Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 Table of Contents 2 Chapter 2: Emissions Trends and Drivers ..................................................................................... 2-1 3 Executive Summary ......................................................................................................................... 2-4 4 2.1 Introduction .......................................................................................................................... 2-8 5 2.2 Past and present trends of territorial GHG emissions .......................................................... 2-9 6 2.2.1 Uncertainties in GHG emissions ................................................................................ 2-11 7 Cross-Chapter Box 2 GHG emission metrics ............................................................................ 2-16 8 2.2.2 Trends in the global GHG emissions trajectories and short-lived climate forcers ..... 2-19 9 2.2.3 Regional GHG emissions trends ................................................................................ 2-25 10 2.2.4 Sectoral GHG emission trends ................................................................................... 2-30 11 2.3 Past and present trends of consumption-based CO2 emissions (CBEs) and emissions 12 embodied in trade........................................................................................................................... 2-33 13 2.3.1 Scope, variability, and uncertainty of CBEs .............................................................. 2-33 14 2.3.2 Trends in global and regional CBEs trajectories........................................................ 2-36 15 2.3.3 Decoupling of emissions from economic growth ...................................................... 2-37 16 2.3.4 Emissions embodied in trade (EET) .......................................................................... 2-39 17 2.4 Economic Drivers and Their Trends by Regions and Sectors ........................................... 2-41 18 2.4.1 Economic Drivers at Global and Regional Levels ..................................................... 2-42 19 2.4.2 Sectoral Drivers.......................................................................................................... 2-44 20 2.4.3 Poverty and Inequality ............................................................................................... 2-53 21 2.4.4 Rapid and Large-scale Urbanisation as a Driver of GHG Emissions ........................ 2-54 22 2.5 Technological Change is Key to Reducing Emissions ...................................................... 2-55 23 2.5.1 Technological Change Has Reduced Emissions ........................................................ 2-55 24 2.5.2 A Low-Carbon Energy Transition Needs to Occur Faster Than Previous Transitions .. 2- 25 56 26 2.5.3 Improvements in Technologies Enable Faster Adoption ........................................... 2-58 27 2.5.4 Rapid Adoption Accelerates Energy Transitions ....................................................... 2-60 28 2.6 Behavioural Choices and Lifestyles ................................................................................... 2-61 29 2.6.1 Introduction ................................................................................................................ 2-61 30 2.6.2 Factors affecting household consumption patterns and behavioural choices............. 2-63 31 2.7 Emissions associated with existing and planned long-lived infrastructure ........................ 2-68 32 2.7.1 Introduction: clarification of concepts ....................................................................... 2-68 33 2.7.2 Estimates of future CO2 emissions from long-lived infrastructures........................... 2-69 34 2.7.3 Synthesis – Comparison with estimates of residual fossil fuel CO2 emissions.......... 2-73 35 2.8 Climate and Non-Climate Policies and Measures and their Impacts on Emissions ........... 2-76 36 2.8.1 Introduction ................................................................................................................ 2-76 37 2.8.2 Comprehensive Multinational Assessments .............................................................. 2-76 Do Not Cite, Quote or Distribute 2-2 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2.8.3 National, Sectoral, and Cross-Sectoral Policies ......................................................... 2-77 2 2.8.4 Emission Impacts of Other Related Policies .............................................................. 2-79 3 2.9 Knowledge Gaps ................................................................................................................ 2-82 4 Frequently Asked Questions (FAQs) ............................................................................................. 2-83 5 References ...................................................................................................................................... 2-85 6 7 8 Do Not Cite, Quote or Distribute 2-3 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 Executive Summary 2 Global net anthropogenic Greenhouse Gas (GHG) emissions during the last decade (2010-2019) 3 were higher than at any previous time in human history (high confidence). Since 2010, GHG 4 emissions have continued to grow reaching 59±6.6 GtCO2eq in 20191, but the average annual growth 5 in the last decade (1.3%, 2010-2019) was lower than in the previous decade (2.1%, 2000-2009) (high 6 confidence). Average annual GHG emissions were 56 GtCO2eqyr-1 for the decade 2010-2019 growing 7 by about 9.1 GtCO2eqyr-1 from the previous decade (2000-2009) – the highest decadal average on record 8 (high confidence). {2.2.2, Table 2.1, Figure 2.5} 9 Emissions growth has varied, but persisted across all groups of greenhouse gases (high 10 confidence). The average annual emission levels of the last decade (2010-2019) were higher than in any 11 previous decade for each group of greenhouse gases (high confidence). In 2019, CO2 emissions were 12 45±5.5 GtCO2,2 CH4 11±3.2 GtCO2eq, N2O 2.7±1.6 GtCO2eq and fluorinated gases (F-gases: HFCs, 13 PFCs, SF6, NF3) 1.4±0.41 GtCO2eq. Compared to 1990, the magnitude and speed of these increases 14 differed across gases: CO2 from fossil fuel and industry (FFI) grew by 15 GtCO2eqyr-1 (67%), CH4 by 15 2.4 GtCO2eqyr-1 (29%), F-gases by 0.97 GtCO2eqyr-1 (250%), N2O by 0.65 GtCO2eqyr-1 (33%). CO2 16 emissions from net land use, land-use change and forestry (LULUCF) have shown little long-term 17 change, with large uncertainties preventing the detection of statistically significant trends. F-gases 18 excluded from GHG emissions inventories such as chlorofluorocarbons and hydrochlorofluorocarbons 19 are about the same size as those included (high confidence). {2.2.1, 2.2.2, Table 2.1, Figure 2.2, Figure 20 2.3, Figure 2.5} 21 Globally, GDP per capita and population growth remained the strongest drivers of CO2 emissions 22 from fossil fuel combustion in the last decade (robust evidence, high agreement). Trends since 1990 23 continued in the years 2010 to 2019 with GDP per capita and population growth increasing emissions 24 by 2.3% and 1.2% yr-1, respectively. This growth outpaced the reduction in the use of energy per unit 25 of GDP (-2% yr-1, globally) as well as improvements in the carbon intensity of energy (-0.3%yr-1). 26 {2.4.1, Figure 2.19} 27 The global COVID-19 pandemic led to a steep drop in CO2 emissions from fossil fuel and industry 28 (high confidence). Global CO2-FFI emissions dropped in 2020 by about 5.8% (5.1% – 6.3%) or about 29 2.2 (1.9-2.4) GtCO2 compared to 2019. Emissions, however, have rebounded globally by the end of 30 December 2020 (medium confidence). {2.2.2, Figure 2.6} 31 Cumulative net CO2 emissions of the last decade (2010-2019) are about the same size as the 32 remaining carbon budget for keeping warming to 1.5°C (medium confidence). Cumulative net CO2 33 emissions since 1850 are increasing at an accelerating rate. 62% of total cumulative CO2 emissions 34 from 1850 to 2019 occurred since 1970 (1500±140 GtCO2), about 43% since 1990 (1000±90 GtCO2), 35 and about 17% since 2010 (410±30 GtCO2). For comparison, the remaining carbon budget for keeping 36 warming to 1.5°C with a 67% (50%) probability is about 400(500)±220 GtCO2. {2.2.2, Figure 2.7; 37 WG1 5.5; WG1 Table 5.8} 38 A growing number of countries have achieved GHG emission reductions longer than 10 years – 39 a few at rates that are broadly consistent with climate change mitigation scenarios that limit FOOTNOTE 1 Emissions of GHGs are weighed by Global Warming Potentials with a 100-year time horizon (GWP100) from the Sixth Assessment Report (Forster et al., 2021). GWP-100 is commonly used in wide parts of the literature on climate change mitigation and is required for reporting emissions under the United Nations Framework Convention on Climate Change (UNFCCC). All metrics have limitations and uncertainties. (Cross- Chapter Box 2, Annex II, Part II, Section 8) FOOTNOTE 2 In 2019, CO2 from fossil fuel and industry (FFI) were 38±3.0 Gt, CO2 from net land use, land-use change and forestry (LULUCF) 6.6±4.6 Gt Do Not Cite, Quote or Distribute 2-4 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 warming to well below 2°C (high confidence). There are about 24 countries that have reduced CO2 2 and GHG emissions for longer than 10 years. Reduction rates in a few countries have reached 4% in 3 some years, in line with rates observed in pathways that likely limit warming to 2°C. However, the total 4 reduction in annual GHG emissions of these countries is small (about 3.2 GtCO2eqyr-1) compared to 5 global emissions growth observed over the last decades. Complementary evidence suggests that 6 countries have decoupled territorial CO2 emissions from Gross Domestic Product (GDP), but fewer 7 have decoupled consumption-based emissions from GDP. This decoupling has mostly occurred in 8 countries with high per capita GDP and high per capita CO2 emissions. {2.2.3, 2.3.3, Figure 2.11, Table 9 2.3, Table 2.4} 10 Consumption-based CO2 emissions in developed countries and the Asia and Developing Pacific 11 region are higher than in other regions (high confidence). In developed countries, consumption-based 12 CO2 emissions peaked at 15 GtCO2 in 2007, declining to about 13 GtCO2 in 2018. The Asia and 13 Developing Pacific region, with 52% of current global population, has become a major contributor to 14 consumption-based CO2 emission growth since 2000 (5.5% yr-1 for 2000-2018); it exceeded the 15 developed countries region, which accounts for 16% of current global population, as the largest emitter 16 of consumption-based CO2. {2.3.2, Figure 2.14} 17 Carbon intensity improvements in the production of traded products have led to a net reduction 18 in CO2 emissions embodied in international trade (robust evidence, high agreement). A decrease in 19 the carbon intensity of traded products has offset increased trade volumes between 2006 and 2016. 20 Emissions embodied in internationally traded products depend on the composition of the global supply 21 chain across sectors and countries and the respective carbon intensity of production processes 22 (emissions per unit of economic output). {2.3, 2.4} 23 Developed countries tend to be net CO2 emission importers, whereas developing countries tend 24 to be net emission exporters (robust evidence, high agreement). Net CO2 emission transfers from 25 developing to developed countries via global supply chains have decreased between 2006 and 2016. 26 Between 2004 and 2011, CO2 emission embodied in trade between developing countries have more 27 than doubled (from 0.47 to 1.1 Gt) with the centre of trade activities shifting from Europe to Asia. 28 {2.3.4, Figure 2.15} 29 Emissions from developing countries have continued to grow, starting from a low base of per 30 capita emissions and with a lower contribution to cumulative emissions than developed countries 31 (robust evidence, high agreement). Average 2019 per capita CO2-FFI emissions in three developing 32 regions - Africa (1.2 tCO2/cap), Asia and developing Pacific (4.4 tCO2/cap), and Latin America and 33 Caribbean (2.7 tCO2/cap) - remained less than half that of developed countries (9.5 tCO2/cap) in 2019. 34 CO2-FFI emissions in the three developing regions together grew by 26% between 2010 and 2019, 35 compared to 260% between 1990 and 2010, while in Developed Countries emissions contracted by 36 9.9% between 2010-2019 and by 9.6% between 1990-2010. Historically, the three developing regions 37 together contributed 28% to cumulative CO2-FFI emissions between 1850 and 2019, whereas 38 Developed Countries contributed 57% and least developed countries contributed 0.4%. {2.2.3, Figure 39 2.9, Figure 2.10} 40 Globally, GHG emissions continued to rise across all sectors and subsectors; most rapidly in 41 transport and industry (high confidence). In 2019, 34% (20 GtCO eq) of global GHG emissions came 2 42 from the energy sector, 24% (14 GtCO eq) from industry, 22% (13 GtCO eq) from AFOLU, 15% (8.7 2 2 43 GtCO eq) from transport and 5.6% (3.3 GtCO eq) from buildings. Once indirect emissions from energy 2 2 44 use are considered, the relative shares of industry and buildings emissions rise to 34% and 17%, 45 respectively. Average annual GHG emissions growth during 2010-2019 slowed compared to the 46 previous decade in energy supply (from 2.3% to 1.0%) and industry (from 3.4% to 1.4%, direct 47 emissions only), but remained roughly constant at about 2% per year in the transport sector (high Do Not Cite, Quote or Distribute 2-5 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 confidence). Emission growth in AFOLU is more uncertain due to the high share of CO2-LULUCF 2 emissions. {2.4.2, Figure 2.13, Figures 2.16 to 2.21}. 3 Average annual growth in GHG emissions from energy supply decreased from 2.3% for 2000– 4 2009 to 1.0% for 2010–2019 (high confidence). This slowing of growth is attributable to further 5 improvements in energy efficiency (annually, 1.9% less energy per unit of GDP was used globally 6 between 2010 and 2019). Reductions in global carbon intensity by -0.2% yr-1 contributed further - 7 reversing the trend during 2000-2009 (+0.2% yr-1) (medium confidence). These carbon intensity 8 improvement were driven by fuel switching from coal to gas, reduced expansion of coal capacity 9 particularly in Eastern Asia, and the increased use of renewables. {2.2.4, 2.4.2.1, Figure 2.17} 10 GHG emissions in the industry, buildings and transport sectors continue to grow, driven by an 11 increase in the global demand for products and services (high confidence). These final demand 12 sectors make up 44% of global GHG emissions, or 66% when the emissions from electricity and heat 13 production are reallocated as indirect emissions to related sectors, mainly to industry and buildings. 14 Emissions are driven by the large rise in demand for basic materials and manufactured products, a global 15 trend of increasing floor space per capita, building energy service use, travel distances, and vehicle size 16 and weight. Between 2010-2019, domestic and international aviation were particularly fast growing at 17 average annual rates of +3.3% and +3.4%. Global energy efficiencies have improved in all three demand 18 sectors, but carbon intensities have not. {2.2.4; Figure 2.18; Figure 2.19; Figure 2.20}. 19 Providing access to modern energy services universally would increase global GHG emissions by 20 at most a few percent (high confidence). The additional energy demand needed to support decent 21 living standards3 for all is estimated to be well below current average energy consumption (medium 22 evidence, high agreement). More equitable income distributions can reduce carbon emissions, but the 23 nature of this relationship can vary by level of income and development (limited evidence, medium 24 agreement). {2.4.3} 25 Evidence of rapid energy transitions exists, but only at sub-global scales (medium evidence, medium 26 agreement). Emerging evidence since AR5 on past energy transitions identifies a growing number of 27 cases of accelerated technology diffusion at sub-global scales and describes mechanisms by which 28 future energy transitions may occur more quickly than those in the past. Important drivers include 29 technology transfer and cooperation, intentional policy and financial support, and harnessing synergies 30 among technologies within a sustainable energy system perspective (medium evidence, medium 31 agreement). A fast global low-carbon energy transition enabled by finance to facilitate low-carbon 32 technology adoption in developing and particularly in least developed countries can facilitate achieving 33 climate stabilisation targets (robust evidence, high agreement). {2.5.2, Table 2.5} 34 Multiple low-carbon technologies have shown rapid progress since AR5 – in cost, performance, 35 and adoption – enhancing the feasibility of rapid energy transitions (robust evidence, high 36 agreement). The rapid deployment and cost decrease of modular technologies like solar, wind, and 37 batteries have occurred much faster than anticipated by experts and modelled in previous mitigation 38 scenarios (robust evidence, high agreement). The political, economic, social, and technical feasibility 39 of solar energy, wind energy and electricity storage technologies has improved dramatically over the 40 past few years. In contrast, the adoption of nuclear energy and CO2 capture and storage in the electricity 41 sector has been slower than the growth rates anticipated in stabilisation scenarios. Emerging evidence 42 since AR5 indicates that small-scale technologies (e.g. solar, batteries) tend to improve faster and be 43 adopted more quickly than large-scale technologies (nuclear, CCS) (medium evidence, medium 44 agreement). {2.5.3, 2.5.4, Figures 2.22 and 2.23} FOOTNOTE 3 Decent Living Standards (DLS) – a benchmark of material conditions for human well-being – overlaps with many Sustainable Development Goals (SDGs). Minimum requirements of energy use consistent with enabling well-being for all is between 20 and 50 GJ cap-1 yr-1 23 depending on the context.