In this chapter, the panel presents the key findings that emerged from its analyses and deliberations. The discussion is centered on the broad themes and conclusions that have implications for research policy and program development and for future directions for research in the behavioral sciences—particularly human factors, industrial engineering, and industrial and organizational psychology. Specific research recommendations can also be found in individual chapters.
ORGANIZATIONAL LINKAGES EXPLAIN THE PRODUCTIVITY PARADOX
Processes at and across individual, group, and organizational levels intertwine and affect one another such that productivity improvements at one level do not translate simply into productivity improvements at higher levels. The introduction of an intervention, such as technology, designed to improve productivity creates a series of trade-offs at different levels of the organization. As discussed by Attewell in Chapter 2, the potential benefits of the intervention might be channeled into one of two alternative directions—either in a direction that is productivity enhancing, that is, enabling the original work to be done more efficiently, or in a direction that is not productivity enhancing, such as improving some attribute of the product, expanding the work to be done, or inhibiting the productivity of other entities in the organization.
Most of the evidence in this report, however, indicates that even when productivity improvement is realized at a lower level (e.g., the individual), there are influences that might inhibit improvement from being realized at a higher level (e.g., the organization). We have examined those influences from the individual to the organization and from the organization to the individual. The concepts that aid in the understanding of these dynamics were presented by Goodman, Lerch, and Mukhopadhyay in Chapter 3. Examinations of those concepts and of how they operate in specific work domains were reported for office automation by Schneider and Klein in Chapter 4, for software engineering by Kiesler, Wholey, and Carley in Chapter 9, and for computer-aided design by Harris in Chapter 10.
Another form of the productivity paradox is provided by the practice of downsizing. Organizational downsizing has been one of the major initiatives undertaken by U.S. firms during the past decade to increase productivity. It is an initiative that is assumed to lead directly to increased productivity by increasing the ratio of output to input. However, according to the evidence summarized by Whetten and Cameron in Chapter 11, this intervention (similar to the introduction of technology) has not had the anticipated effects. They concluded that this productivity paradox is explained by the use of downsizing approaches that typically ignore the effects of organizational linkages.
The concept of organizational linkages provides a useful framework for examining the productivity paradox. It has led the panel to conclude that a major contributor to the paradox has been the common attempt to initiate change through the introduction of a single intervention (technology) at a single level in the organization (the individual). As suggested by Schneider and Klein in Chapter 4 and by Sink and Smith in Chapter 6, changing a single aspect of an organization almost never results in a substantial change in organizational performance. Organizations are too complex, their performance is too multidetermined, and their inertia is too great for a single innovation at the individual level to have a substantial impact on organizational performance. Even if an intervention does in fact augment individual productivity, there may be no resulting improvements in organizational productivity. Multiple, congruent interventions are needed to achieve the desired impact. This leads to the requirement for an organizational systems framework to clarify the multiple reciprocal linkages that determine organizational productivity.
As discussed by Sink and Smith in Chapter 6, making an improvement intervention in one entity and projecting positive performance linkages to other entities at the same or different levels require profound knowledge. Profound knowledge encompasses a theory of sys-
tems, variation, psychology, and knowledge itself. It equates to a sufficient understanding of the organizational system to identify and predict cause-and-effect relationships. When interventions are made without profound knowledge, they are not likely to have their intended effect—subsystem performance may be enhanced, but the performance of the larger system will not be because the linkages are not understood. The consequence is the productivity paradox—extensive investments in enhancing the productivity of individuals and groups that do not lead to expected improvements in larger organizations or in the enterprise.
ORGANIZATIONAL STRUCTURES AND PROCESSES CAN INHIBIT OR FACILITATE LINKAGES
In our examination of linkages, we identified structures and processes that inhibit increases in individual productivity from increasing organizational productivity. These structures and processes are common to organizations engaged in varied activities—office work, software development, postal services, manufacturing, computer-aided design, and others.
A structural inhibitor, for example, can be found in the existence of core and peripheral activities in most organizations. Core activities, such as the production of engineering design specifications, are directly related to the process of transforming inputs into outputs. Peripheral activities, such as updating computer-aided design software, are only indirectly related to this process. Thus, as a consequence of this structure, increases in individual productivity in core activities will be more likely to contribute to organizational productivity than increases in peripheral activities.
Process operates as an inhibitor when an intervention that increases productivity in one set of activities cancels the gain by decreasing productivity in a related set. For example, the introduction of a computer system resulted in increased productivity on routine tasks by individual customer service representatives. However, the change had negative consequences for the functions of supervisors of the customer service representatives. It reduced the visibility of the operations they supervised and thereby reduced their ability to solve problems and coordinate activities. The net result was no gain in the productivity of the work group, even with the intervention of new technology and the improvement in the productivity of the customer service representatives.
