1. Introduction

Carbon pricing mechanisms have become central instruments in global climate policy architecture. As of 2025, carbon taxes and emissions trading systems (ETS) operate in 47 national and 36 subnational jurisdictions, collectively covering approximately 23 percent of global greenhouse gas emissions — a dramatic expansion from merely 5 percent in 2010

[1]

The divergence in findings is not merely a matter of statistical uncertainty. Metcalf and Stock

[2][4][5]

Despite the growing body of individual empirical studies, no prior work has systematically synthesized the evidence on GDP effects of carbon pricing. The most comprehensive existing meta-analysis synthesizes 80 studies and 483 effect sizes — but exclusively for emissions reductions, reporting 5 to 21 percent reductions across carbon pricing schemes

[6][7][8][9]

This paper makes four contributions. First, we present a systematic synthesis of empirical GDP estimates from ten studies employing diverse methodologies — local projections, structural vector autoregressions, vector autoregressions, difference-in-differences, panel econometrics, and computable general equilibrium models. Second, we identify the type of carbon pricing instrument and revenue recycling design as the key moderators explaining the divergence in findings. Third, we develop a simple heterogeneous elasticity model and use it, together with open-source data, to conduct scenario simulations for the Russian Federation — one of the largest economies without comprehensive carbon pricing and the world’s fourth-largest CO2Missing Mark : sub emitter as of 2022. Fourth, we formulate evidence-based policy recommendations for carbon-intensive economies.

2. Methods and Data

The analytical framework of this study comprises three components: a systematic review with narrative synthesis of existing empirical evidence, a simple model of heterogeneous GDP–carbon price elasticity, and a scenario simulation for the Russian Federation using open-source macroeconomic data.

For the review component, we identified empirical studies estimating the impact of carbon pricing on GDP, GDP growth, or aggregate economic output through a combination of targeted searches and citation network analysis. The initial pool of studies was assembled from recent comprehensive reviews of the carbon pricing literature

[8][9][6][16][14][15]

To translate the meta-analytic findings into projections for economies not yet covered by carbon pricing, we develop a simple model of heterogeneous GDP–carbon price elasticity. The model is motivated by the observation, consistent across the reviewed studies, that the macroeconomic impact of carbon pricing varies systematically with the structural characteristics of the economy — particularly energy intensity and the availability of low-carbon substitutes. Consider an economy where aggregate output Y depends on capital, labor, and energy, with energy composed of ‘clean’ and ‘dirty’ (carbon-intensive) inputs aggregated via a constant elasticity of substitution (CES) function. When a carbon price P2Missing Mark : subc is introduced, the relative cost of dirty energy rises, inducing substitution toward clean energy. The magnitude of the output effect depends on the economy’s carbon intensity, the ease of substitution, and the share of renewable energy in the energy balance. Following standard CES production theory

[20]

[1]

where ε is the GDP elasticity to carbon price, ι is the energy intensity of the economy (MJ per dollar of GDP), and ρ is the share of renewable energy in total final energy consumption (percent). The coefficient β1Missing Mark : sub captures the exposure channel: economies with higher energy intensity face larger cost shocks per unit of GDP from a given carbon price. The coefficient β2Missing Mark : sub captures the substitution channel: economies with higher renewable energy shares have greater scope for shifting away from carbon-intensive inputs.

We calibrate this model using a cross-section of 47 jurisdictions with active carbon pricing mechanisms as of 2022. For each jurisdiction, the elasticity estimate is constructed by combining the GDP growth trajectory observed after carbon pricing introduction with the nominal carbon price level, using the local projection method of Metcalf and Stock

[2]

For the Russian simulation, we assemble structural indicators from six open-access databases. GDP per capita in purchasing power parity at constant 2021 international dollars, carbon intensity of GDP (kg CO2Missing Mark : sube per 2021 PPP dollar), energy intensity (MJ per 2021 PPP dollar of GDP), renewable energy consumption as a share of total final energy consumption, trade openness, and industry value added as a share of GDP are all drawn from the World Bank World Development Indicators (WDI). Carbon pricing data come from the World Bank Carbon Pricing Dashboard (updated April 2025) and the ICAP Allowance Price Explorer. Effective carbon rates and carbon pricing scores for Russia are obtained from the OECD

[22][10][11][18]

3. Results

Table 1 presents the core set of empirical estimates identified through our review, organized by estimation method and instrument type. The ten studies span a range of methodological approaches, geographic settings, and carbon pricing instruments, yet reveal a coherent pattern when examined along the dimensions of instrument design and revenue utilization.

The most striking finding is the systematic divergence between carbon tax and ETS effects on GDP. All studies examining carbon taxes — Metcalf and Stock

[2][13][11][12][2]

The evidence for emissions trading systems tells a different story. Känzig

[4][14][15]

Känzig and Konradt

[5][16]

Revenue recycling consistently emerges as the most consequential policy design parameter (see Table 2). Where recycling has been studied, it reduces the negative GDP impact by at least 50 percent relative to lump-sum redistribution or general budget allocation. Goulder and Hafstead

[10][13][18][19]

Table 3 presents Russia’s structural economic indicators. Russia’s energy intensity of 8,46 MJ per dollar of GDP in 2021 exceeds the OECD average by approximately a factor of two, despite a substantial reduction from 12.06 MJ per dollar in 2000. Carbon intensity has likewise improved, falling from 0,570 kg CO2Missing Mark : sube per dollar in 2000 to 0,330 in 2024 — yet remains more than twice the EU average. Most notably, Russia’s renewable energy share of 3,5 percent is far below the approximately 20 percent average among countries with active carbon pricing mechanisms, severely limiting short-run substitution possibilities. The OECD carbon pricing score for Russia’s energy-use sectors stands at 7,7 percent of the emissions-weighted benchmark — indicating near-total absence of effective carbon pricing

