Whether the EU Emissions Trading System (EU ETS) has caused carbon leakage in the cement and steel industries remains empirically contested. A 2016/2017 rolling-cointegration study by Boutabba and Lardic found that the carbon price raised net imports of both sectors in a subset of subperiods between June 2005 and July 2015, but used net imports as a proxy for leakage, did not identify the mechanisms behind this time variation, and covered only the beginning of Phase III. Later work has measured emissions displacement and production relocation directly at firm and manufacturing level (mostly finding little or no leakage) and a 2026 preprint finds no leakage for steel in the newest phases, but no cement-specific plant-level measurement exists, the goods-trade evidence remains mixed according to a 2026 systematic review, and the mechanism of the time-varying carbon-price effect and possible asymmetric responses are untested. The problem is to establish, with direct sector-level evidence, whether, how much, and through which channels the EU ETS causes leakage in these two sectors across its phases.

Background

The EU Emissions Trading System (EU ETS) is the European Union's cap-and-trade program for greenhouse gas emissions, launched in 2005 and organized in successive trading phases: Phase I (2005-2007, trial), Phase II (2008-2012, coinciding with the Kyoto Protocol commitment period), Phase III (2013-2020), which made auctioning the default allocation method and introduced a Union-wide emissions cap - electricity producers have been obliged to buy the allowances they need since 2013, while industrial installations considered to be at significant risk of carbon leakage have continued to receive free, benchmark-based allocation - and Phase IV (2021-2030), which further tightens the cap. Allowances (EUAs) trade at a carbon price that raises production costs of emission-intensive firms. 'Carbon leakage' denotes the situation in which a climate policy in one jurisdiction causes emissions to rise elsewhere, either through imports displacing domestic production of carbon-intensive goods or through relocation of production or investment to unregulated countries. Cement and steel are the canonical leakage-risk sectors: they are energy-intensive, emission-intensive, and exposed to international trade, and they have received most of the free allowances distributed to shield them from leakage.

Ex-ante computable-general-equilibrium and partial-equilibrium models produced widely conflicting predicted leakage rates, from negligible to over 100 percent. Ex-post econometric evidence has also been conflicting. Branger, Quirion and Chevallier (2016) derived an import-demand equation from an analytical model and estimated it with constant-coefficient methods (ARIMA regressions and Prais-Winsten feasible generalized least squares, two standard time-series estimators) for cement and steel over Phases I-II, finding no significant effect of the carbon price on net imports (imports minus exports) and concluding there was no evidence of short-run leakage. Boutabba and Lardic (2017) re-examined the same question allowing the relationship to vary over time: they used a rolling cointegration approach with fully modified OLS (FMOLS), i.e. they re-estimated a long-run equilibrium regression of net imports on the carbon price, demand indicators, energy prices and the exchange rate over rolling subperiods of the June 2005 - July 2015 sample, testing in each window whether the variables share a stable long-run relationship (cointegration). They found the carbon-price coefficient positive and statistically significant in 5 of the evaluated subperiods for cement and 14 for steel, with average magnitudes of roughly 0.02 and 0.06 million tons of additional net imports per USD increase in the EU allowance price respectively, implying a small, time-varying leakage risk that is larger for steel.

Several limitations of this evidence define the present problem: net imports are only a proxy for leakage, since trade flows neither distinguish emissions displacement nor detect production or investment relocation; the reported estimates do not identify why the carbon-price effect appears in some subperiods and not others (candidate mechanisms include capacity constraints, demand fluctuations, and policy announcements); the sample ends in July 2015, barely two years into Phase III, so the effect of the revised Phase III rules could not be assessed; and possible nonlinearities or asymmetric responses of trade to carbon-price increases versus decreases are not addressed by the reported linear rolling estimates. Sector- and plant-level data are available for these two sectors: verified installation-level emissions are recorded in the European Union Transaction Log, the official EU ETS registry recording allowance transactions and annual verified emissions, alongside plant-level production and firm ownership and trade data.