{5.2.2, 5.2.2, Box 5.3, Figure 5.6} Do Not Cite, Quote or Distribute 2-6 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 Robust incentives for investment in innovation, especially incentives reinforced by national policy 2 and international agreements, are central to accelerating low-carbon technological change (robust 3 evidence, medium agreement). Policies have driven innovation, including instruments for technology 4 push (e.g., scientific training, R&D) and demand pull (e.g., carbon pricing, adoption subsidies), as well 5 as those promoting knowledge flows and especially technology transfer. The magnitude of the scale-up 6 challenge elevates the importance of rapid technology development and adoption. This includes 7 ensuring participation of developing countries in an enhanced global flow of knowledge, skills, 8 experience, equipment, and technology itself requires strong financial, institutional, and capacity 9 building support (robust evidence, high agreement). {2.5.4, 2.5, 2.8} 10 The global wealthiest 10% contribute about 36-45% of global GHG emissions (robust evidence, 11 high agreement). The global 10% wealthiest consumers live in all continents, with two thirds in high- 12 income regions and one third in emerging economies (robust evidence, medium agreement). The 13 lifestyle consumption emissions of the middle income and poorest citizens in emerging economies are 14 between 5-50 times below their counterparts in high-income countries (medium evidence, medium 15 agreement). Increasing inequality within a country can exacerbate dilemmas of redistribution and social 16 cohesion, and affect the willingness of rich and poor to accept lifestyle changes for mitigation and 17 policies to protect the environment (medium evidence, medium agreement) {2.6.1, 2.6.2, Figure 2.25} 18 Estimates of future CO2 emissions from existing fossil fuel infrastructures already exceed 19 remaining cumulative net CO2 emissions in pathways limiting warming to 1.5°C with no or 20 limited overshoot (high confidence). Assuming variations in historic patterns of use and 21 decommissioning, estimated future CO2 emissions from existing fossil fuel infrastructure alone are 660 22 (460-890) GtCO2 and from existing and currently planned infrastructure 850 (600-1100) GtCO2. This 23 compares to overall cumulative net CO2 emissions until reaching net zero CO2 of 510 (330-710) Gt in 24 pathways that limit warming to 1.5°C with no or limited overshoot, and 880 (640-1160) Gt in pathways 25 that limit likely warming to 2°C (high confidence). While most future CO2 emissions from existing and 26 currently planned fossil fuel infrastructure are situated in the power sector, most remaining fossil fuel 27 CO2 emissions in pathways that limit likely warming to 2°C and below are from non-electric energy – 28 most importantly from the industry and transportation sectors (high confidence). Decommissioning and 29 reduced utilization of existing fossil fuel installations in the power sector as well as cancellation of new 30 installations are required to align future CO2 emissions from the power sector with projections in these 31 pathways (high confidence). {2.7.2, 2.7.3, Figure 2.26, Table 2.6, Table 2.7} 32 A broad range of climate policies, including instruments like carbon pricing, play an increasing 33 role in GHG emissions reductions. The literature is in broad agreement, but the magnitude of the 34 reduction rate varies by the data and methodology used, country, and sector (robust evidence, high 35 agreement). Countries with a lower carbon pricing gap (higher carbon price) tend to be less carbon 36 intensive (medium confidence). {2.8.2, 2.8.3} 37 Climate-related policies have also contributed to decreasing GHG emissions. Policies such as taxes 38 and subsidies for clean and public transportation, and renewable policies have reduced GHG emissions 39 in some contexts (robust evidence, high agreement). Pollution control policies and legislations that go 40 beyond end-of-pipe controls have also had climate co-benefits, particularly if complementarities with 41 GHG emissions are considered in policy design (medium evidence, medium agreement). Policies on 42 agriculture, forestry and other land use (AFOLU) and AFOLU sector-related policies such as 43 afforestation policies can have important impacts on GHG emissions (medium evidence, medium 44 agreement). {2.8.4} 45 Do Not Cite, Quote or Distribute 2-7 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2.1 Introduction 2 As demonstrated by the contribution of Working Group I to the Sixth Assessment Report (IPCC, 3 2021a), greenhouse gas 4 (GHG) concentrations in the atmosphere and annual anthropogenic GHG 4 emissions continue to grow and have reached a historic high driven mainly by continued fossil fuels 5 use (Peters et al., 2020; Jackson et al., 2019; Friedlingstein et al., 2020). Unsurprisingly, a large volume 6 of new literature has emerged since the AR5 on the trends and underlying drivers of anthropogenic 7 GHG emissions. This chapter provides a structured assessment of this new literature and establishes the 8 most important thematic links to other chapters in this report. 9 While AR5 has mostly assessed GHG emissions trends and drivers between 1970 and 2010, this 10 assessment focusses on the period 1990–2019 with the main emphasis on changes since 2010. 11 Compared to Chapter 5 in the contribution of WG III to the AR5 (Blanco et al., 2014), the scope of the 12 present chapter is broader. It presents the historical background of global progress in climate change 13 mitigation for the rest of the report and serves as a starting point for the assessment of long-term as well 14 as near- and medium-term mitigation pathways in Chapters 3 and 4, respectively. It also provides a 15 systemic perspective on past emission trends in different sectors of the economy (Chapters 6–12), and 16 relates GHG emissions trends to past policies (Chapter 13) and observed technological development 17 (Chapter 16). There is also a greater thrust into the analysis of consumption-based sectoral emissions 18 trends, empirical evidence of emissions consequences of behavioural choices and lifestyles, and the 19 social aspects of mitigation (Chapter 5). Finally, a completely new section discusses the mitigation 20 implications of existing and planned long-lived infrastructure and carbon lock-in. 21 Figure 2.1 presents the road map of this chapter. It is a simplified illustration of the causal chain driving 22 emissions along the black arrows. It also highlights the most important linkages to other chapters in this 23 volume (blue lines). The logic of the figure is the following: the main topic of this chapter is trends of 24 GHG emissions (discussed only in this chapter at such level of detail), hence they are at the top of the 25 figure in yellow-shaded boxes. The secondary theme is the drivers behind these trends, depicted in the 26 second line of yellow-shaded boxes. Four categories of drivers highlight key issues and guide readers 27 to chapters in which more details are presented. Finally, in addition to their own motivations and 28 objectives, climate and non-climate policies and measures shape the aspirations and activities of actors 29 in the main driver categories, hence shown in the yellow-shaded box below. FOOTNOTE4 Greenhouse gases are gaseous constituents of the atmosphere that absorb and emit radiation at specific wavelengths within the spectrum of radiation emitted by the Earth's surface, by the atmosphere itself, and by clouds. This property causes the greenhouse effect. Water vapour (H 2O), carbon dioxide (CO2), nitrous oxide (N2O), methane (CH4), and ozone (O3) are the primary GHGs in the Earth's atmosphere. Human-made GHGs include sulphur hexafluoride (SF6), hydrofluorocarbons (HFCs), chlorofluorocarbons (CFCs), and perfluorocarbons (PFCs); see Annex I. Do Not Cite, Quote or Distribute 2-8 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 Figure 2.1 Chapter 2 road map and linkages to other chapters 3 Black arrows show the causal chain driving emissions, blue lines indicate key linkages to other chapters in this 4 report. 5 Accordingly, the yellow-shaded boxes at the top of Figure 2.1 show that the first part of the chapter 6 presents GHG emissions from two main perspectives: their geographical locations and the places where 7 goods are consumed and services are utilised. A complicated chain of factors called drivers underlie 8 these emissions. They are linked across time, space, and various segments of the economy and society 9 in complex non-linear relationships. Sections shown in the second row of yellow-shaded boxes assess 10 the latest literature and improve the understanding of the relative importance of these drivers in 11 mitigating GHG emissions. A huge mass of physical capital embodying immense financial assets and 12 potentially operating over a long lifetime produces vast GHG emissions. This long-lived infrastructure 13 can be a significant hindrance to fast and deep reductions of emissions, it is therefore also shown as an 14 important driver. A large range of economic, social, environmental, and other policies has been shaping 15 these drivers of GHG emissions in the past and are anticipated to influence them in the future, as 16 indicated by the yellow-shaded policies box and its manifold linkages. As noted, blue lines show 17 linkages of sections to other chapters discussing these drivers and their operating mechanisms in detail. 18 19 2.2 Past and present trends of territorial GHG emissions 20 Total anthropogenic greenhouse gas (GHG) emissions as discussed in this chapter comprise CO2 21 emissions from fossil fuel combustion and industrial processes5 (FFI), net CO2 emissions from land use, FOOTNOTE5 Industrial processes relate to CO2 releases from fossil fuel oxidation and carbonate decomposition. Do Not Cite, Quote or Distribute 2-9 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 land-use change, and forestry (CO2-LULUCF) (in previous IPCC reports often named FOLU: forestry 2 and other land-use), methane (CH4), nitrous oxide (N2O) and fluorinated gases (F-gases) comprising 3 hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), sulphur hexafluoride (SF6) as well as nitrogen 4 trifluoride (NF3). There are other major sources of F-gas emissions that are regulated under the Montreal 5 Protocol such as chlorofluorocarbons (CFCs) and hydrochlorofluorocarbons (HCFCs) that also have 6 considerable warming impacts (see Figure 2.4), however they are not considered here. Other substances 7 including ozone and aerosols that further contribute climate forcing are only treated very briefly, but a 8 full chapter is devoted to it in the Working Group I contribution to AR6 (Naik et al., 2021a; b). 9 A growing number of global GHG emissions inventories have become available since AR5 (Minx et 10 al., 2021). However, only a few are comprehensive in their coverage of sectors, countries and gases – 11 namely EDGAR (Emissions Database for Global Atmospheric Research) (Crippa et al., 2021), 12 PRIMAP (Potsdam Real-time Integrated Model for probabilistic Assessment of emissions Paths) 13 (Gütschow et al., 2021a), CAIT (Climate Analysis Indicators Tool) (WRI, 2019) and CEDS (A 14 Community Emissions Data System for Historical Emissions) (Hoesly et al., 2018). None of these 15 inventories presently cover CO2-LULUCF, while CEDS excludes F-gases. For individual gases and 16 sectors, additional GHG inventories are available, as shown in Figure 2.2, but each has varying system 17 boundaries leading to important differences between their respective estimates (Section 2.2.1). Some 18 inventories are compiled bottom-up, while others are produced synthetically and are dependent on other 19 inventories. A more comprehensive list and discussion of different datasets is provided in the Chapter 20 2 Supplementary Material (SM2.1) and in Minx et al. (2021). 21 Across this report version 6 of EDGAR (Crippa et al., 2021) provided by the Joint Research Centre of 22 the European Commission is used for a consistent assessment of GHG emission trends and drivers. It 23 covers anthropogenic releases of CO2-FFI, CH4, N2O, and F-gas (HFCs, PFCs, SF6, NF3) emissions by 24 228 countries and territories and across 5 sectors and 27 subsectors. EDGAR is chosen, because it 25 provides the most comprehensive global dataset in its coverage of sources, sectors and gases. For 26 transparency and as part of the uncertainty assessment EDGAR is compared to other global datasets in 27 Section 2.2.1 as well as in the Chapter 2 Supplementary Material (SM2.1). For individual country 28 estimates of GHG emissions, it may be more appropriate to use inventory data submitted to the 29 UNFCCC under the common reporting format (CRF) (UNFCCC, 2021). However, these inventories 30 are only up to date for Annex I countries and cannot be used to estimate global or regional totals. As 31 part of the regional analysis, a comparison of EDGAR and CRF estimates at the country-level is 32 provided, where the latter is available (Figure 2.9). 33 Net CO2-LULUCF estimates are added to the dataset as the average of estimates from three 34 bookkeeping models of land-use emissions (Houghton and Nassikas, 2017; Hansis et al., 2015; Gasser 35 et al., 2020) following the Global Carbon Project (Friedlingstein et al., 2020). This is different to AR5, 36 where land-based CO2 emissions from forest fires, peat fires, and peat decay, were used as an 37 approximation of the net-flux of CO2-LULUCF (Blanco et al., 2014). Note that the definition of CO2- 38 LULUCF emissions by global carbon cycle models, as used here, differs from IPCC definitions (IPCC, 39 2006) applied in national greenhouse gas inventories (NGHGI) for reporting under the climate 40 convention (Grassi et al., 2018, 2021) and, similarly, from FAO estimates of carbon fluxes on forest 41 land (Tubiello et al., 2021). The conceptual difference in approaches reflects different scopes. We use 42 the global carbon cycle models’ approach for consistency with Working Group I (Canadell et al., 2021) 43 and to comprehensively distinguish natural from anthropogenic drivers, while NGHGI generally report 44 as anthropogenic all CO2 fluxes from lands considered managed (see Section 7.2.2 in Chapter 7). 45 Finally, note that the CO2-LULUCF estimate from bookkeeping models as provided in this chapter is 46 indistinguishable to the CO2 from Agriculture, Forestry and other Land Use (AFOLU) as reported in 47 Chapter 7, because the CO2 emissions component from agriculture is negligible. Do Not Cite, Quote or Distribute 2-10 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 The resulting synthetic dataset used here has undergone additional peer-review and is publicly available 2 (https://doi.org/10.5281/zenodo.5566761). Comprehensive information about the dataset as well as 3 underlying uncertainties (including a comparison with other datasets) can be found in the 4 Supplementary Material to this chapter and in Minx et al. (2021). 5 In this chapter and the report as a whole, different greenhouse gases are frequently converted into 6 common units of CO2 equivalent (CO2eq) emissions using 100-year Global Warming Potentials 7 (GWP100) from WGI of IPCCs Sixth Assessment Report (AR6) (Forster et al., 2021a). This reflects 8 the dominant use in the scientific literature and is consistent with decisions made by Parties to the Paris 9 Agreement for reporting and accounting of emissions and removals (UNFCCC, 2019). Other GHG 10 emissions metrics exist, all of which, like GWP100, are designed for specific purposes and have 11 limitations and uncertainties. The appropriate choice of GHG emissions metrics depends on policy 12 objective and context (Myhre et al., 2013; Kolstad et al., 2015). A discussion of GHG metrics is 13 provided in a Cross-Chapter Box later in the chapter (see Cross-Chapter Box 2) and, at length, in the 14 Chapter 2 Supplementary Material. Throughout the chapter GHG emissions are reported (in GtCO2eq) 15 at two significant digits to reflect prevailing uncertainties in emissions estimates. Estimates are subject 16 to uncertainty, which we report for a 90% confidence interval. 17 18 2.2.1 Uncertainties in GHG emissions 19 Estimates of historical GHG emissions – CO2, CH4, N2O and F-gases – are uncertain to different 20 degrees. Assessing and reporting uncertainties is crucial in order to understand whether available 21 estimates are sufficiently robust to answer policy questions; for example, if GHG emissions are still 22 rising, or if a country has achieved an emission reduction goal (Marland, 2008). These uncertainties can 23 be of scientific nature, such as when a process is not sufficiently understood. They also arise from 24 incomplete or unknown parameter information (e.g. activity data, or emission factors), as well as 25 estimation uncertainties from imperfect modelling techniques. There are at least three major ways to 26 examine uncertainties in emission estimates (Marland et al., 2009): 1) by comparing estimates made by 27 independent methods and observations (e.g. comparing atmospheric measurements with bottom-up 28 emissions inventory estimates) (Saunois et al., 2020; Petrescu et al., 2020b; a; Tian et al., 2020); 2) by 29 comparing estimates from multiple sources and understanding sources of variation (Andrew, 2020; 30 Macknick, 2011; Ciais et al., 2021; Andres et al., 2012); 3) by evaluating estimates from a single source 31 (Hoesly and Smith, 2018), for instance via statistical sampling across parameter values (e.g. Robert J. 32 Andres et al., 2014; Monni et al., 2007; Solazzo et al., 2021; Tian et al., 2019). 33 Uncertainty estimates can be rather different depending on the method chosen. For example, the range 34 of estimates from multiple sources is bounded by their interdependency; they can be lower than true 35 structural plus parameter uncertainty or than estimates made by independent methods. In particular, it 36 is important to account for potential bias in estimates, which can result from using common 37 methodological or parameter assumptions, or from missing sources (systemic bias). It is further crucial 38 to account for differences in system boundaries, i.e. which emissions sources are included in a dataset 39 and which are not, otherwise direct comparisons can exaggerate uncertainties (Macknick, 2011; 40 Andrew, 2020). Independent top-down observational constraints are, therefore, particularly useful to 41 bound total emission estimates, but are not yet capable of verifying emission levels or trends (Petrescu 42 et al., 2021a; b). Similarly, uncertainties estimates are influenced by specific modelling choices. For 43 example, uncertainty estimates from studies on the propagation of uncertainties associated with key 44 input parameters (activity data, emissions factors) following the IPCC Guidelines (IPCC, 2006) are 45 strongly determined by assumptions on how these parameters are correlated between sectors, countries, 46 and regions (Solazzo et al., 2021; Janssens-Maenhout et al., 2019). Assuming (full) covariance between 47 source categories, and therefore dependence between them, increases uncertainty estimates. Estimates Do Not Cite, Quote or Distribute 2-11 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 allowing for some covariance as in Sollazo et al. (2021) also tend to yield higher estimates than the 2 range of values from ensemble of dependent inventories (Saunois et al., 2016, 2020). 3 For this report, a comprehensive assessment of uncertainties is provided in the Supplementary Material 4 (SM2.2) to this chapter based on Minx et al. (2021). The uncertainties reported here combine statistical 5 analysis, comparisons of global emissions inventories and an expert judgement of the likelihood of 6 results lying outside a defined confidence interval, rooted in an understanding gained from the relevant 7 literature. This literature has improved considerably since AR5 with a growing number of studies that 8 assess uncertainties based on multiple lines of evidence (Petrescu et al., 2021a; b; Tian et al., 2020; 9 Saunois et al., 2016, 2020). 10 To report the uncertainties in GHG emissions estimates, a 90% confidence interval (5th-95th percentile) 11 is adopted, i.e., there is a 90 % likelihood that the true value will be within the provided range if the 12 errors have a Gaussian distribution, and no bias is assumed. This is in line with previous reporting in 13 IPCC AR5 (Ciais et al., 2014; Blanco et al., 2014). Note that national emissions inventory submissions 14 to the UNFCCC are requested to report uncertainty using a 95% confidence interval. The use of this 15 broader uncertainty interval implies, however, a relatively high degree of knowledge about 16 the uncertainty structure of the associated data, particularly regarding the distribution of uncertainty in 17 the tails of the probability distributions. Such a high degree of knowledge is not present over all regions, 18 emission sectors and species considered here. 19 Based on this assessment of relevant uncertainties above, a constant, relative, global uncertainty 20 estimates for GHGs is applied at a 90% confidence interval that range from relatively low values for 21 CO2-FFI (±8%), to intermediate values for CH4 and F-gases (±30%), to higher values for N2O (±60%) 22 and CO2-LULUCF (±70%). Uncertainties for aggregated total GHG emissions in terms of CO2eq 23 emissions are calculated as the square root of the squared sums of absolute uncertainties for individual 24 gases (taking F-gases together), using 100-year Global Warming Potentials (GWP100) to weight 25 emissions of non-CO2 gases but excluding uncertainties in the metric itself. 