Network formation is often characterized by homophily—the tendency that agents connect with
others who have similar attributes. However, while most agents are homophilous, others may be
heterophilous, namely, aiming to create ties with dissimilar agents. This study finds evidence
supporting this hypothesis for the first time in the literature by applying random coefficient
models to information-sharing network data for Vietnamese small and medium-sized enterprises
(SMEs). One possible interpretation for this heterophily is that firms can obtain more useful and
performance-improving information from those dissimilar—as opposed to similar—to
themselves, as suggested by certain social network studies.