It all started with Lester Thurow, then dean of MIT Sloan School of Management. In May 1986 he gave a speech at the American Association for the Advancement of Science, showing that while the number of blue-collar workers in the US was reduced by 6% between the 1978 and 1985, total output rose 15%. Fewer people produced more--a growth in productivity comparable to Europe and Japan during the same period. At the same time, however, the number of white-collar workers (including managers) grew 21%, swallowing the productivity growth in the blue-collar industry and making the United States less competitive (Thurow, 1986). This period of sagging productivity coincided with a period of high investments in office automation, technologies that intuitively should have increased, not decreased, productivity of white-collar workers.
Stephen S. Roach, an economist of Morgan Stanley, found through an elaborate analysis of labor statistics that productivity1 in the US service sector had not increased since the late 1960s (Roach, 1987). Martin Neil Baily, an economist at the Brookings Institution, showed that the investments in technology in the white collar industries has been immense, particularly in banks (Baily & Gordon, 1988, p. 390) where information technology has gone from representing 3.9% of capital during 1970-79 to 34.1% as of 1987. For insurance carriers the figures are 7.2% to 63.8%, respectively (see Exhibit 2). Much of this investment is in personal computers: The number of PCs used in Fortune 1000 companies rose from 2.5 million in 1983 to 18.4 million in 1985, and the percent of white-collar staff using them from 7% to 56% (Jeffery, 1991).
Paul A. Strassmann (1990, pp. 31-58) has shown that there is no correlation between financial return (as measured in profitability, sales per employee etc.) and the various measures used by the computer trade press for technological excellence (IT investments as a percentages of revenues, PCs per employee, etc.).
Books and articles like these, or rather the references to them in business journals, is causing difficult questions from CEOs to CIOs all over: Are they really getting any value out of having information systems--or should they maybe spend their money on something else instead--like advertising?2
Unfortunately for the CIO, there are vast differences between what economic analysis and overall statistics tell us, and what is found when we look at individual companies. The picture is far from clear, partly because the data is lacking and faulty for large parts of the economy, but also because the role of information technology is not well understood. In this note, I will try to bring some clarity to the picture by examining the role of IT in providing business value on two steps: First, I define a framework to discuss the role of IT in supporting different types of tasks. Within this framework, I try to show what effects use of IT can produce, and how these effects can provide business value. Second, I try to assess how successful the use of information technology been so far. Third, I try to give a strategy for the plagued CIO: how to respond to the challenges posed by the debate over the value of information technology in business.
PROCESSThe framework is imprecise in the sense that it is not readily apparent in which quadrant a task falls: most work contains elements of all these categories5, the distinction between physical and information product may be dubious6, and furthermore, the axes are continuous rather than dichotomic. Still, using this rather crude categorization enables us to focus on the effect of information technology and on singular types of tasks, a more precise viewpoint than looking at computers in management, the service industries or productivity statistics. I believe information technology can be used to improve productivity in the traditional sense only in some of these quadrants, either by increasing the capacity within the quadrant or by creatively moving tasks, or portions of tasks, from one quadrant to another. This implies that the kinds of systems that improve performance in one quadrant, as well as the measures used to gauge its success, might be useless or even counterproductive in other quadrants.
* * *
Physical * Manu- * Design and *
goods * facturing * diagnosis *
* * *
* 1 * 2 *
* 3 * 4 *
* Back office/ * Knowledge *
Information * office * worker *
* factory * *
* * *
In this quadrant, information technology can increase capacity by improving quality and quantity of products through faster, more reliable, more precise and more continuous operations. In- dustrial robots, for example, do not need coffee breaks, sanitary facilities and fringe benefits, and can be put to use in dangerous or dirty environments where humans prefer not to work. Information technology, being electronic, is less prone to wear than mechanical devices--there is no need to grease a microprocessor. The effects of IT translates into business value through lower cost and increasing flexibility in the production process.
Although the products in this quadrant are complex, there is little doubt about what they are. Quantity is readily observable, quality is measured by comparing to standards or to equivalent products from other sources. Although errors in design and diagnosis often are traceable, there may be considerable disagreement as to which inputs provide increases in quality. Although quality of output may be measurable, the value of increased quality may be difficult to measure: how much more secure should a car be, how much should be spent on "hopeless" injuries or sicknesses, etc. Even though the production process is unstructured, there are often standards for what constitutes a good process and what does not.