26 This assessment of uncertainties is broadly in line with WGIII AR5 (Blanco et al., 2014), but revises 27 individual uncertainty judgements in line with the more recent literature (Friedlingstein et al., 2020; 28 Janssens-Maenhout et al., 2019; Solazzo et al., 2021; Tian et al., 2020; Saunois et al., 2016, 2020) as 29 well as the underlying synthetic analysis provided here (e.g. Figure 2.2, Figure 2.3, Minx et al. (2021)). 30 As such, reported changes in these estimates do not reflect changes in the underlying uncertainties, but 31 rather a change in expert judgement based on an improved evidence base in the scientific literature. 32 Uncertainty estimates for CO2-FFI and N2O remain unchanged compared to AR5. The change in the 33 uncertainty estimates for CH4 from 20% to 30% is justified by larger uncertainties reported for EDGAR 34 emissions (Solazzo et al., 2021; Janssens-Maenhout et al., 2019) as well as the wider literature (Tubiello 35 et al., 2015; Kirschke et al., 2013; Saunois et al., 2020, 2016). As AR6 – in contrast to AR5 - uses CO2- 36 LULUCF data from global bookkeeping models, the respective uncertainty estimate is based on the 37 reporting in the underlying literature (Friedlingstein et al., 2020) as well as Working Group I (Canadell 38 et al., 2021). The 70% uncertainty value is at the higher end of the range considered in AR5 (Blanco et 39 al., 2014). 40 Finally, for F-gas emissions top-down atmospheric measurements from the 2018 World Meteorological 41 Organisation’s (WMO) Scientific Assessment of Ozone Depletion (see Engel and Rigby, 2018; 42 Montzka and Velders, 2018) are compared to the data used in this report (Minx et al., 2021; Crippa et 43 al., 2021) as shown in Figure 2.3. Due to the general absence of natural F-gas fluxes, there is a sound 44 understanding of global and regional F-gas emissions from top-down estimates of atmospheric 45 measurements with small and well-understood measurement, lifetime and transport model uncertainties 46 (see Engel and Rigby, 2018; Montzka and Velders, 2018). However, when species are aggregated into 47 total F-gas emissions, EDGARv6 emissions are around 10% lower than the WMO 2018 values 48 throughout, with larger differences for individual f-gas species, and further discrepancies when Do Not Cite, Quote or Distribute 2-12 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 comparing to older EDGAR versions. Based on this, the overall uncertainties for aggregate F-gas 2 emissions is judged conservatively at 30% - 10 percentage points higher than in AR5 (Blanco et al., 3 2014). 4 Aggregate uncertainty across all greenhouse gases is approximately ±11% depending on the 5 composition of gases in a particular year. AR5 applied a constant uncertainty estimates of ±10% for 6 total GHG emissions. The upwards revision applied to the uncertainties of CO2-LULUCF, CH4 and F- 7 gas emissions therefore has a limited overall effect on the assessment of GHG emissions. 8 9 Figure 2.2 Estimates of global anthropogenic greenhouse gas emissions from different data sources 1970- 10 2019. 11 Top-left panel: CO2 FFI emissions from: EDGAR - Emissions Database for Global Atmospheric Research 12 (this dataset) (Crippa et al., 2021); GCP – Global Carbon Project (Friedlingstein et al., 2020; Andrew and 13 Peters, 2021); CEDS - Community Emissions Data System (Hoesly et al., 2018; O’Rourke et al., 2021); 14 CDIAC Global, Regional, and National Fossil-Fuel CO2 Emissions (Gilfillan et al., 2020); PRIMAP-hist - 15 Potsdam Real-time Integrated Model for probabilistic Assessment of emissions Paths (Gütschow et al., 16 2016, 2021b); EIA - Energy Information Administration International Energy Statistics (EIA, 2021); BP - 17 BP Statistical Review of World Energy (BP, 2021); IEA - International Energy Agency (IEA, 2021a; b); 18 IPPU refers to emissions from industrial processes and product use. Top-right panel: Net anthropogenic 19 CO2-LULUCF emissions from: BLUE – Bookkeeping of land-use emissions (Hansis et al., 2015; 20 Friedlingstein et al., 2020); DGVM-mean – Multi-model mean of CO2-LULUCF emissions from dynamic 21 global vegetation models (Friedlingstein et al., 2020); OSCAR – an earth system compact model (Gasser 22 et al., 2020; Friedlingstein et al., 2020); HN – Houghton and Nassikas Bookkeeping Model (Houghton and 23 Nassikas, 2017; Friedlingstein et al., 2020); for comparison, the net CO2 flux from FAOSTAT (FAO Tier 24 1) is plotted, which comprises net emissions and removals on forest land and from net forest conversion 25 (Tubiello et al., 2021; FAOSTAT, 2021), emissions from drained organic soils under cropland/grassland 26 (Conchedda and Tubiello, 2020), and fires in organic soils (Prosperi et al., 2020), as well as a net CO2 flux 27 estimate from National Greenhouse Gas Inventories (NGHGI) based on country reports to the UNFCCC, 28 which include land use change, and fluxes in managed lands (Grassi et al., 2021). Bottom-left panel: 29 Anthropogenic CH4 emissions from: EDGAR (above); CEDS (above); PRIMAP-hist (above); GAINS - 30 The Greenhouse gas – Air pollution Interactions and Synergies Model (Höglund-Isaksson et al., 2020); 31 EPA-2019: Greenhouse gas emission inventory (US-EPA, 2019); FAO –FAOSTAT inventory emissions 32 (Tubiello et al., 2013; Tubiello, 2018; FAOSTAT, 2021); Bottom-right panel: Anthropogenic N2O 33 emissions from: GCP – global nitrous oxide budget (Tian et al., 2020); CEDS (above); EDGAR (above); Do Not Cite, Quote or Distribute 2-13 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 PRIMAP-hist (above); GAINS (Winiwarter et al., 2018); EPA-2019 (above); FAO (above). Differences in 2 emissions across different versions of the EDGAR dataset are shown in the Supplementary Material (Fig. 3 SM2.2). 4 Source: Minx et al. (2021) 5 6 Figure 2.3 Comparison between top-down estimates and bottom-up EDGAR inventory data on GHG 7 emissions for 1980-2016 8 Left panel: Total GWP-100-weighted emissions based on IPCC AR6 (Forster et al., 2021a) of F-gases in 9 Olivier & Peters (2020) [EDGARv5FT] (red dashed line, excluding C4F10, C5F12, C6F14 and C7F16) and 10 EDGARv6 (purple dashed line) compared to top-down estimates based on AGAGE and NOAA data from 11 WMO (2018) (blue lines; Engel and Rigby (2018); Montzka and Velders (2018)). Right panel: Top-down 12 aggregated emissions for the three most abundant CFCs (-11, -12 and -113) and HCFCs (-22, -141b, - 13 142b) not covered in bottom-up emissions inventories are shown in green and orange. For top-down 14 estimates the shaded areas between two respective lines represent 1σ uncertainties. 15 Source: Minx et al. (2021). 16 GHG emissions metrics such as GWP-100 have themselves uncertainties, which has been largely 17 neglected in the literature so far. Minx et al. (2021) report the uncertainty in GWP-100 metric values as 18 ±50% for methane and other SLCFs, and ±40% for non-CO2 gases with longer atmospheric lifetimes 19 (specifically, those with lifetimes longer than 20 years). If uncertainties in GHG metrics are considered, 20 and are assumed independent (which may lead to an underestimate) the overall uncertainty of total GHG 21 emissions in 2019 increases from ±11% to ±13%. Metric uncertainties are not further considered in this 22 chapter (but see Cross-chapter Box 2 and Chapter 2 Supplementary Material on GHG metrics (SM2.3)). 23 The most appropriate metric to aggregate GHG emissions depends on the objective (see Cross-chapter 24 Box 2). One such objective can be to understand the contribution of emissions in any given year to 25 warming, while another can be to understand the contribution of cumulative emissions over an extended Do Not Cite, Quote or Distribute 2-14 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 time period to warming. In Figure 2.4 the modelled warming from emissions of each gas or group of 2 gases is also shown - calculated using the reduced-complexity climate model FAIRv1.6, which has been 3 calibrated to match several aspects of the overall WGI assessment (Forster et al., 2021a; specifically 4 Cross-Chapter Box 7 in Chapter 10 therein). Additionally, its temperature response to emissions with 5 shorter atmospheric lifetimes such as aerosols, methane or ozone has been adjusted to broadly match 6 those presented in Naik et al. (2021a). There are some differences in actual warming compared to the 7 GWP-100 weighted emissions of each gas (Figure 2.4), in particular a greater contribution from CH4 8 emissions to historical warming. This is consistent with warming from CH4 being short-lived and hence 9 having a more pronounced effect in the near-term during a period of rising emissions. Nonetheless, 10 Figure 2.4 highlights that emissions weighted by GWP-100 do not provide a fundamentally different 11 information about the contribution of individual gases than modelled actual warming over the historical 12 period, when emissions of most GHGs have been rising continuously, with CO2 being the dominant and 13 CH4 being the second most important contributor to GHG-induced warming. Other metrics such as 14 GWP* (Cain et al., 2019) offer an even closer resemblance between cumulative CO2eq emissions and 15 temperature change. Such a metric may be more appropriate when the key objective is to track 16 temperature change when emissions are falling, as in mitigation scenarios. 17 18 Figure 2.4 Contribution of different GHGs to global warming over the period 1750 to 2018 19 Top row: contributions estimated with the FaIR reduced-complexity climate model. Major GHGs and 20 aggregates of minor gases as a timeseries in a) and as a total warming bar chart with 90% confidence 21 interval added in b). Bottom row: contribution from short-lived climate forcers as a time series in c) and 22 as a total warming bar chart with 90% confidence interval added in d). The dotted line in c) gives the net 23 temperature change from short-lived climate forcers other than CH4. F-Kyoto/Paris includes the gases 24 covered by the Kyoto Protocol and Paris Agreement, while F-other includes the gases covered by the 25 Montreal Protocol but excluding the HFCs. 26 Source: Minx et al., 2021 27 Do Not Cite, Quote or Distribute 2-15 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 START CROSS CHAPTER BOX 2 HERE 2 3 Cross-Chapter Box 2 GHG emission metrics 4 5 Authors: Andy Reisinger (New Zealand), Alaa Al Khourdajie (United Kingdom/Syria), Kornelis Blok 6 (the Netherlands), Harry Clark (New Zealand), Annette Cowie (Australia), Jan S. Fuglestvedt 7 (Norway), Oliver Geden (Germany), Veronika Ginzburg (the Russian Federation), Celine Guivarch 8 (France), Joanna House (United Kingdom), Jan Christoph Minx (Germany), Rachid Mrabet (Morocco), 9 Gert-Jan Nabuurs (the Netherlands), Glen P. Peters (Norway/Australia), Keywan Riahi (Austria), 10 Roberto Schaeffer (Brazil), Raphael Slade (United Kingdom), Anders Hammer Strømman (Norway), 11 Detlef van Vuuren (the Netherlands) 12 13 Comprehensive mitigation policy relies on consideration of all anthropogenic forcing agents, which 14 differ widely in their atmospheric lifetimes and impacts on the climate system. GHG emission metrics6 15 provide simplified information about the effects that emissions of different GHGs have on global 16 temperature or other aspects of climate, usually expressed relative to the effect of emitting CO2 (see 17 glossary). This information can inform prioritisation and management of trade-offs in mitigation 18 policies and emission targets for non-CO2 gases relative to CO2, as well as for baskets of gases expressed 19 in CO2-eq. This assessment builds on the evaluation of GHG emission metrics from a physical science 20 perspective by Working Group I (Forster et al., 2021b). For additional details and supporting references, 21 see Chapter 2 Supplementary Material (SM2.3) and Annex II 8. 22 The Global Warming Potential (GWP) and the Global Temperature change Potential (GTP) were the 23 main metrics assessed in AR5 (Myhre et al., 2013; Kolstad et al., 2014). The GWP with a lifetime of 24 100 years (GWP100) continues to be the dominant metric used in the scientific literature on mitigation 25 assessed by WGIII. The assessment by Working Group I (Forster et al., 2021) includes updated values 26 for these metrics based on updated scientific understanding of the response of the climate system to 27 emissions of different gases, including changing background concentrations. It also assess new metrics 28 published since AR5. Metric values in the AR6 include climate-carbon cycle feedbacks by default; this 29 provides an important update and clarification from the AR5 which reported metric values both with 30 and without such feedbacks. 31 The choice of metric, including time horizon, should reflect the policy objectives for which the metric 32 is applied (Plattner et al., 2009). Recent studies confirm earlier findings that the GWP is consistent with 33 a cost-benefit framework (Kolstad et al., 2014), which implies weighting each emission based on the 34 economic damages that this emission will cause over time, or conversely, the avoided damages from 35 avoiding that emission. The GWP time horizon can be linked to the discount rate used to evaluate 36 economic damages from each emission. For methane, GWP100 implies a social discount rate of about 37 3-5% depending on the assumed damage function, whereas GWP20 implies a much higher discount rate, 38 greater than 10% (medium confidence; Mallapragada and Mignone 2019; Sarofim and Giordano 2018). 39 The dynamic GTP is aligned with a cost-effectiveness framework, as it weights each emission based on 40 its contribution to global warming in a specified future year (e.g. the expected year of peak warming 41 for a given temperature goal). This implies a shrinking time horizon and increasing relative importance 42 of SLCF emissions as the target year is approached (Johansson, 2011; Aaheim and Mideksa, 2017). 43 The GTP with a static time horizon (e.g. GTP100) is not well-matched to either a cost-benefit or a cost- 44 effectiveness framework, as the year for which the temperature outcome is evaluated would not match FOOTNOTE6 Emission metrics also exist for aerosols, but these are not commonly used in climate policy. This assessment focuses on GHG emission metrics only. Do Not Cite, Quote or Distribute 2-16 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 the year of peak warming, nor the overall damages caused by each emission (Mallapragada and 2 Mignone, 2017; Edwards and Trancik, 2014; Strefler et al., 2014). 3 A number of studies since the AR5 have evaluated the impact of various GHG emission metrics and 4 time horizons on the global economic costs of limiting global average temperature change to a pre- 5 determined level (e.g. Strefler et al. 2014; Harmsen et al. 2016; Tanaka et al. 2021; see SM2.3 for 6 additional detail). These studies indicate that for mitigation pathways that likely limit warming to 2°C 7 above pre-industrial levels or lower, using GWP100 to inform cost-effective abatement choices between 8 gases would achieve such long-term temperature goals at close to least global cost within a few percent 9 (high confidence). Using the dynamic GTP instead of GWP100 could reduce global mitigation costs by 10 a few percent in theory (high confidence), but the ability to realise those cost savings depends on the 11 temperature limit, policy foresight and flexibility in abatement choices as the weighting of SLCF 12 emissions increases over time (medium confidence; van den Berg et al. 2015; Huntingford et al. 2015). 13 Similar benefits as for the dynamic GTP might be obtained by regularly reviewing and potentially 14 updating the time horizon used for GWP in light of actual emission trends compared to climate goals 15 (Tanaka et al., 2020). 16 The choice of metric and time horizon can affect the distribution of costs and the timing of abatement 17 between countries and sectors in cost-effective mitigation strategies. Sector-specific lifecycle 18 assessments find that different emission metrics and different time horizons can lead to divergent 19 conclusions about the effectiveness of mitigation strategies that involve reductions of one gas but an 20 increase of another gas with a different lifetime (e.g. Tanaka et al. 2019). Assessing the sensitivity of 21 conclusions to different emission metrics and time horizons can support more robust decision-making 22 (Levasseur et al. 2016; Balcombe et al. 2018; see SM2.3 for details). Sectoral and national perspectives 23 on GHG emission metrics may differ from a global least-cost perspective, depending on other policy 24 objectives and equity considerations, but the literature does not provide a consistent framework for 25 assessing GHG emission metrics based on equity principles. 26 Literature since the AR5 has emphasized that the GWP100 is not well suited to estimating the warming 27 effect at specific points in time from sustained SLCF emissions (e.g. Allen et al. 2016; Cain et al. 2019; 28 Collins et al. 2019). This is because the warming caused by an individual SLCF emission pulse 29 diminishes over time and hence, unlike CO2, the warming from SLCF emissions that are sustained over 30 multiple decades to centuries depends mostly on their ongoing rate of emissions rather than their 31 cumulative emissions. Treating all gases interchangeably based on GWP100 within a stated emissions 32 target therefore creates ambiguity about actual global temperature outcomes (Fuglestvedt et al., 2018; 33 Denison et al., 2019). Supplementing economy-wide emission targets with information about the 34 expected contribution from individual gases to such targets would reduce the ambiguity in global 35 temperature outcomes. 36 Recently developed step/pulse metrics such as the CGTP (Combined Global Temperature Change 37 Potential; Collins et al. 2019) and GWP* (referred to as GWP-star; Allen et al. 2018; Cain et al. 2019) 38 recognise that a sustained increase/decrease in the rate of SLCF emissions has a similar effect on global 39 surface temperature over multiple decades as a one-off pulse emission/removal of CO2. These metrics 40 use this relationship to calculate the CO2 emissions or removals that would result in roughly the same 41 temperature change as a sustained change in the rate of SLCF emissions (CGTP) over a given time 42 period, or as a varying time series of CH4 emissions (GWP*). From a mitigation perspective, these 43 metrics indicate greater climate benefits from rapid and sustained methane reductions over the next few 44 decades than if such reductions are weighted by GWP100, while conversely, sustained methane increases 45 have greater adverse climate impacts (Lynch et al., 2020; Collins et al., 2019). The ability of these 46 metrics to relate changes in emission rates of short-lived gases to cumulative CO2 emissions makes 47 them well-suited, in principle, to estimating the effect on the remaining carbon budget from more, or Do Not Cite, Quote or Distribute 2-17 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 less, ambitious SLCF mitigation over multiple decades compared to a given reference scenario (high 2 confidence; Collins et al. 2019; Forster et al. 2021). 