Does Carbon Pricing Impose Unequal Economic Costs?
Känzig (2026) asks how carbon pricing affects emissions, the broader economy, and households at different income levels. He examines European Union carbon-market regulations, high-frequency carbon-price data, macroeconomic indicators, and household data. He finds that a carbon-policy shock large enough to raise energy prices by 1% reduces emissions by about 0.75%, but also lowers GDP by roughly 0.3%, reduces industrial production by nearly 1%, and raises unemployment by 0.15 percentage points. Lower-income households experience larger reductions in spending and income than higher-income households. Indirect effects through employment and income account for about two-thirds of the decline in consumption.
Why This Article Was Selected for The Policy Scientist
Carbon pricing has become increasingly important as governments seek to reduce emissions without imposing excessive economic costs or widening existing economic disparities. Understanding who ultimately bears these costs is especially timely as carbon markets expand and governments consider more ambitious climate policies. Känzig makes an important contribution by showing that the burden extends beyond higher energy bills to employment, wages, and household consumption. The unusually rich combination of high-frequency market, macroeconomic, and household data is a major strength. Particularly compelling is the causal identification strategy, which uses unexpected regulatory changes as an instrument rather than relying on conventional regression. Published in the American Economic Review, one of economics’ premier journals, the study provides unusually strong evidence, although the European setting warrants caution when generalizing to jurisdictions with different energy markets, labor institutions, or carbon-pricing systems.
Full Citation and Link to Article
Känzig, D. R. (2026). The unequal economic consequences of carbon pricing. American Economic Review. Advance online publication.
Central Research Question
The study asks how carbon pricing affects emissions, economic activity, and the distribution of economic costs across households. Carbon pricing is intended to make greenhouse-gas emissions more expensive, encouraging firms and consumers to reduce carbon-intensive activity. Yet its broader consequences are difficult to establish because carbon prices and economic conditions influence one another. The central empirical challenge is therefore to isolate changes in carbon prices caused by policy rather than by underlying economic conditions.
Känzig examines three related questions. First, does tighter carbon pricing actually reduce greenhouse-gas emissions and stimulate low-carbon innovation? Second, what happens to output, employment, inflation, consumption, investment, and wages? Third, are these costs distributed evenly across households? The final question is especially important because conventional analyses often emphasize the direct burden of higher energy prices. The study investigates whether indirect effects operating through employment, wages, and income substantially increase the economic burden and whether these effects differ by household income.
Previous Literature
The paper builds on several strands of environmental and macroeconomic research. Traditional assessments of carbon pricing have relied heavily on structural models, including Nordhaus (1992) and Goulder (1995), as well as the macroeconomic framework developed by Golosov et al. (2014). These models establish the fundamental tradeoff between reducing emissions and the economic resources required to achieve those reductions, but their quantitative conclusions depend on assumptions about technology, household behavior, policy design, and the disposition of carbon-pricing revenue.
A second literature uses observed carbon-pricing policies to estimate effects on regulated firms. Important contributions include Fowlie, Holland, and Mansur (2012), Martin, De Preux, and Wagner (2014), Marin, Marino, and Pellegrin (2018), Dechezleprêtre et al. (2019), and Cui et al. (2021). This research generally finds substantial reductions in emissions but comparatively modest effects on firm employment or profitability. Other studies, including Andersson (2019), Metcalf (2019), and Metcalf and Stock (2020, 2023), examine broader economic consequences and similarly find limited evidence of major adverse effects on output or employment.
Känzig departs from much of this literature by estimating economy-wide effects and explicitly incorporating indirect responses transmitted through prices, wages, employment, and aggregate demand. Methodologically, the identification strategy draws on high-frequency approaches developed for monetary policy, including Kuttner (2001), Gürkaynak, Sack, and Swanson (2005), Gertler and Karadi (2015), and Nakamura and Steinsson (2018). The study also extends distributional research that generally finds modest regressive effects when concentrating primarily on households’ direct exposure to higher energy prices.
Data
The empirical setting is the European Union Emissions Trading System (EU ETS), established in 2005 and covering roughly 40 percent of EU greenhouse-gas emissions. Känzig identifies 126 regulatory events between 2005 and 2019 involving such matters as emission caps, allowance allocations, auctions, and international credits. After excluding events coinciding with potentially confounding major economic news, the analysis uses 114 regulatory events.
High-frequency futures prices for EU emission allowances provide the critical information for identifying unexpected carbon-policy changes. These data are combined with European macroeconomic measures including energy and consumer prices, greenhouse-gas emissions, industrial production, unemployment, interest rates, stock prices, and oil prices. The principal macroeconomic sample extends from 1999 through 2019, ending before the COVID-19 pandemic.
