Cutting Free Allocation Cuts Emissions: Evidence from Phase IV of the EU ETS (Working Paper)

Phase IV (2021 to 2030) of the EU ETS tightened the cap and reformed free allocation. In 2021, a narrower carbon leakage list and benchmark updates produced a one-off drop in realized free allocation. Using 2015 to 2023 data on Phase III and Phase IV incumbents from the EUTL, we construct a time-invariant, installation-level dose from the 2020 to 2021 change and estimate a continuous difference in differences with installation and year fixed effects. The event study shows stable pre trends, and allocation levels are essentially flat after 2021. Cutting free allocation reduces emissions. In our preferred specification, the slope after 2021 is −0.359, implying about −3.5% lower emissions for a +10 percentage-point larger loss of free allocation. From the perspective of the independence property, this dose response pattern indicates that, in Phase IV manufacturing with activity-linked free allocation, changes in the allocation schedule are not neutral for verified emissions. Among exposed firms with positive pre-2020 banks, we find no detectable attenuation. Aggregating to the BvD-entity perimeter, the emissions response replicates in direction and timing. In standard Orbis accounting aggregates over 2021–2023 we do not detect short-run movements within the ETS-linked perimeter. Non-ETS plants of the same entity are unobserved, so we neither establish nor rule out carbon leakage. The evidence is consistent with stronger effective carbon pricing at the installation margin under activity-linked allocation in Phase IV and provides quantitative input on the environmental effect of phasing down free allocation and the transition to CBAM.