The project advanced along three tightly connected scientific streams: (i) mapping the knowledge and innovation landscape of carbon dioxide removal (CDR) technologies, (ii) analysing firms and financial dynamics in the carbon removal sector, and (iii) developing macroeconomic models to assess the systemic implications of large-scale deployment of negative emission technologies.
First, we constructed and integrated several original datasets to study how scientific advances in CDR translate into technological innovation, policy uptake, and public attention. We systematically retrieved scientific publications on multiple CDR pathways and linked them to patents, policy documents, and media coverage. This enabled a quantitative assessment of knowledge spillovers across domains. The results show substantial heterogeneity across technologies: some methods, such as Direct Air Capture, display strong science-to-innovation linkages, while others are more visible in policy and media debates but less connected to patenting activity. These findings highlight coordination gaps between research, innovation, and policy.
To improve the identification of CDR-related innovation, we developed a novel “chemistry-proof” patent classification methodology. Instead of relying on keyword searches or standard patent codes, we first identified patents in which CO2 is chemically assigned a removal or disposal function. We then applied Large Language Models to classify inventions into detailed technological categories. This approach was applied to a high-resolution global dataset of innovations and revealed patterns that contradict earlier claims on the dynamics of technological change in negative emission technologies.
Second, the project is constructing and keeping up-to-date novel datasets. We built the most comprehensive firm-level dataset currently available on companies developing carbon removal technologies, covering more than 1,300 technology developers and more than 3000 firms across the value chain. The dataset integrates information on technology type, patents, financial accounts, ownership structures, country of origin, and more. Empirical analyses show that the CDR industry is highly fragmented and that major fossil fuel companies remain largely disconnected from most removal technologies except point-source CCS. We also constructed a transaction-level dataset of voluntary carbon removal credit purchases from 2010 to 2025, merging several registers. Event-study evidence further indicates that financial markets react negatively to firms’ announcements of carbon removal credit purchases.
Third, we worked on the development of climate-economy numerical models. We extended the Dystopian Scumpeter meeting Keynes model (DSK) to evaluate a large ensemble of policy packages combining rapid decarbonization with growth, job creation, and stability. Further, we coupled DSK and a detailed, process-based integrated assessment model (WITCH)—for the first time. This integration allows us to link detailed mitigation pathways with macroeconomic and financial dynamics, including employment, investment allocation, and carbon price transmission. The extended modelling framework provides a rigorous basis to evaluate how different policy mixes and the diffusion of NETs interact with mitigation and the macroeconomy.
