Michael E. D. Koenig
Graduate School of Library and Information Science
River Forest, IL 60305, USA
1. THE PRODUCTIVITY PARADOX
The productivity paradox is the observation, or more accurately the charge, that despite massive investments in information technology, there have been almost no productivity increases attributable to that investment.
The most compelling articulation of this hypothesis
is Stephen Roach's (1991) Harvard Business Review article, with
the innocuous title of "Services Under Siege -- The Restructuring Imperative"
but with the provocative subtitle of "Foreign competition and deregulation
are exposing the U.S. service sector's hidden vulnerabilities -- over investment
in information technology and chronic inefficiency." A more popular treatment
of the same theme is Gary Loveman's (1991) near simultaneous article in
Computerworld entitled "Cash Drain, No Gain" with the subtitle "U.S.
Companies have spent billions on powerful technology to do things faster
and better. Why, then, it is asked, have productivity gains been so puny?"
2. BUSINESS PROCESS REDESIGN
Business Process Redesign is the argument that dramatic improvements in business effective-ness can be achieved by redesigning business processes, taking advantage of the capabilities of information technology. The whole business process it is argued, or almost any significant compo-nent thereof, is likely to be a candidate for thorough overhaul (Davenport & Short, 1990; Keen, 1991; Knorr, 1991; Short & Venkatramen, 1992; Davenport, 1993).
3. THE APPARENT CONTRADICTION
These two hypotheses appear to be quite contradictory, one argues that we have over-invested in information technology, and that little or nothing has improved, and the other argues that great benefits can result from the judicious use of information technology.
How can we reconcile these views. In fact, these two themes are linked in a very significant and only imperfectly realized fashion and that linkage is ultimately a reconciliation. That linkage, that relationship, can best be examined by looking at the responses to and the attempts to explain the productivity paradox.
The most obvious response is simply that there are many intangible or at least functionally intangible benefits that are not measured by productivity calculations. What is the utility of higher definition television, or a vastly increased number of channels?
The second response is that there are many benefits that could, at least in principle, be mea-sured but are not. A good example of this is the productivity of the pharmaceutical industry. Suppose there were no Chemical Abstracts or Beilstein online, no structure-activity-relationship software, no substructure searching, no synthetic chemical pathway analysis, no receptorsite modeling. It is quite reasonable to postulate that we might well be at least half a decade or so behind where we are now in the development of effective new therapeutic agents. But, would the productivity calculations for the pharmaceutical industry look any different? Probably not, pharma-ceutical industry sales would not differ markedly from those at present, consumers would simply be buying less effective therapeutic agents. However, the conventional productivity calculation make no provision for the efficacy of therapeutic agents, much less for issues such as extension of life expectancy or increases in the quality of life.
Similarity the point is made that for financial services, the productivity calculations take no notice of the quality of service to the consumer, or of productivity and efficiency as perceived by the consumer. Certainly few of us would wish to return to a world with no ATMS, no automated banking centers at the airport. Yet, the productivity calculations for the financial services industry behave (are calculated) as though this dramatic improvement in customer service and system produc-tivity had never taken place (Banks, 1989). We could attempt to measure many of these things, but we don't.
A third response is simply that many of the capabilities and outputs of information technology are used simply to gain or maintain competitive advantage, that competitive advantage is often a sponge soaking up the benefits of information technology. Loveman (1991) provides the example of the airline industry, where immense information technology resources are utilized to support fre-quent flyer programs and an enormously Byzantine fare structure, with little discernible overall effect on productivity. Inherent in the nature of information is the capability for it to be modified and varied to a degree far in excess of what is achievable with more concrete conventional goods. It is therefore not surprising that a large component of information technology costs seem to be consumed for the purposes of creating product differentiation without enhancing productivity. That may just be the inevitable nature of the beast.
