Integrated Manufacturing: Redesign the Organization before Implementing Flexible Technology

I response to international competitive pressures, Western manufacturing organizations have focused a great deal of attention on new techniques and technologies for improving manufacturing activities.1 Two dominant manufacturing strategies have emerged. One is the just-in-time (JIT) manufacturing system, originally developed by Toyota, which includes a range of techniques aimed at simplification and waste reduction within the manufacturing system. The other is the computer-integrated manufacturing (CIM) approach, which uses computer-based information systems to link islands of automation, islands of information, and advanced flexible production technologies throughout the manufacturing organizational system.

Although these two strategies have been developed and adopted more or less independently and their compatibility is not well understood, managers generally assume that both approaches are advantageous for improving the productivity and competitiveness of manufacturing operations. For example, both systems are supposed to increase productivity by improving organizational integration, product quality, and manufacturing flexibility and responsiveness. As such, future factories can be expected to combine some of the characteristics of both the JIT and CIM approaches.

Empirically, however, what is known about the success of the two manufacturing approaches within industry is rather limited. JIT has been shown by Toyota and other Japanese firms to be an effective strategy for improving productivity when implemented appropriately.2 CIM is largely unproven; few well-documented case studies of the system are available. Most of the literature describing CIM considers the system from a purely hypothetical perspective and tends to consist mainly of predictions about its success based on the theoretical potential of CIM technology.3

Despite the limited amount of empirical evidence, there are some indications that the JIT approach is more likely to increase productivity than the CIM system. Two separate international surveys that attempted to examine the determinants of manufacturing productivity drew similar conclusions about the relative contribution the two strategies might make toward firm productivity.4 Both found, for instance, that JIT-type approaches aimed at reducing manufacturing throughput time and simplifying the production system were systematically related to higher productivity levels, whereas investments in advanced technologies showed no such direct relationship. Another study, which examined Italian manufacturing firms, found that JIT implementation was strongly correlated with an overall factory performance indicator as well as a wide range of individual manufacturing performance indicators (fifteen out of the seventeen indicators examined). The implementation of a technology-driven approach, however, was found to have no systematic relationship with overall factory performance and contributed to fewer individual performance indicators (only seven out of seventeen).5

Regardless of the contributions the two systems might make in theory, it is worth noting that Western manufacturers have had quite limited success in implementing either strategy. Often this is because managers make assumptions about the organizational context in which the systems are implemented. For example, managers often assume they can implement JIT without modifying their organizational structure.6 Similarly, with respect to CIM, incorrect assumptions about human resources have been shown to result in suboptimal performance of advanced technological systems.7

The objective of this paper is to examine JIT and CIM as integrated manufacturing systems and to consider their relative impact on the manufacturing organizations in which they are implemented. We use a conceptual framework of the JIT system that we have developed through our past research as well as other organizational theory in order to (1) point out critical shortcomings in some of the assumptions underlying the CIM approach and (2) highlight some of the organizational issues that must be addressed for future factories to effectively utilize either JIT or CIM.

Conceptual Framework

A Common Goal: Throughput Time Reduction

Before considering the differences between JIT and CIM, we note one important similarity. Both systems share the idea that reduced system throughput time is a key factor leading to improved manufacturing productivity. A major focus of JIT is on reducing production throughput time, primarily by reducing or eliminating inventory buffers throughout the manufacturing system and supply chain. Such techniques as kanban, leveled schedules, and quick die change allow manufacturers to operate with low inventory levels. The CIM system addresses the concept more broadly to include not only production throughput time but also administrative throughput time, such as order processing time and product development time. It is argued that the CIM system can dramatically reduce both production and administrative throughput time by using both computer-based information technologies, which allow for instantaneous information transmission and data sharing, and flexible manufacturing technologies capable of producing in very small batches.8

The theoretical relationship between throughput time and manufacturing productivity has been identified in the literature.9 Wacker has shown, for instance, that when the organization makes reduced production throughput time the overriding goal for its manufacturing system, other goals that have often been viewed as contradictory, such as low costs, demand responsiveness, and high quality, become quite compatible.10 Consequently, underlying both the JIT and CIM systems is a common and apparently valid assumption that reducing throughput times can lead to improved overall factory productivity.

