How Far do Network Effects Spill Over? Evidence from an Empirical Study of Performance Differentials in Interorganizational Networks

Organizations join interorganizational networks in the hope of gaining exposure to learning opportunities, and accessing valuable extramural resources and knowledge. In this paper we argue that participation in interorganizational networks also reduces performance differentials among organizational nodes. We examine three alternative mechanisms capable of sustaining this prediction. The first (strength of ties) operates at a strictly local level defined in terms of dyadic relations linking organizations. The second mechanism (social proximity) operates at an intermediate – or meso level of interdependence defined in terms of membership in overlapping cliques into which interorganizational networks are typically organized. The third mechanism (structural equivalence) is global and pertains to jointly occupied network positions. The objective of this paper is to examine at which of these levels network effects operate to reduce performance differentials among members of interorganizational networks. Our empirical analysis of performance differentials between hospitals in a regional community supports the following conclusions: (i) performance spillover effects are highly differentiated and vary significantly across network levels; (ii) organizations occupying similar positions within the network are more similar in terms of performance; (iii) joint membership in multiple sub-groups (or cliques) reduces performance differentials up to a limit; after this limit is reached, the performance of organizational partners begins to diverge; (iv) the strength of direct collaboration between organizational partners does not necessarily reduce interorganizational performance differentials. The results of the study are new because available research on interorganizational networks says little about the range of network effects, i.e., about how far the performance spillover effects that operate through networks propagate throughout organizational fields and communities. These results are also consequential because they suggest that network effects on performance differentials are sensitive to the specification of network boundaries.