Another response, perhaps the most compelling, is
the argument that our present calculations of productivity in the service
sector are hopelessly flawed. The argument is that since in the case of
services, the measure of output is simply the amount paid for service,
as our economics become increasingly service industry based, productivity
calculations will inevitably and meaninglessly converge. Both the cost
of providing services and the amount paid for services will asymptotically
converge on the amount of total wages paid. The ratio of two numbers converging
on the same thing (the amount paid for services divided by the cost of
providing services) will inevitably con-verge on unity; anything divided
by itself is one. Thus it can be argued that the standard measure of productivity
in a service environment is fundamentally doomed to be increasingly meaningless,
and that to have meaningful measures we must totally overhaul how the productivity
of services is calculated.
4. THE RESOLUTION
However, the most intriguing and provocative response to the productivity paradox is one that has been put forward, in very cautious academic fashion by David (1990) -- that there is likely to be a considerable lag when a new "general purpose engine" or enabling technology is introduced. David draws a parallel with the introduction of the dynamo (the electric generator/motor) and the electrical system or network that it enabled and ultimately created, to computers and information technology. He points out that the U.S. and the U.K experienced a pronounced slowdown in economic growth in the late nineteenth and early twentieth centuries (roughly 1890-1913) precisely when the new technology of the dynamo was being installed. The likely potential reasons that have been put forward for the slowness with which electrification had an impact upon productivity are many, ranging from the necessity of often running both old and new systems simultaneously, to the fact that many of the benefits of electrification derived from the "unit drive" principle, i.e. that a machine could be powered by its own electric engine quite independent of belts and pulleys and shafts. Those belts and pulleys and shafts required sturdy massive structures with machines rather permanently located, and the buildings were apt to be vertical, both to provide access to natural light and to minimize lengthy shaft runs and friction losses. The dynamo, accompanied by electronic illumination, allowed, indeed to a large extent required horizontal buildings of comparatively flimsy construction that provided a flexibility, unknown in the massive buildings of the belt and shaft era. In short, the benefits of the unit drive principle could not be adequately achieved until literally the whole shop floor was redesigned, and that in turn was often a function of the ability and willingness to design, and build a completely new type of building. That sort of change, however, is a major paradigm shift that does not happen overnight.
It is clear from David's discussion that he views the need to redesign the whole factory, what one might best call "manufacturing process redesign" as probably the single most compelling of the reasons put forward as to why this dynamic new technology of the dynamo (double entendre in-tended) was so slow to have an effect upon manufacturing productivity.
David's principal thesis is to draw attention to the parallel between the dynamo and the com-puter, and to caution us that a noticeably long delay before any major effect upon overall produc-tivity is evidenced is not surprising, and that it is perhaps even to be expected in the introduction of a new enabling technology of vast potential. As David points out, if there is indeed a parallel between the dynamo and the computer, then an awareness of that parallel may save us, at least in part from "both the pitfall of undue sanguinity and the pitfall of unrealistic impatience" (David, 1990, p. 359).
Even more interesting and perhaps more enlightening, and the theme of this essay, is the extension that can be made to the David hypothesis. At almost precisely the same time that the litera-ture revealed concern about the productivity paradox, the notion of business process redesign appeared. The extension to the David hypothesis is this, that manufacturing process redesign bears the same relationship to the old technology, the dynamo, as business process redesign does to the new technology, the computer. Expressed in a ratio form this analogy becomes:
business process redesign manufacturing process redesign
---is to --- as --- is to ---
the computer and information the dynamo and electrical
This has exciting ramifications. If this is true,
and if we are at the beginning of the era of business process redesign,
then we are also at the beginning of an era of dramatic productivity increases.
That is an exciting prospect.
5. WHAT EVIDENCE CAN BE PUT FORWARD TO SUPPORT THIS EXTENDED
Consider what is perhaps the classic story of successes
in business process redesign -- Ford Motor Company's redesign of their
receiving, accounting, and payment procedures (Keen, 1991, p.130). Another
print-on-paper shafts, pulleys and belts
-- is to -- as -- are to --
business process manufacturing process
In short, as has been pointed out, but is still not yet adequately realized, many of our present procedures and processes are still based on the constraints of print-on-paper technology. Just as introducing unit drive devices into the old factory doesn't change much until the whole manufac-turing process was redesigned, so the computer has had far less impact than we would have expected -- we haven't yet redesigned the whole business process. Just as engineers in the late nineteenth century probably didn't realize how constrained their thinking was by the shaft and belt world they grew up in, so we don't realize how constrained our thinking is by the print-on-paper world we grew up in. Business process redesign, it can be argued, is the attempt to free ourselves from those constraints.
