Carbon emissions and economic impacts of an EU embargo on Russian fossil fuels

The Russia–Ukraine conflict lays bare the dependence of the European Union (EU) on fossil fuel imports from Russia. Here, we use a global computable general equilibrium model, C3IAM/GEEPA, to estimate CO2 emission and gross domestic product (GDP) impact of embargoing fossil fuels from Russia. We find that embargoes induce more than 10% reduction of CO2 emissions in the EU and over 5% increases of emissions in Russia, while both regions experience GDP losses (around 2% (US $486 billion) for the EU and about 5% (US $149 billion) for Russia, ignoring the relative impact of other sanctions). Reacting to increasing energy prices with demand-side response inside the EU would increase CO2 emission savings, and ease GDP losses; however, the world would continue to suffer economic damage (US $655 billion). An EU embargo on Russian fossil fuels would lead to a rapid decrease in fossil fuel combustion, GHG emissions reductions and potential economic losses. This analysis quantifies such effects, while also demonstrating how demand-side responses would impact the shock.

The Russia-Ukraine conflict lays bare the dependence of the European Union (EU) on fossil fuel imports from Russia. Here, we use a global computable general equilibrium model, C 3 IAM/GEEPA, to estimate CO 2 emission and gross domestic product (GDP) impact of embargoing fossil fuels from Russia. We find that embargoes induce more than 10% reduction of CO 2 emissions in the EU and over 5% increases of emissions in Russia, while both regions experience GDP losses (around 2% (US $486 billion) for the EU and about 5% (US $149 billion) for Russia, ignoring the relative impact of other sanctions). Reacting to increasing energy prices with demand-side response inside the EU would increase CO 2 emission savings, and ease GDP losses; however, the world would continue to suffer economic damage (US $655 billion).
The Russia-Ukraine conflict has intensified geopolitical frictions, created turmoil in global energy markets and laid bare the dependence of Europe on fossil fuels from Russia. Russia is one of the world's top three oil producers, vying for the top spot with Saudi Arabia and the United States, and is the second largest producer of natural gas, behind the United States 1 . Russia also has extensive oil export pipeline capacity and a wide-reaching gas export pipeline network, in particular to Europe. It is the world's largest natural gas exporter, the second largest crude oil and condensate exporter after Saudi Arabia and the third largest coal exporter after Indonesia and Australia in 2021. Europe is a key destination for energy exports from Russia. Almost 50% of oil exports, 70% of natural gas exports and one-third of coal exports from Russia are sent to Europe 2 . Germany, the Netherlands and Turkey are the main importers.
Given the important position of Russia in global energy markets, concerns about future disruptions in Russian energy supply have increased. Since CO 2 emissions come mainly from the burning of fossil fuels, a disruption in Russian energy supply not only threatens energy security and economic stability but also has implications for resulting fossil fuel combustion and GHG emissions. In this situation, large-scale geopolitical disruption may be related to the last possible timing to rapidly reduce GHG emissions in alignment with the goals of the Paris agreement 3 .
Considering the close relations and possible disruptions between Russia and the European Union (EU), existing studies have attempted to discuss the effects of energy disruptions. Most of them focused on the partial or one-sided impacts on energy importers or Russia 4-6 . For example, an investigation of a potential cut-off of the German economy from Russian energy imports finds that German gross domestic product (GDP) would fall by 0.5% and 3% in the short-run 4 . Another study investigates the effects of an import ban on Russian oil and gas by the EU and G7 countries and concludes that the most severe trade sanction scenario would lead to a permanent GDP reduction of 1.06% and 522,000 job losses for Russia 5 . The economy-wide impact analysis Analysis https://doi.org/10.1038/s41558-023-01606-7 liquified natural gas (LNG) facilities in principle. But considering the use of the LNG facilities of the EU can be transferred from Russia to the United States and Africa, the corresponding trade transfer limitation has been released. The releasing ratio is calculated by dividing EU LNG imports from Russia into EU gas imports from the region (the related trade data refers to Statistical Office of the European Communities); similarly, because nearly all oil exports are dependent on ships and 80% of oil tank ships insurances are issued in London, a substitution limit on exports of oil to other regions for Russia has been considered and the model assumes that oil imports of other regions from Russia can only increase by at most 20%.
