that the same dollar investment at a later time purchases substantially more power or capability. Competitors who invest later are able to buy more capacity per dollar than the original innovator, and competitors end up with either an equal capacity at less cost (reducing their fixed costs relative to those of the innovator) or a greater capacity for the same cost (resulting in a greater potential for business expansion relative to the innovator). The net result is that the ease of access to productivity-enhancing hardware (usually available to all com petitors) forces many service producers to pass benefits through to others and often lowers rather than raises their prices while increasing their break- even volume. Given that the benefits of IT-enabled innovations in services are not fully realized by the innovator, who does capture them? Three beneficiaries stand out: (1) the company's direct customers, (2) its suppliers, and (3) the general public. For example, wholesalers such as McKesson or SuperValu are linked to retail drugstores and grocery stores. As these wholesalers improved their service capabilities for customers, they actually lost percentage in margins. In the case of McKesson, currently a $7.8 billion firm, the percentage of wholesaling margins has dropped from about 7 percent to 3 percent over the last 15 years, making another 4 percentage points available for the margins of its retailers and producers. McKesson passed savings on to others by lowering prices and by providing higher levels of service to its suppliers and customers (e.g., McKesson's central maintenance of virtually all price stickers and price files for its drug retailers). As David Malmberg, vice president of McKesson's inventory management and systems services, said, "These used to be extremely time-consuming and expensive things for the retailer. They are done by our technology now. But this doesn't show up in our margins, and our competitors have to match the service. In effect, we've upped the investment ante for everyone." The experience of SuperValu, a $9.4 billion distributor to independent grocers, is comparable. SuperValu has developed extensive inventory and product-handling technologies that allow it to manage thousands of product lines and assist suppliers and retail customers with their distribution func- tions, including architectural design, construction, and space management. IT systems provide services upstream, downstream, and geographically. In

128 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY addition to controlling inventories in transit, SuperValu also offers exten- sive host-support services for the electronic point-of-sale systems of its supermarket customers, permitting them to manage their operations better through instant access to price files, shelf tag printing, and detailed reports about their products' turnover, gross profits, category profits, and so on. So effective is SuperValu at providing these services that several major super- market chains now buy from SuperValu in order to obtain the economies of its system. No single food product manufacturer could compete with these services. This is even truer for food producers for whom SuperValu handles the complex functions of coordinating field-produce contracting and trans- portation, fresh-meat delivery and marking, and in-store shelf-price verifi- cation. Yet because of competitive forces, SuperValu's total aggregate margins have not changed significantly. Firms outside the service sector that purchase services from an innova- tor e.g., manufacturers buying communication services from AT&T or MCI- are often major beneficiaries. Strategic alliances or major "outsourcing" arrangements- can present even greater potentials for pass-throughs. A large downstream partner may radically reduce its internal costs by cooper- ating with an efficient service provider whose own costs may not be re- duced by the partnership. In these cases, the service producer may show no gain in productivity, while service users reap substantial benefits. An ex- ample is the alliance between State Street Bank and Trust Company and Ford Motor Company. State Street has had a program with Ford for several years to connect Ford's U.S. treasury and trading operations with State Street's. In the words of State Street's Marshall Carter, "Ford feels pretty strongly that they've managed to increase quality yet reduce their staff substantially as a result of this alliance and its electronic connections. You wouldn't necessarily see a productivity gain here at State Street, but you would see a dramatic productivity gain at Ford." A third major category of beneficiary may be end users or the general public. When a transaction's marginal benefit to a customer is high and its cost drops precipitously, transaction volumes are likely to increase; this has been especially true in financial services, banking, and communications. For interstate toll calls, volume has increased since deregulation at a rate of 10.6 percent a year, while prices have decreased at a rate of 7.1 percent a year.~9 New York Stock Exchange transactions have risen from a daily average of 10 million to 12 million shares in the early 1970s to an average of 183 million shares in the third quarter of 1992.2° Both of these service industries were deregulated during the 1980s, eliminating the monopoly profits of existing producers. Customers benefited from lower costs and subsequently a greater selection of services. Often the volume of transac- tions increased so rapidly that companies could not reduce the number of total employees or costs despite much more efficient transaction-level auto