3 The potential application of GWP* in wider climate policy (e.g. to inform equitable and ambitious 4 emission targets or to support sector-specific mitigation policies) is contested, although relevant 5 literature is still limited (Rogelj and Schleussner, 2019; Schleussner et al., 2019; Cain et al., 2021; 6 Rogelj and Schleussner, 2021; Allen et al., 2021). Whereas GWP and GTP describe the marginal effect 7 of each emission relative to the absence of that emission, GWP* describes the equivalent CO2 emissions 8 that would give the same temperature change as an emissions trajectory of the gas considered, starting 9 at a (user-determined) reference point. The warming based on those cumulative CO2-equivalent 10 emission at any point in time is relative to the warming caused by emissions of that gas before the 11 reference point. Because of their different focus, GWP* and GWP100 can equate radically different CO2 12 emissions to the same CH4 emissions: rapidly declining CH4 emissions have a negative CO2-warming- 13 equivalent value based on GWP* (rapidly declining SLCF emissions result in declining temperature, 14 relative to the warming caused by past SLCF emissions at a previous point in time) but a positive CO2- 15 equivalent value based on GWP or GTP (each SLCF emission from any source results in increased 16 future radiative forcing and global average temperature than without this emission, regardless whether 17 the rate of SLCF emissions is rising or declining). The different focus in these metrics can have 18 important distributional consequences, depending on how they are used to inform emission targets 19 (Reisinger et al., 2021; Lynch et al., 2021), but this has only begun to be explored in the scientific 20 literature. 21 A key insight from WGI is that for a given emissions scenario, different metric choices can alter the 22 time at which net zero GHG emissions are calculated to be reached, or whether net zero GHG emissions 23 are reached at all (see SM2.3 for details). From a mitigation perspective, this implies that changing 24 GHG emission metrics but retaining the same numerical CO2-equivalent emissions targets would result 25 in different climate outcomes. For example, achieving a balance of global anthropogenic GHG 26 emissions and removals as stated in Article 4.1 of the Paris Agreement could, depending on the GHG 27 emission metric used, result in different peak temperatures and in either stable or slowly or rapidly 28 declining temperature after the peak(Tanaka and O’Neill, 2018; Allen et al., 2018; Fuglestvedt et al., 29 2018; Schleussner et al., 2019). A fundamental change in GHG emission metrics used to monitor 30 achievement of existing emission targets could therefore inadvertently change their intended climate 31 outcomes or ambition, unless existing emission targets are re-evaluated at the same time (very high 32 confidence). 33 The WGIII contribution to the AR6 reports aggregate emissions and removals using updated GWP100 34 values from AR6 WGI unless stated otherwise. This choice was made on both scientific grounds (the 35 alignment of GWP100 with a cost-benefit perspective under social discount rates and its performance 36 from a global cost-effectiveness perspective) and for procedural reasons, including continuity with past 37 IPCC reports and alignment with decisions under the Paris Agreement Rulebook (see Annex II.8 for 38 further detail). A key constraint in the choice of metric is also that the literature assessed by WGIII 39 predominantly uses GWP100 and often does not provide sufficient detail on emissions and abatement 40 of individual gases to allow translation into different metrics. Presenting such information routinely in 41 mitigation studies would enable the application of more diverse GHG emission metrics in future 42 assessments to evaluate their contribution to different policy objectives. 43 All metrics have limitations and uncertainties, given that they simplify the complexity of the physical 44 climate system and its response to past and future GHG emissions. No single metric is well-suited to 45 all applications in climate policy. For this reason, the WGIII contribution to the AR6 reports emissions 46 and mitigation options for individual gases where possible; CO2-equivalent emissions are reported in 47 addition to individual gas emissions where this is judged to be policy-relevant. This approach aims to Do Not Cite, Quote or Distribute 2-18 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 reduce the ambiguity regarding mitigation potentials for specific gases and actual climate outcomes 2 over time arising from the use of any specific GHG emission metric. 3 4 END CROSS CHAPTER BOX 2 HERE 5 6 2.2.2 Trends in the global GHG emissions trajectories and short-lived climate forcers 7 2.2.2.1 Anthropogenic greenhouse gas emissions trends 8 Global GHG emissions continued to rise since AR5, but the rate of emissions growth slowed (high 9 confidence). GHG emissions reached 59±6.6 GtCO2eq in 2019 (Table 2.1 and Figure 2.5). In 2019, CO2 10 emissions from FFI were 38 (±3.0) Gt, CO2 from LULUCF 6.6±4.6 Gt, CH4 11±3.2 GtCO2eq, N2O 11 2.7±1.6 GtCO2eq and F-gases 1.4±0.41 GtCO2eq. There is high confidence that average annual GHG 12 emissions for the last decade (2010-2019) were the highest on record in terms of aggregate CO2eq 13 emissions, but low confidence for annual emissions in 2019 as uncertainties are large considering the 14 size and composition of observed increases in the most recent years (Minx et al., 2021; UNEP, 2020a). 15 2019 GHG emissions levels were higher compared to 10 and 30 years ago (high confidence): about 16 12% (6.5 GtCO2eq) higher than in 2010 (53±5.7 GtCO2eq) (AR5 reference year) and about 54% (21 17 GtCO2eq) higher than in 1990 (38±4.8 GtCO2eq) (Kyoto Protocol reference year and frequent NDC 18 reference). GHG emissions growth slowed compared to the previous decade (high confidence): From 19 2010 to 2019 GHG emissions grew on average by about 1.3% per year compared to an average annual 20 growth of 2.1% between 2000 and 2009. Nevertheless the absolute increase in average annual GHG 21 emissions for 2010-2019 compared to 2000-2009 was 9.1 GtCO2eq and, as such, the largest observed 22 in the data since 1970 (Table 2.1) – and most likely in human history (Friedlingstein et al., 2020; 23 Gütschow et al., 2021b). Decade-by- decade growth in average annual GHG emissions was observed 24 across all (groups of) gas as shown in Table 2.1, but for N2O and CO2-LULUCF emissions this is much 25 more uncertain. 26 27 Table 2.1 Total anthropogenic GHG emissions (GtCO2eq yr-1) 1990-2019 28 CO2 from fossil fuel combustion and industrial processes (FFI); CO 2 from Land Use, Land Use Change 29 and Forestry (LULUCF); methane (CH4); nitrous oxide (N2O); fluorinated gases (F-gases: HFCs, PFCs, 30 SF6, NF3). Aggregate GHG emission trends by groups of gases reported in Gt CO2eq converted based on 31 global warming potentials with a 100-year time horizon (GWP-100) from the IPCC Sixth Assessment 32 Report. Uncertainties are reported for a 90% confidence interval. 33 Source: Minx et al. (2021) Average annual emissions (GtCO2eq) CO2 FFI CO2 CH4 N2O Fluorinated GHG LULUCF gases 2019 38±3.0 6.6±4.6 11±3.2 2.7±1.6 1.4±0.41 59±6.6 2010-2019 36±2.9 5.7±4.0 10±3.0 2.6±1.5 1.2±0.35 56±6.0 2000-2009 29±2.4 5.3±3.7 9.0±2.7 2.3±1.4 0.81±0.24 47±5.3 1990-1999 24±1.9 5.0±3.5 8.2±2.5 2.1±1.2 0.49±0.15 40±4.9 1990 23±1.8 5.0±3.5 8.2±2.5 2.0±1.2 0.38±0.11 38±4.8 34 35 Reported total annual GHG emission estimates differ between the Working Group III contributions in 36 AR5 (Blanco et al., 2014) and AR6 (this chapter) mainly due to differing global warming potentials Do Not Cite, Quote or Distribute 2-19 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 (high confidence). For the year 2010, total GHG emissions were estimated at 49±4.9 Gt CO2eq in AR5 2 (Blanco et al., 2014), while we report 53±5.7 Gt CO2eq here. However, in AR5 total GHG emissions 3 were weighted based on GWP-100 values from IPCC SAR. Applying those GWP values to the 2010 4 emissions from AR6 yields 50 Gt CO2eq (Forster et al., 2021a). Hence, observed differences are mainly 5 due to the use of most recent GWP values, which have higher warming potentials for methane (29% 6 higher for biogenic and 42% higher for fugitive methane) and 12% lower values for nitrous oxide (see 7 Cross-Chapter Box 2 in this chapter). 8 Emissions growth has been persistent but varied in pace across gases. The average annual emission 9 levels of the last decade (2010-2019) were higher than in any previous decade for each group of 10 greenhouse gases: CO2, CH4, N2O, and F-gases (high confidence). Since 1990, CO2-FFI have grown by 11 67% (15 GtCO2eq), CH4 by 29% (2.4 GtCO2eq), and N2O by 33% (0.65 GtCO2eq), respectively (Figure 12 2.5). Growth in fluorinated gases (F-gas) has been by far the highest with about 250% (1.0 GtCO2eq), 13 but it occurred from low levels. In 2019, total F-gas levels are no longer negligible with a share of 2.3% 14 of global GHG emissions. Note that the F-gases reported here do not include chlorofluorocarbons 15 (CFCs) and hydrochlorofluorocarbons (HCFCs), which are groups of substances regulated under the 16 Montreal Protocol. The aggregate CO2eq emissions of HFCs, HCFCs and CFCs were each 17 approximately equal in 2016, with a smaller contribution from PFCs, SF6, NF3 and some more minor 18 F-gases. Therefore, the GWP-weighted F-gas emissions reported here (HFCs, PFCs, SF6, NF3), which 19 are dominated by the HFCs, represent less than half of the overall CO2eq F-gas emissions in 2016 20 (Figure 2.3). 21 The only exception to these patterns of GHG emissions growth is net anthropogenic CO2-LULUCF 22 emissions, where there is no statistically significant trend due to high uncertainties in estimates (Figure 23 2.2, Figure 2.5; for a discussion see Chapter 2 Supplementary Material). While the average estimate 24 from the bookkeeping models report a slightly increasing trend in emissions, NGHGI and FAOSTAT 25 estimates show a slightly decreasing trend, which diverges in recent years (Figure 2.2). Similarly, trends 26 in CO2-LULUCF estimates from individual bookkeeping models differ: while two models (BLUE, 27 OSCAR) show a sustained increase in emissions levels since the mid 1990s, emissions from the third 28 model (HN) declined (see Figure 2.2; Friedlingstein et al., 2020). Differences in accounting approaches 29 and their impacts CO2 emissions estimates from land use is covered in Chapter 7 and in the Chapter 2 30 Supplementary Material (SM2.2). Note that anthropogenic net emissions from bioenergy are covered 31 by the CO2-LULUCF estimates presented here. Do Not Cite, Quote or Distribute 2-20 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 Figure 2.5 Total anthropogenic GHG emissions (Gt CO2eq yr-1) 1990-2019 3 CO2 from fossil fuel combustion and industrial processes (FFI); net CO 2 from Land Use, Land Use 4 Change and Forestry (LULUCF); methane (CH 4); nitrous oxide (N2O); fluorinated gases (F-gases: HFCs, 5 PFCs, SF6, NF3). Panel a: Aggregate GHG emission trends by groups of gases reported in Gt CO2eq 6 converted based on global warming potentials with a 100-year time horizon (GWP-100) from the IPCC 7 Sixth Assessment Report. Panel b: Waterfall diagrams juxtaposes GHG emissions for the most recent 8 year 2019 in CO2 equivalent units using GWP-100 values from the IPCC’s Second, Fifth, and Sixth 9 Assessment Report, respectively. Error bars show the associated uncertainties at a 90% confidence 10 interval. Panel c: individual trends in CO2-FFI, CO2-AFOLU, CH4, N2O and F-gas emissions for the 11 period 1990-2019, normalised to 1 in 1990. 12 Source: Data from Minx et al., 2021 13 The CO2-FFI share in total CO2eq emissions has plateaued at about 65% in recent years and its growth 14 has slowed considerably since AR5 (high confidence). CO2-FFI emissions grew at 1.1% during the 15 1990s and 2.5% during the 2000s. For the last decade (2010s) - not covered by AR5 - this rate dropped 16 to 1.2%. This included a short period between 2014-2016 with little or no growth in CO2-FFI emissions 17 mainly due to reduced emissions from coal combustion (Peters et al., 2017a; Qi et al., 2016; Jackson et 18 al., 2016; Canadell et al., 2021). Subsequently, CO2-FFI emissions started to rise again (Peters et al., 19 2017b; Figueres et al., 2018; Peters et al., 2020). 20 Starting in the spring of 2020 a major break in global emissions trends was observed due to lockdown 21 policies implemented in response to the COVID-19 pandemic (high confidence) (Quéré et al., 2020; Le 22 Quéré et al., 2021; Liu et al., 2020d; Forster et al., 2020; Bertram et al., 2021). Overall, global CO2-FFI 23 emissions are estimated to have declined by 5.8% [5.1%-6.3%] in 2020, or about 2.2 (1.9-2.4) GtCO2 Do Not Cite, Quote or Distribute 2-21 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 in total (Crippa et al., 2021; Friedlingstein et al., 2020; Liu et al., 2020d; IEA, 2021a; BP, 2021). This 2 exceeds any previous global emissions decline since 1970, both in relative and absolute terms (Figure 3 2.6). Daily emissions, estimated based on activity and power-generation data, declined substantially 4 compared to 2019 during periods of economic lockdown, particularly in April 2020 –as shown in Figure 5 2.6 – but rebounded by the end of 2020 (Le Quéré et al., 2021; Liu et al., 2020d; Quéré et al., 2020). 6 Impacts were differentiated by sector, with road transport and aviation particularly affected. Inventories 7 estimate the total power sector CO2 reduction from 2019 to 2020 at 3% (IEA, 2021a) and 4.5% (Crippa 8 et al., 2021). Approaches that predict near real-time estimates of the power sector reduction are more 9 uncertain and estimates range more widely, between 1.8% (Le Quéré et al., 2021; Quéré et al., 2020), 10 4.1% (Liu et al., 2020d) and 6.8% (Bertram et al., 2021); the latter taking into account the over- 11 proportional reduction of coal generation due to low gas prices and merit order effects. Due to the very 12 recent nature of this event, it remains unclear what the exact short and long-term impacts on future 13 global emissions trends will be. 14 15 Figure 2.6 Global CO2 emissions from fossil fuel combustion and industry (FFI) in 2020 and the impact of 16 COVID-19 17 Panel a depicts CO2-FFI emissions over the past 5 decades (GtCO2yr-1). The single year declines in 18 emissions following major economic and geopolitical events are shown, as well as the decline recorded in 19 5 different datasets for emissions in 2020 (COVID-19) compared to 2019 (no COVID-19). Panel b depicts 20 the change in global daily carbon emissions (MtCO2 per day) in 2020 compared to 2019, showing the 21 impact of COVID-19 lockdown policies. 22 Source: Crippa et al. (2021), Friedlingstein et al. (2020), BP (BP, 2021), IEA (IEA, 2021a), Carbon Monitor 23 (Liu et al., 2020d), Le Quéré et al. (Quéré et al., 2020). 24 From 1850 until around 1950, anthropogenic CO2 emissions were mainly (>50%) from land-use, land 25 use change and forestry (Figure 2.7). Over the past half-century CO2 emissions from LULUCF have 26 remained relatively constant around 5.1±3.6 GtCO2 but with a large spread across estimates (Le Quéré 27 et al., 2018a; Friedlingstein et al., 2020, 2019). By contrast, global annual FFI-CO2 emissions have Do Not Cite, Quote or Distribute 2-22 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 continuously grown since 1850 and since the 1960s from a decadal average of 11±0.9 GtCO2 to 36±2.9 2 GtCO2 during 2010-2019 (see Table 2.1). 3 Cumulative CO2 emissions since 1850 reached 2400±240 GtCO2 in 2019 (high confidence)7. More than 4 half (62%) of total emissions from 1850 to 2019 occurred since 1970 (1500±140 GtCO2), about 42% 5 since 1990 (1000±90 GtCO2) and about 17% since 2010 (410±30 GtCO2) (Friedlingstein et al., 2020; 6 Canadell et al., 2021; Friedlingstein et al., 2019) (Figure 2.7). Emissions in the last decade are about 7 the same size as the remaining carbon budget of 400±220 (500, 650) GtCO2 for limiting global warming 8 to 1.5°C and between one third and half the 1150±220 (1350, 1700) GtCO2 for limiting global warming 9 below 2°C with a 67% (50%, 33%) probability, respectively (Canadell et al., 2021). At current (2019) 10 levels of emissions, it would only take 8 (2-15) and 25 (18-35) years to emit the equivalent amount of 11 CO2 for a 67th percentile 1.5°C and 2°C remaining carbon budget, respectively. Related discussions of 12 carbon budgets, short-term ambition in the context of NDCs, pathways to limiting warming to well 13 below 2°C and carbon dioxide removals are mainly discussed in Chapters 3, 4, and 12, but also Section 14 2.7 of this chapter. 15 16 17 Figure 2.7 Historic anthropogenic CO2 emission and cumulative CO2 emissions (1850-2019) as well as 18 remaining carbon budgets for limiting warming to 1.5°C and 2°C 19 Panel a shows historic annual anthropogenic CO2 emissions (GtCO2 yr-1) by fuel type and process. Panel 20 b shows historic cumulative anthropogenic CO2 emissions for the periods 1850-1989, 1990-2009, and 21 2010-2019 as well as remaining future carbon budgets as of 1.1.2020 to limit warming to 1.5°C and 2°C at 22 the 67th percentile of the transient climate response to cumulative CO 2 emissions. The whiskers indicate a 23 budget uncertainty of ±220 GtCO2eq for each budget and the aggregate uncertainty range at 1 standard 24 deviation for historical cumulative CO2 emissions, consistent with Working Group 1. 25 Sources: Friedlingstein et al. (2020) and Canadell et al. (2021). 26 Comparisons between historic GHG emissions and baseline projections provide increased evidence that 27 global emissions are not tracking high-end scenarios (Hausfather and Peters, 2020), and rather followed 28 “middle-of-the-road” scenario narratives in the earlier series, and by combinations of “global- 29 sustainability” and “middle-of-the-road” narratives in the most recent series (SRES and SSP-baselines) 30 (Strandsbjerg Tristan Pedersen et al., 2021; Pedersen et al., 2020). As countries increasingly implement 31 climate policies and technology costs continue to evolve, it is expected emissions will continually shift 32 away from scenarios that assume no climate policy but remain insufficient to limit warming to below 33 2°C (Hausfather and Peters, 2020; Vrontisi et al., 2018; UNEP, 2020b; Roelfsema et al., 2020). FOOTNOTE 7 For consistency with WG1, uncertainties in this paragraph are reported at a 68% confidence interval. This reflects the difficulty in the WG1 context of characterizing the uncertainty in the CO2 fluxes between the atmosphere and the ocean and land reservoirs individually, particularly on an annual basis, as well as the difficulty of updating the emissions from land use change. Do Not Cite, Quote or Distribute 2-23 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 The literature since AR5 suggests that compared to historical trends baseline scenarios might be biased 2 towards higher levels of fossil fuel use compared to what is observed historically (Ritchie and 3 Dowlatabadi, 2017; Ritchie, 2019; Ritchie and Dowlatabadi, 2018; Creutzig et al., 2021). Ritchie and 4 Dowlatabadi (2017) show that per-capita primary energy consumption in baseline scenarios tends to 5 increase at rates faster than those observed in the long-term historical evidence – particularly in terms 6 of coal use. For example, SSP5 envisions a 6-fold increase in per capita coal use by 2100 – against flat 7 long-term historical observations – while the most optimistic baseline scenario SSP1-Sustainability is 8 associated with coal consumption that is broadly in line with historical long-term trends (Ritchie and 9 Dowlatabadi, 2017). In contrast, models have struggled to reproduce historical upscaling of wind and 10 solar and other granular energy technologies (Creutzig et al., 2017; Wilson et al., 2020b; Sweerts et al., 11 2020; Wilson et al., 2013; van Sluisveld et al., 2015; Shiraki and Sugiyama, 2020). 