Climate policies tend to focus on measures that affect territorial greenhouse gas (GHG) emissions – the accounting framework of the United Nations Framework Convention on Climate Change (UNFCCC)1. The influence of these policies on emissions abroad is therefore often overlooked2. Carbon pricing initiatives enforced so far have been insufficient to drive deep decarbonization3,4 and are far from unilateral. Differences in climate policy commitments between countries and regions also tend to induce carbon leakage risks5–9, where emissions increase elsewhere due to domestic climate policy. The mere risk of carbon leakage has also played a significant role historically by discouraging politicians in the United States10 and European Union11 from enforcing adequate climate policies.
Sweden has relatively low territorial GHG emissions – 5 tons of carbon dioxide equivalents per capita (tCO2epc) in 201912 – and has adopted a net-zero target for 2045 to address these emissions13. Swedish consumption-based GHG emissions were 9.8 tCO2epc in 201914 (incl. non-CO2 effects for aviation), showing a large discrepancy between the accounting frameworks. Further, 42% of Swedish consumption-based emissions in 2018 occurred in countries where climate policies are considered weak15.
Consumption-based emission accounting could potentially incentivize abatement along the whole supply chain and reduce carbon leakage risks1,8. Emissions are here attributed to end-use products and services7. However, a complete shift to consumption-based accounting is neither feasible nor an attractive alternative5,8. In 2022, a Swedish parliamentarian committee unanimously proposed12 a compromise – adopting a consumption-based net-zero GHG emission target alongside the territorial net-zero target. Consumption-based accounting is not new from a scientific perspective1,5,8. However, the proposal signifies a new policy approach to be implemented in practice and the theoretical basis for adopting consumption-based targets alongside territorial commitments to the UNFCCC has not yet been explored in the scientific literature.
Moreover, assessing the ambition level of the proposed targets in relation to other policy targets, potentials for emission reductions, and global climate policy development, can emphasize the added value of the targets. Scenario analysis is an important tool in this context, especially for assessing the impact of both supply- and demand-side mitigation measures16,17. Prospective lifecycle assessment18 can be used for this purpose15,19–22, where the future carbon footprint is estimated based on assumptions of how future emerging technologies, consumption patterns, and production systems, evolve.
In this work, we scrutinize the theory behind adopting consumption-based emission accounting alongside the UNFCCC’s territorial framework. We also assess the consequences of adopting the proposed consumption-based targets for Sweden by comparing them to different emission pathways. We use scenario analysis for domestic mitigation options, both technological and behavioral, based on a prospective lifecycle assessment framework. The framework considers the inertia of the transition towards a low-carbon society – in Sweden as well as abroad.
Viability of complementary consumption-based targets
Here, we scrutinize the theoretical basis for adopting a complementary consumption-based accounting framework in a national setting. For a complementary consumption-based target to be justified it should (i) be consistent with current international frameworks and agreements, (ii) have sound economic underpinning, and (iii) provide added value by addressing weaknesses in the territorial framework.
Consistency with UNFCCC. The transparency framework of the Paris Agreement24 is built on territorial accounting of GHG emissions. It has the dual purpose of generating national GHG inventories that can be summed up to track global progress, and tracking progress with the countries’ nationally determined contribution (NDC). Consumption-based accounting comes with risks of double-counting, high complexity, and significant uncertainties, which could be detrimental to the transparency of the Paris Agreement if consumption-based accounting were to replace territorial accounting or be adopted as a formal part of countries’ NDCs5, see Supplementary Note 1.
However, the adoption of the Paris Agreement marked a shift23 from negotiations on territorial targets under the Kyoto Protocol to a framework where each country determines its own contribution. This also shifted responsibility to the individual country on clarifying how their NDC adheres to the fairness principles of the UNFCCC. Consumption-based accounting could play an important role in assessing one aspect of a fair NDC by tracking how consumption-based emissions change when territorial mitigation measures are put in place. Hence, while consumption-based targets should not be included explicitly in a country’s NDCs, they could enhance the fairness of the NDC.
Economic theory. A central concept in policy design is the polluter pays principle1, where policies should target the polluting source directly (as opposed to proxies such as inputs or consumption activities) to stimulate abatement and avoid unnecessary distortions to the economy. This principle is consistent with the territorial accounting framework. Thus, regulations of polluting activities are limited to nations or an equivalent jurisdictional body.