[22]

Applying the calibrated model to Russia’s parameters yields a predicted GDP elasticity of ε = -0,028 + (-0,0065 × 8,46) + (0,0008 × 3,5) = -0,080. This value is approximately 1,9 times the sample mean elasticity of -0,043 for the 47 jurisdictions with existing carbon pricing, reflecting Russia’s high energy intensity only partially offset by a negligible renewable energy effect. Table 4 presents scenario projections based on this elasticity, cross-checked against independent CGE estimates.

The projections in Table 4 are broadly consistent with independent CGE estimates. Makarov, Chen, and Paltsev

[17][18][21]

4. Discussion

The central finding of this review is that the macroeconomic impact of carbon pricing depends more on policy design than on price level. The apparent contradiction between studies finding zero GDP effects and those documenting significant output declines dissolves once the instrument type and revenue utilization are accounted for. Carbon taxes with revenue recycling through reductions in distortionary taxes represent a fundamentally different policy intervention than cap-and-trade systems where revenue allocation is uncertain, delayed, or directed toward general government expenditure. This distinction has immediate practical implications: for economies contemplating the introduction of carbon pricing, a revenue-neutral carbon tax is the instrument most likely to achieve emissions reductions while preserving macroeconomic stability.

The Russian case illustrates the stakes involved. With an estimated GDP elasticity approximately twice the average for carbon-pricing countries, Russia faces substantially larger adjustment costs than early adopters of carbon pricing. At the same time, Russia’s high labor tax burden creates an unusually large reservoir of distortionary costs that carbon revenue could offset, making the double dividend more achievable than in economies with more efficient tax systems, as Orlov and Grethe

[18]

The findings also speak to the broader empirical puzzle regarding firm-level versus aggregate effects. Dechezleprêtre et al.

[14][15][4]

Several limitations merit acknowledgment. First, the synthesis is necessarily narrative rather than statistical, as ten studies — each employing different methods, samples, and effect definitions — do not provide sufficient degrees of freedom for formal meta-regression. The patterns we identify are consistent across studies but cannot be assigned conventional statistical significance at the meta-level. Second, the heterogeneous elasticity model is a linear approximation that abstracts from nonlinearities, threshold effects, and feedback mechanisms. Its R2Missing Mark : sup of 0,41 indicates that more than half the cross-jurisdictional variation in elasticity remains unexplained by energy intensity and renewable energy share alone. Third, the Russian simulation is based on model extrapolation from economies with active carbon pricing to one without it. Russia has not implemented comprehensive carbon pricing, and our projections cannot be validated against observed outcomes. Natural experiments in other carbon-intensive economies (Kazakhstan’s ETS, launched in 2013, or South Africa’s carbon tax, introduced in 2019) could provide valuable validation evidence as data accumulate. Fourth, our framework focuses on aggregate output effects and does not address distributional consequences across regions, industries, or household income groups. Känzig

[4]

5. Conclusion

This study has presented the first systematic synthesis of empirical evidence on the macroeconomic effects of carbon pricing, drawing on ten studies spanning local projections, structural vector autoregressions, vector autoregressions, difference-in-differences, panel econometrics, and computable general equilibrium models. Four principal conclusions emerge.

First, carbon taxes with revenue recycling show no statistically significant negative effect on GDP across all econometric studies reviewed. This consensus — spanning European carbon taxes and British Columbia’s revenue-neutral implementation — challenges the prevalent assumption of an inevitable growth-environment tradeoff. Second, emissions trading systems generate larger and statistically significant GDP declines, on the order of 0,3 percent for a shock normalized to a 1 percent energy price increase. The divergence is attributable to differences in revenue recycling, sectoral coverage, price volatility, and monetary policy interactions. Third, revenue recycling is the single most consequential design parameter, reducing negative GDP impacts by 50 percent or more across all settings where it has been studied. This finding validates the “weak” double dividend hypothesis and argues strongly for earmarking carbon pricing revenues for reductions in distortionary taxation. Fourth, Russia’s structural characteristics — energy intensity approximately twice the OECD average and a renewable energy share far below that of carbon-pricing economies — imply adjustment costs approximately double those experienced by early adopters, but revenue recycling through labor tax reductions could reduce the projected impact at 50 dollars per ton of CO2Missing Mark : sub from approximately 0,8–1,0 percent to 0,4–0,5 percent of GDP.

For policymakers in carbon-intensive economies, these findings suggest a clear design hierarchy: a carbon tax is preferable to an ETS on macroeconomic grounds; revenue neutrality through distortionary tax reduction is essential; phased introduction (beginning at 10 to 15 dollars per ton and rising predictably) allows capital stock adjustment; and implementation during economic expansion minimizes adaptation costs. For the research community, the absence of a formal meta-regression of GDP effects represents both a limitation of the current study and an opportunity: as the empirical base continues to grow with new natural experiments and longer time horizons, quantitative meta-analysis will become feasible. The integration of household-level distributional analysis, region-specific labor market effects, and long-run innovation dynamics into the macroeconomic assessment framework constitutes the most pressing direction for future research.

The additional file for this article can be found as follows: