Special Issue: Knowledge, Knowing, and Organizations: Knowledge as a Contingency Variable: Do the Characteristics of Knowledge Predict Organization Structure?

This paper examines the validity of knowledge as a contingency variable. Building on recent advances in thinking about the dimensions of knowledge assets (Winter 1987, Zander and Kogut 1995), we argue that such dimensions might have an important influence on organization structure. More specifically, we focus on two dimensions of knowledge—observability and system embeddedness—and their influence over the level of unit autonomy and interunit integration in an intemational network of R&D units. Statistical analysis of questionnaire responses from 110 R&D unit managers show strong association between the dimensions of knowledge and organization structure. It also indicates partial support for the "fit" hypothesis in contingency theory. The paper makes two important contributions to the knowledge management literature. First, we find support for the contingency logic, suggesting that effective organization design has to take into account the underlying characteristics of the firm's knowledge base. Second, we shed light on a relatively neglected dimension of knowledge that we call system embeddedness. This is the extent to which knowledge is a function of the social and physical system in which it exists. In the statistical analysis it emerges as a strong predictor of organization structure. Moreover, it also appears to be conceptually distinct from the tacit-articulate dimension that is normally emphasized. This allows us to speculate on four generic forms that a firm's knowledge might take, that we label integrated, isolated, opaque, and transparent. These are discussed using examples from the data, (Knowledge Management; Contingency Theory; R&D Management) This paper uses a contingeticy theory framing to exatnine the proposition that the characteristics of a firm's knowledge base have an important influence on its choice of organization structure. Traditionally, contingency theory has focused on such contingency variables as environmental uncertainty, firm size, and firm technology. Our approach is to build on recent advances in knowledge management to establish whether "knowledge" can be considered a useful contingency variable in its own right. Using questionnaire data from 110 R&D units in 15 multinational firms, we find strong support for this assertion. We also find evidence that the "fit" between knowledge characteristics and organization structure is related to performance, though this finding is less clear-cut. The research has important implications for the measurement of different dimensions of knowledge. Working with the questions developed by Udo Zander and Bruce Kogut (Zander 1991, Kogut and Zander 1993, Zander and Kogut 1995), we focus on two aspects of knowledge, observability, and system embeddedness. Observability is defined as how easy it is to understand the activity by looking at and examining different aspects of the process or final product. System embeddedness is the extent to which the knowledge in question is a function of the system or context in which it is embedded (see next section for a discussion of these definitions). Our analysis suggests that system embeddedness is the stronger predictor of organization structure and, moreover, that the two dimensions are orthogonal. This insight suggests some important implications regarding the management of knowledge in large firms. In terms of the empirical context of the research, we collected data from 110 R&D unit managers in 15 Swedish multinational firms. All these firms have their R&D activities dispersed among multiple locations, so it is a matter of some importance to decide how R&D work should be configured and coordinated. Some firms, such ORGANIZATION SCIENCE, © 2002 INFORMS Vol, 13, No, 3, May-June 2002, 274-289 1047-7039/02/1303/0274/$05,0