Double agency and co-evolution for two-mode networks, with an application to corporate interlocks and firms’ environmental performance

Three extensions of the Stochastic Actor-oriented Model for two-mode networks are proposed. The first represents tie changes based on agency for both node sets. The second is a model for selection and influence in co-evolution of two two-mode networks, where the first is a membership network which serves to represent connections between groups making choices of activities in the second network. The third consists of new effects for categorical nodal covariates, representing the heterogeneity between nodes that often occurs in two-mode networks. As an illustration, selection and influence processes are studied in the co-evolution of the director interlock network of a set of 141 large European companies and the two-mode network of environmental policies of these companies, for the period from 2018 to 2022. Similarity between environmental policies is expressed in two ways: endorsing the same policy items, and correlation between the number of items endorsed. The findings show that declared pro-environmental practices spread through directorate interlocks and directors’ selection is based on peer referral.