Projected Effects of the Foreign Pollution Fee Act of 2025

Carbon–intensity-based border measures, in which a country imposes tariffs on imported goods according to their carbon emissions from each unit of production, have emerged as a key element of the trade and climate policy conversation in the United States and abroad. Proponents of such measures in the US Congress have cited multiple potential benefits, including supporting domestic competitiveness, reducing emissions in US-consumed goods, and reducing the emissions intensity of domestic manufacturing.

There have been few detailed studies of the effects of US border measures based on carbon intensity, despite their current policy relevance and long history in the carbon pricing literature. In part, this is due to inherent challenges in the analysis of such border measures, including limited data on the carbon intensities of products worldwide, the complexity of trade relationships, and the myriad potential responses to such measures by actors throughout the global economy.

Here, we use a global economic model to assess the effects of a border measure stylized after the Foreign Pollution Fee Act of 2025 (FPFA) introduced to the 119th Congress by Senators Bill Cassidy (R-LA) and Lindsey Graham (R-SC). The FPFA would impose tariffs on a set of covered products including iron and steel, aluminum, cement, glass, fertilizer, hydrogen, solar products, and long-duration storage, based on their relative carbon intensities compared to US production.

We find that the FPFA would:

• Shift US imports toward countries with lower carbon intensity manufacturing: Imports for covered products are reduced from countries facing the carbon tariffs (e.g. China, Mexico, and India) and increase from countries exempt from the tariffs (e.g. the European Union, United Kingdom, and Japan) due to their lower carbon intensity of manufacturing for those products.
• Increase US manufacturing of covered products: The carbon tariffs protect US manufacturing of covered products, thereby raising their output: cement (+9.1 percent), aluminum (+7.9 percent), iron and steel (+7.4 percent), metal products (+3.7 percent).
• Raise revenue: Annual revenues from the policy are projected to be $2.8 billion (in 2024$) in the first year and total $33.3 billion over ten years.
• Have a minimal effect on global emissions: Changes in trade patterns and increased US manufacturing would reduce the embodied emissions of US imported goods, but reshuffling of global trade for covered products offsets the reduction of US imported emissions. US emissions increase due to greater domestic manufacturing. The net effect is that global emissions are relatively unchanged by the policy.
• Reduce output in downstream industries: Industries such as construction and transportation equipment manufacturing use covered products as inputs, exposing them to higher costs. US production from such downstream industries is projected to fall by 0.2–2 percent.

Border measure policies have been featured in climate policy proposals for decades. They have generally been included alongside carbon pricing proposals to deter the “leakage” of carbon emissions from a country imposing a carbon policy to countries without the same level of policy. In 2023, the European Union (EU) imposed the Carbon Border Adjustment Mechanism (EU CBAM), which provides such a symmetric approach to border measures: for certain energy-intensive goods (e.g. steel and cement), EU importers are required to surrender carbon certificates under obligations that mirror those imposed on EU producers by the EU’s Emissions Trading System (EU ETS).

The US does not have a climate policy analogous to the EU ETS. However, for many energy-intensive, heavily traded goods, the carbon intensity of US manufacturing is lower than major US trading partners like China, Mexico, and India (David et al., 2025; Rorke et al., 2025; DeFilippo and Wise, 2025). This US “carbon advantage” has led to multiple congressional trade proposals using carbon intensity as the basis for tariffs to maintain US competitiveness, reduce emissions in imported goods, and incentivize domestic manufacturing with lower greenhouse gas intensities. Import tariffs could also provide a source of revenue to pay for other policy priorities, adding to their relevance in current US legislative conversations.

Analyses of border measures based on carbon intensity have generally taken one of two approaches to address the complexity of global trade and specific data limitations related to carbon intensity of manufacturing. The first is a bottom-up, “partial equilibrium” modeling approach (e.g. David et al., 2025) that uses detailed, product-level information to assess price and trade effects for each product. Partial equilibrium approaches do not account for interactions across sectors and may not be capturing the full substitution among suppliers from the different countries facing different tariff rates.