Facilitators are equivalent in importance to inhibitors. Facilitators are processes that can function either to remove conditions that impede linkages or to create conditions that facilitate linkages. Several critical facilitative processes were defined and examined in vari-
ous chapters of the report. Examples include coordination, problem solving, focus of attention, organizational evolution, and motivation. These facilitators were examined and illustrated by Goodman, Lerch, and Mukhopadhyay in Chapter 3.
At a somewhat more specific level, the following facilitators were defined for the domain of software engineering: coordination through team design and coordination through communication. Team design is the development of appropriate structure and procedures that provide built-in solutions to coordination—task decomposition, lines of authority, centralization of control, and standard operating procedures. Communication is the process provided to permit team members to interface with each other, to permit the negotiation of goals and processes by people of different skills and perspectives, and to share information and integrate work outputs. These facilitators might become inhibitors if they are misapplied and lead to costs that outweigh benefits. Coordination issues in software engineering were addressed by Kiesler, Wholey, and Carley in Chapter 9. Additional examples of inhibitors and facilitators, along with descriptions of how they operate in other work domains, were provided by Schneider and Klein in Chapter 4 and Harris in Chapter 10.
There is much to be learned about organizational structures and processes and how they inhibit or facilitate linkages. For example, how do different linkage situations generate inhibitors? What types of interventions or productivity changes will more likely evoke negative consequences under different linkage situations? We have emphasized the role of processes in facilitating individual-organizational linkages. However, in our analyses, processes such as problem solving and organizational evolution were treated as though they were independent of other variables. This gives rise to the question, will these processes affect linkages in the same manner and to the same degree under conditions of complexity and uncertainty as under conditions of stability? In the specific area of software development, a large portion of the variance in the productivity of technical teams is known to derive from how the teams coordinate their work. However, there are many unknowns and questions in this domain alone, such as, what are the side effects and outcomes of linkages over time? The research direction required to provide answers to these and related questions will provide valuable insights into how to realize productivity gains from technology and other organizational interventions.
LINKAGE INFLUENCES ARE SUBJECT TO LEVEL OF ANALYSIS AND DIFFERENTIAL SCALING
Many linkage explanations are likely to be specific to the level of analysis. That is, what accounts for the relationship between individual and group productivity might be different from what accounts for the relationship between group and organizational productivity. For example, the role of the supervisor might be more influential in transforming increases in individual productivity to group productivity than in transforming group increases to organizational increases.
Researchers must also recognize that explanations might scale differentially for different dimensions of an organization. Differential scaling is analogous to the problem faced in designing boats. Estimating hull performance by using a small but accurate scale model may not accurately predict the performance of a full-size version. Differential scaling will occur because different features or dimensions of a hull scale up at different rates, causing relationships between features found for a small hull to be different when scaled up for a large one.
Translated into an organizational setting, this analogy suggests that a change in scale of one feature of a firm may lead to unexpected effects because other dimensions of the organization will not change proportionately. In downsizing, for example, a retrenching firm might shrink its managerial staff by 10 percent. An unintended consequence of this action might be an increase in managerial work load, because each manager now has, on average, a larger number of units reporting than before. Differential scaling occurs because it is unlikely that the number of units reporting to managers will decrease as fast as the number of managers, even if employment in the firm as a whole is cut by an equivalent 10 percent. Thus, span of control does not scale down at the same rate as number of managers, which has potentially negative consequences for organizational productivity. This and other problems associated with attaining productivity gains from downsizing were addressed by Whetten and Cameron in Chapter 11.
HORIZONTAL AND VERTICAL LINKAGES DO NOT NECESSARILY OPERATE IN THE SAME MANNER
In this report, influences on organizational linkages have been invoked as both inhibitors and facilitators of the spread of productivity from one work unit in an organization to other work units. The point has been made in Chapter 9, for example, that coordination and communication mechanisms are important determinants of the extent and degree of this spread. It is less clear whether horizontal and vertical
linkages operate in the same manner—horizontal linkages being between units at the same level and vertical linkages being between units at different levels.
