Internal Network Structure As a Knowledge-protection Mechanism

Nicholas Argyres (Washington University in St. Louis)
Luis A. Rios (University of Pennsylvania)
Brian S. Silverman (University of Toronto)

Abstract : Organizational economists have explicated several mechanisms by which firms can protect proprietary knowledge, including patents, a litigious reputation, non-compete agreements, and equity-based governance in interfirm technology alliances. However, the role of within-firm organization of R&D in facilitating knowledge protection has received little attention. How do patterns of interaction among employees—notably, firm inventors—influence a firm’s ability to prevent rivals from appropriating returns to its innovations? We exploit a new measure of knowledge spillovers, and explore how knowledge protection is related to the structure of the co-invention network within a firm. Our study of patented innovation at several hundred corporations over a 20-year period reveals a positive relationship between the extent to which a firm’s inventors are connected to each other and the strength of that firm’s knowledge protection. We explore several explanations for this finding. We find no evidence that firms with more connected networks produce different kinds of innovations. We do find evidence that more connected inventor networks facilitate knowledge protection by enabling faster, pre-emptive patenting. We speculate that in connected internal networks, more inventors participate (directly or indirectly) in the creation of a given innovation, and are thus better situated to pursue follow-on research along multiple trajectories, as well as to construct more defensive “patent thickets.”

A Monitoring Theory of the Underclass: with Examples from Outcastes, Koreans, and Okinawans in Japan

J. Mark Ramseyer (Harvard University)

Abstract : When members of a minority group can monitor and constrain each other, they can leverage their internal social capital to financial gain. When they live within dense networks of personal contacts, they will more often have the information necessary to learn whether potential trade partners have kept their word and to punish those who have not. When members of a minority group lack that social capital, they not only lose these advantageous transactions but become vulnerable to their own self-appointed leaders as well. Lacking a network of close ties, they can neither monitor nor constrain others in the group. This vacuum creates an opening for opportunists to purport to act on behalf of their behalf (perhaps to obtain ethnic subsidies or other group preferences), but actually to generate hostility toward the group and divert rents to themselves. Arrovian statistical discrimination and selective out-migration follow. The opportunists raise the level of dysfunction within the group. Faced with an outside majority that treats minority members by the observed group mean, those minority members with the highest outside options will now leave and abandon the group to the opportunists. Any ethnic subsidies will offset the discrimination in part, of course. The higher the level of subsidies, the fewer the number of minority members who will find it advantageous to leave; the higher the level of subsidies, the slower the pace at which the dysfunctional minority will merge into the mainstream I illustrate these dynamics with examples from the burakumin outcastes in Japan, the Japan-resident Koreans, and the Okinawans.

Team Network and Performance: Renovating a Classic Experiment to Identify Network Effects on Team Problem Solving

Ray Reagans (MIT Sloan)
Hagay Volvovsky (MIT Sloan)
Ronald Burt (Chicago Booth)

Abstract : Available research findings illustrate a contingent association between a team’s network and performance. For a basic task, the ideal network is organized around a central individual. For a complex task, the ideal network is more decentralized and democratic. The contingent network effect has been documented for structured problems, problems with defined solutions sets. When a solution set is ill-defined, the ideal team network is unknown. To solve an unstructured problem, team members must identify and evaluate a diverse set of solutions. A team working in a decentralized network could excel at evaluating solutions, but fail to consider enough solutions, while a team working in a centralized network could discover better solutions, but struggle to evaluate their relative merits. In each case, the team could end up selecting and implementing an inferior solution to the assigned problem. It is unclear which of these suboptimal outcomes is more likely to occur and therefore which team network should be preferred. We analyze the performance of seventy-seven teams working on an unstructured problem. Teams are randomly assigned to different network conditions. Our research findings indicate centralized teams do better than decentralized teams. We also estimate the performance of teams working in networks that combine elements of centralized and decentralized networks. Teams that combine both network features are the best teams.