COVID-19, Green Deal and recovery plan permanently change emissions and prices in EU ETS Phase IV

The paper investigates how recent shocks and policy changes affect cumulative emissions and allowance prices in the EU Emissions Trading System (EU ETS) Phase IV. Three major developments are examined: (i) a temporary negative demand shock for allowances due to COVID-19 lockdowns; (ii) uncertain positive or negative demand changes from overlapping policies financed by the NextGenerationEU recovery package; and (iii) a permanent tightening of allowance supply via a more ambitious 2030 ETS reduction target (−61% vs 2005) under the EU’s Fit for 55 Package. Under a fixed cap, such shocks would only affect prices, not cumulative emissions (the waterbed effect). However, since 2018, the Market Stability Reserve (MSR) with an invalidation rule ties supply to the total number of allowances in circulation (TNAC), making cumulative emissions endogenous. The research question is how the timing of shocks, their announcement, and the endogenous year when the ‘waterbed’ is sealed (TNAC < 833 MtCO2) interact to change cumulative emissions and prices. The study formalizes these mechanisms with the concept of waterbed leakage and quantifies impacts under the 2018 design and the proposed Fit for 55 design.

Prior work established the classic waterbed effect under cap-and-trade and how the 2018 MSR quantitatively altered EU ETS dynamics (e.g., Perino 2018; Bruninx et al. 2020). Studies have analyzed the 2009 recession’s impact on prices, the role of Clean Development Mechanism credits, and renewable policies in depressing EUAs (Koch et al. 2014). Literature on overlapping policies shows that concurrent measures can shift ETS outcomes through banking and invalidation, potentially leading to backfiring (Rosendahl 2019; Perino, Ritz & van Benthem 2020). Theoretical contributions discuss ‘green paradox’ mechanisms and carbon leakage across time and space (Jensen et al. 2015; Eichner et al. 2011). Recent analyses assess the MSR under exogenous shocks like COVID-19 (Azarova & Mier 2021; Gerlagh et al. 2020) and propose alternative stability mechanisms, including price-based triggers akin to California’s system (Perino et al. 2021), while highlighting emissions uncertainty in market design (Borenstein et al. 2019). This paper builds on and extends these strands by quantifying direct and indirect effects of overlapping policies under both current and proposed EU ETS/MSR designs and by introducing numerical estimates of waterbed leakage across timing scenarios.

The authors develop a stylized partial-equilibrium EU-ETS-MSR simulation model with a representative, rational, price-taking, and risk-neutral firm that optimizes intertemporal abatement and allowance banking with no borrowing. Given a price path, the firm emits until marginal abatement cost (MAC) equals the EUA price; it minimizes discounted allowance procurement costs subject to banking constraints and allowance market balance. Equilibrium EUA prices are found via an iterative price-search algorithm (ADMM-based), updating prices based on imbalances between demand and MSR-adjusted net supply until convergence. Abatement is represented by time-invariant marginal abatement cost curves: p_t = β (E_t − q_t)^γ with baseline emissions E_t = 1900 MtCO2, discount rate 8%, γ ∈ [0.5, 4.2] (step 0.05). For each γ, β is calibrated to match the average 2019 EUA price (24.7 €/tCO2), given observed 2019 emissions and ETS state. The MSR rules are fully implemented, including intake rates, TNAC thresholds, invalidation, and timing of supply adjustments (4/12 of intake realized Sep–Dec of year t and 8/12 Jan–Aug of year t+1). Waterbed leakage is defined as the ratio of the change in cumulative emissions to a marginal change in allowance demand. It is estimated by adding a 1 MtCO2 demand change in specific years (2020–2050) to calibrated equilibria and comparing cumulative emissions, ensuring the marginal shock does not change the duration of the punctured waterbed. Direct and indirect effects are separated: the direct effect is computed by applying MSR rules to a hypothetical emissions path where the demand shock directly shifts emissions one-for-one, while the indirect effect arises from price-path adjustments due to anticipation and banking. Analytical approximations for direct invalidation are provided for both the 2018 and Fit for 55 MSR designs, with special handling when TNAC is within 833–1096 MtCO2 in the latter, which can yield more-than-proportional invalidation (up to 1.67 for certain timings). To assess real-world shocks, five policy scenarios are simulated across the MACC set: (1) No pandemic, −40% target in 2030 (ETS −43% vs 2005), rebased for Brexit and including intra-EEA aviation; (2) Scenario (1) plus a COVID-19 negative demand shock totaling 720 MtCO2 (2020–2025 linearly declining); (4) Scenario (2) plus Fit for 55 design changes (cap rebasing incl. maritime, LRF 4.2% to zero supply, 24% intake through 2030, dynamic intake when 833 ≤ TNAC ≤ 1096, invalidation threshold 400 MtCO2); (3) and (5) add overlapping policies to (4) changing allowance demand by −100 or +100 MtCO2/year in 2021–2030, respectively (e.g., renewables support vs electrification). Outputs include cumulative emissions, timing of waterbed sealing, and EUA prices (e.g., in 2021).