IMPACTS OF INFORMATION TECHNOLOGYAT THE ENTERPRISE LEVEL 129 mation. In other nonregulated situations, many professional companies in- terviewed by the committee reported that they could not increase prices to reflect the greater value added and the lower costs enabled by their IT systems because of overcapacity, competition, industry practice, or legal constraints on pricing. This phenomenon appears paramount in evaluating the impact of word processing and spreadsheets on report writing. The cost of making a change in a document is quite small; the result is that authors or analysts revise their manuscripts with much higher frequency (compared to pre-PC days). The customer or user of the report receives the benefit of a thoroughly massaged final product, but the total effort that went into the production of that draft may not have declined. A similar situation obtains in the use of automated teller machines (ATMs). The cost to customers of using ATMs is low; hence they engage in many more transactions than would be the case in the absence of ATMs. The ATM user thus benefits from the machine, but the expense incurred in deploying the ATM was at first generally borne by the bank. Later, banks sought to recoup some of these costs through usage fees. On the other hand, the use of IT sometimes allows companies to pass costs through to customers as well especially in the form of personal labor or waiting times. For example, IT enables telephone customers to dial long-distance telephone calls themselves and enter their own accounting information (credit card numbers) at pay telephones; bank customers per- form for themselves many of the functions of bank tellers at ATMs (e.g., checking account balances or processing deposit and withdrawal transac- tions). While customers may prefer to perform these activities for them- selves because doing so is more convenient or faster the actual labor required to provide a service has not been altered much by the introduction of IT. However, it is customers who are now doing much of the work. In a similar manner, IT has enabled airlines to manage more efficient "hub-and- spoke" routing systems, and retailers to provide "retail warehouse" sales of goods in bulk. But customers also bear some of the costs from these inno- vations, in the for of additional waiting times or a less friendly shopping atmosphere. Very few firms if any track costs borne by the customer in any systematic manner, and the overall effects on society are impossible to estimate. Time Lags and the Need for Higher Skills Timing and implementation factors also complicate measures of IT's impact on performance at the enterprise level.22 Various studies indicate that, as in other technology areas, there are long average time lags before highly innovative new IT systems may achieve expected payoffs.23 Work

130 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY ers' productivity constantly lags behind the systems' potentials as workers learn new tasks, and there are costs associated with confusion that arises as the systems are debugged. Sometimes, workers may barely reach full capa- bilities before the next generation of equipment requires changes to keep up with customer demands for quality and responsiveness. One implication is that if the benefits of a system are evaluated too early in its life cycle, they are likely to be underestimated. The other is that learning effects, training requirements, backup systems, extra supervision, and delays must be pro- vided for in any complete plan for introducing new IT systems. The time needed for workers to learn how to interact with a new system may be compounded by concurrent changes in ways outputs are produced. Some systems may facilitate or may require the use of more expensive or more highly trained knowledge workers in the place of less skilled workers. A now familiar example is the movement away from centralized typing pools, first to distributed word processing by secretaries and then to direct preparation of texts by their professional or managerial authors. An analy- sis of word processing's payoffs in the early 1980s would differ consider- ably from today's reality. If the complexity of a business problem also increases sufficiently, both the number of workers and their skill require- ments may go up24 (Box 3.4~. In other cases, the use of IT may simplify tasks, so that the required skill levels may drop, with the effect of deterring increases in wages.25 This pattern has been particularly evident in repetitive tasks such as product sorting, or in retail sales. A combination of lower skill requirements and wanes in some cases may encourage higher staffing levels than in the past, ,, ~e, _ ~ ~ .. _ _ and hence reduce measured productivity.26 On the other hand, the capacity BOX 3~4 Better Cancer Treatments Nay Cost Fore New cancer treatment procedures (e.