Tasks in the second quadrant are often taught through apprenticeship: since the process of, say, blowing glass or taking out appendixes is not as easily conveyed verbally as it is shown. The student of these arts learn by watching experienced performers and copying their techniques through a closely monitored process of trial and error. Apprenticeship time may be long, often because the product is costly or dangerous and the student cannot proceed to more difficult levels of performance before mastering the basic skills. The skill level necessary is high, often not only because of the dexterity necessary but because making decision on what to do is part of the task.
Information technology's main effect in this quadrant is to increase quality of the product by allowing experimentation, increasing information available for decision making, and shortening process time. A flight simulator's main advantage is that it allows trial and error without risk to life and property: through crashing many simulated planes the pilot-in-being learns how not to crash the real ones. And the experienced pilot may try out hair-raising new maneuvers without risking his or her pension. Information technology-based exploratory tools like computer tomography can guide a surgeon's knife in real time.
This translates into business value by increasing the demand for the product, because the product is of higher quality, or is designed or processed faster and more accurately than without the technology.
Increases in capacity in this quadrant is achieved by increasing the number of operations performed per employee - much like in the first quadrant. Information technology may support this process through moving the recording of information out of specialized back office departments to where the transaction takes place. POS (point of sale) systems, which registers all information about a sale right at the customer checkouts, is an example of this. "Business process redesign"--a fashionable approach to information systems analysis that uses IT heavily--aims at vastly improving the productivity in this quadrant (Hammer, 1990).
The business value should come through record-keeping tasks needing relatively fewer people than before, through allowing vast increases in transaction volume, and in providing aggregated information to decision-makers, most of whom reside in the fourth quadrant.
The effect of IT in this quadrant should be to increase the knowledge workers capacity to process and communicate information. This comes from two sources: automation what can be automated in the knowledge worker's job, freeing him or her to do what the computer cannot, and by increasing amount and quality of information available through technological media. Here we find the end users of DSSs and EISs, as well as personal computers and electronic communication services. The knowledge worker often designs the production process him- or herself, including the question of how and when to use information technology. This means that the frequency of use is often an important measure for the quality of the services IT provides.
The automated parts of a knowledge workers job, such as word processing and spreadsheet calculations, are a much smaller part of the time at work than the time spent simply communicating with people. Extensive use of electronic mail and conferencing systems thus should have a great productivity potential, especially if an organization is geographically dispersed. Strassmann (1990, pp. 357-359) argues that the largest potential is found in electronic communications, since 30-70% of an employee's time is spent in face-to-face communication, in addition to time spent in the telephone.
The business value provided should be products of higher quality and quantity: primarily better and faster decisions.
Transfer of tasks from the fourth to the third quadrant is the industrialization of clerical work. This shift increases productivity for two reasons: the tasks, when automated, can in many cases be performed by less trained (and presumably cheaper) personnel; and the tasks become more measurable, meaning that the effects of changes in procedure can be documented better. This process is similar to industrialization in many respects. Unstructured processes are automated by reducing the process to identifiable, replicable steps, and then redesigning the process so it takes advantage of the speed and reliability of information technology. This constitutes an explication of the knowledge formerly contained in the minds and bodies of the people performing the tasks (Zuboff, 1988). Explicating and standardizing tasks through specialization has been the traditional way of increasing productivity: the most visible effect being "electronic sweatshops" (Garson, 1988); data entry departments in large organizations, typically banks and insurance companies, where large numbers of employees with low education enter data into mainframes.
Transfer of tasks from the third to the fourth quadrant is seen in companies that use information technology to automate the routine aspects of clerical work, thereby orienting the job of the employee towards things that are less automatable, such as dealing with customers. Employees shift from specialization in one repetitive step of a complex task to handling all aspects of it. USAA corporation, a large insurance company based in San Antonio, Texas, has used this approach to advantage: They use imaging technology and a job distribution system to banish paper from its operations. This frees up insurance clerks to communicate with customers, and concentrate on the judgement and relationship aspects of cases, rather than spending most of their time keeping the files in order (Harvey & Ryan, 1991; Henkoff, 1991). A problem in this process is that the necessary skill level normally is higher in the fourth than in the third quadrant: although a task may move, the worker performing it may not.
