Firm complexity and carbon emissions: Evidence from textual analysis

Utilizing a novel measure of firm complexity derived from textual analysis by Loughran and McDonald (2024), we examine the relationship between firm complexity and carbon emissions in the United States from 2005 to 2021, considering the roles of AI investment levels and firm leverage as mediating factors. We find that more complex firms are inclined to invest in AI technologies associated with increased carbon emissions. Additionally, these firms tend to have higher leverage, further amplifying their carbon emissions. Our results indicate that this relationship is more pronounced when oil price uncertainty is low and in manufacturing and high-tech industries. The relationship is also stronger among smaller firms, as they face more significant challenges in adopting ESG policies. Finally, firms with higher returns on assets, long-term debt, and gross profit demonstrate a more substantial positive relation between complexity and carbon emissions.