~., bone marrow transplants) omen require inure ,~rec~se and expensive equipment, greater training, higher wages, more laboratory facilities, and more people to operate (per procedure) than the older chemotherapy treatn~ents dads At the same time, the new procedures nosy be significantly more eRective in terms clinically measured patient outcomes' including morbidity, modality, and fong-term care costs. However, while the costs of new procedures may well be incurred by treatment providers, none of these benefits may show up in their balance sheets or income statements. Actual outcomes may be uncertain for years, and prices established through d~agnostic-related-group com- pensation caps may prevent prices far medical services from rejecting the higher value added

IMPACTS OF INFORMATION TECHNOLOGYAT THE ENTERPRISE LEVEL 131 to use less skilled people may have the benefit of increasing employment opportunities. This appears to have been the pattern in low-wage services such as retailing or fast foods. Such complexities obviate the use of any simple measures of productivity at the enterprise level. Shifts in the Basic Nature of the Business As noted earlier in this chapter, the use of IT may change the very nature of a firm's business. In this case, comparisons of a firm's productiv- ity or performance in two different time periods may not be a true "apples- to-apples" comparison. It may even be the case that performance metrics designed for the old business lag in reflecting changes at the enterprise evel.27 For example, in many large professional service organizations, tasks that used to be the very core of professional practices (e.g., doing audit checks in public accounting or preparing bubble charts for portfolio analy- ses in financial services or consulting firms) have become so routine and automated that they are almost loss leaders, much of whose value is passed through in order to sell clients other services that are now the professionals' core competencies. In many cases, the basic business of professional firms has become the capture, analysis, communication, and leveraging of higher- value current information for clients. Such information-intensive tasks re- quire a high level of specialized intellectual expertise from sophisticated knowledge workers. Professional firms often obtain a competitive edge by leveraging their high-priced talent through automation of most routine tasks (e.g., literature searches, table look-ups, and routine or standardized design calculations) and development of very powerful analytical algorithms, ex- pert systems, and extensive networking capabilities. Investment companies provide a high-profile example. High-perfor- mance computing and electronic data systems have enabled development of complex, customized financial instruments that are difficult to create and support but that are generating totally new kinds of business. State Street Bank-managing assets of over $1 trillion and enjoying a 20 percent growth rate is typical of those working on this complex business frontier (specifi- cally portfolio and trust management). State Street's Carter noted: There are people out there all the time trying to create financial instruments that give a better yield on a portfolio. We are working on the edges of a system creating complex instruments that right at the beginning will be inefficient operations because of their new characteristics. Yet these are expanding the capital markets. The old instruments have become very sta- ble, efficient, and easy to serve. The average statistics may say you're not getting "productivity" increases. If you disaggregate it you'd see more productivity on the older (lower-margin) instruments than on the newer, higher-margin ones. But as a businessman, where do you want to be?

32 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY SUMMARY AND CONCLUSIONS Executives invest in IT programs with the intent of increasing expected returns as compared to other alternatives, for example, not investing or investing in other programs. The returns they derive from these decisions may be vital to the success of the enterprise but may not show up readily in standard measures of industry revenues or sector productivity. Information technology has enabled service firms not just to do things better in the old way (efficiency) but also to do entirely new things that provide greater benefits to outside parties (effectiveness). Many of these improvements in performance may not be reflected in increased ROI, sales, or margins, or in macroeconomic statistics. Thus, many of the difficulties in measuring the productivity and effectiveness of IT in services at the macroeconomic level have their origin at the enterprise level. Impacts of IT that may not show up in standard financial or data reports include maintaining market share in a relatively mature marketplace, avoid- ing catastrophic losses due to process failures, avoiding losses of market position