The second is a top-down modeling approach that captures inter-industry “general equilibrium” effects, such as how changes in the prices and quantities of steel affect other sectors such as motor vehicles, aircraft, and construction, as well as the effect on aggregate GDP and growth over time. The ability to capture such effects comes at the expense of sectoral detail.

For this study, we have adopted the general equilibrium approach, using the Global Economic Model (GEM) described in Cao et al. (2024). GEM is built on the Global Trade Analysis Project (GTAP) database covering 160 countries and 65 industries. To reduce the complexity of the GEM, each of the G20 countries are represented individually—while the remaining countries are grouped into nine regions (Table S1a Citations which include the letter “S” refer to figures and tables in the supplementary document. View this document here. )—and GEM distinguishes 30 economic sectors (Table S1b). The model simulates how the tariffs change all prices in the economy (not just covered products) and how producers change their input mix to respond to the changed costs.

The imposition of US tariffs based upon carbon intensity would have myriad effects. Of particular relevance for the policy discussion are effects on: 1) Patterns of US imports; 2) US and foreign output for each sector; 3) US government revenue; 4) Overall economic output; and 5) Country-level and global emissions. Here, we use GEM to assess these metrics for a policy stylized after the FPFA.

The FPFA was reintroduced by Senators Cassidy and Graham for the 119th Congress on April 8, 2025, marking a major revision from its original 2023 introduction. In brief For a more detailed discussion of the FPFA’s legislative approach see https://www.rff.org/publications/issue-briefs/foreign-pollution-fee-act-design-elements-options-and-policy-decisions/. , the FPFA would assess an ad valorem fee on imports of covered products from countries with a carbon intensity greater than 110 percent of the US performance benchmark for that product. The ad valorem fee is calculated based upon a formula, which increases across three tiers of relative carbon intensity (Figure 1). The US benchmark is calculated as the average carbon intensity for covered products produced domestically. Outside of specific circumstances, such as within international partnership agreements, importers are assigned the national average carbon intensity for a given sector for the country of manufacture.

Specific covered products are identified in the legislation by Harmonized Tariff Schedule (HTS) code, which include iron and steel, aluminum, cement, glass, fertilizer, hydrogen, solar products, and battery inputs. Greenhouse gas emissions included in carbon intensity calculations include direct emissions from manufacturing (often referred to as Scope 1 emissions) as well as indirect emissions from consumed electricity, steam, heating, or cooling (Scope 2); and emissions linked to so-called precursors (i.e. inputs into a manufacturing process, such as iron ore) and transportation (Scope 3). The FPFA does not include any compliance obligation for domestic manufacturers.

3.1. Calculated Carbon Intensities and Ad Valorem Rates

We use the GTAP dataset to calculate country- and sector-specific carbon intensities for each of the relevant GTAP sectors based on scope 1 and scope 2 emissions. Covered products are first mapped into the corresponding six GTAP sectors which encompass those products (e.g. cement and glass products are mapped to the non-metallic mineral products sector, Table S2a). For each of the six GTAP sectors containing an FPFA-covered product, we next calculate the percentage of US imports of covered products For example, the GTAP sector ‘non-ferrous metals’ used to represent aluminum also includes additional metals such as copper and nickel. as a percentage of total imports from that GTAP sector: ferrous metals (82 percent); non-metallic mineral products (46 percent); fabricated metal products (41 percent), non-ferrous metals (32 percent); chemicals (4 percent); and machinery and electrical equipment (2 percent) (Table S2b). Based on this analysis, we do not model tariffs for two GTAP sectors (chemicals, machinery and electrical equipment) even though they encompass fertilizer, solar panels, and batteries, which are covered products under the FPFA. These products represent a very small percentage of their respective total GTAP sector, so applying the tariffs to the full sector would substantially overstate the effect of the tariffs levied on those products.

The highly aggregate nature of GTAP sectors encompassing the list of covered products, as well as the data being from 2017, means that the estimated carbon intensities are averages which will deviate from more detailed analyses of specific covered products using more recent data. For each covered sector, a relatively small set of countries provides most of the imports to the United States (e.g. Canada, Mexico, and China; Figure 2).

