W-267 ONLINE FILE 14.1 EXPLANATIONS FOR THE PRODUCTIVITY PARADOX DATA AND ANALYSIS PROBLEMS HIDE PRODUCTIVITY GAINS Productivity numbers are only as good as the data used in their calculations. Therefore, one possible explanation for the productivity paradox is that the data, or the analysis of the data, is actually hiding productivity gains. For manufacturing, it is fairly easy to measure outputs and inputs. Toyota, General Motors, Ford, and DaimlerChrysler, for example, produce motor vehicles, relatively well-defined products whose quality changes gradually over time. It is not difficult to identify, with reasonable accuracy, the inputs used to produce these vehicles. However, the trend in the United States and other developed countries is away from manufacturing and toward services. In service industries, such as finance or health care delivery, it is more difficult to define what the products are, how they change in quality, and how to allocate to them the corresponding costs. For example, banks now use IT to handle a large proportion of deposit and withdrawal transactions through automated teller machines (ATMs). The ability to withdraw cash from ATMs 24 hours per day, 7 days per week is a substantial quality increase in comparison to the traditional 9 A.M. to 4 P.M. hours for live tellers. But what is the value of this quality increase in comparison with the associated costs? If the incremental value exceeds the incremental costs, then it represents a productivity gain; otherwise the productivity impact is negative. Similarly, the productivity gains may not be apparent in all processes supported by information systems. Mukhopadhyay et al. (1997), in an assessment of productivity impacts of IT on a toll-collection system, found that IT had a significant impact on the processing of complex transactions, but not on simple transactions. Another important consideration is the amount of time it takes to achieve the full benefits of new technologies. Economists point out that it took many decades to start achieving the full productivity impacts of the Industrial Revolution. Productivity actually may decrease during the initial learning period of new software and then increase over a period of a year or longer. Hitt and Brynjolfsson (1996) point out that answers to questions about the value of IT investments depend on how the issue is defined. They emphasize that productivity is not the same thing as profitability. Their research indicates that IT increases value to consumers and organizational productivity but does not increase organizational profitability. Brynjolfsson and Hitt (1998) suggest using alternate measures, other than traditional measures, to measure productivity. GAINS ARE OFFSET BY LOSSES IN OTHER AREAS Another possible explanation of the productivity paradox is that IT produces gains in certain areas of the economy, but that these gains are offset by losses in other areas. One company s IT usage could increase its share of market at the expense of the market share of other companies. Total output in the industry, and thus productivity, remains constant even though the competitive situation may change. Offsetting losses can also occur within organizations. Consider the situation where an organization installs a new computer system that makes it possible to increase output per employee. If the organization reduces its production staff but increases employment in unproductive overhead functions, the productivity gains from information technology will be dispersed.
W-268 GAINS ARE OFFSET BY IT COSTS OR LOSSES The third possibility is that IT in itself really does not increase productivity. This idea seems contrary to common sense: Why would organizations invest tremendous amounts of money in something that really does not improve performance? On the other hand, there are considerations that support this possibility. Strassmann (1997) compared relative IT spending at a sample of corporations and found little or no relationship between IT spending and corporate profitability. To determine whether IT increases productivity, it is not enough simply to measure changes in outputs for a new system. If outputs increase 40 percent but inputs increase 50 percent, the result is a decline in productivity rather than a gain. Or consider a situation where a new system is developed and implemented but then, because of some major problems, is replaced by another system. Even though the second system has acceptable performance, an analysis that includes the costs of the unsuccessful system could indicate that IT did not increase productivity, at least in the short run. Therefore, productivity evaluations must include changes in inputs, especially labor, over the total life cycle, including projects that are not implemented. These inputs need to include not just the direct labor required to develop and operate the system, but also indirect labor and other costs required to maintain the system. Examples of factors that, under this broader perspective, reduce productivity are provided next. FACTORS THAT COULD REDUCE Factors that could reduce IT productivity are the following: Support costs. The GartnerGroup estimates the total cost of a networked PC can be as high as $13,000 per year (Munk, 1996). Technical support accounts for 27 percent of this cost, and administration for another 9 percent. The additional employees required for these support activities could offset a significant portion of the productivity benefits from the hardware and software. Wasted time. Personal computers make it possible to work more productively on some tasks but also result in nonproductive activities. A survey of 6,000 workers indicated the average PC user loses 5 hours per week waiting for programs to run, reports to print, tech support to answer the phone, and so on or futzing with the hardware or software (Munk, 1996). The GartnerGroup estimates that businesses lose 26 million hours of employee time per year to these nonproductive activities, and that these activities account for 43 percent of the total cost of a personal computer on a network. The cartoon in Figure W14.1.1 highlights the issue of wasted time. Employees also use the Internet and e-mail for private purposes, wasting even more time. Software development problems. Some information systems projects fail and are not completed. Others are abandoned completed but never used. Others are Figure W14.1.1 Dilbert analyzes the productivity paradox. (Source: Dilbert; reprinted by permission of United Feature Syndicate, Inc.)
W-269 runaway projects, systems that are eventually completed but require much more time and money than originally planned. Software development problems are not uncommon: One survey (King, 1997) found that 73 percent of software projects at 360 U.S. corporations were canceled, over budget, or were late. Labor hours associated with these projects can offset productivity gains from more successful projects. Software maintenance. The expense of software maintenance, which includes fixing bugs and modifying or enhancing system functionality, now accounts for up to 80 percent of IS budgets (see Murphy, 2003). Many of the modifications for example, updates to payroll systems to reflect tax law changes do not increase outputs. They are necessary just to keep the system at the same level of performance, so productivity declines because labor increases while output volumes do not. The Year 2K Problem is a notable example of software maintenance that did not add to productivity and actually cost hundreds of billions of dollars to the economy. Most global organizations are also required to incur additional costs for acquiring and maintaining domain name registrations. GartnerGroup (2000) estimated that the average global organization has to register a total of at least 300 name variants, which may amount to $75,000. Many additional names are required because companies want to avoid cyberbashing. Incompatible systems and workarounds. Although individual systems produce productivity gains, the increased labor required to get them to work together (or at all) could offset these benefits. Other possible explanations of the productivity paradox have been noted. A number of researchers have pointed out, for example, that time lags may throw off the productivity measurements (Qing and Plant, 2001). Many IT investments, especially those in CRM, for example, take five or six years to show results, but many studies do not wait that long to measure productivity changes. Another possible explanation was suggested by Devaraj and Kohli (2003), who tried to relate the actual use of a system, rather than the potential use, to the paradox. A list of other explanations of the paradox proposed by Devaraj and Kohli (2002) is presented next. OTHER EXPLANATIONS OF THE PRODUCTIVITY PARADOX Devaraj and Kohli (2002) provide some other possible explanations for the productivity paradox. These include: Time lag. Many IT investments pay off only after a long time. One such case is investment in CRM. According to Reichheld and Scheffer (2000), it takes at least 5 6 years before CRM benefits are large (e.g., due to loyalty of customers). Unfortunately, many studies take only a snapshot and not after sufficiently long duration. For an empirial study of time lag, see Qing and Plant (2001). Anecdotal evidence. Several large IT failures, especially the ERP ones (see Chapter 8), received lots of publicity. On the other hand, there are many unpublicized success stories. It is difficult to generalize from few cases. Aggregate level. Technology is usually implemented at a project or process level, whereas results are measured at an organizational level. There is frequently only an indirect relationship between the two. Difficulties in isolating the effect of IT. Improvements in results at organizational levels (e.g., improvement in profitability) may be the result of several factors, some of which are external. It is difficult to isolate the effect of IT from the others.
