Alternative carbon price trajectories can avoid excessive carbon removal

The study addresses how the shape of carbon price trajectories affects temperature overshoot and the scale of carbon dioxide removal (CDR) required to meet the Paris Agreement’s goal of holding warming well below 2 °C (and implications for 1.5 °C). Many existing mitigation scenarios assume exponentially growing carbon prices (Hotelling rule), which, combined with availability of CDR, lead to peak-and-decline temperature pathways and large net-negative emissions later in the century. The authors hypothesize that once CDR is available, the assumption of a finite exhaustible carbon budget underpinning the Hotelling rule is invalid, and that alternative price paths can achieve targets with lower reliance on CDR and reduced risks. The purpose is to derive and evaluate economically reasonable alternative price trajectories that maintain compliance with temperature limits while minimizing long-term economic, technical, and institutional risks associated with massive CDR deployment.

Prior work highlighted the prevalent reliance on large-scale CDR in stringent scenarios and associated sustainability and feasibility concerns. Studies proposed pathways with lifestyle changes, demand reduction, and lower energy demand to reduce BECCS needs, but without directly addressing the economic drivers of CDR reliance. Research showed that lower discount rates can reduce budget overshoot and hence net-negative emissions, and that applying budgets only until carbon neutrality aligns with peak-warming budget concepts. Cost-benefit analyses (e.g., Golosov et al.; Dietz & Venmans) suggest optimal carbon taxes rising at GDP growth rates (below the discount rate) or near-linear profiles (Nordhaus), contrasting with exponentially rising Hotelling paths. The authors build on this by focusing on the cost-effectiveness framework with explicit temperature-consistent carbon budgets and by examining how alternative carbon price shapes affect overshoot and CDR demands.

The analysis uses the global multi-regional energy–economy–climate integrated assessment model REMIND (version 2.1.0). REMIND models each region as a Ramsey-type macroeconomy (maximizing utility from consumption) coupled to a detailed energy system with fossil and renewable resources, conversion technologies (including CCS), and trade of energy carriers, a composite good, and emission permits. Annual regional CCS deployment is limited to 0.5% of total storage capacity, capping global CCS use near ~20 Gt CO2/yr. Temperature outcomes are computed using MAGICC. Scenarios: Six global scenarios assume current policies until 2020 and a uniform global carbon price thereafter, adjusted to meet a global CO2 budget of 1070 Gt CO2 (2018 onward), consistent with a 67% chance of limiting warming to 2 °C.

• OPT (Optimal): Endogenous carbon price path ensuring the cumulative CO2 budget is never exceeded; carbon price rises at the discount rate until net-zero, then adjusts to maintain the target without unnecessary net negatives.
• H2C (Hotelling to Constant): Carbon price increases at the discount rate (Hotelling) until net-zero, then remains constant.
• HOS (Hotelling Overshoot): Exponential Hotelling price throughout with a constraint on end-of-century cumulative CO2 (budget met by 2100), allowing interim overshoot.
• HBL (Hotelling Below): Exponential Hotelling price with a constraint that the carbon budget is never exceeded (peak cumulative constraint); no overshoot in cumulative emissions.
• GDP (GDP growth): Carbon price increases at the GDP growth rate.
• LIN (Linear): Approximately linear carbon price path. The starting price equals the OPT price in 2025; the annual increment is iteratively set so the carbon budget is never exceeded, yielding slightly higher prices than OPT in the first half of the century and lower later, with similar 2025 and 2100 levels. Implementation details: Hotelling price pathways increase at 5% per year; starting levels (and in LIN, annual increments) are iteratively calibrated to meet the budget constraints. For the exponential price path, both peak-cumulative and end-of-century cumulative constraints are tested (HBL vs HOS). Additional analysis considers a tighter 1.5 °C-compatible budget (requiring net-zero around 2050) to assess generality. CDR representation: Three CDR options are included alongside fossil/industrial CCS: (i) Afforestation/reforestation (A/R) from MAgPIE 4.0 with land-use dynamics, incentivized via a GHG price consistent with SSP2-2.6; only 50% of the energy-system carbon price is applied to A/R to reflect permanence risks. (ii) BECCS for electricity, hydrogen, gas, or liquid fuels with technology-specific capture rates; bioenergy supply from MAgPIE; an additional 100% tax on bioenergy prices reflects unmodeled sustainability externalities (e.g., biodiversity, water). (iii) DACCS parameterized from literature (Broehm et al.): 10 GJ/tCO2 heat and 2 GJ/tCO2 electricity; heat supplied by natural gas or H2; if gas is used, 90% capture is assumed for resulting CO2; capital costs of ~100 $/tCO2 (excluding energy and storage).

