The research reviewed in this monograph suggests that the relationship between learning processes and outcomes at different levels of analysis needs to be better articulated (e.g., see March, 1991; Miner & Haunschild, 1995). Whether, when and how learning at different levels of analysis complements each other, substitutes for each other or are independent of each other are important questions that would benefit from further research. These issues are developed here in the context of understanding the relationship between group and organizational learning.
1. Group and Organizational Learning
This section explores the relationship between learning at the level of the work group or organizational subunit versus learning at the level of the organization or larger system in which subunits are embedded. When are benefits of promoting learning at the level of the work group or subunit particularly strong? When can benefits be realized by transferring knowledge across groups and thus promoting learning at the level of the organization? Can organizations simultaneously promote group learning as well as organizational learning or does a focus on one form of learning preclude the other? Does group learning complement or substitute for organizational learning? The section begins with an example that illustrates the ten- sion between group and organizational learning. The conditions under which it will be especially advantageous to foster local learning and those under which it will be advantageous to foster organization-wide learning are then developed.
Adler and Cole (1993) described the tension between learning at different levels of analysis in their comparison of the “democratic Taylorism” approach to produc- tion used at the Toyota-GM Nummi plant in California versus the autonomous work group approach once used at Volvo’s Uddevalla plant. Adler and Cole presented data suggesting that the Nummi plant was more effective than the Uddevalla plant and argued that the difference in performance was due to an emphasis on organiza- tional (rather than local) learning at Nummi. At Nummi, there is much emphasis on standardization and transferring knowledge across groups and departments so that productivity gains made in one area of the plant can benefit another. By contrast, the Uddevalla plant fostered individual or group learning at the expense of organiza- tional learning (Adler & Cole, 1993). Work groups were given a great deal of auton- omy; little effort was made at transferring knowledge across them. Thus, a productivity improvement made in one group did not benefit another. As evidence in support of this point, Adler and Cole (1994) noted the unevenness in performance of teams at Uddevalla.
In a rejoinder to the Adler and Cole article, Berggen (1994) argued that the autonomous work groups used at Uddevalla were effective and that a major benefit of those systems was their adaptability and flexibility. There has been considerable discussion in the literature of the conditions under which autonomous work groups improve organizational performance (e.g., see Goodman, Devadas, & Hughson, 1988). For the purpose of this monograph, the question is: under what conditions is it most beneficial to promote learning at the group level versus at the organizational level?
In answering this question, this section specifically addresses how the nature of the environment in which groups and organizations operate and the tasks they per- form affect the most appropriate balance between group and organizational learn- ing. Although dimensions of the organization’s environment and its task are socially determined to some extent, these factors are more “given” or exogenous than dimen- sions of an organization’s structure. Environmental and task factors are more critical in determining the appropriate balance between group and organizational learning than structural factors because structural factors are more under the control of the organization. Structural factors can be designed by the organization to promote or retard group or organizational learning.
The similarity of customer preferences affects the relative balance between group and organizational learning. Do the groups or organizational subunits face environ- ments with customers with very different needs and preferences or are their prefer- ences similar? To the extent that customer preferences are similar across groups and organizational units, there are great benefits of transferring knowledge because an innovation made in one group will be relevant for another. Thus, when groups face environments with similar customer preferences, organizational learning and its emphasis on transfer of knowledge across groups or establishments is favored. As customer preferences diverge, local learning at the level of the work group is prefer- able because these groups are better positioned to adapt to local conditions.
Similarity in customer preferences translates into a more uniform product. Building on the Nummi versus Uddevalla example, discussions of when autono- mous work groups will be most effective emphasize that the extent to which the product is uniform is a determining factor. For example, Williams (1994) argued that although autonomous work groups have many benefits, these groups might not be suitable for the production of large numbers of uniform products, such as the final assembly of automobiles.
Thus, when groups face similar customers or produce similar products, there are many advantages to transferring knowledge across groups. Under these conditions, organizational learning is favored. Conversely, when groups face customers with different needs and preferences and/or produce a nonuniform product, there are benefits to providing groups with much autonomy and flexibility so that they can adapt to local conditions. Under these conditions, group learning is favored.
Another major factor indicating the relative balance between group and organiza- tional learning is the degree of interdependence of the tasks involved. If groups per- form tasks that are highly interdependent, organizational learning is favored. Conversely, if groups perform tasks that are low in interdependence and require little interaction, group learning is favored. In systems high in interdependence, the output of one group greatly affects another, whereas in systems low in interdependence, the output of one group is decoupled from that of another (e.g., see Thompson, 1967).