Throughput Time Reduction in the JIT System

Through our past research on JIT,11 we have developed a conceptual framework for understanding the system and its implications for manufacturing organizations based on the theory of cybernetic systems.12 In particular, the framework considers the impact of reductions in the inventory level (i.e., reduced throughput times) on the operation and structure of manufacturing organizations.

Theoretically, inventory within a manufacturing system is a buffer that absorbs variability between interrelated manufacturing processes.13 The inventory buffer decouples interrelated processes from one another, thereby preventing the variability of one process from having an immediate impact on another. Suppliers shipping an incorrect part, machines breaking down in production, and unexpectedly absent workers are disruptions that can be absorbed by high levels of inventory so that production operations can continue. There may be enough stock on hand to give the supplier time to send another shipment, to allow time to repair a malfunctioning machine, or for other workers to continue production even if one worker does not show up for work.

Other sources of manufacturing variability include organizational functions, such as design engineering or marketing. For example, product designs that incorporate a large number of nonstandard components or a marketing department that generates frequent changes in the production schedule both constitute sources of variability that have traditionally been absorbed by high inventory levels.

Significant inventory reductions under the JIT approach decrease the time it takes for variability in one process or organizational function to have an impact on another and thereby increase the degree of interdependence between related activities. Small amounts of inventory mean that the wrong part, machine breakdowns, absent workers, nonstandard design components, and sudden schedule changes will rapidly disrupt the manufacturing system.

If the organization cannot use inventory to handle variability, it must take other courses of action to avoid these kinds of disruptions. Essentially, there are two available options. The first option is to reduce variability at the source; the second is to increase variability handling mechanisms at the point of impact within production. Variability reduction might include such activities as pressuring suppliers to ensure consistently correct shipments, performing preventive maintenance to reduce the likelihood of machine breakdowns, modifying the employee reward structure to discourage absenteeism, ensuring the use of standardized components in design, and leveling production schedules to filter out market fluctuations. Strategies aimed at increasing variability handling within the system include developing emergency shipping procedures to handle late supplier shipments, having backup equipment available in case of machine breakdowns, developing a multifunctional work force such that workers can be reassigned in the case of absenteeism, and increasing the flexibility of manufacturing processes to cope with high levels of component variety or unstable production schedules.

Organizational Design and Manufacturing System Integration

There is a direct relationship between the preceding concepts and the idea of an integrated manufacturing system. An integrated manufacturing system would be one in which interrelated organizational tasks and activities are effectively coordinated with each other. Therefore, a measure of the system’s overall integration would be its manufacturing and administrative throughput times. A well-integrated system would have short throughput times, and a poorly integrated system would have long throughput times.

As we have noted, short throughput times depend on a coordinated effort of variability reduction and handling. Managers making this effort should not overlook the connections between manufacturing and other functional units. Functions like marketing generate a tremendous amount of variability for manufacturing, simply because functionally designed organizations typically lead to poor communication and coordination.14 Manufacturing thinks the variability coming from marketing is unavoidable and stocks up on inventory.

To improve coordination, activities that create high levels of variability, such as marketing, purchasing, and design engineering, must be brought into closer contact with the activities, such as manufacturing, that must cope with the impact of this variability. One approach is to change the organizational structure to group individuals according to the products they work on. Under conditions of reduced inventory or throughput time, interdependence increases among individuals working on the same products. Thompson suggests that activities with the highest degree of interdependence ought to be grouped together within the same units since they require a higher degree of communication and coordination.15 Similarly, Galbraith suggests that when there is a high degree of uncertainty among interrelated tasks, organizations should try to reduce uncertainty as well as associated information processing and coordination costs by grouping these tasks together.16

The principle of grouping highly interdependent activities applies at various organizational levels. Manufacturing work cells represent such coordination within a product grouping at the level of individual manufacturing processes. In this case, “product” could mean an individual component or subassembly produced within a single work cell. At higher levels, the meaning of product becomes more generalized. For instance, at the factory level, it may be appropriate to group units and departments according to more broadly defined product families.