A commonly proposed first-best solution for meeting a global climate target would be that all countries impose economy-wide carbon prices on the polluting sources24. However, since a unilateral carbon price is not likely to materialize in the near term8, complementary policy measures may be required to achieve the global climate targets25–28. Consumption-oriented policies may be preferable when monitoring costs are high, technological abatement options are limited, uncertain, or unavailable in the short-term, and where demand is elastic and/or substitutions through alternative products or services are possible23. Further, consumption-oriented policies that incentivize technology innovation among domestic and foreign producers alike may be cost-effective and deliver faster emission reductions (thereby generating lower cumulative emissions) even if they imply high monitoring costs31,32.
Addressing weaknesses. Consumption-based targets could address carbon leakage risks and incentivize additional mitigation measures by targeting domestic activities where emissions occur outside the country’s territory. In addition, consumption-based targets adopted alongside territorial targets could introduce a consumer pays perspective1, incentivizing both demand- and supply-side emission reductions irrespective of where the emissions occur. Demand-side measures are instrumental in some scenarios29 that strive to limit the global mean temperature increase to 1.5°C without large dependency on carbon dioxide removal. These scenarios often foresee substitution away from carbon-intensive products and services to low-carbon options – e.g., car-free mobility solutions, plant-based diets – or absolute reductions in demand – e.g., air travel, space heating/cooling – that could be incentivized by consumption-based policies.
Examples of such policies are differentiated consumption taxes on basic material or animal-based food. For basic materials (e.g., cement, steel, and petrochemicals), abatement cost per unit tends to be high while the downstream consumer (e.g., a car buyer) typically would see little of the carbon cost30. Differentiated consumption charges have been proposed for these sectors to both reduce the risk of carbon leakage31 and reinstate the demand-side carbon price signal32. In the case of animal-based food, high monitoring costs are paired with comparatively low technological potentials for reducing ruminant methane emissions and nitrous oxide emissions from fertilizers. Hence, consumption charges could be a cost-effective policy for mitigating agricultural GHG emissions33–37. Emerging plant-based protein sources and the increasing number of vegetarians and flexitarians in the EU38 also indicate opportunities for substitution.
Ambition level of proposed Swedish consumption-based targets
As shown in the previous section, complementary consumption-based targets and policies can be supported theoretically. Here, we compare the net-zero target trajectories for consumption-based emissions proposed by the Swedish parliamentarian committee12 with four different scenarios, see Figure 1. The proposal12 shows two alternative trajectories of emission reductions for specific years in relation to the consumption-based emissions of 2010 (i.e., linearly interpolated): (i) the main trajectory with reductions of 26% by 2030, 57% by 2040, and 66% by 2045, and (ii) the alternative trajectory with reductions of 47% by 2030, 69% by 2040, and 77% by 2045.
The scenario analysis is based on a hybrid approach. Bottom-up simulation models are used for estimating future emissions for passenger travel, construction and housing, and food consumption. Emissions related to remaining consumption categories are estimated using a top-down approach based on an expanded Kaya identity39 and data from Exiobase40, see Online Methods.
Reference scenario. Consumption-based emission decrease by 9% from 2019 to 2020, see Figure 2, due to reduced air travel in response to the Covid-19 pandemic. Air travel emissions recover by 2025 and reach 0.9 tCO2epc, close to pre-pandemic levels of 1.0 tCO2epc, see details for each consumption category in Extended Data Figure 2. Changes in other consumption categories counteract the trend, especially emissions from passenger cars that decrease by 23% compared to 2019 due to electrification. While air travel emissions increase after 2025, reaching 1.2 tCO2epc in 2045, emissions decrease in all other consumption categories. Passenger car travel stands out with a reduction of 60% compared to 2019 due to the electrification of the fleet. Emission reductions in other consumption categories are related to current efficiency trends. Subsequently, the carbon intensity of Swedish electricity generation and district heating is assumed to reach zero by 2058, when the EU Emission Trading System’s (EU ETS) CO2 cap reaches zero41 (i.e., not including recent provisional decisions related to the Fit-for-55-proposal). Consumption-based emissions reach 7.6 tCO2epc in 2045. This is equivalent to a reduction of 23% compared to 2010 (referring to emissions for the total population and excluding non-CO2 climate impacts of aviation, both in line with the proposal). The reduction can be compared to the proposed Swedish consumption-based targets of 66-77% for the main or alternative trajectories, respectively.
