Carbon Taxes and CO2 Emissions: Sweden as a Case Study (Andersson)

Summary

Julius Andersson's "Carbon Taxes and CO₂ Emissions: Sweden as a Case Study" (American Economic Journal: Economic Policy 11(4), November 2019, pp. 1–30) is a landmark: it is, in the author's words, "the first to find a significant causal effect of carbon taxes on emissions." Sweden was among the first countries in the world to price carbon, introducing a carbon tax in 1991 at about US$30/tonne CO₂ (since raised to today's ~US$132/tonne, one of the highest rates anywhere) and, in March 1990, extending its 25% value-added tax to transport fuels. Andersson uses the synthetic control method — constructing a "synthetic Sweden" from a weighted combination of comparable OECD countries that did not price carbon — to estimate what Swedish transport emissions would have been absent the reform. It is the methodological benchmark for the whole carbon-tax evaluation literature and a natural flagship for this outcome, distinct from (and complementary to) the British Columbia evidence already on the claim page.

Key Findings

• Transport CO₂ fell almost 11%. "After implementation, carbon dioxide emissions from transport declined almost 11 percent… relative to a synthetic control unit constructed from a comparable group of OECD countries," with the largest share attributable to the carbon tax itself.[1]
• The carbon tax alone did most of it. Disentangling the tax from the VAT, Andersson estimates a post-treatment reduction of 6.3% from the carbon tax alone, or about 1.5 million metric tons of CO₂ in an average year (roughly −0.31 tonnes per capita).[1]
• Carbon taxes are more potent than their price tag suggests. "The carbon tax elasticity of demand for gasoline is three times larger than the price elasticity" — i.e., a krona of visible, policy-labelled carbon tax cuts fuel use about three times as much as an equivalent movement in the market price. This is the same "salience" phenomenon found for British Columbia (Rivers & Schaufele), now documented for a long-running European tax.[1]
• The methodological warning. Because of that salience gap, "policy evaluations of carbon taxes, using price elasticities to simulate emission reductions, may thus significantly underestimate their true effect."[1] Simulated (ex-ante) estimates that plug ordinary price elasticities into models systematically understate what carbon taxes actually achieve.
• Robustness. The result survives an extensive battery of placebo tests — "in-time" (introducing a fake tax a decade early), "in-space," "leave-one-out," and full-sample permutations — supporting a causal, not merely correlational, reading.[1]

What It Supports

• Carbon pricing cuts emissions — modestly, and without wrecking growth — this is the flagship quasi-experimental case for the emissions half of the claim: a significant, causally-identified reduction (≈11% transport CO₂; 6.3% from the tax alone) using the cleanest available method, complementing the British Columbia natural experiment with a second, longer-running, higher-rate example.
• Pigouvian Taxation — direct empirical vindication of the Pigouvian claim that pricing an externality reduces it, and evidence that the policy-labelled charge outperforms an equal untagged price.

What It Cuts Against / Honest Limits

• Magnitudes are modest — and the honesty stays. An ~11% transport-sector reduction (6.3% from the tax alone) over the study window is real but not transformative, and it is a transport-sector figure, not economy-wide — consistent with this wiki's insistence that carbon prices "bend the curve" without, at the rates tried, being sufficient alone.
• Sector-specific. The clean identification is for transport fuels; it does not establish an economy-wide counterfactual, echoing the Pretis (2022) caution already on the claim page that BC's aggregate effect is weaker than its transport-fuel effect.
• Below the social cost of carbon. The 1991 US$30/tonne launch rate (and even much of the subsequent path) sat below most estimates of the social cost of carbon, so the measured effect reflects modest-to-moderate prices, not a full Pigouvian correction.

Bears On

• Outcome: Carbon pricing cuts emissions
• Concept: Pigouvian Taxation
• Concept: Ecological Georgism — the atmospheric-commons framing the evidence underwrites
• Research: The EU ETS reduced CO2 emissions despite low prices (Bayer & Aklin) — the cap-and-trade counterpart

See Also

• Carbon pricing cuts emissions
• Pigouvian Taxation
• The EU ETS reduced CO2 emissions despite low prices (Bayer & Aklin)
• The Macroeconomic Impact of Europe's Carbon Taxes (Metcalf & Stock)

Sources

1. Julius J. Andersson (2019), "Carbon Taxes and CO₂ Emissions: Sweden as a Case Study," American Economic Journal: Economic Policy 11(4), 1–30. DOI 10.1257/pol.20170144 — used for every key finding on this page; abstract verified verbatim and body text (the 6.3%-carbon-tax-alone figure, 1.5 Mt/yr, −0.31 t/capita, the US$30→US$132 rate path, the 1990 VAT extension, the three-times-price-elasticity salience result, and the placebo-test battery) verified against the AEA-hosted text and the author's LSE PhD-thesis chapter this session; the aeaweb PDF returned HTTP 403 to this wiki's egress, so the full text was read via the open AEA article record and the thesis. AEA · LSE thesis (open)