Although many companies acknowledge the work cell as an appropriate way of organizing manufacturing tasks, they often have difficulty applying the same principle to other functional activities. However, in our research on successful JIT organizations, we found that lower throughput times correlated with companies that were structured along product lines at three different levels of analysis: the shop-floor level, at which functions such as production, quality control, maintenance, and material handling were grouped by product; the plant or factory level, at which functions like production control, manufacturing engineering, purchasing, and information systems were grouped by product families; and the corporate level, where each plant might operate as a focused factory with responsibility for a narrow range of products.17 Hence, system integration is achieved in successful JIT organizations not only by redesigning interrelated processes and functional activities to reduce or handle variability, but also through the basic redesign of the organizational structure.

The main issue in designing low throughput time manufacturing systems, therefore, is addressing the way in which variability is reduced or handled within the organizational system. Whether it is accomplished using technology, people, or any other method is not the crucial issue. Therefore, we will now apply this framework to the CIM system. How does CIM achieve organizational integration and handle system variability?

CIM as an Integrated Manufacturing Strategy

The CIM approach to manufacturing system integration is quite different from the JIT approach. Proponents argue that CIM integrates the organization by automating the flow of information among interrelated processes and organizational functions (islands of automation) using advanced information technologies. In addition, flexible manufacturing technologies — such as robotics, NC or CNC machines, automated guided vehicles, and automated storage and retrieval systems — reduce production throughput times by quickly processing a broad range of products in small batches.18 Thus the main approach used in the CIM system for dealing with organizational variability is to increase the level of flexibility in order to handle variability at the point of impact (within manufacturing). Both integration and variability handling within the CIM system are essentially assumed to be purely technological issues rather than organizational issues. In the following sections, we examine these assumptions in detail.

Assumptions about Integration

Underlying the concept of technological integration are several assumptions about the nature of organizations and manufacturing information that seem to be at odds with actual practice.

Manufacturing Information Can Be Handled by Computer Systems.

CIM assumes that most relevant organizational information can be effectively coded into a form computers can handle. This may not be the case. Computer systems are adept at handling large amounts of simple numerical data, but little of the information used within manufacturing organizations can be coded into such a format. Manufacturing organizations rely to a large extent on “soft” data about such issues as future customer demand, human performance, and the expected output of a particular machine. For example, in one study, researchers compared the existing operations research and artificial intelligence approaches to modeling production scheduling with the actual approaches used by schedulers in manufacturing settings. They found that the models captured only a fraction of the reality of the scheduling task. They were too simplistic and based on unreasonable assumptions about the nature of the manufacturing environment.19

Furthermore, even when manufacturing data generated in one unit of the organization can be coded into forms computers can handle, often they cannot be appropriately translated into forms that are relevant to another unit’s needs. Many of the difficulties traditionally encountered in the design of management information systems are related to an inability to appropriately transform information from a common database to meet the specific needs of different decision makers. Manufacturing managers commonly complain, for example, that their performance is judged on the basis of irrelevant accounting data. Thus, simply putting financial or production data on-line does not necessarily make that data relevant to all decision-making situations within the organization.

Reduced Information Transmission Time Implies Reduced Throughput Times.