Achieving territorial targets scenario. If Sweden's territorial targets are met and global industries develop in line with current trends and policies, consumption-based emissions could decrease to 6.4 tCO2epc by 2045 (cf. 7.6 tCO2epc in the reference scenario). The major differences between this scenario and the reference depend on assumptions for the construction sector (i.e., 54% emission reduction by 2045 compared to the reference scenario), mainly due to domestically produced fossil-free steel and asphalt, and cement production using carbon capture and storage; passenger car travel (i.e., 39% emission reduction in 2045 compared to the reference scenario) due to full fleet electrification and phase-out of fossil fuels; and food production (i.e., 24% reduction in 2045 compared to the reference scenario), due to domestic use of fossil-free fertilizers and feed additives that reduce methane occurrence in cattle42. The assumed measures are detailed in the first column of Table 1. A higher pace in the decarbonization of electricity and district heating (by 2045 instead of 2058) also contributes to reducing emissions. These reductions are offset by growth in air travel emissions due to an assumed doubling of flights per capita. Air travel emissions level out around 2035. The share of sustainable aviation fuels (SAFs) is assumed to approach 100% by 2045 for domestic flights43 and EU-bound flights taking off from Sweden12.
Global climate transition. Here, we assume global action, where other countries match the ambition of the Paris Agreement. This implies reaching global net-zero GHG emissions by 207044. For the bottom-up simulated categories, faster global decarbonization mainly impacts Swedish air travel, where emissions start to decrease from 0.8 tCO2epc in 2025 to 0.6 tCO2epc in 2045 due to an increased share of SAFs for international air travel. Decarbonized global construction materials and vehicle manufacturing also contribute to lower emissions as well as fossil-free fertilizers and feed additives in global food production.
A global climate transition also has a significant impact on emissions related to the top-down modeled consumption category, which decrease by 49% in 2045 compared to the reference scenario. However, the rate of energy efficiency (i.e., the energy use per unit of economic activity) and carbon intensity improvements assumed in the top-down methodology are highly uncertain, see Extended Data Figures 3-4. On the one hand, the energy efficiency assumption used – the historic trend at the global level45 – may overestimate the improvement potential given that the energy efficiency of Swedish economic activities is already relatively high46. On the other hand, behavioral changes could enable higher rates of energy efficiency improvements through, e.g., pushing towards buying high-quality products with longer lifetimes or increasing the degree of servitization – where users pay for a service instead of buying equipment47.
Swedish consumption-based emissions reach 3.8 tCO2epc by 2045 if territorial targets are achieved in combination with a global climate transition. Hence, achieving the territorial emission reduction targets would reduce consumption-based emissions by 35-62% in 2045 compared to 2010 (referring to emissions for the total population and excluding non-CO2 climate impacts of aviation, both in line with the proposal), depending on the development in the rest of the world.
Table 1. Categorization of mitigation actions. Actions implemented in the mitigation scenarios according to the avoid-shift-improve framework. Note that the scenario Achieving territorial targets only covers measures in the improve category. Quantifications of these assumptions and additional details are provided in Online Methods as well as Supplementary Notes 2-8.
Consumption category
|
Achieving territorial targets
|
Advanced technology and behavioral changes
|
|
Improve
|
Avoid
|
Shift
|
Improve (additional)
|
International air travel
|
Increased use of sustainable aviation fuels, measures to reduce non-CO2 effects, electrification
|
Sharp reduction in air travel per capita
|
Increased train and bus trips to continental Europe
|
|
Domestic air travel
|
Increased use of sustainable aviation fuels, measures to reduce non-CO2 effects, electrification
|
Sharp reduction in air travel per capita
|
Increased domestic train and bus trips
|
|
Passenger car travel
|
Electrification following phase-out of internal combustion engines in new car sales by 2030, and phase out of fossil fuels by 2040
|
Less commuting
|
Increased cycling and public transport for commuting as well as leisure trips
|
Self-driving cars enabling car sharing in cities
|
Construction, space heating and household energy use
|
More energy efficient buildings and reduced emissions in electricity generation and district heating production, decarbonized domestic building materials
|
Reduced housing and commercial space per capita
|
Commercial and office space converted to housing; renovations instead of new construction
|
Increased materials efficiency in construction
|
Food
|
Fossil-free fertilizers and feed additives
|
|
Plant-based protein-rich products replace animal products
|
|
The emission impact of demand-side mitigation measures
Here, we explore how demand-side measures impact Swedish consumption-based emissions and compare those impacts with the net-zero target trajectories proposed by the Swedish parliamentarian committee12. Two scenarios are analyzed: (i) a scenario combining technologies assumed to be implemented in response to the territorial targets, additional advanced technologies, and behavioral changes that shift or fully avoid consumption, see categorization along the avoid-shift-improve spectrum17,48 in Table 1, and (ii) a reference scenario combined with only avoid or shift measures that highlight the effect of behavioral changes alone. The latter scenario also highlights how far sufficiency actions16,17,51 can take us in reducing the emissions.