Under the above disruption scenarios, production and consumption in the EU, especially energy-intensive sectors will be negatively affected (Supplementary Information 1 gives the complete results of all sectors) and its GDP would fall by 1.5-2.5% in 2022 (US $286-486 billion) (Fig. 1). Aiming for import source substitution, the EU tends to increase its imports from other regions, except for Russia, so that the EU import share from Russia will decrease substantially and correspondingly increase its import share from other regions. For example, the import share of the EU to the Middle East and Africa (MAF) for oil and natural gas in 2022 would increase by about 12-22 percentage points and 14-24 percentage points, respectively (Fig. 2). Even so, the total fossil fuel consumption of the EU will decrease notably due to the significant rise in domestic fossil energy prices (increasing by 12-20% in oil price and 39-64% in gas price), causing 12.3-18.3% (530-787 MtCO 2 ) of CO 2 emissions decline (Fig. 1). Meanwhile, local air pollutants have also been reduced, with the ranges of 4.9-7.6% for fine particulate matter (PM 2.5 ), 7.9-12.4% for sulfur dioxide (SO 2 ) and 9.0-14.0% for nitrogen oxide (NO x ). Also, the renewable electricity of the EU has grown to some extent (the share increases by 5.3-8.3 percentage that can incorporate the ripple-through effect is still limited. Also, if most countries in the Organisation for Economic Co-operation and Development (OECD) put major restrictions on the energy exports of Russia, Russia will suffer the largest economic losses, with the EU being the second most affected region but on a much smaller extent 7 . As another key response measure, a comment paper points to the potential to substitute for Russian fossil fuel by ad hoc energy saving measures in the building, transport and food sector 8 . However, no study quantitatively evaluates the possible response measures to energy disruptions considering general equilibrium effects. To this end, we here use the Global Energy and Environmental Policy Analysis model in the China Climate Change Integrated Assessment Model (C 3 IAM/ GEEPA), a global computable general equilibrium model, to simulate and analyse the potential effects of a disruption of exports from Russia to the EU, for example, as a result of a comprehensive embargo and of possible response measures.
We develop two disruption scenarios, as follows.
(1) Moderate disruption scenario. Considering the difficulty of all EU countries in imposing sanctions on Russia and the possibility that economic sanctions will have a substantial deterrent effect, we assume that the main importing countries (3-5 countries) of the EU are more likely to implement the embargo and set the corresponding ratio. In detail, the main oil importers include the Netherlands, Germany and Poland (total 61%); the main natural gas importers include Germany, Turkey, France and Poland (total 70%); and the main coal importers include Germany, the Netherlands and Turkey (total 64%) 2,9 . (2) Strong disruption scenario. A stronger disruption with 90% fossil fuels trade between Russia and the EU has been designed. Under both disruption scenarios, we prohibit immediate world market alignments in natural gas trade as the limitation of natural gas for pipeline transport and Analysis https://doi.org/10.1038/s41558-023-01606-7 points), thereby resulting in an increase in its electrification rate of 1.8-2.5 percentage points.
However, because Russia relies heavily on revenues from oil and natural gas, which in 2021 made up 45% the federal budget 10 , the decreasing exports from Russia would hurt the economy more severely, with a GDP loss of 3.7-5.5% (US $100-149 billion) (as Fig. 1) (not considering the effect of other non-fossil fuel sanctions). Other relevant quantitative studies, such as refs. 4,5,7 , also found that the economies of both Russia and the EU will be hurt by the embargo; some of the studies conclude that the shock for Russia is greater than that for the EU. For example, ref. 7 found that the economic shock of energy disruption on Russia is 11.7% (measured with loss of real income), which is also larger than loss of the EU (1.7%). Moreover, such a disruption would lead to ample domestic supplies and lower prices, making the positive substitution effects clearly outweigh the negative income effects caused by GDP losses, which in turn would stimulate the domestic fossil energy consumption, especially the energy-intensive sector and other manufacturing, thus leading to a 5.9-8.5% increase in CO 2 emissions (160-231 MtCO 2 ) ( Fig. 1). At the same time, such a shock would ripple through the rest of the world, knocking 0.5-0.8% off global GDP and decreasing global CO 2 emissions by 1.1-1.8% (498-810 MtCO 2 ).