• COVID-19 negative demand shock (total ~0.72 GtCO2 over 2020–2025) largely translates into lower cumulative emissions, with limited effect on 2021 EUA prices when firms continue banking beyond 2021.
• Raising ambition to −55% by 2030 (Fit for 55) increases EUA prices to about 45–94 €/tCO2 “today” (e.g., 2021), consistent with observed all-time highs (~86 €/tCO2 in Dec 2021), and reduces cumulative ETS emissions to 14.2–18.3 GtCO2 versus 23.5–33.1 GtCO2 under the −40% target and 2018 design.
• Cumulative cap (2020–end ETS) without invalidation: 35.5 GtCO2 under −40%; falls to 21.7 GtCO2 under −55% (despite adding maritime). Invalidation ensures cumulative emissions are always below the cap; in a pre-pandemic scenario, invalidation ranges from ~2.4 GtCO2 (waterbed seals 2022) up to ~12.5 GtCO2 (seals 2050). In the reference case, moving to −55% increases invalidation by ~2.6 GtCO2 (from ~4.8 to ~7.5 GtCO2).
• Waterbed leakage (WL) depends on timing and announcement: under the 2018 MSR, WL for marginal policies is <1 and indirect effects always oppose direct effects; under Fit for 55, direct WL can equal or exceed 1 (up to 1.67) for pre-sealing shocks when TNAC enters 833–1096 MtCO2, due to timing of MSR supply adjustments and TNAC definition.
• Under Fit for 55, pre-announced policies before sealing can have positive indirect effects that amplify the direct effect (total WL up to ~2.71). Pre-announced policies executed after sealing backfire strongly (negative WL down to ~−2.36), as indirect effects dominate and can be reinforced.
• Scenario illustrations of overlapping policies (±100 MtCO2/year, 2021–2030 under −55%): when the waterbed remains punctured longer, most of the shock maps into cumulative emissions (WL near 1). WL can reach ~1.11 (negative demand) or ~1.22 (positive demand) for certain sealing years (2033 and 2030, respectively). When sealing occurs sooner, indirect backfiring reduces effects on cumulative emissions and increases price sensitivity.
• EUA prices in 2021 under raised ambition (−55%) span roughly 44.9–93.6 €/tCO2; price impacts of overlapping policies are larger when sealing occurs sooner because less of the shock is absorbed via invalidation.
• The timing of sealing (endogenous) is pivotal and shaped by MSR design, shocks, overlapping policies, and the convexity of MAC curves; higher expected future abatement costs induce more current abatement and banking, prolonging puncture and increasing invalidation (self-reinforcing effect).

The findings show that with the MSR invalidation rule, cumulative emissions in the EU ETS are endogenous and shaped by the timing and anticipation of shocks and overlapping policies. The three determinants—when a policy affects allowance demand, when it is announced, and when the waterbed seals—jointly govern the direct invalidation and indirect price/banking responses. Under the 2018 MSR, indirect effects typically counteract direct effects and can cause backfiring for policies enacted near or after sealing. Under the Fit for 55 MSR design, more-than-proportional supply adjustments can make direct effects exceed one-to-one and can flip the sign of indirect effects before sealing, amplifying overall impacts. This implies that overlapping policies implemented while TNAC remains positive can materially change cumulative emissions, not just prices, and that pre-announced measures scheduled for after sealing can backfire without compensating actions (e.g., voluntary cancellation). The results are policy-relevant: (i) they align rising EUA prices with tighter 2030 targets, informing stakeholders including those affected by the proposed carbon border adjustment; (ii) they provide a quantitative map (waterbed leakage) for assessing marginal policies by timing; and (iii) they highlight how design choices in the MSR (intake thresholds, timing of supply adjustments, invalidation thresholds) critically influence the magnitude and direction of effects on both emissions and prices.

The paper contributes a quantitative framework for assessing how shocks (COVID-19), increased ambition (Fit for 55), and recovery-plan-induced overlapping policies affect cumulative emissions and allowance prices under EU ETS Phase IV. It introduces waterbed leakage to summarize the emissions impact of marginal demand changes and shows that, under Fit for 55, direct effects can exceed unity and pre-announced measures may be amplified or backfire depending on timing relative to sealing. Key outcomes include: limited price impact but lower cumulative emissions from COVID-19; substantially higher EUA prices and lower cumulative emissions under −55%; and strong timing sensitivity of overlapping policies with potential for backfiring without allowance cancellation. Policy design suggestions include refining the MSR to reduce threshold-induced oscillations (e.g., lowering the 1096 MtCO2 threshold post-2030) and adjusting the timing or accounting of supply adjustments (frontloading or including future adjustments in TNAC) to cap direct effects at one-to-one. Future research should examine price-based stability mechanisms, incorporate myopic behavior and uncertainty (risk aversion), explore richer abatement representations, and link ETS sectors with non-ETS sectors and the broader economy.

The analysis uses a deterministic, perfect-foresight, partial-equilibrium model with a representative, risk-neutral, price-taking firm and stylized, time-invariant MAC curves calibrated to 2019 prices. It does not predict exact price/emissions paths but presents ranges contingent on when sealing occurs. Behavioral frictions (myopia), uncertainty, and risk aversion are not modeled and could shorten or lengthen puncture duration and alter invalidation volumes. The timing of the one-off supply adjustment needed to implement the 4.2% LRF in Phase IV is not analyzed. Scenario representations of overlapping policies are simplified (generic ±100 MtCO2/year). Parameter choices (e.g., baseline emissions, MAC curvature range) and institutional assumptions (e.g., MSR timing, TNAC accounting) influence quantitative outcomes.