Problem Statement

Does the EU Emissions Trading System cause carbon leakage in the cement and steel industries, measured directly rather than through net-import proxies? The target is this causal question itself: whether, how much, and through which channels the EU ETS carbon price causes leakage in each of the two sectors across its phases. Three aspects of that question remain unresolved in the current evidence: (1) direct measurement and causal attribution - measuring emissions displacement, production and investment relocation, and embodied-carbon trade (trade flows measured by the CO2 emissions embodied in the traded goods rather than their physical tonnage) for the two sectors using direct sector- or plant-level data, covering Phase III (2013-2020) and Phase IV for as much of that period as data allow rather than only the 2005-2015 sample, and attributing the measured displacement to the EU ETS carbon price rather than to other trends, so that the result is a causal estimate of the leakage the policy causes and not only a descriptive account of emissions and trade aggregates; (2) mechanisms - why the carbon price significantly raises net imports in some subperiods and not others in the rolling-cointegration evidence (capacity constraints, demand fluctuations, policy announcements, or other channels), of which the disagreement with the constant-coefficient null results is part; and (3) functional form - whether the response of trade and production to the carbon price is nonlinear or asymmetric between carbon-price increases and decreases. The question therefore asks for direct evidence produced for each aspect rather than assembled from the existing literature: for (1), measurement of all three channels in both sectors, including the cement measurement that no study has yet provided; for (2), a mechanism identified from that evidence rather than a list of untested candidates; and for (3), a test of nonlinearity or asymmetry rather than an assumed linear response.

References

  1. Mohamed Amine Boutabba and Sandrine Lardic, "EU Emissions Trading Scheme, competitiveness and carbon leakage: new evidence from cement and steel industries", Annals of Operations Research 255(1-2), 47-61 (print 2017; first published online June 2016), DOI: 10.1007/s10479-016-2246-9, https://doi.org/10.1007/s10479-016-2246-9
  2. Frederic Branger, Philippe Quirion and Julien Chevallier, "Carbon Leakage and Competitiveness of Cement and Steel Industries Under the EU ETS: Much Ado About Nothing", The Energy Journal 37(3), 109-136 (2016), DOI: 10.5547/01956574.37.3.fbra, https://doi.org/10.5547/01956574.37.3.fbra
  3. Helene Naegele and Aleksandar Zaklan, "Does the EU ETS cause carbon leakage in European manufacturing?", Journal of Environmental Economics and Management 93, 125-147 (2019), DOI: 10.1016/j.jeem.2018.11.004, https://doi.org/10.1016/j.jeem.2018.11.004
  4. Antoine Dechezleprêtre, Daniel Nachtigall and Frank Venmans, "The joint impact of the European Union emissions trading system on carbon emissions and economic performance", Journal of Environmental Economics and Management 118, 102758 (2023), DOI: 10.1016/j.jeem.2022.102758, https://doi.org/10.1016/j.jeem.2022.102758
  5. Jonathan Colmer, Ralf Martin, Mirabelle Muûls and Ulrich J. Wagner, "Does Pricing Carbon Mitigate Climate Change? Firm-Level Evidence from the European Union Emissions Trading System", Review of Economic Studies 92(3), 1625-1660 (2025; first published online May 2024), DOI: 10.1093/restud/rdae055, https://doi.org/10.1093/restud/rdae055
  6. Sean Healy, Katja Schumacher and Wolfgang Eichhammer, "Analysis of Carbon Leakage under Phase III of the EU Emissions Trading System: Trading Patterns in the Cement and Aluminium Sectors", Energies 11(5), 1231 (2018), DOI: 10.3390/en11051231, https://doi.org/10.3390/en11051231
  7. Gianluca Bidinot, Alfonso Dufour and Simone Varotto, "Carbon Leakage in the EU Steel Sector", SSRN preprint (2026), DOI: 10.2139/ssrn.6135206, https://doi.org/10.2139/ssrn.6135206
  8. Filippo Maria D'Arcangelo and Marzio Galeotti, "Environmental policy and investment location: The risk of carbon leakage in the EU ETS", Energy Policy 202, 114587 (2025), DOI: 10.1016/j.enpol.2025.114587, https://doi.org/10.1016/j.enpol.2025.114587 (earlier version: SSRN, DOI: 10.2139/ssrn.4240927)
  9. Han Chen, Monica Di Gregorio and Jouni Paavola, "Carbon leakage in emissions trading systems: a systematic review and meta-analysis of ex-ante and ex-post evidence", Climate Policy, 1-17 (published online 12 February 2026), DOI: 10.1080/14693062.2026.2627750, https://doi.org/10.1080/14693062.2026.2627750
  10. Clara Ulmer, "Free allowances and the risk of carbon leakage in the beginning of the third phase of the EU ETS", SSRN preprint (2021), DOI: 10.2139/ssrn.3880946, https://doi.org/10.2139/ssrn.3880946