In the long run, if the level of energy disruption between Russia and the EU is halved from 2025 in both disruption scenarios, the economic development of the EU would rebound to be around 0.6-1.1% lower than business-as-usual (BAU) in 2025 and 2030. Because of the energy transition and substitution effect that has taken place within the EU, its trade dependence on fossil fuels and CO 2 emissions would still be notably lower than BAU. The EU Green New Deal requires that the renewable energy share of total energy consumption reaches 40% in 2030 and CO 2 emissions reduce by 55% in 2030 relative to the 1990 level. Also, the REPowerEU plan presented by the European Commission is trying to phase out its dependency on Russian fossil fuels through energy savings, diversification of energy supplies and accelerated roll-out of renewable energy. Russian fossil fuel substitution caused by this disruption seems to align with the EU Green Deal policies and the REPowerEU plan. The supply shock increases the economic incentive for renewable deployments (even as interest rates are rising). In addition, our model also shows that the supply shock will reduce the marginal abatement costs (the carbon price level required to achieve a certain emission reduction target) of achieving the EU Green New Deal by 25-40% in 2030. The costs of using fossil fuels would increase by 0.4-1.3% for coal (equivalent to the carbon price of US$2.6-4.0 per tCO 2 ), 6.7-10.8% for oil (equivalent to the carbon price of US$16-25 per tCO 2 ) and 9.7-12.7% for gas (equivalent to the carbon price of US$14-19 per tCO 2 ), the share of renewable energy would increase by 1.8-2.6 percentage points and the energy costs for transportation and commercial buildings (approximately represented by energy use in other tertiary sectors) would increase by 4.4-6.9% and 3.7-5.6%, respectively. Comparatively, the incentives and monetary effects for energy savings increase. For Russia, although the level of embargo shock is assumed to be halved in 2025, the previous cumulative economic effects just make its GDP losses slightly smaller (3.7-5.5% in 2022 versus 3.3-5.1% in 2025). However, the released export transfer limitation makes the substitution effects notably weaker (for example, the ratio of domestic prices of natural gas to foreign prices would fall by 20-24% relative to BAU in 2022, while it would only fall by 6.9-12.6% in 2025). Finally, the total effects would decrease the fossil fuel consumption and CO 2 emissions of Russia. After that, GDP losses of Russia would gradually expand (5.1-8.3% in 2030) if the disruption is not further released. The negative income effects would exceed the substitution effects more and more obviously, so that Russia's fossil fuel consumption and CO 2 emissions would further decline.
In addition, on the basis of the embargo simulation method implemented by the model, we can estimate that the corresponding importing tariff/fine revenue if the EU imposes sanctions on Russian fossil fuels through economic measures such as by raising importing tariffs or imposing heavy fines. For example, under the moderate disruption scenario, the increasing production costs and living costs caused by increasing domestic fossil fuel prices, as well as the tariff revenue reduction caused by decreasing imports, would reduce total revenue of the EU government by 1.7% in 2022. However, if the embargo is implemented by imposing tariffs, the corresponding revenue would increase by US$98 billion, which would offset the negative shocks and the final total government revenue would fall by 0.1%. In the case of strong disruption, the original government revenue reduction would be 2.9%, while implementing the partial embargo via tariffs/fines would produce US$50 billion revenue increment, resulting in a reduced government revenue loss rate of 2.0%. Hence, if EU embargoes are not implemented via tariffs, GDP losses for the EU would be magnified due to greater negative impacts on revenues and investments.
On the basis of the disruption situation, this study also modelled the two response scenarios as the probable complementary measures associated with the EU embargo. (1) Supply-chain shift: the EU will increase the imports of fossil fuels from other main alternative   Under the product disruption scenario, oil, gas and coal represent single energy disruption. c,d, For the labour supply market assumptions, 'fe' represents fullemployment, 'ls' represents labour surplus and 'ss' represents sector-segmented labour market. e,f, For trade substitutions effects, 'st' represents strict trade substitution limit, 'mt' represents moderate trade substitution limit and 'nt' represents no trade substitution limit. a-f, According to the results: for single product disruption, the EU and Russia would mainly encounter economic losses in the case of oil disruption (0.8-1.5% for the EU and 1.4-2.6% for Russia) and natural gas disruption (0.5-0.9% for the EU and 1.4-1.8% for Russia) (a,b); for labour supply market assumption, economic losses of the EU and Russia under moderate disruption (stronger disruption) would be 0.5-1.5% (0.9-2.5%) and 2.7-6.1% (4. (2) Demand-side response: the EU and national government change the demand patterns of the transportation and building sectors in two ways: (i) modal shift away from fossil fuel powered transport, including transport electrification; and (ii) change in consumption patterns of households and tertiary industries to reduce end-use energy demand. We assume that aggressive measures can reduce demand by 10% (ref. 11 ), which is also consistent with the EU regulation to reduce 15% of gas between 1 August 2022 and 31 March 2023 (ref. 12 ). In addition, since the model does not subdivide the specific transport modes of residents, the corresponding simulation is realized by adjusting residents' consumption structure, to reduce the share of oil consumption and increase the share of electricity consumption. We found that if the EU reduces energy-related tariffs for other major trading countries, its GDP losses could be eased but the overall effects would be very weak owing to very low original tariffs (<1%), as shown in Fig. 3. This shows, for the EU, that the true bottleneck is not trade tariffs but availability of fossil fuels supply. Moreover, this small variation in GDP would make CO 2 emissions of the EU higher than the disruption scenarios after 2022, with a rise of more than 0.5 percentage points in 2025 and 2030.