Also bolstering the thesis is its congruence with
several of the stage hypotheses of information systems development that
have been developed. To illustrate this point, it is necessary to review
those stage hypotheses.
6. STAGE HYPOTHESES
By way of background, it needs to be mentioned that there are two classes of predictive stage hypotheses for information systems. The first class are the life cycle hypotheses that posit that there is a predictable life cycle of development as each major new information technology comes along. The best known of these repetitive stage hypotheses is that of Nolan (1973, 1974). He posited that as new information technologies emerge, they would go through a series of predictable stages. Initially there were four stages in his schema, later increased to six: initiation, contagion, control, integration, data administration, and maturity.
McFarlan and company (McFarlan, McKenney, & Pyburn, 1983) recast the Nolan schema into four stages (identification and initial investment, experimentation and learning, control, and widespread technology transfer) that place greater emphasis upon the management styles appropriate to each state.
These cyclical theories are not uniformly accepted (King & Kraemer, 1984; Sullivan, 1985). Nonetheless, these concepts have had a great impact, and Nolan's article is one of the most highly cited that the Harvard Business Review has ever published.
The second class of stage hypotheses are the noncyclical nonrepetitive hypotheses that try to predict broad spectrum maturational stages of overall information technology application growth. It is this second class of stage hypotheses that are converging and that in their convergence support our hypothesis.
Rather than describe and compare them in detail, which has already been done elsewhere (Koenig, 92), we will simply list them in the chronological order of their origin, and describe their salient features which will be sufficient to establish both their own internal consistency, and their consistency with the productivity paradox versus business process redesign relationship that we have hypothesized.
• Marchand -- Based on extensive work in analyzing the use of information systems in govern-ment agencies, Marchand (1983) postulated a stage hypothesis for the development of integrated information management, a structure that formed the basis for an extensive body of work by Marchand and Horton (1985, 1986). Marchand posited four stages:
- Physical Control,
- Management of Automated Technology,
- Information Resources Management, and
- Knowledge Management.
• Rockart -- Only a year later Rockart (1984), who had done extensive systems work in the corporate environment, popularized the thesis that information systems were developing through three clearly observable stages: first clerical, second operational, third managerial (Rockart & Scott Morton, 1984). Rockarts' observational work was the basis of the later more analytical work by Gibson and Jackson (1987).
• Koenig -- In 1986 Koenig posited (1986a,1986Bb) a stage hypothesis quite orthogonal to the others. This hypothesis derives from a pair of very fundamental observations:
- The doubling periods of the components of information technology are very rapid, on the order of a year or two, a phenomenon known as Moores' Law, (Noyce, 1977) and
- There are three fundamental components of information systems -- computation, storage, and communication (Sullivan, 1985).
From this derive three stages:
- Stage 1 (pre 1971) characterized by exponential Moores' Law growth of computational capability, and relative stayis in storage and communication,
- Stage 2 (1971-1989?) characterized by exponential growth of computation and storage, and relative statis in communication, and
- Stage 3 (post 1989?) characterized by exponential growth of all three components.
Koenig has extended his stage hypothesis, adding a fourth state derived from Mooers' Law (Mooers, 1960). Mooers' Law is the now classic observation that "An information retrieval system will tend not to be used whenever it is more painful and troublesome for a customer to have infor-mation than for him not to have it." This fourth stage will be ushered in by the achievement of continuous speech recognition and voice rather than keyboard input.
• Gibson and Jackson -- Gibson and Jackson (1987) elaborated and refined the work of Rockart. They developed a benefit/beneficiary matrix, and from that they derived three stages which they termed domains:
- Domain I was the automation of traditional back office operations, yielding benefits to the functional unit.
- Domain II, emerging with the advent of the minicomputer to be followed shortly by the microcomputer, resulted in greater efficiency and effectiveness, particularly for end-user computing.
