Moving beyond the Dyad: Addressing network endogeneity in strategic alliance formation using exponential random graph models (ERGM)

Gulati and Gargiulo's (1999) examination of alliance formation established the now taken-for-granted argument that endogenous characteristics of network structure predict subsequent tie formation. Using dyadic-level regression of alliance ties on composite dyadic measures of alliance ties and firm-specific characteristics, their results suggested that alliance formation is explained by strategic interdependence, structural differentiation, and also relational, positional, and structural embeddedness. Among the many studies in strategic management that have built upon this seminal work, Ahuja, Polidoro & Mitchell (2009) generated similar baseline analyses. We replicate both studies with alliance data from the chemical and semiconductor industries between 1991-2000, and then apply exponential random graph modeling (ERGM) to clarify endogenous predictors. We illustrate that network results are susceptible to industry context as well as methodological approach.