The analysis supplements these aggregate data with measures of GDP, consumption, investment, wages, and low-carbon patenting. Patent applications come from PATSTAT and identify technologies related to climate-change mitigation and adaptation. Cross-country data permit analysis of differences among advanced European economies. Detailed United Kingdom household data are then used to determine how expenditure and income responses differ across the income distribution. The breadth of the data is important because it permits the study to connect policy shocks simultaneously to environmental, macroeconomic, technological, and household outcomes.
Methods
The central methodological innovation is a high-frequency identification strategy designed to isolate plausibly exogenous carbon-policy shocks. Carbon prices ordinarily cannot simply be treated as causes of subsequent economic changes because regulators may adjust climate policy in response to economic conditions. Känzig instead measures changes in carbon futures prices within narrow windows surrounding regulatory announcements. Because prevailing economic information should already be incorporated into market prices immediately before an announcement, unexpected price movements around the announcement provide a measure of policy news.
The author further removes variation predictable from prior macroeconomic, financial, oil-market, and climatic information. The resulting refined policy-surprise measure is then used as an external instrument in a structural vector autoregression. This allows the study to estimate dynamic causal effects rather than simple correlations between carbon prices and economic outcomes. The first-stage robust F-statistic for the baseline instrument is 16.85, providing evidence that the instrument has sufficient predictive strength.
For additional outcomes, Känzig extracts the estimated carbon-policy shock and employs local projections. Extensive sensitivity analyses test alternative event selections, identifying assumptions, sample periods, lag structures, estimators, and specifications. Alternative methods relaxing some assumptions of the baseline VAR yield broadly similar conclusions.
Findings/Size Effects
Carbon pricing produces substantial environmental effects. A restrictive carbon-policy shock normalized to increase energy prices by 1 percent reduces greenhouse-gas emissions by approximately 0.75 percent at the peak. The emissions decline persists longer than the contraction in economic activity. Low-carbon patenting also increases, indicating that higher carbon prices encourage technological responses rather than producing emissions reductions solely through lower current production.
These environmental gains accompany measurable macroeconomic costs. Industrial production falls by nearly 1 percent, while real GDP declines by approximately 0.3 percent at its peak. Investment falls by roughly 1 percent and consumption by approximately 0.3 percent. The unemployment rate increases by about 0.15 percentage points, real wages decline, and consumer prices rise by nearly 0.2 percent. Stock prices fall by more than 2 percent. Economic activity subsequently recovers more quickly than emissions, producing a longer-term reduction in the emissions intensity of production.
A particularly important result concerns the mechanism producing the consumption decline. Energy accounts for somewhat less than 10 percent of European household expenditures. Consequently, even if energy demand were completely unresponsive to price, a 1 percent increase in energy prices should directly reduce consumption by no more than about 0.1 percent. The estimated decline is approximately 0.3 percent. Thus, roughly two-thirds of the aggregate consumption response arises from indirect economic effects rather than the direct increase in household energy bills.
The distributional analysis shows that these indirect effects are also unequal. Lower-income households reduce expenditures substantially more than higher-income households. Part of this difference reflects their larger energy expenditure shares, but income effects are also important. Lower-income households experience larger income losses because they are more likely to work in sectors whose activity contracts following carbon-policy shocks. These are not necessarily the most carbon-intensive industries; many produce discretionary goods and services whose demand declines as higher energy costs and falling incomes reduce household spending.
The estimated economy-wide marginal cost of eliminating an additional ton of carbon dioxide is approximately €107. This substantially exceeds the roughly €12 average EU ETS allowance price during the sample, suggesting that market allowance prices capture firm-level abatement costs more readily than broader economic adjustments involving consumption, employment, wages, and output.
Conclusion
The study concludes that carbon pricing successfully reduces greenhouse-gas emissions and promotes low-carbon innovation, but its economic consequences extend considerably beyond the direct cost of more expensive energy. Changes in household income, employment, wages, consumption, and aggregate demand constitute central mechanisms through which carbon policy affects the economy.
The distributional findings alter the interpretation of carbon pricing's incidence. Lower-income households face greater direct exposure because energy constitutes a larger portion of their budgets, but this channel alone explains only part of their larger response. Indirect income and employment effects substantially amplify the differences across households. Consequently, analyses limited to changes in energy expenditures can materially understate both the aggregate economic effects and their distribution across income groups.
More broadly, the findings demonstrate the importance of analyzing climate policy within an economy-wide framework. Carbon pricing simultaneously affects emissions, technological innovation, prices, production, labor markets, and household income. The evidence therefore connects environmental policy with macroeconomics and household inequality rather than treating these as separate outcomes. By identifying unexpected regulatory changes and tracing their consequences across multiple levels of the economy, the study provides evidence that the full economic incidence of carbon pricing depends not simply on who consumes energy, but also on how the broader economy adjusts after the price of carbon changes.
Comments