or profits if a competitor adopts a new technology and the firm does not, increasing flexibility to respond to unknown future market or process changes, improving employee relations or the work environment, improving the quality of customer service, handling increasing complexity, and improving the scope and responsiveness of the firm's service outputs. These are key elements of service performance, although they may not affect productivity. In the committee's sample of companies, techniques used for decisions about investments in IT are comparable to those used for corporate deci- sions about investments in R&D or other complex technologies. As in R&D investments, none of the sampled companies could separate the ef- fects of a single input (R&D or IT) from other factors that might affect the ultimate success of a given project. Nor could they always estimate or calculate financial returns accurately. For this reason many firms use engi- neering metrics rather than financial measures to evaluate the potential and actual impacts of an investment in IT. But these metrics also do not permit aggregation at the industry or macroeconomic levels. Within enterprises, investments tend to fit into a few basic categories: those supporting basic infrastructure, mandated requirements, cost reduc- tion, new products, improvements in quality, new strategic potentials, and new desktop or workstation information exchange networks. In some of these cases (e.g., cost-reducing systems, systems to support new products), financial returns are relatively easy to evaluate through the use of well- understood formal investment analyses. In others (e.g., systems like those for basic infrastructure, improvements in quality, and new strategic uses), other metrics like engineering metrics or results of customer surveys

IMPACTS OF INFORMATION TECHNOLOGY AT THE ENTERPRISE LEVEL 133 may be used to estimate impacts. Even though managers may use the best techniques available, some investments in IT still are based to a large extent on the intuition or judgment of management. Some of the more important problems in measuring impacts of IT at the firm level are (1) defining meaningful units of output that are consistent over time, (2) defining the contributions of IT as a component in the total investment package needed to create a result, (3) capturing or measuring benefits passed through to customers or suppliers, (4) establishing financial measures of the impact of intangible benefits (such as faster cycle times, greater reliability, or a broader selection of service products), (5) estimating the "opportunity costs" of what would have happened without use of a technology, (6) dealing with payoffs that are delayed due to learning factors and with (sometimes extensive) lags until actual outcomes are known, and (7) measuring changes in productivity when IT has changed the basic nature of a business and the competitive environment. Granted all of these complications, a large majority (80 percent) of re- spondent companies felt that their investments in IT overall had paid off well. A majority said that they had made special studies of IT's payoff, although none had made routine measurements of overall IT returns. A large majority indicated that their special studies confirmed a positive overall payoff from IT (see Appendix D). Most expressed concern that their earlier management techniques had not been as efficient or effective as they should have been. Most were trying to improve these currently (see Chapter 5~. NOTES AND REFERENCES 1See, for example, Keen, P. 1988, Competing in Time: Using Telecommunications for Competitive Advantage, Ballinger Publishing Company, Cambridge, Mass.; U.S. Congress, Office of Technology Assessment, 1985, Automation of America's Offices 1985-2000, Govern- ment Printing Office, Washington, D.C.; and Watts, L., 1986, "What Price Automation?," Northeastern University Magazine, December. 2Denison, E. 1989. Estimates of Productivity Change by Industry: An Evaluation and an Alternative, Brookings Institution, Washington, D.C. Denison makes a strong case for the last explanation, though there is disagreement in the community on this point. 3Brynjolfsson, Erik, and Lorin Hilt. 1993. "Is Information Systems Spending Produc- tive? New Evidence and New Results," MIT Sloan School of Management, Working Paper 3571-93, September 24. (To appear in Proceedings of the 14th International Conference on Information Systems.) 4The work of Sassone suggests that downsizing secretarial pools may have the unin- tended result that professionals end up doing many more non-IT-based support tasks that were previously done by secretaries. Thus, on balance, the result may be lower measured productiv- ity. See Sassone, Peter. 1992. "Survey Finds Low Office Productivity Linked to Staffing Imbalances," National Productivity Review, Spring, pp. 147-158. SLevitt, Theodore. 1976. "Industrialization of Service," Harvard Business Review, Sep- tember-October, pp. 63-74. 6Brynjolfsson, Erik, and Bruce Bimber. 1991. "Information Technology and the Produc- tivity Paradox," Working Paper, Brookings Institution, Washington, D.C., February 7.