The advanced technology transition includes self-driving cars that enable car sharing in cities, and to some extent ride sharing. We assume that policy measures minimize any induced travel demand49 implied by these technologies. The behavioral changes include assumptions of reduced driving by 20% per capita compared to 2019 by increased cycling and use of public transport as well as reductions in travel. Total air travel per capita is halved, and domestic flights are reduced even more. Increased rail travel is assumed to compensate for some flights, both domestic and – to a lesser extent – international. Trips to continental Europe make use of the potential for shifting to night trains50. Construction of new buildings and infrastructure are assumed to decrease significantly by increased maintenance and renovation, and by converting offices and commercial space into housing. The average living space is assumed to fall by 10% compared to 2019. Additional material efficiency measures in construction and renovations contribute to further reducing emissions51. A shift towards consumption of plant-based protein-rich products enable a decrease in beef, lamb, and liquid dairy products by 75% compared to 2019 and cheese consumption is halved. Demand changes are detailed in Supplementary Table 1 and the rationale is discussed in the respective Supplementary Notes 2-8.
Reference scenario with behavioral changes. Significant emission reductions in 2045 compared to the reference scenario in response to behavioral changes alone are seen in air travel (73%), food consumption (35%), passenger car travel (36%), and construction (31%). The scenario reaches an emission level of 5.8 tCO2epc in 2045, see Figure 3. Total emissions are reduced by 39% in 2045 compared to 2010 (referring to emissions for the total population and excluding non-CO2 climate impacts of aviation, both in line with the proposal). However, the emissions are larger than the proposed targets, indicating that behavioral measures alone are not enough to achieve the targets.
Advanced technology and behavioral scenario. Further emission reductions are expected compared to the reference scenario in 2045 for air travel (74-86%), food consumption (61-66%), passenger car travel (68-83%), and construction (75-83%), where ranges depend on the global decarbonization pathway. Compared to the scenario achieving territorial targets, the effect is largest for air travel emissions (0.2-0.3 tCO2epc in 2045 compared to 0.6-1.0 tCO2epc, where ranges depend on the global decarbonization pathway) and food consumption (0.5 tCO2epc in 2045 compared to 1.0 tCO2epc – the difference between the global decarbonization pathways is smaller than 0.1 tCO2epc). These effects can be explained by both categories lacking options for close-to-zero decarbonization technologies in the near term. For passenger car travel, construction, space heating, and household electricity use, behavioral shifts in combination with technological advances result in earlier emission reductions, but only have minor effects on the level of emissions by 2045, see Extended Data Figure 2. The scenario combining advanced technologies with behavioral changes could reach 2.7-4.8 tCO2epc by 2045, see Figure 3, reducing consumption-based emissions by 49-71% in 2045 compared to 2010 (referring to emissions for the total population and excluding non-CO2 climate impacts of aviation, both in line with the proposal).
Results for top-down modeled categories. Consumption-based emissions for the top-down modeled category are in our analysis not affected by any explicit behavioral changes. Demand-side measures for consumer products could potentially reduce emissions further52. Figure 4 shows that these consumption categories make up a significant share of remaining emissions in 2045, especially if the global pathway develops according to current trends and policies. The sensitivity analysis of rate of change in the energy efficiency of economic activities, see Extended Data Figure 3-4, suggests that decreasing energy use per economic unit could result in significant emission reductions. Such changes could include shifts towards more expensive, high-quality products and shifts from products to services. However, more detailed studies are needed on how the manufacturing of consumer products may change, what the potentials are for shifts in consumption, and how increased circularity as well as rebound effects could affect those trends.