12 2.2.2.2 Other short-lived climate forcers 13 There are other emissions with shorter atmospheric lifetimes that contribute to climate changes. Some 14 of them like aerosols, sulphur emissions or organic carbon reduce forcing, while others like black 15 carbon, carbon monoxide or non-methane organic compounds (NMVOC) contribute to warming (also 16 see Figure 2.4) as assessed in Working Group I (Forster et al., 2021c; Naik et al., 2021a). Many of these 17 other short-lived climate forcers (SLCFs) are co-emitted during combustion processes in power plants, 18 cars, trucks, airplanes, but also during wildfires and household activities such as traditional cooking 19 with open biomass burning. As these co-emissions have implications for net warming, they are also 20 considered in long-term emission reduction scenarios as covered in the literature (Smith et al., 2020; 21 Rauner et al., 2020b; Vandyck et al., 2020; Harmsen et al., 2020) as well as Chapter 3 of this report. 22 These air pollutants are also detrimental to human health (e.g. Lelieveld et al., 2015, 2018; Vohra et al., 23 2021). For example, Lelieveld et al. (2015) estimates a total of 3.3 (1.6-4.8) million pre-mature deaths 24 in 2010 from outdoor air pollution. Reducing air-pollutants in the context of climate policies therefore 25 lead to substantial co-benefits of mitigation efforts (Rauner et al., 2020a; Rao et al., 2017; Von Stechow 26 et al., 2015; Lelieveld et al., 2019). Here we only briefly outline the major trends in emissions of short- 27 lived climate forcers. 28 Conventional air pollutants that are subject to significant emission controls in many countries include 29 SO2, NOx, BC and CO. From 2015 to 2019, global SO2 and NOx emissions have declined, mainly due 30 to reductions in energy systems (Figure 2.8). Reductions in BC and CO emissions appear to have 31 occurred over the same period, but trends are less certain due to the large contribution of emissions 32 from poorly quantified traditional biofuel use. Emissions of CH4, OC and NMVOC have remained 33 relatively stable in the past five years. OC and NMVOC may have plateaued, although there is 34 additional uncertainty due to sources of NMVOCs that may be missing in current inventories 35 (McDonald et al., 2018). 36 Do Not Cite, Quote or Distribute 2-24 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 Figure 2.8 Air pollution emissions in by major sectors from CEDS (1970-2019) and EDGAR (1970-2015) 3 inventories 4 Source: Crippa et al., 2019a, 2018; O’Rourke et al., 2020; McDuffie et al., 2020 5 6 2.2.3 Regional GHG emissions trends 7 Regional contributions to global GHG emissions have shifted since the beginning of the international 8 climate negotiations in the 1990s (high confidence). As shown in Figure 2.9, developed countries as a 9 group have not managed to reduce GHG emissions substantially, with fairly stable levels at about 15 10 GtCO2eq yr-1 between 1990 and 2010, while countries in Asia and the Developing pacific have rapidly 11 increased their share of global GHG emissions – particularly since the 2000s (Jackson et al., 2019; 12 Peters et al., 2020; UNEP, 2020c; Crippa et al., 2021; IEA, 2021b). 13 Most global GHG emission growth occurred in Asia and Developing Pacific, which accounted for 77% 14 of the net 21 GtCO2eq increase in GHG emissions since 1990, and 83% of the net 6.5 GtCO2eq increase 15 since 2010.8 Africa contributed 11% of GHG emissions growth since 1990 (2.3 GtCO2eq) and 10% (0.7 16 GtCO2eq) since 2010. The Middle East contributed 10% of GHG emissions growth since 1990 (2.1 17 GtCO2eq) and also 10% (0.7 GtCO2eq) since 2010. Latin America and the Caribbean contributed 11% 18 of GHG emissions growth since 1990 (2.2 GtCO2eq), and 5% (0.3 GtCO2eq) since 2010. Two regions, 19 Developed Countries, and Eastern Europe and West-Central Asia, reduced emissions overall since 20 1990, by -1.6 GtCO2eq and -0.8 GtCO2eq, respectively. However, emissions in the latter region started 21 to grow again since 2010, contributing to 5% of the global GHG emissions change (0.3 GtCO2eq). 22 Average annual GHG emission growth across all regions slowed between 2010-2019 compared to 23 1990-2010, with the exception of Eastern Europe and West-Central Asia. Global emissions changes 24 tend to be driven by a limited number of countries, principally the G20 group (UNEP, 2020c; Xia et al., 25 2021; Friedlingstein et al., 2020). For instance, the slowing of global GHG emissions between 2010- FOOTNOTE 8 Note that GHG emissions from international aviation and shipping could not be attributed to individual regions, while CO2 emissions from AFOLU could not be attributed to individual countries. Change in GHG emissions that can be easily assigned to regions is 20.3 of 20.8 GtCO2eq for 1990-2019 and 6.3 of 6.5 GtCO2eq for 2010-2019. Do Not Cite, Quote or Distribute 2-25 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2019, compared to the previous decade, was primarily triggered by substantial reductions in GHG 2 emissions growth in China. Two countries (China, India) contributed more than 50% to the net 6.5 3 GtCO2eqyr-1 increase in GHG emissions during 2010-2019 (at 39% and 14%, respectively), while ten 4 countries (China, India, Indonesia, Vietnam, Iran, Turkey, Saudi Arabia, Pakistan, Russian Federation, 5 Brazil) jointly contributed about 75% (Figure 2.9) (see also Minx et al., 2021; Crippa et al., 2021). 6 GHG and CO2-FFI levels diverge starkly between countries and regions (high confidence) (UNEP, 7 2020c; Jackson et al., 2019; Friedlingstein et al., 2020; Crippa et al., 2021). Developed Countries 8 sustained high levels of per capita CO2-FFI emissions at 9.5 t CO2/cap in 2019 (but with a wide range 9 of 1.9-16 tCO2/cap). This is more than double that of three developing regions – 4.4 (0.3-12.8) tCO2/cap 10 in Asia and Developing Pacific, 1.2 (0.03-8.5) tCO2/cap in Africa, and 2.7 (0.3-24) tCO2/cap in Latin 11 America9. Per capita CO2-FFI emissions were 9.9 (0.89-15) tCO2/cap in Eastern Europe and West- 12 Central Asia, and 8.6 (0.36-38) tCO2/cap in the Middle East. CO2-FFI emissions in the three developing 13 regions together grew by 26% between 2010 and 2019, compared to 260% between 1990 and 2010, 14 while in Developed Countries emissions contracted by 9.9% between 2010-2019 and by 9.6% between 15 1990-2010. 16 Least developed countries contributed only a negligible proportion of historic GHG emissions growth 17 and have the lowest per capita emissions. As of 2019 they contribute 3.3% of global GHG emissions, 18 excluding LULUCF CO2, despite making up 13.5% of the global population. Since the start of the 19 industrial revolution in 1850 up until 2019, they contributed 0.4% of total cumulative CO 2 emissions 20 (Figure 2.10). Conversely, Developed Countries have the highest share of historic cumulative emissions 21 (Matthews, 2016; Gütschow et al., 2016; Rocha et al., 2015), contributing approximately 57% (Figure 22 2.10), followed by Asia and developing Pacific (21%), Eastern Europe and West-Central Asia (9%), 23 Latin America and the Caribbean (4%), the Middle East (3%), and Africa (3%). Developed Countries 24 still have the highest share of historic cumulative emissions (45%) when CO2-LULUCF emissions are 25 included, which typically account for a higher proportion of emissions in developing regions (Figure 26 2.10). 27 A growing number of countries have reduced CO2 and GHG emissions for longer than 10 years (medium 28 confidence) (Le Quéré et al., 2019; Lamb et al., 2021a; Wu et al., 2021; Burck et al., 2021). Data up to 29 2018 indicates that about 24 countries have reduced territorial CO2 and GHG emissions (excluding 30 LULUCF CO2), as well as consumption-based CO2 emissions, for at least 10 years (Lamb et al., 2021a). 31 Uncertainties in emissions levels and changes over time prevents a precise assessment of reductions in 32 some cases. Of these 24 countries, 12 peaked emissions in the 2000s; 6 have sustained longer reductions 33 since the 1970s; and 6 are former members of the Eastern Bloc, where emissions dropped rapidly in the 34 1990s and continued declining at a slower pace thereafter. Country emissions reductions have been 35 driven by both climate and non-climate policies and factors, including structural changes. To date, most 36 territorial emissions reductions were realised in the electricity and heat sector, followed by industry and 37 buildings, while in many cases transport emissions have increased since countries reached their overall 38 emissions peak (Climate Transparency, 2021; Lamb et al., 2021a). One estimate of the total reduction 39 in annual GHG emissions – from peak years to 2018 – sums to 3.2 GtCO2eq across all decarbonising 40 countries (Lamb et al., 2021a). These reductions have therefore been far outweighed by recent emissions 41 growth. However, climate policy related reductions may be even larger when compared against a 42 counterfactual case of emissions growth across different sectors (Eskander and Fankhauser, 2020b) 43 (Cross-Chapter Box 1 in Chapter 1; Section 2.8). 44 The recent (2010-2019) emissions changes of some countries are in line with pathways that limit likely 45 warming to below 2°C (e.g. -4% average annual reductions) (Figure 2.10). Overall, there are first FOOTNOTE 9 In all cases, constraining countries within the emissions range to those larger than 1 million population. Do Not Cite, Quote or Distribute 2-26 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 country cases emerging that highlight the feasibility of sustained emission reductions outside of periods 2 of economic disruption (Lamb et al., 2021a). However, such pathways will need to be taken by many 3 more countries for keeping the goals of the Paris Agreement in reach (Höhne et al., 2020; Kriegler et 4 al., 2018a; Roelfsema et al., 2020; den Elzen et al., 2019) as analysed by Chapter 4 of this report. 5 Moreover, observed reductions are not yet consistent and long-term, nor achieved across all sectors, nor 6 fully aligned with country NDC targets (Le Quéré et al., 2019; Lamb et al., 2021a; den Elzen et al., 7 2019; Climate Transparency, 2021; Burck et al., 2021). Do Not Cite, Quote or Distribute 2-27 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 Figure 2.9 Change in regional GHGs from multiple perspectives and their underlying drivers 3 Panel a: Regional GHG emission trends (in GtCO2eq yr-1) for the time period 1990-2019. GHG emissions 4 from international aviation and shipping are not assigned to individual countries and shown separately. 5 Panels b and c: Changes in GHG emissions for the 20 largest emitters (as of 2019) for the post-AR5 6 reporting period 2010-2019 in relative (% annual change) and absolute terms (GtCO2eq). Panels d and e: 7 GHG emissions per capita and per GDP in 2019 for the 20 largest emitters (as of 2019). GDP estimated 8 using constant international purchasing power parity (USD 2017). Emissions are converted into CO 2- 9 equivalents based on global warming potentials with a 100 year time horizon (GWP-100) from the IPCC Do Not Cite, Quote or Distribute 2-28 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 Sixth Assessment Report (Forster et al., 2021a). The yellow dots represent the emissions data from 2 UNFCCC-CRFs (2021) that were accessed through Gütschow et al. (2021a). Net LULUCF CO2 emissions 3 are included in panel a, based on the average of three bookkeeping models (see Section 2.2), but are 4 excluded in panels b due to a lack of country resolution. 5 6 7 Figure 2.10 Different perspectives on historic emissions and equity 8 Panel a shows the regional proportion (%) of total global population or emissions in 2018 or 2019, for five 9 categories: population (persons), consumption-based CO2-FFI emissions (GtCO2), production-based CO2- 10 FFI emissions (GtCO2), production-based GHG emissions excluding CO2-LULUCF (GtCO2eq), and 11 production-based GHG emissions including CO2-LULUCF (GtCO2eq). Panel b shows the regional 12 proportion (%) of total cumulative production-based CO2 emissions from 1850 to 2019, including and 13 excluding CO2-LULUCF (GtCO2). In the lower panels the proportion of each population or emissions 14 category attributable to Least Developed Countries is shown, where available (CO2-LULUCF data is not 15 available for this region). GHG emissions are converted into CO2-equivalents based on global warming 16 potentials with a 100 year time horizon (GWP-100) from the IPCC Sixth Assessment Report (Forster et al., 17 2021a). 18 Source: Data from Friedglinstein et al. (2020) Do Not Cite, Quote or Distribute 2-29 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 3 Figure 2.11 Recent average annual GHG emissions changes of countries (left panel) versus rates of 4 reduction in 1.5°C and 2°C mitigation scenarios 5 Scenario data is taken from Chapter 3 of this report with the scenario categories defined and summarised 6 in Table 3.2 in Chapter 3. Emissions are converted into CO2-equivalents based on global warming 7 potentials with a 100 year time horizon (GWP-100) from the IPCC Sixth Assessment Report (Forster et al., 8 2021a). Circles indicate countries (left panel) or individual scenarios (right panel), the former scaled by 9 total emissions in 2019. Horizontal lines indicate the region average emissions change (left panel), or 10 scenario category average emissions change (right panel). 11 Source: Data from Minx et al., 2021 12 2.2.4 Sectoral GHG emission trends 13 In 2019, 34% (20 GtCO2eq) of the 59 GtCO2eq GHG emissions came from the energy sector, 24% (14 14 GtCO2eq) from industry, 22% (13 GtCO2eq) from AFOLU, 15% (8.7 GtCO2eq) from transport and 6% 15 (3.3 GtCO2eq) from buildings (Figure 2.12). The relative size of each sector depends on the exact 16 definition of sector boundaries (de la Rue du Can et al., 2015; Lamb et al., 2021b). The largest individual 17 sub-sector contributing to global GHG emissions in 2019 was electricity and heat generation at 14 18 GtCO2eq. This subsector can be reallocated to consuming sectors as indirect (Scope 2) emissions to 19 emphasize the role of final energy demand and demand-side solutions in climate change mitigation 20 (Creutzig et al., 2018) (Chapter 5). This increases the emission share of the industry sector to 34% and 21 of the buildings sector to 16%. 22 Average annual GHG emissions growth has been fastest in the transport sector with about 1.8% for the 23 most recent period 2010-2019, followed by direct emissions in the industry sector (1.4%) and the energy 24 sector (1%) (Figure 2.13). This is different to growth patterns observed in the previous decade as 25 reported in AR5 (IPCC, 2014a; Blanco et al., 2014). Between 2000 and 2009 fastest GHG emissions 26 growth was observed for industry with 3.4% followed by the energy sector with 2.3%. GHG emission 27 growth in the transport sector has been stable across both periods at about 1.8%, while direct building 28 emissions growth averaged below 1% during 2010-2019. Ranking of high emitting sectors by direct 29 emissions highlights the importance of the LULUCF CO2 (6.6 GtCO2eq), road transport (6.1 GtCO2eq), 30 metals (3.1 GtCO2eq), and other industry (4.4 GtCO2eq) sub-sectors. Overall, some of the fastest 31 growing sources of sub-sector emissions from 2010 to 2019 have been international aviation (+3.4%)10, FOOTNOTE10 Note that this does not include the additional warming impacts from aviation due to short lived climate forcers, which are assessed in Chapter 10 (Section 10.5) Do Not Cite, Quote or Distribute 2-30 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 domestic aviation (+3.3%), inland shipping (+2.9%), metals (+2.3%), international shipping (+1.7%), 2 and road transport (+1.7%). 3 4 Figure 2.12 Total anthropogenic direct and indirect GHG emissions for the year 2019 (in GtCO 2eq) by 5 sector and sub-sector. 6 Direct emissions estimates assign emissions to the sector in which they arise (scope 1 reporting). Indirect 7 emissions – as used here - refer to the reallocation of emissions from electricity and heat to the sector of 8 final use (scope 2 reporting). Note that cement refers to process emissions only, as a lack of data prevents 9 the full reallocation of indirect emissions to this sector. More comprehensive conceptualisations of 10 indirect emissions including all products and services (scope 3 reporting) are discussed in Section 2.3 of 11 this chapter. Emissions are converted into CO2-equivalents based on global warming potentials with a 12 100 year time horizon (GWP-100) from the IPCC Sixth Assessment Report. Percentages may not add up 13 to 100 across categories due to rounding at the second significant digit. 14 Source: Based on Lamb et al. (2021b); Data: Minx et al., 2021 15 Do Not Cite, Quote or Distribute 2-31 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 Figure 2.13 Total annual anthropogenic GHG emissions by major economic sector and their underlying 3 trends by region 4 Panel a: Trends in total annual anthropogenic GHG emissions (in GtCO 2eq yr-1) by major economic 5 sector. Panel b: Trends in total annual anthropogenic GHG emissions (in GtCO 2eq yr-1) by major 6 economic sector and region. Panels c and d: Largest sub-sectoral changes in GHG emissions for the 7 reporting period 2010-2019 in relative (% annual change) and absolute terms (GtCO2eq yr-1). Emissions 8 are converted into CO2-equivalents based on global warming potentials with a 100 year time horizon 9 (GWP-100) from the IPCC Sixth Assessment Report. Do Not Cite, Quote or Distribute 2-32 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 Source: Based on Lamb et al. (2021b); Data: Minx et al. (2021), Crippa et al. (2021) 2 3 4 2.3 Past and present trends of consumption-based CO2 emissions (CBEs) 5 and emissions embodied in trade 6 2.3.1 Scope, variability, and uncertainty of CBEs 7 Consumption is increasingly met by global supply chains often involving large geographical distances 8 and causing emissions in producing countries (Hubacek et al., 2014, 2016; Wiedmann and Lenzen, 9 2018). Therefore, accounting of emissions of production along the entire supply chain to fulfil final 10 demand, so-called consumption-based emissions (CBEs) is necessary to understand why emissions 11 occur and to what extent consumption choices and associated supply chains contribute to total 12 emissions, and ultimately how to influence consumption to achieve climate mitigation targets and 13 environmental justice (Vasconcellos, 2019). 