134 INFORMATION TECHNOLOGY IN THE SERVICE SOCIETY 7Porter, Michael E., and Victor E. Millar, 1985, "How Information Gives You Competi- tive Advantage," Harvard Business Review, July-August, pp. 149-160; Quinn, J.B., 1992, Intelligent Enterprise, Free Press, New York; and Wriston, Walter B., 1992, The Twilight of Sovereignty, Scribners, New York. 8Loveman, Gary, 1988, "An Assessment of the Productivity Impact of Information Tech- nologies," MIT Management in the 1990's Program, 88-054, July.; and Kendrick, John W., 1988, "Productivity in Services," in Technology in Services: Policies for Growth, Trade, and Employment, Bruce R. Guile and James Brian Quinn (eds.), National Academy Press, Wash- ington, D.C. 9The inadequacy of financial metrics alone for assessing the performance of a firm is discussed in Eccles, Robert, 1991, "The Performance Measurement Manifesto," Harvard Busi- ness Review, January-February, pp. 131 - 137. 10A Bellcore study showed that the number of employees per 1000 access lines one of the operating company's key performance metrics had dropped from 110 to 40 (or about a 10 percent per year compounded improvement) between 1970 and 1990. Roach, Stephen, Morgan Stanley & Co., unpublished data, April 1993. 12Fudenberg, Drew, and Jean Tirole. 1985. "Preemption and Rent Equalization in the Adoption of New Technology," Review of Economic Studies 52:383-401. 13Bresnahan, Timothy F. 1986. "Measuring the Spillovers from Technical Advance: Main- frame Computers in Financial Services," American Economic Review 76(4, September):742-755. 14Institute of Medicine. 1991. The Computer-based Patient Record, National Academy Press, Washington D.C., pp. 85-87. 15The meaning of aggregated statistics is also clouded by the fact that parameters such as relative unit volumes, revenues, costs, timing factors, number of clients, and so on may have little real meaning as indicators of productivity when ultimate outcomes are unknown. For a bank, the face value of loans or the number of loans handled is meaningless without knowing the quality of those loans or the creativity they represent. The number of procedures per- formed at a hospital compared to their cost is useless information, unless one knows their success rate and the relative health of patients at the time they were served. Even the success rate must be qualified by the subsequent quality of life or care costs patients encounter. 16Mark, Jerome, 1986, "Problems Encountered in Measuring Single and Multifactor Pro- ductivity," Monthly Labor Review, December, pp. 3-11; Baily, Martin Neill, and Robert J. Gordon, 1988, "The Productivity Slowdown: Measurement Issues and the Explosion of Com- puter Power," in Brookings Papers on Economic Activity, Vol. 2, William C. Brainard and George L. Perry (eds.), Brookings Institution, Washington, D.C., pp. 347-431; and Brynjolfsson and Bimber, 1991, "Information Technology and the Productivity Paradox." 17Kutscher, R., and J. Mark. 1983. "The Service Sector: Some Common Perceptions Reviewed," Monthly Labor Review, April, pp. 21 -24. 18Bresnahan, 1986, "Measuring the Spillovers from Technical Advance: Mainframe Computers in Financial Services." 19Federal Communications Commission. 1991. Trends in Telephone Service, FCC, Washington, D.C., February, p. 25. 20Standard & Poor's Corp. 1992. Current Statistics, October. 21Both the committee's interviews and independent studies testify to the ubiquity of this phenomenon. See, for example, Brynjolfsson and Bimber, 1991, "Information Technology and  the Productivity Paradox." 22Curley, Kathleen Foley, and Philip J. Pyburn. 1982 "'Intellectual' Technologies: The Key to Improving White-Collar Productivity," Sloan Management Review, Fall, pp. 31-39. 23Guile, Bruce R., and James Brian Quinn (eds.). 1988. Managing Innovation: Cases from the Service Industries, National Academy Press, Washington D.C. Also Brynjolfsson, Erik, Thomas W. Malone, and Vijay Gurbaxani. 1988. "The Impact of Information Technol

IMPACTS OF INFORMATION TECHNOLOGY AT THE ENTERPRISE LEVEL 135 ogy on Markets and Hierarchies," Sloan School of Management, Working Paper 2113-88, Massachusetts Institute of Technology, Cambridge, Mass. 240sterman, Paul. 1986. "The Impact of Computers on the Employment of Clerks and Managers," Industrial and Labor Relations Review 39:175-186. Also, Krueger, A. 1993. "How Computers Have Changed the Wage Structure: Evidence from Microdata, 1984-1989," The Quarterly Journal of Economics 39(February):175-186. 25Bright, J. 1958. Technology in Automation, Harvard Business School Press, Boston, Mass. This is the classic statement of this argument. 26Baily and Gordon, 1988, "The Productivity Slowdown: Measurement Issues and the Explosion of Computer Power." Also, Hunt, H., and T. Hunt. 1986. Clerical Employment and Technological Change, Upjohn Institute for Labor Studies, Kalamazoo, Mich. 27Stalk, G., and T. Hout. 1990. Competing Against Time: How Time-based Competition Is Reshaping Global Markets, Free Press, New York. )