In general, problems in collective action can be resolved by an agency9 (management, for example) enforcing a regime of cooperation while outside the process itself. While an agency might enforce knowledge sharing in the organization, it might have limited concern for the details of the knowledge shared, and so its control might focus more on the structure for sharing, effectiveness of sharing, fairness and distributive justice, punishing free-riders, and so on. For example, a manager of a sales team might try to organize and motivate personnel to share information and experience in building customer relations, in order to improve the overall team sales performance.
The agency can make structural choices where changes are intended to increase points of contact or networks of organization members (Kransdorf, 1998). At the modest end of the scale, as Grant (1996) suggested, knowledge can be shared by forming a group that engages in problem solving and decision making. This alternative is particularly relevant if the task’s complexity is very high and face-to-face interaction is needed to perform it. Moreover, management might create a unit to direct and oversee knowledge sharing. A comparison of manufacturing costs between a firm’s activities in various countries might reveal strong differences. Case studies have shown that in order to exploit these differences, management typically creates a new knowledge management unit, or a knowledge and technology transfer unit whose task is to identify best practices among selected plants, and then foster the transfer and leverage of these practices throughout the firm on a worldwide basis (Dixon, 2000; von Krogh, Ichijo, and Nonaka, 2000; Kriwet, 1997). Typically, such units identify technologies and practices to be shared, document practice, structure knowledge, create manuals for training plant operators and other key people, revamp procedures, and encourage plants to participate in the practice-sharing program, and monitor progress (O’Dell and Grayson, 1998; Lu, Mao, and Wang, 2010).
At the larger end of the scale of organizational changes, Nonaka and Takeuchi’s (1995) case studies show that management can choose a multi-layered organizational structure—what they term the ‘hypertext organization:’ a business system layer that covers the main tasks and business processes of the firm; a project layer where people collaborate to create new knowledge and innovation; and a knowledge-based layer that comprises organizational memory (e.g. positions, databases, organizational culture, etc.), and which organization members can access when needed for projects or operational activities. Employees of such organizations network on at least three different levels when solving operational tasks, working on projects, and storing and structuring knowledge. This networking is also instrumental in knowledge sharing between the various layers, and makes the organization increasingly adaptive to changes in the environment (Nonaka and Takeuchi, 1995; Nonaka, von Krogh, and Voelpel, 2006).
Between these two extreme structural solutions to the sharing problem, we typically find the matrix organizational structures that combine business activities with a regional focus, or a functional focus with an industry or customer focus. The matrix structure is intended to increase the interconnectedness of functions, departments, groups, and regional centers. Decision making and resource allocation are federative rather than centralized (Doz and Prahalad, 1981). Resource allocation results from a bargaining process between the units, and thanks to the federative form, each unit in the matrix has an incentive to specialize, or in other words, create centers of excellence and knowledge that do not overlap, while seeking knowledge from other units as needed to conduct local tasks (Bartlett and Ghoshal, 1986).10 However, as Bartlett and Ghoshal (1986) pointed out in their early writings on the topic, the matrix is more than a structure; it is a ‘frame of mind.’ This implies that performance expectations for both vertical and horizontal activities require extensive knowledge sharing, but that for such structures to work, local individual interests must be reflected in the overall corporate agenda, and effective human resource management must be put in place. In other words, organizational structure must be ‘translated’ into individual behavior (Gibbons, 1998).
The knowledge management literature comes to similar conclusions. Because human interest and tacit knowledge are often involved, there are strong limitations to structural solutions to the knowledge-sharing problem (see also Boisot, 1998; Grant, 1996; Tsai, 2002). Simply increasing people’s exposure to functions, projects, and knowledge—and even to other people—fails to safeguard collective learning in the firm. Making a regional head talk to the supply chain expert at the corporate center will more often than not necessitate a real flow of technical expertise to other regions. Hence, within the agency solution, the knowledge management literature suggests that well-functioning HR management systems are imperative. Management can create management support systems, monetary or career incentives for individual organizational members, and conduct employee performance appraisals that take into account individuals’ interests and attempts to share knowledge and information (e.g. Davenport and Prusak, 1998; Davenport and Probst, 2000; Grant, 1996; Pedler, Burgoyne, and Boydell, 1991). Management’s investment in information and communication technology (ICT), including search capabilities, document management, messaging connectivity, expert systems, and pattern-recognizing technologies, all of which enable knowledge sharing, is strongly related to this emphasis on HR management systems (Allee, 1997; Alavi and Leidner, 2001). Yet, as case studies have shown, in modern organizations, although ICT enables people to connect more effectively, team performance and knowledge sharing are linked to individual and team-based incentives (Govindarajan and Gupta, 2001).
In those instances where tacit knowledge cannot be codified, sharing it between people is slow, costly, and uncertain, as was pointed out in the early work on knowledge in organizations (Kogut and Zander, 1992). Hence, the tacit nature of knowledge and the collective actions of knowledge sharing pose limitations to HR management systems. In an important paper, Osterloh and Frey (2000) discussed whether self-interested individuals are motivated to share knowledge extrinsically, through material incentives such as money, or intrinsically, through the activity itself. Intrinsically motivated individuals act for their immediate need satisfaction, for example, for the inherent joy of performing a task. In the sharing of tacit knowledge in particular, intrinsic motivation outweighs extrinsic motivation. Material incentives cannot make people change their interest in codifying and sharing their tacit knowledge; neither can contracts assure effective and efficient knowledge sharing. The classic option through which organizations can achieve team production (Alchian and Demsetz, 1972) breaks down with tacit knowledge sharing, because there is no way for team members to monitor each other’s sharing behavior.11 If monitoring and sanctioning possible free-riding behavior is challenging for team members themselves, it is even more so for any outside agency that must base disciplinary action on retrieved information about individual team members’ performance. Hence, intrinsic motivation, by which team members realize immediate need satisfaction by working together to solve complex tasks, is a prerequisite for (tacit) knowledge sharing.
However, it is important to know the conditions under which intrinsic motivation operates, one of which relates to cadence. Immediate need satisfaction might be a strong motivation for some organizational members, but the sequencing of activities involved in knowledge sharing12 depends on the type of knowledge to be shared, as well as other individuals’ rates of learning. Experiments have shown that if learning is slow, the rate at which people acquire tacit knowledge differs greatly (Shanks and Johnstone, 1998), and so immediate need satisfaction might fail to satisfy individual needs and provide sufficient stimulation for a change in individual interests. Moreover, the acquisition of tacit knowledge has uncertain outcomes. Since tacit knowledge is associated with both physiological and cognitive structures (Merleau-Ponty, 1995), people may differ greatly in their capacity to act, even after a long period of learning. In fact, it may be years before the apprentice’s task performance comes within reach of the master’s. Worse, some people may never reach the master’s level.