To return to the Nummi versus Uddevalla discussion, car manufacturing is typi- cally characterized by high degrees of interdependence: what happens in one part of the plant (e.g., the body shop) affects what happens in other areas (e.g., the paint shop or final assembly). Tolerances are tight; there is little room for error. Coordination is necessary even in automobile assembly systems organized into autonomous work groups because one group assembles only part of the automobile and that group’s subassembly has to be integrated with other groups’ subassemblies. In systems such as these where interdependence across groups is high and coordina- tion needs are correspondingly great, an emphasis on transferring information across groups is valuable.
Levinthal and March (1993) discussed a task factor closely related to interdepen- dence: decomposability. According to Levinthal and March (1993), when the prob- lems an organization faces can be broken down into decomposable or autonomous units, “local” learning is effective because little interaction is required across units. Decomposability is, in part, a function of the scale and complexity of the operation. Operations where productivity has been boosted dramatically by organizing into autonomous work groups or “manufacturing cells” have this property of decompos- ability (e.g., see Levin, 1994, for a discussion of the assembly of personal comput- ers): all of the work to assemble the product can be accomplished by one group.
Thus, if tasks can be broken down into decomposable units, group or local learn- ing is favored. Conversely, if tasks are highly interdependent, there is a need to transfer knowledge across groups to insure that one group’s activities are coordi- nated with another’s. Under those conditions, organizational learning is favored.
This discussion of when to emphasize group learning and when to emphasize organizational learning is somewhat reminiscent of earlier discussions of when to organize along product lines and when to organize along functional lines (e.g., see Duncan, 1979). A lesson learned from analyzing organizational structures is that they are not static. For example, some organizations centralized after being decen- tralized (e.g., see Stata’s, 1989, discussion of Analog Devices), while others, such as Xerox, decentralized formerly centralized structures.
These structural changes may, in part, be due to changes in environmental or task characteristics that indicate the appropriateness of a different structure. The changes, however, might also be due to an organization’s need to coordinate across both func- tions and products. An organization’s primary grouping along product or functional lines facilitates coordination along the primary dimension. Coordination along the other dimension can be achieved by lateral relations, personal relationships, and the like. Over time, it can become more difficult to coordinate across the nonprimary grouping. Personal relationships atrophy or dissolve through turnover. Proximity, organizational incentives, and the social psychological factors of in-group favorit- ism discussed in the last chapter increasingly favor interacting more with one’s own group and less with other groups. Thus, over time, a product-based organization can find that it is increasingly difficult to coordinate across the different products. An organization might then centralize to transfer knowledge across the different prod- ucts. Conversely, a functional organization can find that it becomes increasingly difficult to coordinate across the different functions for each product. If problems associated with this lack of coordination are severe, the organization might decen- tralize into a product organization.
Thus, for some organizations, a product form of organization is appropriate and likely to remain so for a reasonable time period. For other organizations, a func- tional form of organization is appropriate and likely to remain so. But for many organizations, the forces favoring one form of organization are only slightly stron- ger than the forces favoring another. Matrix organizations are one solution to this dilemma. They are challenging to manage, however, and have a mixed record of success (Davis & Lawrence, 1977). Another solution to the dilemma is to alternate over time between different forms of organizing as a way of achieving the benefits of both product and functional organizations.
The same might be true of group versus organizational learning. Many organiza- tions would be well-served by emphasizing group learning while others would be better served by placing more emphasis on organizational learning. The similarity of environmental and task conditions and the interdependence of groups or subunits are important factors that affect the relative balance between group and organiza- tional learning. For some organizations, however, these factors might not clearly favor one form of learning over another. These organizations need to achieve learn- ing at both levels.
Achieving learning at both organizational and group levels is difficult. Basing his arguments on evolutionary theory, March (1991) argued that strategies that improve the survival of a system’s components might not be the same as strategies that improve the survival of the system as a whole. The social psychological arguments developed in Chap. 6 also suggest that it can be difficult to promote group and orga- nizational learning simultaneously. Providing groups with considerable autonomy and allowing them to develop their own unique task performance strategies under- mines an organization’s ability to transfer knowledge across groups. Conversely, transferring knowledge across groups requires some degree of standardization that can conflict with the emphasis on group autonomy. Kush, Williamson, and Argote (2012) argued that promoting learning through factors that affect motivation, such as rewards, can lead to a trade-off relationship between group and organizational learning while promoting learning through factors that affect opportunities to learn can lead to a positive relationship between group and organizational learning.
Source: Argote Linda (2013), Organizational Learning: Creating, Retaining and Transferring Knowledge, Springer; 2nd ed. 2013 edition.