Differences between horizontal and vertical linkages might be viewed as inhibitors to the transfer of productivity gains from one unit to another. For example, if horizontal and vertical linkages require different coordination modes, that degree of specialization might be difficult or impossible to provide. On the other hand, an understanding of the differential characteristics might be a key to the development and implementation of successful coordination and communication mechanisms. In Chapter 4 Schneider and Klein introduced the concept of open systems and discussed the principle of differentiation, integration, and coordination within this context. They summarized previous research that indicates the need to counterbalance organizational movement toward differentiation with integrating and coordinating mechanisms that bring the system together for unified functioning. An understanding of the differential effects of horizontal and vertical linkages may be required to develop the appropriate mechanisms.
METHODS AND MEASURES ARE NEEDED FOR TRACING LINKAGES
A principal focus of our study has been on methodological and measurement issues surrounding the linkage question. These issues were addressed specifically by Ruch (Chapter 5), Sink and Smith (Chapter 6), Pritchard (Chapter 7), and Campbell (Chapter 8). A principal theme of each of these chapters is that methods are needed to determine, analytically, whether specific productivity changes in one work unit are passed through to others. That is, methods are needed for tracing changes through the system from individual to group to organization.
Empirical research is needed to identify the significant linkage variables and their relative importance. To do such research, multiple links within multiple organizations must be studied using longitudinal research designs. First, improvements in outputs at the most molecular level must be shown to have occurred because of an intervention and those improvements must then be traced through the various linkages to the broadest organizational level. The idea would be to measure each explanatory factor (slack, conflict in objectives, and so on) along with the amount of loss of output across the linkage. Then, the importance of each could be assessed empirically. Ideally, the data would be collected so that the variance accounted for by each factor could be estimated.
The macro models now used to assess the effect of information technology on productivity are informative but not very precise. They do
not do a good job of capturing the nature of the specific form of information technology, determining the degree to which this technology has an impact on individual performance, or revealing the structure of linkages and their influences. Consequently, there are a number of challenges in conducting research on linkage questions. First, one has to develop a reasonable representation of the production function of a firm. Second, one needs a model that can capture changes at different levels of analysis. Third, one needs a strategy to identify the lag structure between changes at the various levels. The lag structure is likely to be a critical issue and to be influenced by the specific technology and by many of the variables discussed earlier.
LINKAGES SHOULD BE EXAMINED RELATIVE TO OTHER CRITERIA OF PERFORMANCE
The performance of an organization is a function of at least seven interrelated criteria: effectiveness, efficiency, quality, productivity, innovation, quality of work life, and profitability. A complete picture of organizational performance over time requires measures of these criteria. Operational definitions of each of these criteria were provided by Sink and Smith in Chapter 6. Productivity has been our main criterion as we have focused on understanding how changes in one work unit might lead to changes in other work units. We have examined factors that facilitate or inhibit the transfer of changes. An important challenge is to extend this type of analysis to the other criteria.
As an example, suppose the introduction of information technology improved quality at the individual level. Are the nature of linkages, and the factors that facilitate and inhibit quality changes at the organizational level, the same as would have been predicted for changes in productivity? Would the analysis of linkages be different if one looked at other performance criteria, such as effectiveness, quality of work life, innovation, or customer satisfaction? An underlying theoretical question is whether there is similarity or dissimilarity in the structure of the criteria. That is, given an understanding of the factors that facilitate linkages between individual-and organizational-level productivity, will those same factors predict linkages for other criteria? The practical issue is that a more complete understanding of facilitators and inhibitors will require the examination of linkages relative to the complete set of performance criteria, not just productivity.
In our study, we did not explore how changes in individual productivity might affect these other criteria of organizational performance. For example, the conditions that enable changes in individual productivity to increase organizational productivity might actually decrease
other organizational criteria, such as quality, innovation, or quality of work life. Ultimately, it will be necessary to address the functional or dysfunctional consequences of how other criteria change as individual and organizational productivity are more strongly linked.
A THEORY OF ORGANIZATIONAL LINKAGES IS NEEDED
A theory of linkages is required and should be developed. We have focused on the productivity paradox and have suggested explanations and solutions. However, the paradox is only an example of the importance of studying and understanding the linkages among organizational subsystems. What is needed at this point is a better understanding of how outputs get combined and transformed across organizational levels, the identification of important factors that facilitate and inhibit those linkages, and the determination of conditions under which facilitators and inhibitors operate.
Panel members agree that a theory of linkages is an important goal. However, they are not all in agreement on the direction to take to reach that goal. Some members (see Chapters 5 and 7) propose to develop a composition theory of linkages—an explication of the variables that affect the translation of outputs from one unit or level of the organization to the next. They claim several advantages for such a theory. First, it would provide an integration of all the factors, as argued in the previous chapters of this report, that define and influence linkages. Second, it would provide a description of how those factors are combined and an explanation of how the factors interact. Finally, if valid mathematical functions could be developed to describe the relationships, one could predict the effects that changes at one unit or level would have on changes at other units or levels. They point out, however, that attempts to aggregate individual productivity measures or to disaggregate organizational measures are thwarted by the dissimilarity in measures of output. At the individual level, for example, output is often counted in units of product produced or service provided. At higher levels of analysis, however, different outputs from different sources are combined by means of some form of accounting scheme that is incompatible with measures at the individual level.
