Beyond interlocks: Board networks and corporate social responsibility performance similarity

While firms largely rely on inter-organizational information and resources to make corporate social responsibility (CSR) decisions, very few studies have delved into the heterogeneous effects of different categories of board networks. Through adopting a large language model (LLM) as an analytical tool to transform unstructured resumes into structured decision-relevant intelligence, we fulfill this research gap by constructing in-depth board networks with three distinct types: current network, past network, and non-profit organization network. Based on a dataset of Chinese listed firms from 2011 to 2020, we empirically estimate whether and how firms connected through board networks would result in CSR performance similarity. The results show that the existence and heterogeneity of board networks are positively associated with CSR performance similarity between interlocked firms. Interestingly, our findings reveal that other than current networks, past networks, and non-profit networks have a higher impact on the CSR performance similarity. Further analysis suggests that the positive association between social networks and CSR performance similarity is attenuated when connected firms face higher media coverage and higher analyst coverage. Overall, this study provides insightful evidence to leverage the effect of board networks, informing decision support for investors, regulators, and corporate strategists by revealing hidden network drivers of CSR-related decisions.