W-270 Level of analysis. The IT payoff was studied at three levels: firm, industry, and economy. When studying at the industry and economy levels, it is difficult to separate the technology-related variables from others. Also it is possible that IT investments benefit one company at the expense of others. Performance is determined by many variables in all levels. Complementary factors. IT may relate to other factors, notably business process restructuring (Chapter 15), but data collectors ignore these relationships. One other major phenomenon that can waste time and reduce productivity is known as junk computing, as described in A Closer Look, below. A Closer Look Controlling Junk Computing Guthrie and Gray (1996) coined the expression junk computing to describe the use of information systems in a way that does not directly advance organizational goals. They created this expression as an analogy to the concept of junk mail. The subjects of their research identified 19 different types of junkcomputing activities, and mentioned seven items most frequently. Two non-work-related items games and doing personal work on company time were at the top of the list, followed by: Junk reporting creating unnecessary reports Excessive computerization automation that is not an improvement over manual systems Excessive detail for example, inputting minor activities into a scheduling system rather than focusing on milestones and higher priority projects Excessive attention to presentation, such as elaborate formatting for simple memos E-mail for example, sending copies to large distribution lists or responding to unimportant messages The causes of junk computing include: Excessiveness. Employees may devote more time to computing than is necessary to achieve the desired result. Physical environment for example, incompatible hardware or systems that require manual effort to transfer or integrate data. Cultural or social pressures. Other employees or the organization may encourage excesses, for example, large reports or very elaborate presentations. Individual behavioral styles. Some employees are very oriented toward computers and try to do as much work as they can with them. Miscommunication and mismanagement. Managers may ask for or indirectly signal employees that they want elaborate work, or try to implement technical solutions instead of dealing with the people or procedures that are causing problems. Guthrie and Gray note that junk computing is not always counterproductive. Experimentation for example, using advanced features of a word-processing program to format a memo may develop skills that the employee can subsequently use in a formal report that requires a more elaborate layout. Or playing a quick game of solitaire may provide a relaxing break to an employee, after which she or he can go back to work with renewed vigor. The following recommendations can help managers control junk computing: Increase awareness. Recognize junk computing as an organizational productivity problem. Control environmental factors. Set policies against inappropriate use, mandate standard formats for documents and presentations to prevent excessive effort, and remove game software from computers. Control behavior through training that covers appropriate and inappropriate computer uses, and through chargeback of computing costs (see Section 14.4). Control mismanagement. Encourage appropriate computer usage through effective managerial communications, and by creating an environment that encourages employees to manage their own computer use consistent with organizational needs. Note that with the advancement of the Web, the problem of junk computing has been accelerating because employees use corporate time for private viewing. Controlling this problem can be a major issue (see Chapter 16). Source: Condensed from Guthrie and Gray (1996).
W-271 References for Online File W14.1 Brynjolfsson, E., and L. M. Hitt, Beyond the Productivity Paradox, Communications of the ACM, August 1998. Devaraj, S., and R. Kohli, The IT Payoff. New York: Financial Times/Prentice Hall, 2002. Devaraj, S., and R. Kohli, Information Technology Payoff Paradox and System Use: Is Actual Usage the Missing Link? Management Science, 49(3), 2003. GartnerGroup, Gartner Says New Domain Names May Cost Up to $75,000 to Maintain, Gartner.com, November 20, 2000, gartner. com/5_about/press_room/pr20001120c.html (accessed July 2006). Guthrie, R., and P. Gray, Junk Computing: Is It Bad for an Organization? Information Systems Management, Winter 1996, pp. 23 28. Hitt, L. M., and E. Brynjolfsson, Productivity, Business Profitability, and Consumer Surplus: Three Different Measures of Information Technology Value, MIS Quarterly, June 1996. King, J., IS Reins in Runaway Projects, Computerworld, February 24, 1997. Mukhopadhyay,T., et al., Assessing the Impact of Information Technology on Labor, Decision Support Systems, Vol. 19, 1997. Munk, N., Technology for Technology s Sake, Forbes, October 21, 1996. Murphy, V., The Exterminator (of Computer Bugs), Forbes Global, May 16, 2003. Qing, H. U., and R. Plant, An Empirical Study of the Causal Relationship Between IT Investment and Firm Performance, Information Resources Management Journal, July September 2001. Reichheld, F., and P. Schefter, E-loyalty Your Secret Weapon on the Web, Harvard Business Review, July August 2000. Strassmann, P. A., The Squandered Computer. New Canaan, CT: Information Economics Press, 1997.