14 Production-based emissions (PBEs) and territorial emissions resulting from the production and 15 consumption of goods and services within a region as well as for export production are often used by 16 authorities to report carbon emissions (Peters, 2008) (see also Section 2.2). PBEs also include emissions 17 from international activities (e.g., international aviation/shipping and non-resident activities), which are 18 excluded from territorial emissions (Karstensen et al., 2018; Shan et al., 2018). In contrast, CBEs refer 19 to emissions along the entire supply chains induced by consumption irrespective of the place of 20 production (Liu et al., 2015b). This reflects a shared understanding that a wider system boundary going 21 beyond territorial emissions is important to avoid outsourcing of pollution and to achieve global 22 decarbonisation. CBEs allow to identify new policy levers through providing information on a country’s 23 trade balance of embodied emissions, households’ carbon implications of their lifestyle choices, 24 companies’ upstream emissions as input for supply chain management, and cities’ footprints outside 25 their administrative boundaries (Davis and Caldeira, 2010; Feng et al., 2013). Kander et al., (2015) 26 proposed a technology-adjusted consumption-based emission accounting (TCBA) approach to address 27 the issue of carbon intensity in exports. TCBA incorporates emissions embodied in trade but also adjust 28 for differences in carbon efficiency in exports of different countries. Unlike PBEs, there are no 29 internationally agreed upon approaches to calculate CBEs, making it a major drawback for 30 mainstreaming the use of this indicator in policy making. 31 There are other proposed emission accounting approaches used in different circumstances. Historical 32 cumulative emissions (HCEs) are used when analysing countries’ historic contribution to emissions and 33 responsibility for emission reduction. HCEs account for a country’s cumulative past emissions, which 34 may be significantly different from the country’s current annual emissions (Botzen et al., 2008; Ritchie 35 2019b), but are sensitive to the choice of cut-off period. For example, the United States and EU-27 36 countries plus the United Kingdom contributed respectively 13.4% and 8.7% to global PBEs in 2019 37 (Crippa et al., 2020), however, they emitted around 25% and 22% of global historical PBEs since 1751 38 (Ritchie, 2019). In contrast, extraction-based emissions (EBEs) accounting allocates all emissions from 39 burning fossil fuels throughout the supply chains to the country where the fuels were extracted 40 (Steininger and Schinko, 2015). EBEs can be calculated by multiplying primary energy extraction of 41 fossil fuels with their respective carbon content adjusting for the fraction of fossil fuels that is not 42 combusted (Erickson and Lazarus, 2013). Another approach for accounting emissions is income-based 43 emission accounting (IBE), which traces emissions throughout all supply chains and allocates emissions 44 to primary inputs (e.g., capital and labour). In other words, IBEs investigates a country’s direct and 45 indirect downstream GHG emissions enabled by its primary inputs (Liang et al., 2017a). All these Do Not Cite, Quote or Distribute 2-33 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 approaches provide complementary information and different angles to assigning responsibility for 2 emissions reductions. 3 4 START BOX 2.2 HERE 5 6 Box 2.2 Policy applications of consumption-based emissions 7 • Consumption-based emissions provide additional or complementary information to 8 production-based emissions that can be used for a variety of policy applications. These 9 include:Complementary national-level emissions accounting and target or budget setting 10 • Raising awareness and increasing understanding of the GHG effects of consumption 11 • Accounting for and understanding of distributional and responsibility issues in GHG emissions 12 mitigation, both nationally and internationally. 13 • Incentives to change consumption patterns or reduce consumption (e.g., through taxation 14 policies) 15 • Accounting for and understanding of carbon leakage and emissions embodied in trade*) 16 • International emissions trading schemes or linked national schemes 17 • Trade policies addressing emissions embodied in trade and international supply chains (e.g., 18 border tax adjustments and clean technology transfers, carbon offsetting or financing, etc.) 19 • Including embodied emissions in product performance standards and labelling 20 • Policies of public and private procurement 21 • Agreements with international suppliers 22 • Discussing the climate impacts of lifestyles and inequalities in consumption and associated 23 emissions 24 Above points are based on a synopsis of studies (Steininger et al., 2014; Afionis et al., 2017; Hubacek 25 et al., 2017b; Wang and Zhou, 2018; Bolea et al., 2020) 26 * Note, however, that comparing embodied emissions in trade between countries is further complicated 27 by the fact that emission intensities differ across countries. Approaches to adjust for these differences 28 and facilitate comparisons have been suggested, e.g., by (Baumert et al., 2019; Dietzenbacher et al., 29 2020; Jakob, 2020; Kander et al., 2015). Many different approaches on how to share responsibility 30 between producers and consumers have been proposed in designing effective integrated global climate 31 policies (Yang et al., 2015; Liu and Fan, 2017; Zhu et al., 2018; Khajehpour et al., 2019; Jakob et al., 32 2021). Ultimately, assigning responsibility is normative. 33 END BOX 2.2 HERE 34 35 The dominant method for calculating CBEs of nations is global multi-region input-output (GMRIO) 36 analysis (Wiedmann and Lenzen, 2018), with other methods playing a minor role, e.g. analysing 37 bilateral trade flows of products and their life-cycle emission factors (Sato, 2014). Generally, the 38 uncertainties associated with CBEs depends on the choice of the dataset/model used for calculation, 39 which differs according to a) the national economic and trade data used, b) the emissions data used, c) 40 the sector or product-level aggregation, d) the regional aggregation, e) the conceptual scope (e.g., 41 residential vs territorial accounting principle) and f) the model construction techniques, which include 42 table balancing algorithms and ways of dealing with missing or conflicting data (Moran and Wood, 43 2014; Owen, 2017; Wieland et al., 2018; Wood et al., 2018b, 2019a). When excluding systematic error Do Not Cite, Quote or Distribute 2-34 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 sources, research has shown that the stochastic relative standard variation (RSD) of total national CBE 2 is not significantly different to that from PBE accounts and in the region of 5-15% (Lenzen et al., 2010; 3 Wood et al., 2018b, 2019a) 4 Six global accounts for consumption-based GHG emissions at the country level are widely used (Table 5 2.2). Each dataset has been constructed by different teams of researchers, covers different time periods 6 and contains CBEs estimates for different sets of countries and regions (Owen, 2017). 7 8 Table 2.2 Features of six global datasets for consumption-based emissions accounts Name of consumption-based account Years available Number of Number of datasets (and references) countries/regions sectors Eora (Lenzen et al., 2013); 1990-2015 190 Varies from 25 (https://worldmrio.com) to >500 EXIOBASE (Stadler et al., 2018); 1995-2016 49 200 products (https://www.exiobase.eu) and 163 industries GTAP (Aguiar et al., 2019; Peters, et al., 2004, 2007, 2011, 140 57 2011b); 2014 (https://www.gtap.agecon.purdue.edu) OECD/ICIO (Yamano and Guilhoto, 2020); 1995-2015 67 36 (http://oe.cd/io-co2) WIOD (Dietzenbacher et al., 2013; Timmer et 2000-2014 44 56 al., 2015); (http://wiod.org) Global Carbon Budget (Friedlingstein et al., 1990-2018 118 N/A 2020) 9 10 Wood et al. (2019b) present the first comprehensive and systematic model intercomparison and find a 11 variation of 5-10% for both PBEs and CBEs accounts of major economies and country groups (e.g., 12 EU28, OECD). The estimates for the US were the most closely aligned, with 3.7% Relative Standard 13 Deviation (RSD). For smaller countries, variability is in the order of 20-30% and can reach more than 14 40% in cases of very small, highly trade-exposed countries such as Singapore and Luxembourg (Wood 15 et al., 2019a). It is recommended to interpret CBE results for such countries with care. 16 Overall, production accounts showed a slightly higher convergence (8% average of RSD) than 17 consumption-based accounts (12%). The variation across model results can be approximately halved, 18 when normalising national totals to one common value for a selected base year. The difference between 19 PBE result variation (4% average RSD after normalisation) and CBEs results (7%) remains after 20 normalisation. 21 In general, the largest contributors to uncertainty of CBEs results are - in descending order of priority - 22 the total of territorial GHG emission accounts, the allocation of emissions to economic sectors, the total 23 and composition of final demand, and lastly the structure of the economy. Harmonising territorial 24 emissions across GMRIO datasets is the single most important factor that reduces uncertainty by about 25 50% (Tukker et al., 2020). More work is required to optimise or even institutionalise the compilation 26 of multi-region input-output data and models to enhance the accuracy of consumption-based accounting 27 (Tukker et al., 2018; Wood et al., 2018). Do Not Cite, Quote or Distribute 2-35 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2.3.2 Trends in global and regional CBEs trajectories 2 In comparison to territorial emissions discussed in Section 2.2, Figure 2.14 shows the trends of global 3 and regional CBEs from 1990 to 2018. This section uses the PBEs and CBEs data from the latest Global 4 Carbon Budget (Friedlingstein et al., 2020), which are slightly different from the PBEs used in Section 5 2.2. The Global Carbon Budget only includes CO2 emissions from fossil fuels and cement production. 6 7 Figure 2.14 Consumption-based CO2 emission trends for the period 1990-2018. The CBEs of countries are 8 collected from the Global Carbon Budget 2020 (Friedlingstein et al., 2020) 9 Source: This figure is modified based on Hubacek et al. (2021). 10 The left two panels in Figure 2.14 show total and per capita CBEs for six regions. The three panels on 11 the right show additional information for the 18 top-emitting countries with the highest CBEs in 2018. 12 In developed countries, consumption-based CO2 emissions peaked at 15 GtCO2 in 2007 with a Do Not Cite, Quote or Distribute 2-36 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 subsequent 16% decline until 2016 (to 12.7 GtCO2) and a slight rebound of 1.6% until 2018 (to 12.9 2 GtCO2). Asia and Developing Pacific has been a major contributor to consumption-based CO2 3 emissions growth since 2000 and exceeded developed countries as the global largest emissions source 4 in 2015. From 1990 to 2018, the average growth rate of Asia and Developing Pacific was 4.8% per year, 5 while in other regions emissions declined by -1.1%-4.3%/year on average. In 2018, 35% of global 6 consumption-based CO2 emissions were from developed countries and 39% from Asia and Developing 7 Pacific, 5% from Latin American and Caribbean, 5% from Eastern Europe and West-Central Asia, 5% 8 from Middle East, and 3% from Africa (Hubacek et al., 2021). Global CBEs kept growing over the 9 period with a short-lived decline in 2008 due to the global financial crisis. In 2020, lockdowns 10 associated with COVID-19 significantly reduced global emissions (Section 2.2.2), including CBEs 11 (Shan et al., 2020). 12 2.3.3 Decoupling of emissions from economic growth 13 There has been a long-standing discussion on whether environmental impacts such as carbon emissions 14 and use of natural resources can be decoupled from economic growth. It is controversial whether 15 absolute decoupling can be achieved at a global scale (Ward et al., 2016; Hickel and Kallis, 2020). 16 However, a number of studies found that it is feasible to achieve decoupling at the national level and 17 have explored the reasons for such decoupling (Ward et al., 2016; Zhao et al., 2016; Schandl et al., 18 2016; Deutch, 2017; Roinioti and Koroneos, 2017; Li et al., 2019; Vadén et al., 2020; Habimana Simbi 19 et al., 2021; Shan et al., 2021). 20 Table 2.3 shows the extent of decoupling of CBEs and GDP of countries based on CBEs from the 21 Global Carbon Budget (Friedlingstein et al., 2020) and GDP data from the World Bank. Table 2.4 also 22 presents countries’ degree of decoupling of PBEs and GDP. These data allow a comparison of 23 decoupling between GDP and both PBEs and CBEs. Absolute decoupling refers to a decline of 24 emissions in absolute terms or as being stable while GDP grows (i.e., a decoupling index11 greater than 25 1); relative decoupling refers to growth of emissions being lower than growth of GDP (a decoupling 26 index between 0 and 1); and no decoupling, which refers to a situation where emissions grow to the 27 same extent or faster than GDP (a decoupling index of less than 0) (Wu et al., 2018). 28 Table 2.3 Country groups with different degree of CBE-GDP decoupling from 2015 to 2018 Absolute Relative No Economic decoupling decoupling decoupling recession Number of countries 23 67 19 6 Total 5.40 25.33 1.93 0.85 CBEs (gigatons) Global share 16.1% 75.6% 5.8% 2.5% Total 4.84 25.73 2.16 0.84 PBEs (gigatons) Global share 14.4% 76.6% 6.4% 2.5% Total 625 5195 768 270 Population (million) Global share 9.1% 75.7% 11.2% 3.9% Total 19,891 54,240 2,300 2,997 GDP (billion) Global share 25.0% 68.3% 2.9% 3.8% Average 31.45 16.29 6.57 17.78 Median 23.55 8.03 2.56 13.12 FOOTNOTE11 The decoupling index can be calculated based on changes of a country’s GDP and CO 2 emissions (Wu et al., 2018; Akizu-Gardoki et al., 2018), see the equation below. 𝐷𝐼 refers to decoupling index; 𝐺1 refers to the GDP of reporting year while 𝐺0 refers to the base year; 𝐸1 refers to emissions of the reporting year while 𝐸0 refers to emissions of the base year. ∆𝐺%−∆𝐸% 𝐺 −𝐺0 𝐸 −𝐸0 𝐺 −𝐺0 𝐷𝐼 = =( 1 − 1 )⁄ 1 ∆𝐺% 𝐺0 𝐸0 𝐺0 Do Not Cite, Quote or Distribute 2-37 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII Per capita GDP Max 110.70 79.23 63.93 33.11 (thousand USD in 1.31 0.49 0.52 5.80 Min 2010 prices) Average 10.27 5.30 4.47 12.55 Per capita CBEs Median 8.87 4.13 1.67 11.33 (tons) Max 37.95 17.65 25.35 23.21 Min 0.64 0.09 0.18 2.33 Average 0.45 0.50 0.93 0.66 CBE intensity (tons Median 0.36 0.42 0.62 0.69 per thousand USD in Max 1.16 2.41 4.10 1.22 2010 prices) Min 0.11 0.10 0.28 0.21 Average 8.20 4.36 5.32 14.15 Median 6.79 3.02 1.19 13.22 Per capita PBEs (tons) Max 19.58 20.13 39.27 27.24 Min 0.49 0.09 0.08 2.23 Average 0.42 0.40 0.94 0.75 PBE intensity (tons Median 0.28 0.31 0.58 0.68 per thousand USD in Max 1.57 1.47 4.83 1.80 2010 prices) Min 0.10 0.05 0.16 0.20 1 Note: CBEs are obtained from the Global Carbon Budget 2020 (Friedlingstein et al., 2020), GDP and population 2 are from the World Bank. One country (Venezuela) does not have GDP data after 2015, so the degree of 3 decoupling was only calculated for 115 countries. This table is modified from Hubacek et al. (2021) 4 5 Table 2.4 Country groups with different degree of PBE-GDP decoupling from 2015 to 2018 Absolute Relative No Economic decoupling decoupling decoupling recession Number of countries 32 41 36 6 Total 6.41 23.43 2.83 0.85 CBEs (gigatons) Global share 19.1% 69.9% 8.4% 2.5% Total 5.33 24.36 3.04 0.84 PBEs (gigatons) Global share 15.9% 72.6% 9.1% 2.5% Total 857 4518 1213 270 Population (million) Global share 12.5% 65.9% 17.7% 3.9% Total 27091 45255 4086 2,997 GDP (billion) Global share 34.1% 57.0% 5.1% 3.8% Average 28.83 19.53 6.00 17.78 Per capita GDP Median 26.36 12.04 3.64 13.12 (thousand USD in Max 79.23 110.70 63.93 33.11 2010 prices) Min 1.09 0.57 0.49 5.80 Average 7.70 6.98 3.99 12.55 Per capita CBEs Median 6.78 6.00 1.95 11.33 (tons) Max 23.22 37.95 25.35 23.21 Min 0.43 0.09 0.18 2.33 Average 0.41 0.50 0.77 0.66 CBEs intensity (tons Median 0.31 0.44 0.52 0.69 per thousand USD in Max 2.41 1.68 4.10 1.22 2010 prices) Min 0.12 0.10 0.20 0.21 Average 6.02 5.69 4.33 14.15 Median 5.36 4.88 1.67 13.22 Per capita PBEs (tons) Max 20.13 16.65 39.27 27.24 Min 0.30 0.09 0.01 2.23 Do Not Cite, Quote or Distribute 2-38 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII Average 0.33 0.45 0.71 0.75 PBEs intensity (tons Median 0.20 0.31 0.44 0.68 per thousand USD in Max 1.47 1.76 4.83 1.80 2010 prices) Min 0.05 0.10 0.13 0.20 1 Note: CBEs are obtained from the Global Carbon Budget 2020 (Friedlingstein et al., 2020), GDP and population 2 are from the World Bank. One country (Venezuela) does not have GDP data after 2015, so the degree of 3 decoupling was only calculated for 115 countries. In order to be consistent with the results of CBEs, we calculate 4 the decoupling of PBE until 2018. The latest PBE data of 2019 may not change the key messages. 5 6 During the most recent three-year period from 2015 to 2018, 23 countries (or 20% of the 116 sample 7 countries) have achieved absolute decoupling of CBEs and GDP, while 32 countries (or 28%) achieved 8 absolute decoupling of PBEs and GDP. 14 of them (e.g., the UK, Japan, and the Netherlands) also 9 decoupled PBEs and GDP. Countries with absolute decoupling of CBEs tend to achieve decoupling at 10 relatively high levels of economic development and high per capita emissions. Most of EU and North 11 American countries are in this group. Decoupling was not only achieved by outsourcing carbon 12 intensive production, but also improvements in production efficiency and energy mix, leading to a 13 decline of emissions. Structural Decomposition Analysis shows that the main driver for decoupling has 14 been a reduction in carbon intensity (that is change in energy mix and energy efficiency) from both 15 domestic production and imports (Hubacek et al., 2021). Similarly, Wood et al., (2019c) found that EU 16 countries have reduced their overall consumption-based GHG emissions by 8% between 1995 and 2015, 17 mainly due to the use of more efficient technology. The literature also shows that changes in the 18 structure of economy with a shift to tertiary sectors of production may contribute to such decoupling 19 (Xu and Ang, 2013; Kanitkar et al., 2015; Su and Ang, 2016). 20 67 (or 58%) countries, including China and India, have relatively decoupled GDP and CBEs between 21 2015 and 2018, reflecting a slower growth in emissions than GDP. It is worth noting that the USA 22 shows relative decoupling of emissions (both CBEs and PBEs) and GDP over the most recent period, 23 although it strongly decoupled economic growth from emissions between 2005 and 2015. Thus 24 decoupling can be temporary and countries’ emissions may again increase after a period of decoupling. 25 Another 19 (or 16%) countries, such as South Africa and Nepal, have experienced no decoupling 26 between GDP and CBEs from 2015 to 2018, meaning the growth of their GDP is closely tied with the 27 consumption of emission-intensive goods. As a result, a further increase of GDP in these countries will 28 likely lead to higher emissions, if they follow the historical trend without substantive improvement in 29 efficiency of production and energy use. 30 It is important to note that a country’s degree of decoupling changes over time. For example, 32 31 countries achieved absolute decoupling from 2010 to 2015 but only 10 of them remained decoupled 32 over the next three years. More importantly, although absolute decoupling has reduced annual emissions, 33 the remaining emissions are still contributing to an increase in atmospheric carbon concentration. 34 Absolute decoupling is not sufficient to avoid consuming the remaining CO2 emission budget under the 35 global warming limit of 1.5°C or 2°C and to avoid climate breakdown (Stoknes and Rockström, 2018; 36 Hickel and Kallis, 2020). Even if all countries decouple in absolute terms this might still not be 37 sufficient and thus can only serve as one of the indicators and steps toward fully decarbonizing the 38 economy and society. 