Transfer of tasks from the fourth to the second quadrant: Although it may seem counterintuitive, information technology can be used to transfer, at least in a limited sense, tasks from being manipulation of information to being handcraft. Some of the research being done under the heading virtual reality asks people do things physically because the information is not readily explicable in any other way9. An example of this is an experimental system developed at the Tsukuba university in Japan, where a camera designer can put his hand into a "steel glove", which lets him "pick up" a camera from a screen. The "steel glove" not only lets him "lift" the image of the camera on the screen, but also lets him feel the balance of the camera -- a feature not easily modelled in a conventional CAD system.
Information technology helps transfer tasks from the second to the fourth quadrant by substituting an electronic representation for a physical product under design. Engineers, architects and designers who once worked with drawings and physical models can now create designs through CAD or mathematical software. Computer-based drawings and 3-D models not only automate the step of actually making drawings and models, but allow designs to be tested electronically. Cars are now safety-tested by computer simulation rather than actual crashes. Electronic designs are more readily communicable than physical drawings and models, and may more easily incorporate or- ganization-wide standards, such as using common piece parts wherever possible.
The picture seems to be that although the manufacturing sector's role in the US economy is decreasing, productivity of blue-collar workers has increased steadily (Drucker, 1991; Thurow, 1986), causing the industrial share of the workforce to drop to 25% of the US total in 1990 (Roach, 1988). The role of information technology in the increased productivity remains unclear, however: several studies (Berndt & Morrison, 1991; Loveman, 1988) indicate that investments in information technology have little or no correlation with increased productivity or profitability in US manufacturing industries. Loveman's study of 250 US manufac- turing business units did find a small, but significant correlation between IT investments and productivity increases three years hence, indicating that it may take time before the investments pay off.
The MIT Commission on Industrial Productivity study (Berger, et al., 1989) shows that US productivity has improved, but productivity in other countries has improved more. They propose that productivity improves not only by adding automation, but in how the automated features are taken advantage of. There is evidence to suggest that US companies are investing in automated factories, but not changing the way production is organized. In particular, flexible manufacturing equipment is not exploited: the machines are used for producing a few products in high volumes rather than taking advantage of the new machines' ability to produce many different products with a very low switching cost (Jaikumar, 1986).
One of CAD/CAM's key benefits is its ability to facilitate integration between the design, production and marketing functions of an industrial enterprise. The problem is often that it is relatively easy to have two of these functions on good terms with each other, but very difficult to have all three pull together--and more often than not, manufacturing gets low priority (Riggs, 1983). Proposed solutions for these problems have been many--most of them not including information technology explicitly, but instead focusing on project management (such as having dedicated projects with high-powered individuals as managers) (Clark & Fujimoto, 1991), using "overlapping" problem solving (Clark & Fujimoto, 1987), and designing for manufacturability (Dean & Susman, 1989).
The research available in this area indicate that there may be a long way to go before this potential is properly exploited. A study by Adler (1988) shows that companies have not found CAD/CAM to be a competitive advantage, mainly because they have failed to take organizational issues into consideration. Some success stories are available: Northrop, in the creation of the B2 Stealth Bomber, went directly from an electronic description of the system to manufacturing, skipping wind tunnel testing. Gleason Manufacturing uses CAD as a communication tool between design and manufacturing, directly feeding CAD drawings into manufacturing processes. And IBM, through its Proprinter project, rendered an automated factory largely obsolete because a better design reduced the number of parts to the point where it was faster assembled manually (Berger, et al., 1989).
Using the framework, we could conclude that productivity improvements within the manufacturing quadrant have been well-mined when it comes to increasing capacity within the quadrant. However, the external environment now values quality and flexibility in addition to low price. This should indicate that there is more productivity to be gained from transferring tasks from the first to the third and fourth quadrant: move information instead of physical product and exploit the specific knowledge gained by those close to the production process by giving them more decision power.