As information technologies are supposed to reduce throughput times by increasing the speed of data transmission, it is worth considering the degree to which transmission time contributes to throughput times. In information terminology, the throughput time between any two processes A and B consists of both the information transmission time and the information processing time at A and B. When one considers that information processing for such activities as design engineering, marketing, and accounting can take weeks, months, or even years (in the case of new product development), it becomes apparent that focusing on high-speed information transmission misses the critical bottleneck entirely. To reduce administrative throughput times, the focus of improvement must be placed on the information processing portion of the equation. To reduce processing times, the actual activities performed by various functions must be coordinated.

Organizations Lack Appropriate Information Transmission Technology.

CIM implies that organizations are not integrated because they lack appropriate information transmission technology. This is clearly not the case. One need only examine manufacturing organizations to recognize the wide range of technological options for transmitting information, including fax machines, electronic mail systems, even telephones. A design engineer who wishes to obtain information about the manufacturability of a potential design has many avenues available, the most direct of which is face-to-face communication with manufacturing engineers and production people. Lack of integration is not, therefore, the result of a lack of information transmission technology or even a lack of available information. It is rather a lack of motivation on the part of individuals to use the available options, as discussed in the next section.

Information Transmission Equals Integration.

The CIM literature assumes that information transmission equals integration, as if organizations have a basic desire to be integrated and will become integrated when the necessary information is available. That is, it assumes that functional units share a higher goal that could be achieved with the availability of the correct information.

The reality is that most large organizations behave as loosely coupled systems in which functional units often have little desire to integrate their activities with one another.20 Each component has its own goals, and the extent to which they overlap or correspond to overall organizational goals is at best questionable. We have identified four reasons for differing functional goals:

1. Different Functional Perspectives. Members of each functional unit may have unique perspectives on their contribution to the organization’s outputs. Design engineers may feel that their main objective is the elegance or performance of a product design. But this goal could conflict with the manufacturability of the product, thus creating difficulties for the production unit.
2. Survival Techniques. Functional units are often interested above all in the well-being and survival of their units; they may pursue their own goals at the expense of organizational goals. For example, one unit may be most interested in departmental growth and may draw resources away from units with more pressing needs.
3. Performance Measurement Systems. Units attend to those aspects of their activities that are measured as performance indicators. If a purchasing department is evaluated on the basis of how inexpensively it can purchase materials or how low it can keep its raw material inventory levels, it will pursue those goals even if they cause problems for production.
4. Means and Ends Inversion.21 The means for achieving an organizational goal can eventually take on the characteristics of a goal in its own right to the detriment of the overall organization. For example, in one organization, design engineers were provided with computer-aided design (CAD) systems to improve design effectiveness. The designers used the systems to incorporate complex design features that had been impossible to draw before. These features added elegance to the designs but also added significant costs to the manufacturing process. The designers had replaced the goal of creating effective designs with the new goal of using the CAD systems to their limit.

Given such goal differences, it is difficult to argue that the mere availability of information will increase organizational integration. In fact, the opposite may be more likely. An increase in information may lead to an increase in its misuse or abuse. For example, freely available information about the difficulties of maintaining product quality on a particular production line could be used against manufacturing by another department to its own political advantage. Given such possibilities, units may begin to withhold or even fabricate information.

In short, we question whether CIM can be implemented without a consideration of organizational implications. Differences in functional goals produce myriad incongruent activities, which are a major source of variability for the production system. Production and administrative throughput times can be reduced only if functional goals are aligned. The most effective means of aligning them is redesigning the organization structure such that members of different functional groups work together more closely.

True organizational integration amounts to correcting the problems that have created poorly integrated, loosely coupled organizational systems in the first place. If these problems — which are essentially organizational rather than technological — are ignored, CIM implementers run the risk of institutionalizing ineffective organizational procedures and communications linkages by automating them rather than correcting them.

Assumptions about Flexibility

The use of flexible manufacturing technology is the main CIM strategy for handling organizational variability. In the following section, we examine some of the assumptions underlying the idea of CIM flexibility.

Infinite Flexibility.

