Abstract

This dissertation investigates the economics of exchange networks, focusing on the mechanisms that drive the association between network position/relationship and economic performance of actors involved in the exchanges. It argues that network position/relationship is endogenously determined—driven by actor-level heterogeneity and the conditions associated with exchanges—rather than a given, and demonstrates that understanding this dynamic can help clarify the position-performance relation. This dissertation begins with a critical review of existing work on exchange networks, pointing to empirical issues in these studies and suggesting research methodologies for addressing them. The empirical part of this dissertation explores two popular topics in the literature—network position and relational embeddedness. The first study examines how past performance influences the relative positions of actors in a network and how accounting for this endogeneity alters the often-demonstrated association between network position and performance. The collaboration network of U.S. biotech inventors shows that inventors with superior track records are more apt to form new collaboration ties, thereby occupying positions that allow them to broker across network boundaries. Controlling for past performance weakens the positive relationship between brokering position and patent performance. Further, when inventor-level heterogeneity is controlled for through inventor-fixed effects, the position-performance correlation all but disappears. These findings suggest that actor-level heterogeneity explains most of the position-performance association. The second study examines the relational embeddedness, focusing on the performance consequences of embedded exchanges and the mechanism behind the embeddedness-performance association. It identifies boundary conditions under which relational embeddedness can fail to deliver the positive outcomes found in prior studies. It also explores why maintaining embedded relationships might be optimal despite the possible hazard associated with the exchanges. The acquirer-advisor relationship data in U.S. M&A market provide evidence of negative returns to embedded exchanges. This negative impact stays significant even after accounting for possible endogeneity in relational embeddedness. Uncertainty associated with the exchanges strengthens the negative correlation. Opportunism appears to prevail when the enabling mechanisms of embeddedness likely fail. Persistent information asymmetry concerning the quality of advisory service seems to foster such malfeasance. Switching costs may partially explain why repeated exchanges nevertheless exist in this market.