Conference Agenda

We document that firm-level emission reductions under the EU ETS program are strongly influenced by climate policy stringency. We define this stringency as the firm-specific expected compliance gap over time, which in turn is defined as the difference between projected allowances and emissions, adjusted by the allowance price and scaled by sales. By capturing both immediate regulatory pressure and firms’ expectations about future compliance obligations, our stringency measure emerges as a key determinant of firms’ decisions to reduce emissions. Sectoral analyses confirm that policy stringency remains influential across diverse industries, irrespective of financial constraints or technological barriers. We use the sectoral model to assess progress toward the EU’s emission goals by 2030. While the phased reduction of free allowances associated with the Carbon Border Adjustment Mechanism (CBAM) drives emissions reductions in covered sectors, it is insufficient to drive the economy-wide reduction required to meet this goal. Meeting the required regulatory pressure across the economy would entail an immediate carbon price of around euro 125 per tonne.

This paper investigates the transmission channel between the primary and secondary markets in the European Union Emissions Trading System (EU ETS), focusing on the behaviour of auction winners. As auctions become the primary mechanism for regulators to inject permits into the system, understanding how auction winners influence secondary markets is increasingly important. Using micro-level data, we analyse auction concentration, identify auction winners and examine their net trading positions and permit banking behaviour. Our findings show that auctions are highly concentrated and dominated by a small group of frequent winners, many acting as net sellers on secondary markets while simultaneously increasing their accumulated holdings of permits. These dynamics suggest potential for the exercise of market power from auction winners and highlight the need for enhanced regulatory oversight as the EU ETS evolves.

Innovation is central to the transition to low-carbon production, yet empirical evidence on how the combination and interaction of multiple climate policy instrument relate to large accelerations in green technological activity remains limited. We study green innovation using patent family data covering 14 green technology fields across 22 countries from 2000 to 2019.

We develop an empirical framework that combines a difference-in-differences–type model with a machine-learning–based variable selection procedure to identify positive structural breaks in green patenting, defined as sustained increases that exceed model-implied trajectories. This approach identifies 108 positive breaks across countries and technologies.

We then link these break episodes to climate mitigation policies drawn from a comprehensive policy database. For individual policy instruments and broader policy categories, we evaluate whether their presence is associated with an increased probability of observing a positive break. In addition, we introduce information-theoretic measures to characterize the cohesion (how typical is the policy mix compared to other countries?) and proximity (novel vs. established combinations) of policy mixes, and show that these measures are predictive of break occurrence.

Across specifications, pricing-based policies are associated with systematically higher probabilities of observing strong increases in green patenting, while subsidy-based and regulatory instruments exhibit more heterogeneous associations across technologies and contexts. The results provide new evidence on the relationship between climate policy design and large, discrete changes in green innovation activity.

We measure the labor orientation of green innovation-whether patented low-carbon technologies are oriented towards automating existing tasks or towards the creation of new work. Using 1.3 million U.S. patents filed between 1980--2022 and CPC climate tags (Y02 and Y04S), we construct a patent-level labor-orientation index from semantic similarity between patent text and occupational task descriptions and newly emerging work titles, following Autor et al. (2024). We find heterogeneity across green technologies: green ICT and smart grids are relatively labor-augmenting, while adaptation is more labor-augmenting. Within renewable and electric vehicle patents, the labor-augmenting share declines over time. We then relate patent orientation to diffusion in a patent-by-technology panel that matches patents to country-level deployment in physical units: higher lagged deployment predicts a lower labor-augmenting share conditional on rich sets of country and technology-specific fixed effects and trends. While these estimates are not causal, they are consistent with learning-by-doing whereby scaling and standardization tilt green inventive activity towards more labor-saving content.