Greater transparency and objectivity in climate financing was the aim of the research project by economist Dr. Anna Stünzi and her team. They developed an NLP model (Natural Language Processing Model) that can identify and classify climate protection projects.
With the HSG Impact Award, the University of St.Gallen honours projects that have a clearly recognisable impact on society - and thus embody the University's motto "From Insight to Impact". Thanks to the newly developed ClimateFinanceBERT model, stakeholders, contributors and NGOs can review climate financing on the basis of standardised criteria and thus create greater transparency.
At the heart of the research project entitled "Consistent and replicable estimation of bilateral climate finance" by Dr. Anna Stünzi and her two research partners from ETH Zürich, Malte Toetze and Florian Egli, is the Natural Language Processing Model (NLP) ClimateFinanceBERT. The model was trained by the research team to replicate the evaluation of climate finance projects and identify additional projects. AI classifies projects into either climate-relevant, (this could be for example a project to install solar panels) and non-climate-relevant (training projects where only a short module relates to climate change but the main focus is not on climate change). If a project is climate-relevant, the model determines the area where it is climate-relevant such as a protective measure against flooding or an investment in solar energy. Machine learning and human judgment complement each other in the research project: "Data alone is useless if we researchers can't draw any conclusions from it. Only in combination with scientific, contextual knowledge can the model achieve greater transparency, trust and more sustainable development in climate protection efforts," says economist Dr. Anna Stünzi.
Anna Stünzi has been a postdoctoral researcher at the SEPS-HSG (School of Economics and Political Science) at the University of St.Gallen since 2021. After studying psychology and economics at the University of Zurich, Stünzi wrote her dissertation at ETH Zurich, which negotiated the fairness principles of contributions to climate change and responsibilities in climate financing. Her research revealed that there is a lack of objective principles in the financing of climate protection projects in international cooperation. This realisation triggered this research project. Both donor and recipient countries, civil society organisations as well as research institutions or world banks could use the classification of data that the NLP model makes possible, says Stünzi. "Climate finance is always political because it is part of global negotiations that are characterised by the different interests of donor and recipient countries. Our research should feed into these negotiations and help to define the new climate financing target," says the researcher, describing her ambitions.
After the research is before the research
Anna Stünzi and her colleagues are currently investigating how the model can be adapted so that it can make even more differentiated statements about the climate relevance of a project. In addition to the classification of climate-relevant or not climate-relevant, the model could also enable a percentage gradation. A current Master's thesis at HSG is also discussing the extent to which the developed language model is specifically applicable to a donor country such as Switzerland.
The paper "Consistent and replicable estimation of bilateral climate finance" was published in the journal Nature Climate Change in fall 2022 and received a broad global response. According to Stünzi, the interest from stakeholders who want to make a difference and be transparent is enormous. The paper received the "Responsible Business Education Award" from the Financial Times in 2024. This academic research prize is awarded by the British newspaper to winners who tackle social and environmental problems in a practical way. IRCAI, an international research centre for Artificial Intelligence supported by Unesco, selected the 2023 publication as one of the ten outstanding AI solutions that are working to achieve the global Sustainable Development Goals (SDGs ). Global awards generate attention and act as a door opener. The research team was invited to speak at an event during the UN Climate Meetings in Bonn about the role of AI for climate action.
Anna Stünzi and her team are currently making the findings more visual and making the results available to the public and interested organisations on a website. "My motivation for research is always its contribution to society. I'm not just interested in publishing papers in specialist journals. I'm interested in concrete and sustainable solutions," emphasises Anna Stünzi. As a lecturer, she introduces Bachelor students to the topic of climate change and lectures on the political governance of climate financing and its negotiation context. The economist repeatedly leaves the "research bubble" in her courses in order to apply the theoretical material in practice. For example, together with her students, they added Wikipedia articles on aspects relating to climate finance in order to make knowledge acquired in the course accessible to the public.
Anna Stünzi's professional career has also led her towards practical solutions: since this summer, she has been applying her knowledge as an environmental specialist at Swiss Federal Railways. The climate finance expert will continue to work as a lecturer at the University of St.Gallen.
Anna Stünzi has been a postdoctoral researcher at the SEPS-HSG (School of Economics and Political Science) at the University of St.Gallen since 2021.