Lately, the term “flexibility” has been used too broadly. Much of the recent manufacturing literature would lead one to believe that new flexible technologies are capable of doing almost anything. However, compared to the average human worker, even the most flexible piece of equipment falls far short in its ability to learn and adapt to new situations, perform tasks requiring a high degree of coordinated motion (including many simple tasks such as picking parts out of bins), or process complex information (such as complicated pattern recognition). Thus, one should not assume that flexible technologies are infinitely flexible. Instead, they are flexible only within a predefined range of possibilities, which may be somewhat wider than that of traditional hard automation but which is still very narrow.

Flexible Technology Equals Organizational Flexibility.

A flexible manufacturing system is only as flexible as its least flexible subsystem, in the same way that a chain is only as strong as its weakest link. Consequently, even though a particular robot is capable of a tremendous range of activities, most of these activities may be completely inaccessible within the constraints of the overall system because of far less flexible material handling equipment or hard tooling.

Furthermore, manufacturing process flexibility represents only part of the chain, which also includes non-manufacturing activities such as purchasing and other organizational functions. An electronics plant, for example, uses automated machines to assemble a wide variety of circuit boards. However, the assembly process is constrained by an inflexible and unresponsive purchasing bureaucracy that is incapable of keeping the equipment supplied with components. This situation forces production to make large batches whenever components are available, thus eliminating any throughput time reductions that the flexible equipment might have been able to provide.

Flexibility Handles Unpredictable Circumstances.

Most flexible manufacturing systems are being justified on the basis that they will be able to adapt to unknown future requirements. However, this assumes that future requirements remain within the range of change envisioned by the system’s designers. When demands change beyond this range, the system becomes obsolete. For example, when the personal computer industry switched from 5.25-inch diskettes to 3.5-inch diskettes, even the most flexible systems for producing the large format diskettes and disk drives were rendered obsolete at factories around the world.

In addition, the investment required to establish flexible technologies can actually reduce the firm’s long-term flexibility and innovation. We know of one case in which a manufacturing plant installed a large-scale, multimillion-dollar automated storage and retrieval system in the early 1980s, when such systems were generally considered to be highly productive investments. A few years later, the company adopted a JIT strategy, which emphasized the elimination of inventory and centralized warehousing. The organization was left with several years worth of outstanding debt and a host of inventory management procedures that now conflicted with the new ideals the company was trying to pursue. Short-term flexibility had been bought at the cost of long-term financial and procedural rigidity.

Flexibility Is Free.

The literature has tended to create the optimistic impression that flexibility is free and has generally ignored any possibilities of additional organizational costs associated with flexible manufacturing technologies. Instead, CIM proponents have suggested that manufacturing flexibility allows organizations to broaden their marketing horizons by competing on the basis of economies of scope rather than economies of scale and by being able to compete effectively against any manufacturing sector, including unit, mass production, or continuous process industries.22 Does CIM really offer something for nothing? What are the potential costs?

One obvious cost is the greater financial investment required to purchase flexible equipment as compared to more traditional, less flexible equipment. We are not aware of any evidence showing that flexible manufacturing technologies can be installed for the same price as inflexible technologies, and there is plenty of evidence to the contrary. All else being equal, a process that is optimized for a narrow range of flexibility will always require less capital investment than one that must be optimized for a broader range. The same holds true for unit production costs. Dedicated equipment that is optimized for a narrow range of processes will always have a lower per unit cost than flexible equipment that is optimized for a broad range of tasks.

A less obvious but potentially enormous cost is that associated with the increased level of complexity within the organization. Advanced production technologies require advanced technical management and support staff, just as traditional data processing and information systems have required management and support staff to operate and maintain them. The literature has generally ignored this additional overhead burden, which represents a sizable cost and organizational risk.

Flexibility Is the Only Option.