- Now, we are entering Domain III which will be characterized by greater efficiency and effec-tiveness for the organization, and by the transformation of jobs, functions, and roles at all levels -- individual, functional unit, and organizational.
• Zachman -- Most recently, Zachman (Koenig,
1992) has posited a three stage process for the development of systems
software. Zachman posits a general similarity in the design of any sophis-ticated
and complex product, whether they be airframes or software. In the first
stage of the development of complex products, let us say airframes or software,
these products are constructed as custom built one-of-a-kind implementations.
Pushed by consumers, who demanded both volume and economy, Stage 1 gives
way to Stage 2, mass production of airframes and software. Customer demand
is still not satisfied, customers want the benefit of mass production,
yet they also want individually configured products, and the result is
Stage 3, that of "assemble to order" airframes and software.
7. THE CONVERGENCE OF THE STAGE HYPOTHESES
What is at first most striking about these stage hypotheses is their extraordinary degree of consistency with each other. If we lay the various hypotheses out side by side, making only one minor change for the sake of consistency, numbering Marchand's four stages as 0-3, rather than 1-4, since his first stage is essentially a pre electronic information technology stage, then we get the following tableau (Figure 1).
The salient point is that these hypotheses concerning the stages through which information technology will pass, although starting from very different points of view, are all converging on the same conclusion -- that we are now at a major transition point, transiting from the second stage of computerized information system development into the third stage. Of course that convergence might possibly be only a coincidence, but it is far more likely that it is corroboratory rather than merely coincidental, and that the theories do reflect some underlying reality.
Equally striking, however, is that these stage hypotheses, derived from very different view-points -- governmental (Marchand), industrial (Rockart, Gibson & Jackson), information techno-logy (Koenig), and information systems development (Zachman) -- are all concluding that the same transition and take-off point that our analysis of the productivity paradox and its relationship to business process redesign argues for, does in fact exist, and that we are in that transition stage now.
Further bolstering this thesis is Beniger's (1986) analysis in his The Control Revolution that accompanying the industrial revolution, and to a large degree enabling it, was a control revolution that allowed complex structures and processes to be administered. This included, of course, the telegraph, but the control revolution was primarily print-on-paper based, and that print-on-paper
paradigm, largely taken for granted, is a paradigm that constrains and limits our thinking and our creativity. But whereas we have been largely aware that we have been living in an industrial age,
and that awareness has influenced how we created
and structured organizations and processes, and we are now aware that in
some fashion we have entered a post industrial age, we have been largely
unaware of how much the print-on-paper paradigm has contributed the substrate
of our industrial world and how much it has constrained our thinking. A
corollary that can be drawn from Beniger's very convincing analysis is
that not only was the industrial age far more dependent upon develop-ment
in largely print-on-paper information technology, as well as upon the more
obvious industrial technologies, then has been generally recognized, but
that there is far more scope for improvement in productivity through effective
use of information technology than has been realized.
If the analogy holds, if
business process redesign manufacturing process redesign
- -- is to --- as --- is to ---
the computer and information the dynamo and electrical
then we are at the beginning of a dramatic new era, and the "productivity paradox" has just been the latency period, a not untypical latency period, before the new enabling technology, information technology, produces dramatic effects. The convergence of the several stage hypotheses with our main thesis would seem to establish that we are, in fact, at some sort of information technology take-off point and that the latency period is over. The effect that the dynamo had upon industrial produc-tivity is indisputable. Is it reasonable to expect business process redesign to have effects of a similar scale? If one combines both Beniger's delineation of how much the industrial age was dependent upon a "control revolution" that was primarily print-on-paper based, with an awareness of how much of the economy is now service industry rather than classic manufacturing industry, and of how dependent upon information processing much of the service economy is, then clearly the answer is yes.
If this analysis is correct, it answers the paradox of the productivity paradox versus the enthu-siasm for business process redesign and informs us that the latency period is over, but it opens up a host of further questions. The conclusion arrived at places us clearly at a dramatic take off point for the effect of information technology upon society. The impact will undoubtedly be far reaching. Further pursuit of this analogy may help shed some light on those likely impacts and consequences.
