Unveiling greenwashing in Colombian manufacturing: A machine learning approach: Revelando el lavado ecológico en la manufactura colombiana: un enfoque de aprendizaje automático

Carolina Henao-Rodríguez, Jenny Paola Lis-Gutiérrez, Harold Delfin Angulo Bustinza

Producción científica: Artículo CientíficoArtículo originalrevisión exhaustiva

Resumen

Greenwashing is the misleading use of environmental claims to market non-eco-friendly products, a growing concern as environmentally conscious consumers seek eco-friendly options. This study aims to identify factors that allow the detection of “greenwashing” practices in Colombian companies across various industrial sectors holding certifications such as ISO14001, the Colombian environmental seal, the district environmental excellence program, or international/regional environmental certifications. We employed a quantitative, cross-sectional methodology using three machine learning algorithms (Decision Tree Classification, Random Forest, Logistic Regression) and the double machine learning technique to identify causal effects and significant variables. Additionally, a greenwashing indicator was formulated, considering two variables: (a) Presence of a comprehensive solid waste management plan, and (b) Presence of a water usage and conservation program. Our findings reveal substantial variations in the adoption of environmental management practices across diverse industrial sectors and regions in Colombia. It is noteworthy that our research emphasizes that: (i) companies with an environmental contingency plan and a monitoring program are less likely to engage in greenwashing; (ii) expenses for personnel dedicated to environmental protection activities and the investment and expenses related to air and climate protection also reduce the likelihood of greenwashing; (iii) companies in the manufacturing of non-metallic mineral products sector are more likely to engage in greenwashing. It was also found that a significant proportion of Colombian industrial companies with environmental certifications do not fully embrace sustainable practices, implying limited effectiveness of environmental certifications in the country (56,09%). This research contributes to the taxonomy of greenwashing by delving into various types identified in the literature. It sheds light on selective disclosure, emphasizing that mere certification without tangible actions can constitute greenwashing. The broadened concept of decoupling now includes not only meeting shareholder expectations but also failing to meet those set by certifications. Attention deflection involves diverting focus toward certifications rather than real investments in environmental practices. Deceptive manipulation occurs when certifications create a misleading perception without concrete actions. Additionally, dubious authorizations and labels are revealed, showcasing the vulnerability of eco-labels to fraud. The study enriches existing categories, viewing certifications as potential instruments for greenwashing, reinforcing the need for substantial, measurable actions to validate sustainability claims.
Idioma originalEspañol (Perú)
PublicaciónResearch in Globalization
Volumen8
N.º100196
EstadoIndizado - jun. 2024

Palabras clave

  • Decision Tree
  • Double machine learning
  • Greenwashing
  • LASSO
  • Logistic regression
  • Machine Learning
  • Random Forest

Citar esto