Preprint Finds EU Likely to Miss 2030 CO2 Benchmark; Transport Is Main Drag

A new arXiv preprint says the European Union is on course to fall well short of a 2030 climate benchmark if current carbon dioxide trends persist, with transport and mobility identified as the biggest drag while power-sector decarbonization remains broadly on track.

The paper projects EU27 CO2 emissions dropping from 2.72 gigatons in 2024 to 2.36 gigatons in 2030, a 13.2% decline. But that would still leave emissions roughly 35% above the paper’s 2030 benchmark, equivalent to an about 620 million metric ton CO2 gap. As the preprint’s abstract puts it, “We project that EU27 emissions will exceed the 2030 target by 35% (620 Mt CO2 shortfall).”

That finding comes with important caveats. The study is a preprint, meaning it has been posted publicly before peer review. It also models CO2 only, while the EU’s legally binding 2030 objective under the European Climate Law is to cut net greenhouse-gas emissions by at least 55% from 1990 levels. The paper therefore is not a legal-compliance forecast. Instead, it uses a CO2-only benchmark as a proxy and asks where emissions are heading under current observed trends.

On that measure, mobility stands out as the weakest area. The paper says the sector is making “minimal progress” and is projected to account for about 36.8% of total CO2 emissions by 2030, making it the bloc’s largest emitting sector in the study’s outlook. That matters because transport has been one of the hardest parts of the economy to decarbonize, even as climate policy has tightened in recent years.

The picture is more favorable in electricity. The paper projects stronger cuts in the power sector and links that performance to the renewable energy transition. In other words, the model suggests that the EU’s emissions trajectory is being pulled down more effectively by cleaner power than by changes in transport and mobility.

The research describes itself as a trend-extrapolation exercise rather than a forecast of how future policy could reshape emissions. It uses annual sector-level emissions and socioeconomic indicators from 2010 through 2023, with 2024 CO2 emissions serving as a starting point for the forward projections. “All indicators are observed annually from 2010 through 2023; 2024 sectoral CO2 emissions are available from Eurostat and used as an anchor for the projection chain…” the paper says.

To produce the projections, the authors used a deep-learning framework that combines a variational autoencoder, a latent-space forecasting model and an emissions predictor. The paper says it ran 10,000 Monte Carlo simulations to generate 90% “model-consistency bands.” But the authors also stress that the exercise assumes no change in policy pace or effectiveness beyond what is already visible in the historical data. In practice, that means it extends existing sectoral momentum forward rather than building in future breakthroughs.

The paper also warns against reading too much certainty into its ranges. “These bands ... are not predictive distributions over future outcomes. In particular, they do not capture uncertainty in the latent forecasting step, future structural shocks, or policy discontinuities,” it says. That limitation is especially relevant for sectors such as transport, where policy may tighten faster than past data alone would suggest, or where technology uptake could shift more abruptly.

Even so, the study’s central message is that on a CO2-only, current-trends basis, the bloc is not moving fast enough overall to close the gap by 2030, despite clearer progress in power generation.

The preprint, “Europe’s Climate Ambition Under Scrutiny: Evidence from Deep Learning Emission Projections,” was posted Aug. 19 to arXiv as arXiv:2608.18690 v1 by Jacopo Ghirri, Carlos Rodriguez-Pardo, Lara Aleluia Reis and Massimo Tavoni, with affiliations including Politecnico di Milano, CMCC Foundation/RFF-CMCC European Institute on Economics and the Environment, and NOVA University Lisbon.