Under the Green Canopy: bringing up to date public climate finance determinants analysis with AI

Pierre Beaucoral*,
UCA CNRS IRD CERDI
pierre.beaucoral@uca.fr
*Indicates Equal Contribution
Paper Supplementary Code arXiv

Abstract

Climate finance is critical for addressing the multifaceted challenges of climate change, encompassing mitigation, adaptation, and environmental sustainability. This study aims to renew the analysis of a critical part of climate finance determinants' allocation across these dimensions and accurately estimate bilateral public climate finance flows using an advanced machine learning approach. ClimateFinanceBERT (Bidirectional Encoder Representations from Transformers) is employed to classify development finance projects, distinguishing those that contribute to climate mitigation, adaptation, and environmental objectives. By examining a comprehensive dataset of development finance projects (OECD CRS) and replicating a recent research on climate public aid determinants, this study identifies key factors influencing the allocation of climate finance. This work updates significant patterns in climate finance distribution. This research contributes to the growing field of climate finance by offering a robust analytical framework for assessing the determinants of climate finance and proposing a scalable solution for monitoring financial flows aimed at addressing climate change in its entirety. The insights gained have important implications for policymakers and stakeholders striving to understand and optimize the allocation of climate finance to support global sustainability and resilience goals

🌍 Key Insights from the Data 📊

🌱 Climate finance plays a crucial role in addressing climate change by supporting mitigation 🌡️, adaptation 🌊, and environmental sustainability 🌿 initiatives. This research employs ClimateFinanceBERT 🤖, an advanced machine learning model, to classify development finance projects and analyze the determinants of bilateral public climate finance allocation.

📄 By leveraging text-based classification of project descriptions from the OECD Creditor Reporting System (CRS), this study refines existing estimates of climate finance flows, offering a more precise alternative to conventional classification methods like Rio markers 📌. The findings suggest that traditional methods overestimate climate finance 💰, potentially due to strategic misclassification ("greenwashing" 🟢 or "aidwashing" 🌍💸).

📈 A key econometric analysis using the double hurdle model 📉 reveals distinct allocation patterns for mitigation and adaptation finance. While mitigation finance follows standardized allocation processes driven by emission reduction targets ♻️, adaptation finance remains highly influenced by recipient vulnerability 🌍 and donor strategic interests 🎯. Interestingly, governance factors 🏛️ play a stronger role in adaptation projects, highlighting the importance of institutional capacity 📚 in climate finance distribution.

Key Contributions

  • ✅ More Accurate Climate Finance Estimates: AI-driven classification provides a restrictive and transparent assessment of climate aid flows.
  • ✅ New Insights on Climate Finance Determinants: Historical ties and self-reported commitments (e.g., NDCs) have less influence than previously thought.
  • ✅ Policy Implications: A data-driven approach to monitoring and optimizing climate finance allocation is crucial for ensuring that funds genuinely support sustainability and resilience goals.

This research contributes to the broader discussion on climate finance transparency and efficiency, providing policymakers with scalable, AI-powered tools for financial accountability and climate action tracking.

BibTeX


        @unpublished{beaucoral2025,
  title={Under the Green Canopy: bringing up to date public climate finance determinants analysis with AI},
  author={Beaucoral, Pierre},
  year={2025}
}