39 2.3.4 Emissions embodied in trade (EET) 40 As global trade patterns have changed over recent decades, so have emissions embodied in trade (EET) 41 (Jiang & Green, 2017). EET refers to emissions associated with production of traded goods and services 42 and is equal to the difference between PBEs and CBEs (Wiebe and Yamano, 2016). EET includes two 43 parts: emissions embodied in imports (EEI) and emissions embodied in exports (EEE). For a given Do Not Cite, Quote or Distribute 2-39 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 country or region with CBEs higher than PBEs, a country is a net importer with a higher EEI than EEE, 2 and vice versa. 3 4 Figure 2.15 Total annual CO2 emissions for 116 countries by global region based on consumption and 5 production-based emissions 6 The shaded areas are the net CO2 trade balances (differences) between each of the regions. Yellow 7 shading indicates that the region is a net importer of embodied CO 2 emissions, leading to consumption- 8 based emission estimates that are higher than traditional territorial emission estimates. Blue shading 9 indicates the reverse. Production-based emissions are collected from EDGAR and consumption-based 10 emissions from the Global Carbon Budget 2020 (Friedlingstein et al., 2020). 11 Source: This figure is modified based on Hubacek et al. (2021). 12 EET have been rising faster since the 1980s due to an increase in trade volume (Xu and Dietzenbacher, 13 2014; Wood et al., 2018b; Zhong et al., 2018). CO2 emissions from the production of internationally 14 traded products peaked in 2006 at about 26% of global CO2 emissions. Since then, international CO2 15 emissions transfers declined but are likely to remain an important part of the climate policy agenda 16 (Wood et al., 2019c). About 24% of global economic output and 25% of global CO2 emissions are 17 embodied in the international trade of goods and services as of 2014 (Hubacek et al., 2021). Do Not Cite, Quote or Distribute 2-40 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2.3.4.1 Net emission transfers 2 Located downstream in global supply chains, developed countries (mostly in Western Europe and North 3 America) tend to be net emission importers, i.e., EEI are larger than EEE. For example, over 40% of 4 national CO2 footprints in France, Germany, Italy, and Spain are from imports (Fan et al., 2017). 5 Developing countries tend to be net emission exporters with higher PBEs than their CBEs (Peters et al., 6 2011b; Le Quéré et al., 2018), especially for Asia and Developing Pacific (as shown in Figure 2.15). 7 That is to say, there is a net emission transfer and outsourcing of carbon-intensive production from 8 developed to developing economies via global trade (Jiang et al., 2018), mainly caused by cheap labour 9 costs (Tate and Bals, 2017) and cheap raw materials (Mukherjee, 2018). Increasing openness to trade 10 (Fernández-Amador et al., 2016) and less stringent environmental legislation (acting as so-called 11 pollution havens) are also possible reasons (Hoekstra et al., 2016; Malik and Lan, 2016; Banerjee and 12 Murshed, 2020). 13 Net emissions transferred between developing and developed countries peaked at 7.3% of global CO2 14 emissions in 2006 and then subsequently decline (Wood et al., 2019c). The main reason for the decline 15 was an improvement in the carbon intensity of traded products of about 40% between 1995 and 2015, 16 rather than a decline in trade volume (Wood et al., 2019c). Despite continued improvements, developing 17 economies tend to have higher emission intensity than developed economies due to less efficient 18 technologies and a carbon-intensive fuel mix (Liu et al., 2015a; Jiang and Guan, 2017). 19 2.3.4.2 Geographical shifts of trade embodied emissions 20 With the rapid growth of developing countries, the geographical centre of global trade as well as trade 21 embodied emissions is changing. The fast growth of Asian countries is shifting the global trade centre 22 from Europe to Asia (Zhang et al., 2019). Asian exports in monetary units increased by 235% from 23 1996 to 2011, and its share in global exports increased from 25% to 46%, whereas Europe’s share in 24 global exports decreased from 51% in 1996 to 39% in 2011. After 2011, global trade has stalled, but 25 Asia's share of global exports further increased to 42% in 2020 (UNCTAD, 2021). 26 In addition to changes in trade volume, trading patterns have also been changing significantly in Asian 27 countries. These countries are replacing traditional trading hubs (such as Russia and Germany) due to 28 the fast growth in trade flows, especially with countries of the global South (Zhang et al., 2019). The 29 largest geographical shifts in trade-embodied emissions between 1995 and 2011 occurred in high-tech, 30 electronics, and machinery (Malik and Lan, 2016; Jiang et al., 2018a). For example, China is shifting 31 its exports to include more low-carbon and higher value-added goods and services. As a result, China’s 32 exported emissions declined by 20% from 2008 to 2015 (Mi et al., 2018). 33 As a result, developing countries are increasingly playing an important role in global trade. Emissions 34 embodied in trade between developing countries, so-called South-South trade, has more than doubled 35 between 2004 (0.47 Gt) and 2011 (1.11 Gt), which is seen as a reflection of a new phase of globalisation 36 (Meng et al., 2018). Developing countries, therefore, have gained importance as global suppliers of 37 goods and services and have also become more relevant as global consumers as they grow their 38 domestic demand (Fernández-Amador et al., 2016). Since 2014, CO2 emission transfer between 39 developing countries has plateaued and then slightly declined and seems to have stabilised at around 40 the same level of transfers between non-OECD and OECD countries at around 2.4 GtCO2 yr-1 (Wood 41 et al., 2019a). In both cases, a decrease in carbon intensity of trade just about offset increased trade 42 volumes (Wood et al., 2019a). 43 44 2.4 Economic Drivers and Their Trends by Regions and Sectors 45 This section provides a summary of the main economic drivers of GHG emissions (mostly territorial) 46 by regions and sectors, including those that are more indirect drivers related to economic activity, such Do Not Cite, Quote or Distribute 2-41 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 as inequality and rapid urbanisation. Trade as a driver of global GHG emissions is described in the 2 Chapter 2 Supplementary Material. Socio-demographic drivers are described in Section 2.6. The Kaya 3 decomposition presented in this section is based on the IEA and EDGAR v6 databases and tracks global, 4 regional, and sectoral GHG emissions from 1990 to 2019 (Crippa et al., 2021; Minx et al., 2021; Lamb 5 et al., 2021b; IEA, 2021c). It shows main contributors to GHG emissions as independent factors, 6 although these factors also interact with each other. 7 2.4.1 Economic Drivers at Global and Regional Levels 8 Economic growth (measured as GDP) and its main components, GDP per capita and population growth, 9 remained the strongest drivers of GHG emissions in the last decade, following a long-term trend (robust 10 evidence, high agreement) (Liddle, 2015; Malik et al., 2016; Sanchez and Stern, 2016; Chang et al., 11 2019; Dong et al., 2019; Liobikiene and Butkus, 2019; Liu et al., 2019a; Mardani et al., 2019; Pan et 12 al., 2019; Dong et al., 2020; Parker and Bhatti, 2020; Xia et al., 2021). Globally, GDP per capita 13 remained by far the strongest upward driver, increasing almost in tandem with energy consumption and 14 CO2 emissions up until 2015, after which some modest decoupling occurred (Deutch, 2017; Wood et 15 al., 2018b) (Section 2.3.3). The main counteracting, yet insufficient, factor that led to emissions 16 reductions was decreased energy use per unit of GDP in almost all regions (-2.0% yr-1 between 2010 17 and 2019 globally (Figure 2.16), see also (Lamb et al., 2021b) (robust evidence, high agreement). These 18 reductions in energy intensity are a result of technological innovation, structural changes, regulation, 19 fiscal support, and direct investment, as well as increased economic efficiency in underlying sectors 20 (Yao et al., 2015; Sanchez and Stern, 2016; Chang et al., 2019; Dong et al., 2019a; Mohmmed et al., 21 2019; Stern, 2019; Azhgaliyeva et al., 2020; Goldemberg, 2020; Gao et al., 2021; Liddle and 22 Huntington, 2021; Xia et al., 2021; Liu et al., 2019b). 23 The decades-long trend that efficiency gains were outpaced by an increase in worldwide GDP (or 24 income) per capita continued unabated in the last ten years (robust evidence, high agreement) 25 (Wiedmann et al., 2020; Xia et al., 2021). In addition, the emissions-reducing effects of energy 26 efficiency improvements are diminished by the energy rebound effect, which has been found in several 27 studies to largely offset any energy savings (robust evidence, high agreement) (Rausch and Schwerin, 28 2018; Bruns et al., 2021; Colmenares et al., 2020; Stern, 2020; Brockway et al., 2021). The rebound 29 effect is discussed extensively in Section 9.9.2. 30 A significant decarbonisation of the energy system was only noticeable in North America, Europe and 31 Eurasia. Globally, the amount of CO2 per unit of energy used has practically remained unchanged over 32 the last three decades (Chang et al., 2019; Tavakoli, 2018), although it is expected to decrease more 33 consistently in the future (Xia et al., 2021). Population growth has also remained a strong and persistent 34 upward driver in almost all regions (+1.2% yr-1 globally from 2010 to 2019, Figure 2.16 , see also Lamb 35 et al., 2021), although per capita emission levels are very uneven across world regions. Therefore, 36 modest population increases in wealthy countries may have a similar impact on emissions as high 37 population increases in regions with low per capita emission levels. Do Not Cite, Quote or Distribute 2-42 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 Figure 2.16 Trends and drivers of global GHG emissions, including a) trends of GHG emissions by 3 sectors 1990–2019, b) share of total and per capita GHG emissions by world region in 2019, and c) Kaya 4 decomposition of CO2 emissions drivers 5 The Kaya decomposition is based on the equation F = P(G/P)(E/G)(F/E), where F is CO 2 emissions, P is 6 population, G/P is GDP per capita, E/G is the energy intensity of GDP and F/E is the carbon intensity of 7 energy. The indicated annual growth rates are averaged across the years 2010–2019 (in panel c, these are 8 for fossil fuel CO2 emissions only, in order to ensure compatibility with underlying energy data). Note 9 that the energy consumption by itself (primary energy supply) is not part of the decomposition, but is 10 listed here for comparison with the Kaya factors. 11 Source: Data from Crippa et al. (2021), IEA (IEA, 2021c), Minx et al. (2021) 12 Developing countries remained major accelerators of global CO2 emissions growth since 2010, mostly 13 driven by increased consumption and production, in particular in East Asia (robust evidence, high 14 agreement) (Jiborn et al., 2020). While energy intensity declined to a similar extent in countries of the 15 OECD (Organisation for Economic Co-operation and Development) and non-OECD countries over the 16 last 30 years, economic growth has been much stronger in non-OECD countries (González-Torres et 17 al., 2021). This led to an average annual growth rate of 2.8% of CO2 emissions in these countries, 18 whereas they decreased by 0.3% yr-1 in OECD countries (UNEP, 2019). The majority of developed 19 economies reduced both production-based and consumption-based CO2 emissions modestly (Jiborn et 20 al., 2020; Xia et al., 2021). This was due to slower economic growth, increased energy efficiency (less 21 energy per unit of GDP), fuel switching from coal to gas (mostly in North America) (Wang et al., Do Not Cite, Quote or Distribute 2-43 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2020b), and the use of less and cleaner energy from renewables in Europe (Peters et al., 2017; 2 Karstensen et al., 2018; Chang et al., 2019; Wood et al., 2019c). 3 Economic growth as the main driver of GHG emissions plays out particularly strong in China and India 4 (robust evidence, high agreement) (Liu et al., 2019b; Ortega-Ruiz et al., 2020; Wang et al., 2020c; Yang 5 et al., 2020; Zheng et al., 2020; Xia et al., 2021), although both countries show signs of relative 6 decoupling because of structural changes (Marin and Mazzanti, 2019). A change in China's production 7 structure (with relatively less heavy industry and lower-carbon manufacturing) and consumption 8 patterns (i.e., the type of goods and services consumed) has become the main moderating factor of 9 emissions after 2010, while economic growth, consumption levels, and investment remain the 10 dominating factors driving up emissions (Wang and Jiang, 2019; Jiborn et al., 2020; Zheng et al., 2020). 11 In India, an expansion of production and trade as well as a higher energy intensity between 2010 and 12 2014 caused growth of emissions (Kanitkar et al., 2015; Wang and Zhou, 2020; (Wang et al., 2020d)). 13 2.4.2 Sectoral Drivers 14 GHG emissions continued to rise since 2010 across all sectors and subsectors, most rapidly in electricity 15 production, industry, and transport. Decarbonisation gains from improvements in energy efficiency 16 across different sectors and worldwide have been largely wiped out by increases in demand for goods 17 and services. Prevailing consumption patterns have also tended to aggravate energy use and emissions, 18 with the long-term trend led by developed regions. Decarbonisation trends in some developed regions 19 are limited in size and geographically. Globally, there are enormous unexploited mitigation potentials 20 from adopting best available technologies. 21 The following subsections discuss main emissions drivers by sector. More detailed analyses of sectoral 22 emissions and mitigation options are presented in Chapters 6–11. 23 2.4.2.1 Energy systems 24 Global energy system emissions growth has slowed down in recent years, but global oil and gas use 25 was still growing (Jackson et al., 2019) and the sector remained the single largest contributor to global 26 GHG emissions in 2019 with 20 GtCO2-eq (34%) (high confidence) (Figure 2.17). Most of the 14 27 GtCO2-eq from electricity and heat generation (23% of global GHG emissions in 2019) were due to 28 energy use in industry and in buildings, making these two sectors also prominent targets for mitigation 29 (Davis et al., 2018; Crippa et al., 2019) (see subsections below). 30 Growth in CO2 emissions from energy systems has closely tracked rising GDP per capita globally 31 (Lamb et al., 2021b), affirming the substantial literature describing the mutual relationship between 32 economic growth and demand for energy and electricity (robust evidence, high agreement) (Khanna 33 and Rao, 2009; Stern, 2011). This relationship has played out strongly in developing regions, 34 particularly in Asia, where a massive scale up of energy supply has accompanied economic growth – 35 with average annual increases of energy demand between 3.8 and 4.3% in 2010–2019 (Figure 2.17). 36 The key driver for slowing the growth of energy systems CO2 emissions has been declining energy 37 intensities in almost all regions. Annually, 1.9% less energy per unit of GDP was used globally between 38 2010 and 2019. Do Not Cite, Quote or Distribute 2-44 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 Figure 2.17 Trends and drivers of global energy sector emissions (see caption of Figure 2.16 for details) 3 Energy is here measured as primary energy supply. 4 5 The carbon intensity of power generation varies widely between (and also within) regions (see also 6 Chapter 6). In North America, both a switch from coal to gas for power generation (Peters et al., 2017, 7 2020; Feng, 2019; Mohlin et al., 2019) as well as an overall decline in the share of fossil fuels in 8 electricity production (from 66% in 2010 to 59% in 2018) (Mohlin et al., 2019) has decreased carbon 9 intensity and CO2 emissions. Since 2007, Europe’s carbon intensity improvements have been driven by 10 the steady expansion of renewables in the share of electricity generation (medium evidence, high 11 agreement) (Peters et al., 2017, 2020; Le Quéré et al., 2019; Rodrigues et al., 2020). Some studies 12 attribute these effects to climate policies, such as the carbon floor price in the UK, the EU emissions 13 trading scheme, and generous renewable energy subsidies across the continent (Dyrstad et al., 2019; 14 Wang et al., 2020a). South-East Asian and Asia-Pacific developed countries stand out in contrast to 15 other developed regions, with an increase of regional carbon intensity of 1.8 and 1.9% yr-1, respectively 16 (Figure 2.17). Generally, the use of natural gas for electricity production is growing strongly in most 17 countries and gas has contributed to the largest increase in global fossil CO2 emissions in recent years 18 (Jackson et al., 2019; Peters et al., 2020). Furthermore, gas brings the risk of increased CH4 emissions 19 from fugitive sources, as well as large cumulative emissions over the lifetime of new gas power plants 20 that may erase early carbon intensity reductions (Shearer et al., 2020). Do Not Cite, Quote or Distribute 2-45 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 The growth of emissions from coal power slowed after 2010, and even declined between 2011 and 2 2019, primarily due to a slowdown of economic growth and fewer coal capacity additions in China 3 (Peters et al., 2020; Friedlingstein et al., 2019). Discussions of a global ‘peak coal’, however, may be 4 premature, as further growth was observed in 2019 (Peters et al., 2020; Friedlingstein et al., 2019). 5 Large ongoing and planned capacity increases in India, Turkey, Indonesia, Vietnam, South Africa, and 6 other countries have become a driver of thermal coal use after 2014 (UNEP, 2017; Steckel et al., 2019; 7 Edenhofer et al., 2018). 8 2.4.2.2 Industry sector 9 When indirect emissions from electricity and heat production are included, industry becomes the single 10 highest emitting sector of GHGs (20.0 GtCO2-eq in 2019) (high confidence). Facilitated by 11 globalisation, East Asia has been the main source and primary driver of global industry emissions 12 growth since 2000 (robust evidence, high agreement) (Lamb et al., 2021). However, while East Asia 13 has emitted 45% of the world’s industry GHG emissions in 2019, a remarkable decrease of 5.0% yr -1 in 14 energy intensity and 1.6% in carbon intensity helped to stabilise direct industrial CO2 emissions in this 15 region (-0.3% yr-1 between 2010 and 2019; Figure 2.18). Direct industry CO2 emissions have also 16 declined in Latin America, Europe and Asia-Pacific developed regions and – to a smaller extent – in 17 North America. In all other regions, they were growing – most rapidly in southern Asia (+4.3% annually 18 for direct CO2 emissions since 2010, Figure 2.18). 19 20 Figure 2.18 Trends and drivers of global industry sector emissions (see caption of Figure 2.16 for details). 