The calculated carbon intensities for each country and sector are used to calculate country- and sector-specific ad valorem fees according to the formula in the legislation (Table 1). According to the tariff formula, country-specific tariffs escalate with increasing carbon intensity based on three tiers, but all countries that are within 10 percent of the performance benchmark are fully exempt. This distinction leads to many European countries, as well as South Korea and Japan, having zero tariffs across all covered sectors in our model. More detailed analysis at the product level could result in non-zero tariffs for specific products (e.g. a specific steel product). The non-market economies of China and Russia would face the maximum 200 percent rate for nearly all covered sectors. India, South Africa, and Indonesia would face 100 percent tariffs for all covered sectors.

3.2. Trade and US Production

The modeled tariffs imposed by the FPFA are projected to increase US production and alter overall import volumes in covered sectors (Figure 3a). The rise in US production is substantial for cement (9.1 percent), aluminum (7.9 percent), iron and steel (7.4 percent), and fabricated metal products (3.7 percent), with a corresponding decrease in aggregate imports for covered sectors. In absolute terms, reductions in imports for covered sectors are generally not fully matched by higher US output in those sectors, indicating an overall decrease in US consumption for those sectors. The exceptions are the iron and steel and cement sectors, which have higher output since they provide inputs to the other expanding covered sectors.

The sharp differences in tariff rates, ranging from zero for some countries and sectors to 200 percent for others, is projected to result in a dramatic rearrangement in trade patterns for covered sectors. Trade shifts toward countries manufacturing with lower carbon intensity under the FPFA (Figure 3b), with US imports increasing from the EU, UK, South Korea, and Japan for all covered sectors. Imports of steel and aluminum from Canada are reduced by 70 percent and 90 percent respectively, while imports of cement and glass nearly double. Imports from Mexico are reduced sharply across all covered sectors, while imports of covered goods from China, India, Russia, and South Africa are virtually eliminated due to the very high tariffs.

Though US manufacturing increases for covered sectors due to the protective effect of the tariffs, downstream US sectors using inputs from covered sectors are negatively affected and reduce output, due to increased input prices driven by the tariffs (Figure 3c). Upstream US sectors, such as mining and electricity production, increase production slightly to meet greater domestic demand from covered sectors.

3.3. US Government Revenues

The tariffs imposed by the FPFA would raise government revenue to the extent that the United States continues to import covered goods from higher carbon intensity countries. The modeled high tariffs imposed by the FPFA on higher carbon intensity countries such as China and Russia virtually eliminate imports of covered products, and with them potential tariff revenues, from those countries. The lack of tariffs on lower carbon intensity countries in the EU and elsewhere provides a strong incentive to increase imports from those countries but would not raise tariff revenue. Given the tiered nature of the FPFA, the tariffs are projected to reduce, but not eliminate, US imports from countries with higher carbon intensity production. The tariffs are projected to raise revenues of $2.8 billion (in 2024$, Table 2) for the first year of the policy, primarily from continued trade in covered goods with Canada and Mexico (Table S4b). With continued global economic growth, tariff revenues increase annually to reach $3.8 billion in year 10, totaling $33.3 billion over the 10-year budget window used by the Congressional Budget Office for estimating revenue effects of proposed legislation. Out of necessity, we apply the tariff to the full GTAP sector in which the covered products fall to understand the effects on trade. One consequence of this simplification is that it overestimates the downstream effects from those aggregate sectors compared to the real-world effects where only the specific HTS codes would be affected and not the entire sector. For calculating revenues, we account in part for this by multiplying by the fraction of US imports for those HTS codes for the year 2017 taken from the US Census Trade Data (see Table S4g). Model calculations are carried out using 2017 data in dollar values from that year for internal consistency. For purposes of reporting revenue here, we have converted to 2024$ using a US GDP deflator of 1.25.

By construction, the model reaches a new trade equilibrium within the first year of the policy. Real world short-run adjustments would depend on existing spare capacity and availability of suitable workers for each product. If there is limited substitution toward imports from zero-tariff countries, then there is a strong incentive for US producers to quickly expand capacity.