Perhaps the most problematic assumption, given our conceptual framework, is that production system flexibility is the most appropriate and cost-effective strategy for dealing with variability within the organization. The CIM approach, as it is defined in the literature, assumes that organizational variability is best handled at the point of impact within production through the use of flexible technology. The literature makes virtually no mention of strategies aimed at reducing variability at the source. Yet variability reduction strategies could eliminate the need for costly flexibility through CIM. For example, by reducing the variety of components used in its designs, an organization can continue to use inexpensive, simple, dedicated machine tools instead of more costly flexible ones. In an automotive parts manufacturing plant, managers dramatically simplified the plant layout, replacing an entire automated guided vehicle line with a manual handcart system. The new system was far simpler to operate and required virtually no maintenance.

Is flexible technology ever the right choice? Certainly. In many situations, the source of variability is beyond the organization’s control. For example, customers will continue to demand a certain amount of product variety. But where the organization does have some control over sources of variability, managers need to decide which strategy is likely to be more cost effective. In some cases, it is indeed more cost effective, at least in the short term, to cope with variability after the fact, rather than attempting to reduce it at the source. It is important to recognize, however, that both options exist. In the long term, strategies that avoid costs by addressing and eliminating problems at their source are superior to strategies that require capital investment yet succeed only in coping with ongoing problems of organizational variability.

This line of reasoning applies regardless of the type of variability a manufacturing organization experiences. That is, although organizations invest in flexibility for a variety of reasons — for volume flexibility to cope with variable demand volume, for product flexibility to quickly bring new products to market, or for process flexibility to produce different products on the same machines with minimal set-up time — the logic holds true. Wherever possible, it is wise to reduce the need for investments in flexibility by reducing the source of variability, rather than simply trying to cope with variability that may be avoidable and unnecessary.

For example, a common way of reducing the need for volume flexibility is schedule leveling, whereby volatile demands are met gradually over time rather than all at once. Similarly, in cases where demand is highly seasonal, companies attempt to manage demand by offering customers reduced off-season pricing. Naturally there are limits to such strategies, and investments in volume flexibility may be appropriate for coping with market variability that cannot be reduced.

Some Japanese automakers have reduced product variability by treating many traditional options as standard features and then grouping the remaining ones into a small number of option packages. By cleverly designing these packages to address actual customer demand patterns, these automakers are able to meet the demands of all but a very small proportion of customers. Such a strategy filters out variability, dramatically reduces the need for process flexibility, simplifies material planning and control activities, and eliminates costs associated with information processing and coordination.

Designing for Integration

To reduce production and administrative throughput times, organizations must create an integrated manufacturing system, which is essentially an issue of organizational system design rather than technological system design. Now we examine some issues that must be addressed as organizations attempt to design truly integrated manufacturing systems, regardless of the techniques or technologies they use.

Reducing the Impact of Environmental Variability

All manufacturing organizations must cope with a certain amount of variability, which is caused by the market, suppliers, competition, government regulations, and so forth. The choices an organization makes for handling variability within these domains have implications for how the organization functions internally. Consider the marketing environment. No single organization can meet all possible market demands; every organization must decide how much market variability it can handle. For example, no car manufacturer can meet all possible customer desires. The Japanese automakers mentioned earlier have reduced variability by limiting the number of choices offered to consumers.

As organizations attempt to handle more market variability, organizational integration becomes increasingly difficult. As the range of options offered to customers increases, so does the difficulty of integrating the activities of materials handling, purchasing, scheduling, design engineering, suppliers, and so on. Therefore, decisions about how much market variability to handle should be made collaboratively, with input from each of these groups, rather than unilaterally by the marketing department, as is too often the case.

The trick in effectively coping with market variability is cleverly balancing the need to satisfy the greatest number of customers while generating the least amount of internal variability for the rest of the organization. This requires a more detailed understanding of real market variability than has often been acquired by marketing departments in the past. In the automotive example, marketing may need to know if it would lose customers by grouping automatic door locks and automatic windows in an option package, rather than offering them separately.23 Do most customers care about such choices? Does the benefit of offering increased choice outweigh the cost of handling the extra variability that will be generated for manufacturing, purchasing, and suppliers? These are fundamental questions that have been ignored too often in the past.

