Corporate carbon emissions and financial performance: A flexible copula-based model to address non-random sample selection

We examine the relationship between corporate carbon emissions and financial performance indicators among European-listed companies in the most polluting sectors from 2010 to 2022. We use a flexible copula-based model within a GAMLSS (Generalised Additive Models for Location, Scale, and Shape) framework to address potential non-random sample selection (Wojtyś et al., 2018). We highlight that neglecting to account for sample selection bias — only analysing firms that disclose their emissions — can lead to biased covariate estimates. Furthermore, the flexibility of the chosen approach in modelling covariate-response relationships helps in identifying complex behavioural patterns that traditional linear methods often overlook. The findings enhance the understanding of the factors affecting corporate carbon emissions. This information is crucial for policymakers and corporate governance, as it helps identify business conditions associated with lower (higher) emissions.