If current trade patterns with higher carbon intensity countries for covered sectors are slow to adjust and persist for a longer timeframe (due perhaps to large sunk costs), real-world revenue estimates would be higher than projected by the model.

3.4. Greenhouse Gas Emissions

The tariffs imposed by the FPFA affect emissions through multiple pathways captured in the model. US emissions, as well as emissions from lower carbon intensity countries, increase because of greater domestic production in covered sectors driven by the tariffs. The amount of embodied carbon in US consumption of covered goods, however, decreases due to greater US production as well as shifts in imports to countries with lower carbon intensities. Direct emissions generally fall in countries producing with higher carbon intensity due to their reduced output of covered products. Total emissions globally also are reduced due to lower aggregate output (i.e. lower GDP, Table S4e) from the distortionary effects of taxes—in this case, tariffs.

Shifts in trade patterns driven by the FPFA result in changes in emissions that are quite small in comparison to annual emissions, both at a country level and globally (Figure 4). For example, the US net increase of 14 MMT of CO₂ is 0.22 percent of overall US emissions. Global emissions are reduced by 32 MMT, an overall decrease of 0.06 percent (EPA, 2024; Rivera et al., 2024).

The stated goal of the FPFA is to “level the playing field for American workers and manufacturers” by applying a variable ad valorem fee to certain imported industrial products based on the greenhouse gas intensity difference between the US average and the country of origin. Our modeling finds that imposing the proposed tariffs would incentivize US manufacturing for covered products and orient US imports in these goods toward countries with lower carbon intensity. Overall, these effects reduce the emissions in US consumption, increase the emissions from US production overall, and raise about $3–4 billion of revenue annually.

The general equilibrium effects captured in our model complement other modeling approaches, yielding important additional insights about the broader effects of these tariffs across the economy. Our model projects that the tariffs would increase prices for covered products (Table S4d), thereby benefiting manufacturers in countries and sectors not facing the tariffs, but would also raise costs for downstream manufacturing sectors and consumers of those goods, reducing output in these sectors (Table S4c). The net effect of the tariffs on aggregate US GDP is slightly negative (Table S4e).

There are two primary means through which any border measure based on carbon intensity could lower global emissions. One is by directly incentivizing individual foreign manufacturers to reduce their emissions intensity and sell goods with lower greenhouse gas emissions to the country imposing the border measure. Incentivizing the reduction of foreign emissions at the facility level, while avoiding a simple reshuffling of exports from higher- to lower-emitting facilities to avoid the tariffs, is a challenging aspect of border measure design. The FPFA’s fee determination method, based on country-average carbon intensities, would reduce the potential for reshuffling but also reduce the incentive for any individual facility to invest in cleaner production, because it wouldn’t recapture the full benefit. For a sector in a given country to reduce its tariff rate meaningfully would require coordinated, collective action across that sector, which would be complicated by incentives for free riding by individual facilities. Similarly, in countries with zero tariffs due to their lower average carbon intensity for a sector, there would be little to no incentive for further reductions by facilities. The international partnership agreement provisions in the FPFA, under which qualifying individual facilities can request using facility-specific data in their tariff determination, could allow for such facility-level incentives, but would need to be widely adopted to have a substantial effect on emissions.

The other mechanism for a border measure to reduce global emissions is by incentivizing other countries to adopt their own domestic greenhouse gas reduction policies. The EU CBAM creates such “policy spillovers” by reducing fees imposed on imports by the amount paid for a carbon price imposed by policies in the country of origin. The imposition of a full domestic carbon price in a given country would likely have a much greater effect on emissions than the effect of tariffs imposed by one trading partner for a subset of goods, thereby amplifying the effect of the border measure on emissions overall. Here, too, the FPFA’s international policy agreements could play a role; though the extent to which the agreements would reduce emissions would depend on their uptake and implementation. Some countries whose exports are exposed to fees may also consider new emissions reduction policies beyond carbon pricing. Insofar as new policies are applied to covered sectors and succeed at lowering greenhouse gas intensities, global emissions will decrease.