On the Geographic Implications of Carbon Taxes
What this paper finds — and why it matters
Standard analyses of unilateral carbon taxes ignore the spatial reallocation of economic activity induced by the policy, leading them to overstate the costs and understate the effectiveness of such taxes. Using a multi-sector dynamic Spatial Integrated Assessment Model (S-IAM) calibrated to over 17,000 locations worldwide, the paper shows that a European Union carbon tax introduced unilaterally — if accompanied by local rebating of tax revenues to the residents of the taxing region — expands the size of the EU economy and improves global welfare. The mechanism: the carbon tax falls disproportionately on non-agricultural, energy-intensive sectors and effectively shifts part of its incidence onto trading partners via higher goods prices, while the rebate accrues only to EU residents, raising EU income per capita and attracting migrants. Under a 40 USD/tCO₂ EU tax with local rebating, EU real income rises by 0.46% in 2021 and EU population rises by 1.1%; without rebating, EU real income falls by 4.96% in 2021. EU CO₂ emissions fall by 41% by 2100, but global emissions fall by only 3% due to carbon leakage — production shifts to US, Japanese, and other unregulated regions, raising US and Japanese emissions by 12% on impact. Global real income per capita declines by 0.63% by 2100 without rebating, while global welfare improves with local rebating as economic activity concentrates in high-productivity non-agricultural regions. Rebating revenues to developing countries instead of locally slows migration to the EU, reduces the spatial efficiency gain, and deteriorates global welfare relative to the local-rebating benchmark.
Summary of a forthcoming paper, AI-assisted and human-reviewed. See the linked original for the authoritative claims and full conditions.
Why does the standard analysis miss the spatial channel, and what formal result does the paper establish?
The paper proves formally that a unilateral carbon tax with local rebating can be expansionary for the implementing region: the tax shifts part of its incidence onto trading partners (by raising the price of goods in which the region has comparative advantage) while the rebate is returned only to locals, increasing local income per capita and attracting migrants. Standard models without trade, migration, and agglomeration externalities predict only contraction. The quantitative magnitude depends on whether the pre-existing spatial equilibrium is efficient; because it is generally not — due to agglomeration externalities and knowledge spillovers — the tax-induced reallocation can improve spatial efficiency.
What is the S-IAM, and what makes it suited to quantifying this channel?
The S-IAM (Conte et al. 2021) features over 17,000 locations with positive land mass, two sectors (agriculture and non-agriculture), multi-sector technology diffusion, trade subject to geography-specific iceberg costs, and migration subject to bilateral moving costs. The model is dynamic (2000–2100) and calibrated to observed sectoral specialization, trade flows, and income levels. Energy use generates CO₂ emissions that cause temperature increases reducing agricultural productivity differentially across latitudes, integrating the climate feedback with the economic geography. Without migration and agglomeration, the expansionary channel is absent.
What happens to EU sectoral specialization under the two rebating regimes?
Without rebating: the carbon tax erodes the EU’s comparative advantage in non-agriculture (which is more energy-intensive), shifting production toward agriculture; EU non-agricultural output falls 3.44% on impact, agricultural output rises 0.86%. With local rebating: the rebate disproportionately benefits non-agricultural regions (which pay more tax), raising their income per capita and drawing workers from the agricultural EU periphery; non-agricultural output grows while agriculture declines. The result is a spatial recentralization around the EU’s non-agricultural core, strengthening the high-productivity cluster.
What are the precise global welfare effects of local versus alternative rebating?
Under a 40 USD/tCO₂ EU tax with local rebating: EU real income rises 0.46% in 2021, EU population rises 1.1%, global welfare improves. Under no rebating: EU real income falls 4.96% in 2021, global real income per capita declines 0.63% by 2100, US and Japanese emissions rise 12% on impact due to carbon leakage, EU emissions fall 41% by 2100 while global emissions fall only 3%. Under rebating to developing countries: migration to the EU slows (developing countries become relatively more attractive), the spatial efficiency gain is smaller, and global welfare declines relative to local rebating.
How does the EU carbon tax affect sub-Saharan Africa and the developing world?
Without rebating: sub-Saharan African real income per capita declines 2.36% by 2100, as the EU’s shift toward agriculture raises agricultural prices while simultaneously directing fewer imports toward agricultural exporters; South and East Asian real income per capita falls 1.35%. With local rebating: the EU’s shift toward non-agriculture reduces demand for agricultural imports, again hurting agricultural exporters. In both scenarios, equatorial and agricultural-exporting regions lose in the short-to-medium run; climate change mitigation benefits these regions in the very long run but the economic geography effect dominates over the 2100 horizon.
How do results for a US unilateral carbon tax compare to the EU case?
A US unilateral carbon tax with local rebating generates qualitatively similar results: the US economy expands, population increases, and activity concentrates in the non-agricultural core. Without an EU tax, the EU incidentally benefits from a US carbon tax — US real income per capita rises 0.17% by 2100 under a US-only tax because the tax shifts activity toward non-agricultural regions including the EU — illustrating that unilateral action generates spatial spillovers beyond standard carbon-leakage accounting.