The increased quality offered by the new technology carries some problems, however. One is the unclear pricing of increased quality: The new technologies, including information technology, now offers the ability to spend enormous amounts of money on marginally prolonging the life or increasing the probability of survival of a patient. The second is the unclear implications of how increased patient care translates into business value: since hospitals profit from using expensive technology, at least as long as public or private insurance is willing to pay for it, the use of technology becomes a quality measurement in itself. The tools, rather than the results, becomes the criterion against which performance is measured.
The combination of available new tools, the use of tools as an indication of quality, and the impossibility of pricing human life, has meant that the minimum acceptable investment level has risen dramatically for hospitals. A "normal" hospital has now become a "labor-intensive and capital-intensive monstrosity", because in this kind of work, capital is not a substitute for labor (Drucker, 1991).
Within the framework, we may conclude that information technology has been successful in increasing quality, which again has increased demand. The problem seem to be that this increased demand has not translated into profitability in a way that shows up in productivity statistics. The exception may be cases where information technology has been able to move tasks to other quadrants: through industrialization or through electronic rather than physical design representations.
Since the effect of information technology, like in the second quadrant, seem to be more to make things possible than to reduce costs, productivity statistics will be misleading. They will be misleading because they measure the wrong things, and because advantage held by investors in information technology is soon competed away. The story of automated teller machines show this: The output per hour in commercial banks grew at only .9% per year from 1973 to 1979, compared to 2.3% per year from 1967 to 1973 (Baily, 1986). This apparent productivity decline is mainly due to a change in what banks do: instead of focusing on increasing the volume of checks processed and loans made (which constituted the basis for measurement by the BLS), banks tried to become more profitable by offering expanded and tailored services, such as ATM's. The investments in new services such as ATMs, and in the people who maintained and expanded them, looked like unnecessary overhead in the productivity statistics. In reality, ATMs have become a requirement for being in the consumer banking market.
The second important effect of information technology has been the ability to vastly expand scale and complexity of operations, since record-keeping is no longer a limit on operations. Without computers to coordinate and keep track of passengers, maintenance and scheduling, New York would probably have had to build one or two new airports to handle the present traffic (Penzias, 1991). Stock Exchanges around the world could not even be close to the transaction volumes they have today without computers. Financial services of the scale and complexity they are today could not have been available. These facts do not show up in productivity statistics, however: the reason being that we have a different world because of computers, but not necessarily a cheaper or more profitable one.
It has been argued that organizations, as a result of reductions in the lower ranks, no longer have the traditional pyramidal shape--instead they are diamond-shaped, with the bulk of the employees in middle management or staff positions (Attewell & Rule, 1984). In terms of the framework, productivity increases in the third quadrant have been eaten up by the fourth: the work on the third quadrant has shifted from doing the transactions to assisting the knowledge workers, mainly by input of more information.
How these effects of information technology translate into business value becomes a function of how well the information is used. May Department stores offers a good example: buying decisions are taken by top management, who knows by Monday morning how every product has sold compared to budget the week before (Pliner & Springer, 1990). This translates to business value because in retailing buying decisions are both extremely important and easy to centralize by computer.
There is little question that the ability for knowledge workers to process and communicate information has been vastly expanded because of information technology the last ten years. Portable telephones, fax, personal and portable computers have all meant that information can be sought faster and from a wider range of sources than before. There is considerable doubt, however, about to what extent this has contributed to business value. Strassmann (1990, pp. 117-120) argues that management is the real heavy user of information technology in most industries: "mission critical systems are few and inexpensive as compared with costly management information systems." While a number of successful case stories are available11, the overall picture is that of "staff infection" (Thurow, 1986). Instead of fulfil- ling the demand for information, computers instead create an ever larger demand for data of dubious incremental value.
There are indications that many investments in IT are not economically motivated. Some organizational theorists have argued that organizations make decisions based not on what is economically best, but what is considered good practice at the time (Meyer & Rowan, 1977). From this it can be argued that many IT investments are "me too"-investments, related more to fad, fashion and available funds than crisp economic analysis. One manufacturing firm traced its IT spending over a decade and found that it was closely correlated to what it called "spending appetite"--last years profitability.
