Geographical clusters, alliance network structure, and innovation in the United States biopharmaceutical industry

I examine the effects of firms' cluster membership on their alliance network structure, and how firms' absorptive capacity moderates the relationship between alliance network structure and innovation. Little is known regarding the inter-relationship between cluster membership, network structure and innovation. This study bridges this gap by first establishing the endogenous nature of network structure with respect to cluster membership and then by studying the moderating effect of absorptive capacity for the alliance network structure and innovation relationship.; I contribute to the strategic management literature in several important ways. First, I clarify the implications of cluster membership on network structure by including two competing explanations: complementary and substitution mechanisms. Contrary to the popular belief that cluster membership does not matter, I find that it does matter in the study of the US biopharmaceutical industry. My findings show that firms' location within a cluster area does not substitute for their strategic choices specifically for their alliance strategies. Second, I theoretically argue and then empirically demonstrate that network structure is an endogenous phenomenon with respect to cluster membership. Third, I demonstrate that when controlled for endogeneity with respect to cluster membership, alliance network structure and innovation relationship is positively moderated by firms' absorptive capacity. In contrast to prior literature, I find that the main effect of firms' structural holes on innovation is not significant when controlled for endogeneity. This finding is important given the mixed findings for structural holes and innovation relationship in previous studies. Finally, to the best of my knowledge, in the strategic management literature this study is the first study to introduce an exponential regression model with Generalized Methods of Moments (GMM) estimation that accounts for both the endogenous nature of independent variables and the count nature of dependent variable.