The demand-side response measure is likely to have positive impacts on both economy and environment in the EU in the long run. The demand-side response would reverse the negative economic tide. Compared with the above disruption scenarios, EU GDP losses would decrease around 0.8 percentage points in 2022 (Fig. 4). Moreover, EU GDP would eventually increase by 0.3% (if moderate disruption happens) or fall by 0.1% (if strong disruption happens) relative to BAU in 2025. In particular, due to the increasing consumption of most non-energy products, the EU economy would recover in the long term, going up by 1.3% of GDP relative to above disruptions in 2030. Under the demand-side response scenario, CO 2 emissions in the EU are expected to decrease further due to the reduction in final energy demand and the change in residents' transport mode during the current year and in the long term. GHG emissions would be reduced 1.5-1.7 percentage points more compared to above disruption scenarios in 2022.
In addition, we found that the demand-side response would cushion the rise in the importing oil prices of each region. Compared with the only embargo situation, the increase of demand-side response would reduce international importing oil price by around 0.2% (Eastern European CIS, EES) to 1.0% (Russia). In this case, GDP losses of all countries would decrease and total losses in the world economy would be 0.2 percentage points lower (global residual economic damage are US $314-655 billion), except that the economy and environment of Russia would suffer slightly more losses than the above disruption scenarios (Figs. 3 and 4).
As another contribution to existing research, different aspects of uncertainty analysis are performed in this study by examining each disruption product separately, assuming different labour markets and limiting different degrees of trade substitution effects. First, considering the uncertainty of the disruptions, our simulation of separate product disruption is helpful to examine the main fossil source of effects. Second, considering the possible characteristics of wage and labour, our modelling of different labour markets helps to identify the effect of assumptions about wage and labour. Third, we vary the magnitude of trade substitution effects to investigate the uncertainty of trade transfer in reality. Supplementary Information 3 provides detailed settings. Results show that the above conclusions still hold, which means that the directions of the impacts on economy and environment remain consistent but their extent will change slightly, as shown in Fig. 5. In addition, the economic losses of the EU and Russia are mainly from the disruption of oil and natural gas with less impacts from coal disruption. On the other hand, the labour market has been assumed to be important for overall economic effects 13,14 and we assume rigid wage and mobile labours among sectors in the basic scenario. On the basis of this, two additional labour adjustment scenarios are tested here: a more optimistic scenario with flexible wage and a more pessimistic scenario without labour mobility among sectors. We found that, the negative economic effects would be reduced in the former case and the latter case would worsen the economic losses of Russia and the EU. Finally, the examination about trade substitution effect shows that the tighter the trade substitution restrictions, the greater the GDP losses of Russia and the EU. For example, with moderate disruption in 2022, GDP losses of Russia are 1.3% (no limit) to 3.7% (strict limit) and GDP losses of the EU are 1.2% (no limit) to 1.5% (strict limit), respectively. Thus, our conclusions are strongly dependent on the factors of the labour market assumptions. All indicators of the uncertainty analysis results are available in Supplementary Information 4. To conclude, if the EU embargoes Russian fossil fuels, in particular of oil and natural gas, would hurt both sides and global economy. The embargo would increase CO 2 emissions for Russia, though accelerating the EU Green Deal. Importing tariffs are similar to a partial or full embargo, depending on the height of tariffs. Moreover, such an economic sanction could generate extra revenue, which would obviously offset the negative shocks on government deficits, otherwise the negative impact on the long-term economic growth of the EU would be greater. For the EU, the supply-chain shift measure through simply improving unimpeded trade has little effects. Demand-side response measure (for example, reducing residents' heating demand, speed limits and car-free Sundays) would, however, mitigate overall economic impacts in synergy with further decreasing GHG emissions. Finally, we acknowledge the complexity of the disruption development, for example, if Russia completely cuts off natural gas supplies. The effects of 100% complete disruptions would be the same even if the disruptions are initiated by a broken supply for Russia or the EU embargo. However, when partial disruptions are imposed by economic means (for example, by raising tariffs), the outcomes would be different for Russia and the EU due to different revenue sharing. For another example, this study did not analyse the impacts of damages in the global free trade order that such an embargo might trigger. Also, energy disruptions are considered in this study but currently the EU has adopted a package of sanctions targeting the financial sector, energy and transport sectors or dual-use goods. Therefore, these can be discussed in future studies.