Since the use of information technology in many cases is a personal decision for the knowledge worker, the impact on productivity often is, too. Shelby Foote, historian, novelist and noted commentator of the TV serial The Civil War, writes everything with an inkwell pen -- to ensure that every sentence is thought out before it gets on paper and that there is no temptation to change anything once it is down there12. On a smaller scale, this has been called the First Law of Word Processing: "Time saved in using a word processor is spent fiddling with layout and fonts." While word processing and desktop publishing have made many writers more prolific, there are no indication that the quality of what is written is higher. Similarly, since information technology has produced a tremendous increase in the amount of readily available information, more information will be considered--with uncertain implications for decision quality.
It may be argued that a lax attitude to spending and profitability of the technology is necessary in a period of technology introduction: since many of the new technologies cannot provide business value until they are adopted by many users, adoption should be encouraged. The rapid adoption of PCs, for one thing, has had an effect of lacking interoperability13: moving information between different kinds of computer systems has been difficult and laborious. The result is a situation where IT is locally efficient, but a lack of standards and planning results in "people running errands between machines" (Penzias, 1991).
Most executives are not terribly concerned about the productivity of their white-collar workers (Erdmann, 1991), a finding consistent with Lester Thurow's attribution of the sagging white-collar productivity to an incapacity in the American managerial culture. According to Thurow, there can be no productivity gains until American managers shed some of their entrenched attitudes:
These [attitudes] have to do with power (American bosses exist to boss); style (a good boss should know everything and, in principle, have the knowledge to make all decisions); institutions (most middle-level managers get paid based on the number of people who report to them); peer pressure (it is harder to fire those who directly work with you than those at a distance); and beliefs (if the system is based solely upon individual effort, there is no need for group motivation, voluntary cooperation, or teamwork). (Thurow, 1986)In terms of the framework: IT may provide a beneficiary effect, as illustrated by numerous cases of "successful systems". However, the translation into business value is dubious, because we have not learned how to do this, and because the main problems in doing it is organizational, not a feature of the technology.
The movement from the design to the knowledge worker quadrant is merely a deduction from the fact that a lot more is done on electronic representations of reality than what is done with physical representations of information. The movement from manufacturing to design/diagnosis is due to the quality movement: since products are getting more complex, more knowledge is required by the individual worker.
There are other movements, of course, automation still being a key reason for introducing information technology into tasks. Pure automation of existing tasks, however, is no longer a very attractive alternative for IT introduction: most of what could be automated now is. Although there are many ways to increase the efficiency of these automated solutions, I will hold that it is unlikely to move many tasks over from the fourth to the third quadrant.
If you don't know where you are going,Arno Penzias14
computers will take you there faster
and more expensively.
From what we have seen so far, what conclusions should a CIO draw, and what can he or she do about the situation? Possible measurement errors aside, the discrepancies between what IT has promised and what it has delivered so far seems substantial.
The best strategy would probably be to acknowledge that, yes, there is a discrepancy, but that there are good reasons for it. These reasons are:
|Short name||manufacturing||design||back office||knowledge worker|
|Productivity measures||quantity, quality of output||time in process, quality of design, successful final outcome||quantity of tasks performed, accuracy, time in process||quality of outcome|
|Necessary skill level||low to medium, fairly specialized||high, specialized||low to medium, fairly general||high, both specialized and general|
|Compensation level||low to medium||medium to high||low||medium to high|
|Measurability of output||high||low in the short run||medium||low|
|Measurability of activity||high||low to medium||high||medium to high|
|Size and growth (Based on Roach (1988) and Drucker (1991)).||substantial but declining||small, stable||large, stable after high growth||substantial, growing|
|Effect of IT||Increasing capacity and flexibility||Increasing quality of product||Increasing detail capacity and information overview||Increasing information availability and communication capacity|
|....Finance and insurance||3.5||4.7||38.6||234.73||90.51|
|Agriculture & extractive||50%||28%||14%||4%||2%|
|Manufacturing, commerce, industry||36||53||37||29||22|
|Information, knowledge, education||2||9||29||50||66|
On the other hand, when the users of the IBM 438L expert system used by the Strategic Air Command in the 1960s turned out not to be used, top managment solved this by ordering personnel on duty to ask at least two questions of the system on each shift. The number of questions asked consequently rose to twice the number of officers on duty (Earnest, 1990). In this case, frequency of use went up because of top management involvement, making the system seem more successful.