The author would like to acknowledge Professor Tom
Wilson of the University of Sheffield whose comments were a key catalyst
for this essay.
Banks, Howard, "The Productivity Paradox," Forbes, 144 (2): 15, (July 24, 1989).
Beniger, James R. The Control Revolution: Technologies and Economic Origins of the Information Society. Cambridge, MA: Harvard University Press, 1986.
Davenport, Thomas H. Process Innovation: Reengineering Work through Information Technology. Boston, MA: Harvard Business School Press, 1993.
Davenport, Thomas H. & Short, James E. (1990). "The New Industrial Engineering, Information Technology and Business Process Redesign." Sloan Management Review, 31(4): 11-27.
David, Paul A., "The Dynamo and the Computers: An Historical Perspective on the Modern Productivity Paradox," American Economic Review, 80 (2): 355-361, (May 1990).
Gibson, Cyrus F., & Jackson, Barbara B. The Information Imperative: Managing the Impact of Information Technology on Business and People. Lexington, MA: D.C. Heath, 1987.
Keen, Peter G.W. Shaping the Future: Business Design through Information Technology. Boston, MA: Harvard Business School Press, 1991.
King, J.L., & Kraemer, K.L., "Evolutions and organization information systems: An Assessment of Nolan's stage model," Communications of the ACM, 27(5), 466-475 (1984).
Koenig, Michael E.D., "Entering Stage III The Convergence of the Stage Hypotheses," Journal of the American Society for Information Science, 43 (3): 204-209, (April 1992).
Koenig, Michael E.D., "The Convergence of Computers and telecommunications: Information Management Implications," Information Management Review, 1 (3): 23-33 (1986). [1986a]
Koenig, Michael E.D., "Stage III of Information Systems Technology," in Intelligent Information Systems for the Information Society, edited by B. C. Brookes. A collection of papers presented at the 6th International Research Forum on Information Science, September 16-18, Frascati, Italy. Amsterdam: Elsevier, 1986. [1986b]
Knorr, Robert O., "Business Process Redesign: Key to Competitiveness," Journal of Business Strategy, 12 (6): 4851 (1991).
Loveman, Gary, "Cash Drain, No Gain," Computerworld, 25 (47): 69-72, (November 1991).
Marchand, Donald A. & Horton, Forrest W. Infotrends: Profiting from your Information Resources. New York: Wiley, 1986.
Marchand, Donald A., "Information Management: Strategies and Tools in Transition?" Information Management Review, 1 (1): 27-34 (1985).
Marchand, Donald A., "Strategies and Tools in Transition?" Business and Economic Review, 29 (5): 4-8 (1983).
McFarlan, F. Warren; McKenney, James L. & Pyburn, Philip, "The Information Archipelago -- Plotting a Course," Harvard Business Review, 61 (1): 145-156 (1983).
Mooers, Calvin N., "Mooers' Law: Or Why Some Retrieval Systems are used and Others are not," Zator Technical Bulletin, 136. (Reprinted as editorial (same title), American Documentation, 11 (3): i, 1960 (1959).
Nolan, Richard L., "Managing the Crisis in Data Processing," Harvard Business Review, 57 (2): 115-126 (1979).
Nolan, Richard L., "Managing the Computer Resource: A Stage Hypothesis," Communications of the ACM, 16: 399-405 (1973).
Noyce, Robert N., "Microelectronics," Scientific American, 237 (3): 62-69, (September 1977).
Roach, Stephen S., "Services Under Siege-The Restructuring Imperative," Harvard Business Review, 69 (5): 82-91, (September-October, 1991).
Rockart, John F. & Scott Morton, M., "Implications of Changes in Information Technology for Corporate Strategy," Interfaces, 14 (1): 84-95 (1984).
Short, James E. & Venkatramen, N., "Beyond Business Process Redesign: Redefining Baxter's Business Network," Sloan Management Review, 34 (1): 7-21 (1992).
Sullivan, Cornelius H., Jr., "Systems Planning in the Information Age," Sloan Management Review, 26 (2): 3-12 (1985).