Reducing and Handling Variability Internally

The same logic applies internally. At every organizational level, each unit’s activities may generate undesirable variability for other units. Design engineering generates variability in the form of engineering changes, the number of components used in a design, and the manufacturability of design features — all of which must be handled downstream. Similarly, machines and work units within production can create variability for other work units, in the form of breakdown delays, changes in production rate, or inconsistent product quality. Organizations must make sure that the amount of variability generated by one unit is matched by the capability of other units to handle that variability.24

As we have stated, although such a match can be obtained either by reducing variability at the source or by increasing variability handling at the point of impact, variability reduction is often the preferred solution. This is because every variability handling mechanism comes at a cost and tends to generate additional undesirable variability. It is widely accepted that reducing quality variability at the source using statistical process control (SPC) or similar methods is a more effective strategy than after-the-fact inspection and rework. However, with respect to variability generated elsewhere in the organization, most manufacturers have not applied the same logic.

Thus, one basic principle in designing an integrated manufacturing system is to keep the overall level of variability within the organization to a manageable level. The variability generated by each functional and work unit must be kept under control. The lesson of SPC needs to be applied throughout the manufacturing organization, so that all forms of variability are controlled at their source.

Each unit must balance its needs with the variability it generates by meeting those needs. Manufacturing must weigh the need for more sophisticated production processes against the potential for increased breakdowns and maintenance associated with increased process complexity. Design must balance the need for enhanced product design features against the impact design changes will have on production and other functions. One organization we have worked with has limited the frequency of engineering changes to one a year in order to reduce logistical confusion.

Appropriate Organizational Structure

Effective variability management requires a strong degree of communication and coordination among interdependent units. As we have stated, such coordination is quite difficult when interdependent units are separated organizationally. The impact of their variability on one another is not obvious, and they receive little feedback from one another. Such isolated units tend to work toward their own goals with no regard for their impact on other units. To close the feedback loop, units that have the greatest impact on one another should be grouped as close together as possible. In the context of JIT, product-based organizational structures provide better coordination than functional structures.

At the extreme, this line of reasoning suggests that it is desirable to group interdependent tasks not only within the same units but also within the same jobs. Many organizations tend to underutilize the potential of human beings for handling variability. They invest in flexible technologies and then restrict human operators to simple, narrowly defined tasks. However, as individuals typically possess the most accurate information about the variability that affects their activities, it makes sense that they also be given the responsibility to find appropriate ways of dealing with it. Incidentally, this is quite apart from any intrinsic benefits that may be gained from traditional sorts of job enrichment.

Implications for Future Factories

We have argued that in order to take best advantage of CIM and JIT techniques and technologies, it is necessary to examine these systems within the organizational context. Such an examination not only can improve the design of such systems, it may also increase the likelihood of successful implementation. CIM implies many assumptions about the capabilities of information and flexible manufacturing technology as well as the outcomes of its usage. Most of these assumptions have been unquestioned in the literature, although many of them are difficult to defend. Thus, we believe that researchers and manufacturing managers should rethink the use of advanced technologies.

It is important to recognize that major changes in manufacturing systems, such as those represented by CIM and JIT, have direct implications for the overall organizational design. Thus, it may be more productive to redesign the organizational structure before implementing available technology than to hope the technology will bring about manufacturing effectiveness. In summary, JIT emphasizes variability reduction; it creates an environment in which the level of variability is inherently more manageable. On the other hand, CIM emphasizes variability handling, suggesting that manufacturers can use flexible technology to handle an unmanageable situation.

We advocate an approach that begins by examining the interdependencies that exist among functional units within the overall manufacturing organization. The sources of production system variability generated by different functions need to be identified and corrected before variability handling is addressed. Investments in flexible technologies should then be considered as a last resort and used to handle variability that cannot be reduced through more efficient means.