Leveraging natural language processing techniques to explore the potential impact of the EU’s Carbon Border Adjustment Mechanism (CBAM)

Debates on controversial policies often stimulate extensive discourse, which is difficult to interpret objectively. Political science scholars have begun to use new textual data analysis tools to illuminate policy debates, yet these techniques have been little leveraged in the international business literature. We use a combination of natural language processing, network analysis and trade data to shed light on a high-profile policy debate—the EU’s recently enacted Carbon Border Adjustment Mechanism (CBAM). We leverage these novel techniques to analyze business inputs to the EU’s public consultation, differentiating between different types of organizations (companies, trade associations, non-EU actors) and nature of impact (direct, indirect, potential). Although there are similarities in key concerns, there are also differences, both across sectors and between collective and individual actors. Key findings include the fact that collective actors and indirectly affected sectors tended to be less concerned about the negative impacts of the new measure on international relations than individual firms and those directly affected. Firms’ home country also impacted on their positions, with EU-headquartered and foreign-owned companies clustering separately. Our research highlights the potential of natural language processing techniques to help better understand the positions of business in contentious debates and inform policy making.