• Optimal price shape: In a cost-effectiveness setting with a binding peak cumulative CO2 budget, the optimal carbon price increases at the discount rate until net-zero emissions and then flattens (even dips) to avoid unnecessary net-negative emissions. Approximations (H2C, GDP, LIN) yield similar mid-century outcomes to OPT for the 2 °C budget, whereas Hotelling paths (HOS, HBL) drive excessive late-century CDR and higher prices.
• Near-term policy signal: Peak temperature constraint determines near-term emission reductions largely independent of price-path shape. To stay well below 2 °C, 2025 carbon prices must be high: OPT ~36 $/tCO2 vs HOS ~23 $/tCO2 (55% higher in OPT), inducing stronger early abatement.
• CDR scale-up timing: Early higher prices induce earlier CDR deployment without increasing its peak: 2030–2050 CDR scale-up averages ~115 MtCO2/yr in OPT (reaching ~2.7 GtCO2/yr in 2050) vs ~60 MtCO2/yr in HOS.
• Hotelling’s long-run effects: By 2100, Hotelling paths yield carbon prices 2.4× (HOS) to 3.8× (HBL) higher than OPT and roughly twice the CDR deployment, producing large net-negative emissions and peak-and-decline temperatures.
• Economic costs (2 °C budget): Cumulative discounted consumption loss (2020–2100) relative to continued current policies: OPT 0.85%; HBL 1.06% (more costly due to high late-century prices and CDR); HOS 0.66% but with temporary overshoot of the 2 °C limit and associated risks.
• Financial risks of net negatives: If CDR is rewarded at the full contemporaneous carbon price, net-negative emissions could require up to ~10.5 trillion US$ (NPV) in financing, with annual needs reaching ~8.7% of gross world product in 2100 in Hotelling scenarios (notably HBL). Capping CDR remuneration (e.g., via auctioning) to ≤250 $/tCO2 could reduce NPV needs to <3 trillion US$ and ~1.2% of GWP in 2100.
• 1.5 °C budget: With net-zero required by ~2050, only the optimal path avoids net-negative emissions and limits CDR to <10 GtCO2/yr; alternative paths reach up to ~20 GtCO2/yr and show peak-and-decline temperatures. Shapes other than optimal perform worse on costs (typically 15–37% higher than OPT) or temperature, and the optimal price drops sharply after net-zero, which other shapes cannot approximate.
• Robustness: Alternative non-Hotelling paths (H2C, GDP, LIN) approximate OPT well under the 2 °C budget, achieving similar emissions, peak temperature, and consumption losses while avoiding excessive long-run CDR and extreme carbon prices.

Findings demonstrate that when CDR is available, the classical Hotelling logic of exponentially rising carbon prices ceases to be economically optimal for meeting temperature-consistent peak carbon budgets. Hotelling paths drive very high late-century prices and large net-negative emissions, creating technical, institutional, and financial risks and inducing peak-and-decline temperature profiles. In contrast, a high early carbon price that ensures rapid emissions cuts, followed by only moderate increases (or stabilization) after net-zero, maintains well-below-2 °C warming with smaller CDR, lower long-term prices, and reduced risks. Near-term emissions reductions are primarily set by the peak temperature constraint, making required 2030 reductions robust across reasonable non-Hotelling price shapes. Policy implications include prioritizing strong near-term pricing, designing mechanisms to limit CDR remuneration (e.g., auctions) to actual costs, and avoiding policy designs that implicitly require massive net-negative emissions later. The results align with cost-benefit insights favoring slower-than-discount-rate price growth and provide a cost-effective blueprint that reduces implementation risk while respecting temperature limits.

Choosing carbon price trajectories that start high and then rise only moderately after reaching net-zero can keep warming well below 2 °C while substantially reducing reliance on CDR, long-run carbon prices, and associated economic, technological, and institutional risks. For 2 °C budgets, several simple alternative price shapes (constant after net-zero, GDP-growth, linear) approximate the optimal path closely in terms of emissions, peak temperature, and costs. For tighter 1.5 °C budgets, only the optimal path avoids substantial net-negative emissions. Future research should broaden scenario spaces beyond Hotelling paths, further integrate damages below target temperatures and co-benefits (e.g., air quality) into cost-effectiveness assessments, explore risk and uncertainty in setting near-term prices, and improve endogenous treatment of lifestyle and demand-side transformations. Policy design should also consider caps or auction mechanisms for CDR remuneration to limit financial exposure from net-negative emissions.

The optimal carbon price path derived is cost-optimal only for achieving the 2 °C target under the modeled assumptions and excludes explicit valuation of damages below 2 °C and co-benefits from mitigation (e.g., air pollution). Model-based results depend on REMIND’s structure and parameterizations, including assumptions about CDR potentials, CCS deployment caps, bioenergy sustainability costs, and DACCS performance and costs. The analysis assumes globally uniform carbon pricing and idealized implementation, which may not capture real-world political, institutional, or heterogeneity constraints. Under tighter (1.5 °C) budgets, non-optimal price shapes perform poorly, indicating sensitivity to target stringency. Uncertainty in climate response and socio-economic pathways remains and could justify higher near-term prices than suggested by pure cost-effectiveness.