Other panel members (see Chapter 6, for example) conclude that composition approaches would not be fruitful. They would emphasize the application of systems theory and statistical approaches to the measurement of total system performance. Rather than pursuing the aggregation or disaggregation of measures at various levels, they would attempt to construct cause-and-effect relationships among measures. For example, they claim that modeling organizational linkages and ana-
lyzing the productivity paradox for a selected set of specific examples could generate tangible theories about cause-and-effect relationships and permit the framing of the problem in a manner susceptible to a solution. Their view is that the paradox of unrealized productivity improvements is an example of incomplete systems thinking and failure to understand the nature of organizational linkages. Their approach would provide answers to such questions as, at which level would one expect performance to improve as a result of an intervention? What aspects of organizational performance will be quantifiable, and which will not? To what extent has what is known versus what is believed about cause and effect relative to linkages been clarified? Thus, an important issue to be resolved is which direction to take in developing a theory of organizational linkages.
We believe that the findings presented in this report have significant implications for policy and research. Our study has focused on the widely reported productivity paradox—successful interventions at individual and group levels have not led to increased productivity at higher organizational levels. We have studied this paradox by examining linkages among individuals, groups, and organizations and have found that those linkages can not only explain the paradox, but also provide the systems framework necessary for determining much of what needs to be learned about improving organizational performance. A key implication of our study is that organizational change, from the introduction of information technology to the downsizing of the enterprise, is an extremely complex endeavor. Recognizing this complexity and addressing it by means of systems thinking are the first steps in successful intervention programs. A successful intervention to increase performance requires that a number of actions be taken at different organizational levels and that they be congruent in their goals, strategies, actions, and measures. As stated earlier, organizations are too complex, their performance too multidetermined, and their inertia too great for a single innovation at the individual or group level to have any substantial impact on organizational performance.
We have described in this report many of the inhibitors and facilitators that influence the extent to which changes in the productivity of one entity result in corresponding changes in the productivity of another entity. We have also identified specific organizational structures and processes that can inhibit or facilitate linkages. Moreover, we have examined and illustrated, in detail, how those structures and processes
affect organizational linkages in several specific domains—office operations, software engineering, and computer-aided design.
The major contribution of our effort, however, lies in our identifying what is known about organizational linkages and in providing a framework for designing and conducting needed research. Although most of our specific research recommendations have been provided in the individual chapters in which the issues related to them were introduced, we complete our report by listing what we consider to be the three most compelling research opportunities.
Theory and methods are required for the measurement of performance across work units of different sizes at different levels in an enterprise. For example, methods are needed that will permit changes to be traced through the system, from individual to group to organization. Because the tools now available are at such an immature stage of development, research on organizational linkages and their influences is difficult to conduct at the level of precision required.
Much more needs to be learned about how structures and processes inhibit and facilitate organizational linkages. The dynamics of these interactions are very complex, and enhancing linkages is likely to be more than simply a function of removing inhibitors and activating facilitators. At this point, possible inhibitors and facilitators have been identified, but a theory is lacking for relating the structures and processes involved to organizational linkages. For example, it is not known when certain structures will inhibit the transfer of productivity gains and when they will not. Nor is it known under what conditions certain processes will facilitate productivity gains and under what conditions they will not.
The analysis of productivity linkages should be expanded to encompass other aspects of organizational performance—effectiveness, efficiency, quality, quality of work life, innovation, and profitability. Focusing on a single performance criterion, productivity, does not provide a complete and true picture of the effects of an intervention, even if the goal of the intervention was solely to improve productivity. Do the conditions that increase productivity also increase or decrease product quality or the level of innovation in the organization? A different question to be addressed by an expanded analysis is, will a theory of organizational linkages developed for productivity apply as well to other criteria of organizational performance?
As we look ahead, we cannot help but conclude that attaining a more productive society and higher overall standard of living will require a better understanding of organizational linkages. The United
States and the rest of the world will inevitably continue to invest more heavily in technological solutions. However, as this report has demonstrated, implementation of new technology is not enough. Organizational linkages must be understood well enough to permit the creation of conditions that will ensure that investments in technology provide the returns of which they are capable.