Methods
The disruption effects in this study are simulated through the C 3 IAM/ GEEPA we developed 15 . C 3 IAM/GEEPA is a multiregional recursive dynamic computable general equilibrium (CGE) model, which is composed of production, income/expenditure, investment and foreign trade. It generally follows the style of the other CGE global models that target energy and environment issues and its basic framework is shown in Extended Data Fig. 1 (ref. 16 ). An important advantage of such a model is that it can extract the separate impacts of external shocks (energy disruption here), which is very useful for targeted policies or measures. Currently, C 3 IAM/GEEPA has been used to analyse the socio-economic and environmental impacts of a variety of energy and environmental policies worldwide, such as Sino-US trade friction 17 , green recovery policies in the postpandemic era 18 20,21 . The base-year GHGs are derived from the database of greenhouse gas and air pollution interactions and synergies (GAINS) 22 .
Future BAU development is generated following the shared socio-economic pathway 2 (SSP2) narrative-a middle-of-the-road pathway. The world follows a path in which social, economic and technological trends do not shift markedly from historical patterns, with some progress towards achieving development goals, reductions in resource and energy intensity at historical rates and slowly decreasing fossil fuel dependency 23 . Besides, the energy development is calibrated following the trends projected by EIA (US Energy Information Administration) and the environmental emissions development is calibrated on the basis of the trends in the Climate Modelling Intercomparison Project Phase 6 emissions 24 .
A set of complementary constraints is used to simulate both disruption and non-disruption referring to refs. 25 (1) Where, χ i,s,r represents the importing windfall profit tax rate for product i from region r to region s; Ms disruption i,s,r and Ms i,s,r represent the importing cap and import volume, respectively. If the second constraint is a strict equality, the windfall profit tax rate χ i,s,r will be the shadow price of this constraint and be zero or strictly positive, when the first and third constraints are satisfied; if the second constraint is strict inequality with no effective disruption restrictions, the third constraint has to hold with strict equality to satisfy the first constraint.
The windfall profit WP i,s,r is determined by equation (2).
WP i,s,r = χ i,s,r × PMs i,s,r × Ms i,s,r (2) Where WP i,s,r is the windfall profit and PMs i,s,r is the import price. The import demand function can be modified as shown in equation (3).
Where M i,r represents aggregate import goods i in region r; PM i,r represents aggregate import price i in region r; Am i,r represents the scaling coefficient in the Armington commodities function; αm i,s,r is the share parameter in the Armington commodities function; η i represents the substitution parameter in CES function between imports; and tm i,s,r represents import tariff rate i in region r from region s. The income balance for the government is modified as shown in equation (4).
Where G Income,r is government income in region r; TOTITAX r is total indirect taxes from production in region r, TOTTARIFF r is total import taxes in region r, TOTEXSUB r is total subsidies for exports in region r, TOTHTAX r is total direct taxes from households in region r.

Data availability
The data that support the plots within this paper and other findings of this study are available from the corresponding authors upon reasonable request. https://doi.org/10.1038/s41558-023-01606-7 Extended Data Fig. 1 | The framework of C 3 IAM/GEEPA. C 3 IAM/GEEPA is a multiregional recursive dynamic computable general equilibrium (CGE) model, which is composed of production, income/expenditure, investment and foreign trade. When producing one commodity, labour, capital, energy and other intermediate products are all inputs in each sector, which are assumed to follow a nested constant elasticity of substitute (CES) function. Household income mainly comes from labour income and capital returns; Government income is composed of tariff, indirect tax, household income tax and transfers from other countries/regions. C 3 IAM/GEEPA adopts Armington assumption, assuming there is imperfect substitutability between imports and domestic output sold domestically. The commodity that supplied domestically is composed of domestic and imported commodities following a CES function. A constant elasticity transformation (CET) function is used to allocate total domestic output between exports and domestic sales. The commodity, capital, and labour markets are cleared in C 3 IAM/GEEPA. The model adopts the recursive dynamic mechanism and is pushed forward through capital accumulation, population growth, and improvement of total factor productivity.