21 Energy is here measured as total final energy consumption. Do Not Cite, Quote or Distribute 2-46 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 The main global driver of industry emissions has been a massive rise in the demand for products that 3 are indirectly used in production, such as cement, chemicals, steel, aluminium, wood, paper, plastics, 4 lubricants, fertilizers, and so on. This demand was driven by economic growth, rising affluence, and 5 consumption, as well as a rapid rise in urban populations and associated infrastructure development 6 (robust evidence, high agreement) (Krausmann et al., 2018). There is strong evidence that the growing 7 use of concrete, steel, and other construction materials is particularly tightly coupled to these drivers 8 (Cao et al., 2017; Pauliuk et al., 2013; Plank et al., 2018; Haberl et al., 2020; Krausmann et al., 2017). 9 Per capita stocks of cement and steel show a typical pattern of rapid take-off as countries urbanise and 10 industrialise, before slowing down to low growth at high levels of GDP. Hence, in countries that have 11 recently been industrialising and urbanising – that is Eastern, Southern and South-Eastern Asia – a 12 particularly strong increase of emissions from these subsectors can be observed. Selected wealthy 13 countries seem to stabilise at high per capita levels of stocks, although it is unclear if these stabilizations 14 persist and if they result in significant absolute reductions of material use (Wiedenhofer et al., 2015; 15 Cao et al., 2017; Krausmann et al., 2018). Opportunities for prolonging lifetimes and improving end of 16 life recycling in order to achieve absolute reductions in extraction activities are as yet unexploited 17 (Krausmann et al., 2017; Zink and Geyer, 2017). 18 On the production side, improvements in the efficiency of material extraction, processing, and 19 manufacturing have reduced industrial energy use per unit of output (Wang et al., 2019b). These 20 measures, alongside improved material substitution, lightweight designs, extended product and 21 servicing lifetimes, improved service efficiency, and increased reuse and recycling will enable 22 substantial emissions reductions in the future (Hertwich et al., 2019). In absence of these improvements 23 in energy intensity, the growth of population and GDP per capita would have driven the industrial CO2 24 emissions to rise by more than 100% by 2017 compared with 1990, instead of 56% (Lamb et al., 2021b). 25 Nonetheless, many studies point to deep regional differences in efficiency levels and large globally 26 unexploited potentials to improve industrial energy efficiency by adopting best available technologies 27 and practices for metal, cement, and chemical production (Gutowski et al., 2013; Talaei et al., 2018; 28 Schulze et al., 2016; Hernandez et al., 2018). 29 2.4.2.3 Buildings sector 30 Global direct and indirect GHG emissions from the buildings sector reached 9.7 GtCO2-eq in 2019, or 31 16% of global emissions). Most of these emissions (66%, or 6.4 GtCO2-eq) were upstream emissions 32 from power generation and commercial heat (Figure 2.19). The remaining 33% (3.3 GtCO2-eq) of 33 emissions were directly produced in buildings, for instance by gas and coal boilers, and cooking and 34 lighting devices that burn kerosene, biomass, and other fuels (Lamb et al., 2021). Residential buildings 35 accounted for the majority of this sector’s emissions (64%, 6.3 GtCO2-eq, including both direct and 36 indirect emissions), followed by non-residential buildings (35%, 3.5 GtCO2-eq) (high confidence). 37 Global buildings sector GHG emissions increased by 0.7% yr-1 between 2010 and 2019 (Figure 2.19), 38 growing the most in absolute terms in East and South Asia, whereas they declined the most in Europe, 39 mostly due to the expansion of renewables in the energy sector and increased energy efficiency (Lamb 40 et al., 2021). North America has the highest per capita GHG emissions from buildings and the second 41 highest absolute level after East Asia (Figure 2.19). Do Not Cite, Quote or Distribute 2-47 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 Figure 2.19 Trends and drivers of global buildings sector emissions (see caption of Figure 2.16 for 3 details). Energy is here measured as total final energy consumption. 4 5 Rising wealth has been associated with more floor space being required to service growing demand in 6 the retail, office, and hotel sectors (medium evidence, high agreement) (Daioglou et al., 2012; Deetman 7 et al., 2020). In addition, demographic and social factors have driven a cross-national trend of increasing 8 floor space per capita. As populations age and decrease in fertility, and as individuals seek greater 9 privacy and autonomy, households declined in size, at least before the COVID-19 pandemic (Ellsworth- 10 Krebs, 2020). These factors lead to increased floor space per capita, even as populations stabilise. This 11 in turn is a key driver for building sector emissions, because building characteristics such as size and 12 type, rather than occupant behaviour, tend to explain the majority of energy use within dwellings 13 (Guerra Santin et al., 2009; Ürge-Vorsatz et al., 2015; Huebner and Shipworth, 2017) (see Chapter 9). 14 Energy activity levels further drive regional differences. In Eurasia, Europe and North America, thermal 15 demands for space heating dominate building energy use, at 66%, 62% and 48% of residential energy 16 demand, respectively (IEA, 2020a). In contrast, cooking has a much higher share of building energy 17 use in regions of the global South, including China (Cao et al., 2016). And despite temperatures being 18 on average warmer in the global South, electricity use for cooling is a more prominent factor in the 19 global North (Waite et al., 2017). This situation is changing, however, as rapid income growth and Do Not Cite, Quote or Distribute 2-48 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 demographic changes in the global South enable households to heat and cool their homes (Ürge-Vorsatz 2 et al., 2015, 2020). 3 Steady improvements in building energy intensities across regions can be attributed to baseline 4 improvements in building fabrics, appliance efficiencies, energy prices, and fuel shifts. Many countries 5 have adopted a mix of relevant policies, such as energy labelling, building energy codes, and mandatory 6 energy performance requirements (Nie and Kemp, 2014; Nejat et al., 2015; Economidou et al., 2020). 7 Efforts towards buildings refurbishments and retrofits have also been pursued in several nations, 8 especially for historical buildings in Europe, but evidence suggests that the recent rates of retrofits have 9 not made a significant dent on emissions (Corrado and Ballarini, 2016). The Chinese central 10 government launched various policies, including command and control, economic incentives, and 11 technology measures, but a big gap remains between the total rate of building green retrofit in the nation 12 and the future retrofit potential (Liu et al., 2020a, 2020b). Still, one major global factor driving down 13 energy intensities has been the global transition from inefficient coal and biomass use in buildings for 14 heating and cooking, towards natural gas and electricity, in part led by concerted policy action in Asian 15 countries (Kerimray et al., 2017; Thoday et al., 2018; Ürge-Vorsatz et al., 2015). As developing 16 countries construct new buildings, there is sizable potential to reduce and use less carbon-intensive 17 building materials and adopt building designs and standards that lower life cycle buildings energy use 18 and allow for passive comfort. Chapter 9 describes the mitigation options of the buildings sector. 19 2.4.2.4 Transport sector 20 With a steady, average annual growth of +1.8% yr-1 between 2010 and 2019, global transport GHG 21 emissions reached 8.9 GtCO2-eq in 2019 and accounted for 15% of all direct and indirect emissions 22 (Figure 2.20). Road transport passenger and freight emissions represented by far the largest component 23 and source of this growth (6.1 GtCO2-eq, 69% of all transport emissions in 2019) (high confidence). 24 National plus international shipping and aviation emissions together accounted for 2.0 GtCO2-eq or 25 22% of the sector’s total in 2019. North America, Europe and Eastern Asia stand out as the main 26 regional contributors to global transport emissions and together account for 50% of the sector’s total. 27 The proportion of total final energy used in transport (28%) and its fast expansion over time weighs 28 heavily on climate mitigation efforts, as 92% of transport energy comes from oil-based fuels (IEA, 29 2020b). These trends situate transport as one of the most challenging sectors for climate change 30 mitigation – no country has so far been able to realise significant emissions reductions in the sector. 31 North America’s absolute and per capita transport emissions are the highest amongst world regions, but 32 those of South, South-East and East Asia are growing the fastest (between +4.6% and +5.2% yr -1 for 33 CO2 between 2010 and 2019, Figure 2.20) (high confidence). Do Not Cite, Quote or Distribute 2-49 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 Figure 2.20 Trends and drivers of global transport sector emissions (see caption of Figure 2.16 for 3 details). Energy is here measured as total final energy consumption. 4 5 More so than any other sector, transport energy use has tracked GDP per capita growth (Figure 2.20), 6 (Lamb et al., 2021). With the exception of road gasoline demand in OECD countries, the demand for 7 all road fuels generally increases at least as fast as the rate at which GDP per capita increases (Liddle 8 and Huntington, 2020). Developments since 1990 continue a historical trend of increasing travel 9 distances and a shift from low- to high-speed transport modes that goes along with GDP growth (Schäfer 10 et al., 2009; Gota et al., 2019). Modest improvements in energy efficiency have been realised between 11 2010 and 2019, averaging -1.5% yr-1 in energy intensity globally, while carbon intensities of the 12 transport sector have remained stable in all world regions (Figure 2.20). Overall, global increases in 13 passenger and freight travel activity levels have outpaced energy efficiency and fuel economy 14 improvements, continuing a long-term trend for the transport sector (Gucwa and Schäfer, 2013; Grübler, 15 2015; McKinnon, 2016) (medium evidence, high agreement). 16 Despite some policy achievements, energy use in the global transport system remains to the present 17 deeply rooted in fossil fuels (robust evidence, high agreement) (IEA, 2019; Figueroa et al., 2014). In 18 part this is due to the increasing adoption of larger, heavier combustion-based vehicles in some regions, 19 which have tended to far outpace electric and hybrid vehicle sales (Chapter 10). Yet, stringent material Do Not Cite, Quote or Distribute 2-50 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 efficiency and lightweight design of passenger vehicles alone would have the potential to cut cumulative 2 global GHG emissions until 2060 by 16–39 GtCO2-eq (Pauliuk et al., 2020). 3 While global passenger activity has expanded in all world regions, great disparities exist between low 4 and high income regions, and within countries between urban and rural areas (ITF, 2019). While private 5 car use is dominant in OECD countries (EC, 2019), the growth of passenger-km (the product of number 6 of travellers and distance travelled) has considerably slowed there, down to an increase of just 1% yr -1 7 between 2000 and 2017 (SLoCaT, 2018) (Chapter 10). Meanwhile, emerging economies in the global 8 South are becoming more car-dependent, with rapidly growing motorisation, on-demand private 9 transport services, urban sprawl, and the emergence of local automotive production, while public 10 transport struggles to provide adequate services (Dargay et al., 2007; Hansen and Nielsen, 2017; Pojani 11 and Stead, 2017). 12 Freight travel activity grew across the globe by 68% in the last two decades driven by global GDP 13 increases, together with the proliferation of online commerce and rapid (i.e., same-day and next-day) 14 delivery (SLoCaT, 2018). Growth has been particularly rapid in heavy-duty road freight transport. 15 While accounting for a small share of total GHG emissions, domestic and international aviation have 16 been growing faster than road transport emissions, with average annual growth rates of +3.3% and 17 +3.4%, respectively, between 2010 and 2019 (Crippa et al., 2021; Minx et al., 2021;). Energy efficiency 18 improvements in aviation were considerably larger than in road transport, but were outpaced by even 19 larger increases in activity levels (SLoCaT, 2018; Lee et al., 2021) (Chapter 10). 20 2.4.2.5 AFOLU sector 21 GHG emissions from agriculture, forestry and land use reached 13 GtCO2-eq globally in 2019 (Figure 22 2.21) (medium confidence). AFOLU trends, particularly those for CO2-LULUCF, are subject to a high 23 degree of uncertainty (Section 2.2.1). Overall, the AFOLU sector accounts for 22% of total global GHG 24 emissions, and in several regions – Africa, Latin America, and South-East Asia – it is the single largest 25 emitting sector (which, at the same time, is also significantly affected itself by climate change; see WGI 26 Chapters 8, 11, and 12, and WGII Chapter 5). Latin America has the highest absolute and per capita 27 AFOLU GHG emissions of any world region (Figure 2.21). CO2 emissions from land-use change and 28 CH4 emissions from enteric fermentation together account for 74% of sector-wide GHGs. Note that 29 CO2-LULUCF estimates included in this chapter are not necessarily comparable with country GHG 30 inventories, due to different approaches to estimate anthropogenic CO2 sinks (Grassi et al., 2018) 31 (Chapter 7). Do Not Cite, Quote or Distribute 2-51 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 Figure 2.21 Trends and drivers of global AFOLU sector emissions, including a) trends of GHG emissions 3 by subsectors 1990–2019, b) share of total sector and per capita GHG emissions by world region in 2019, 4 and c) Kaya decomposition of GHG emissions drivers 5 Based on the equation H=P(A/P)(L/A)(H/L), where P is population, A/P is agricultural output per capita, 6 L/A is the land required per unit of agricultural output (land efficiency), and H/L is GHG emissions per 7 unit of land (GHG intensity) (Hong et al., 2021). GHG emissions H comprise agricultural CH4 and N2O 8 emissions from EDGAR v6.0. The indicated annual growth rates are averaged across the years 2010–2019 9 (LULUCF CO2 emissions are excluded in panel c). (Note: due to different datasets, the population 10 breakdown for AFOLU emissions is slightly different than that in the other sector figures above). 11 12 Unlike all other sectors, AFOLU emissions are typically higher in developing compared to developed 13 regions (medium confidence). In Africa, Latin America, and South-East Asia, CO2 emissions associated 14 with land-use change and management predominate, dwarfing other AFOLU and non-AFOLU sources 15 and making AFOLU the single largest sector with more than 50% of emissions in these regions (Lamb 16 et al., 2021b). Land-use and -management emissions there is associated with the expansion of 17 agriculture into carbon-dense tropical forest areas (Vancutsem et al., 2021), where large quantities of 18 CO2 emissions result from the removal and burning of biomass and draining of carbon rich soils 19 (Pearson et al., 2017; IPCC, 2018; Hong et al., 2021). Ruminant livestock rearing takes place on vast 20 tracts of pasture land worldwide, contributing to large quantities of CH4 emissions from enteric 21 fermentation in Latin America (0.8 GtCO2-eq in 2018), Southern Asia (0.6 GtCO2-eq), and Africa (0.5 Do Not Cite, Quote or Distribute 2-52 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 GtCO2-eq), while also playing a sizable role in the total AFOLU emissions of most other regions (Lamb 2 et al., 2021b). 3 In all regions, the amount of land required per unit of agricultural output has decreased significantly 4 from 2010 to 2019, with a global average of -2.5% yr-1 (land efficiency metric in Figure 2.21). This 5 reflects agricultural intensification and technological progress. However, in most regions this was 6 mirrored by an increase in output per capita, meaning that absolute GHG emissions in most regions 7 increased over the last decade. A significant increase in total AFOLU emissions occurred in Africa, 8 driven by both increased GHG emissions per unit of land and increased populations (Figure 2.21). 9 The AFOLU sector and its emissions impacts are closely tied to global supply chains, with countries in 10 Latin America and South-East Asia using large portions of their land for agricultural and forestry 11 products exported to other countries (see Chapter 7). The strong increases in production per capita and 12 associated GHG emissions seen in these regions are at least partly attributable to growing exports and 13 not national food system or dietary changes. At the same time, efforts to promote environmental 14 sustainability in regions like the EU and the USA (but also fast-growing emerging economies such as 15 China) can take place at the cost of increasing land displacement elsewhere to meet their own demand 16 (Meyfroidt et al., 2010; Yu et al., 2013; Creutzig et al., 2019). 17 Global diets are a key driver of production per capita, and thus land pressure and AFOLU emissions 18 (Chapter 7). As per capita incomes rise and populations urbanise, traditional, low-calorie diets that 19 emphasise starchy foods, legumes, and vegetables transition towards energy-intensive products such as 20 refined sugars, fats, oils, and meat (Tilman and Clark, 2014; Pradhan et al., 2013). At a certain point in 21 national development, affluence and associated diets thus override population growth as the main driver 22 of AFOLU emissions (Kastner et al., 2012). Very high calorie diets have high total GHG emissions per 23 capita (Heller and Keoleian, 2015) and are common in the developed world (Pradhan et al., 2013). Over 24 the last few decades, a “westernisation” of diets has also been occurring in developing countries 25 (Pradhan et al., 2013). Low- and middle-income countries such as India, Brazil, Egypt, Mexico, and 26 South Africa have experienced a rapid dietary shift towards western-style diets (De Carvalho et al., 27 2013; Pradhan et al., 2013; Popkin, 2015). Another driver of higher food requirements per capita is food 28 waste, the amounts of which increased more or less continuously since the 1960s in all regions but 29 Europe (Porter and Reay, 2016). 30 2.4.3 Poverty and Inequality 31 Increasing economic inequality globally has given rise to concern that unequal societies may be more 32 likely to pollute and degrade their environments (Chancel, 2020; Hailemariam et al., 2020; Millward- 33 Hopkins and Oswald, 2021; Masud et al., 2018). The nature of this relationship has important 34 implications for the design of income redistribution policies aiming to reduce inequalities (Section 2.6 35 presents evidence on how affluence and high consumption relate to emissions). Income inequality and 36 carbon intensity of consumption differs across countries and individuals (Baležentis et al., 2020) 37 (Section 2.3.3). Reduced income inequality between nations can reduce emissions intensity of global 38 income growth, if energy intensity reductions from income growth in some nations offset increases in 39 energy and emissions from higher growth in other nations (Rao and Min, 2018). Increasing income 40 inequality between individuals can translate into larger energy and emissions inequality if higher 41 incomes are spent on more energy-intensive consumption and affluent lifestyles (Oswald et al., 2020; 42 Wiedmann et al., 2020) (Section 2.6). 43 Literature shows that more equitable income distributions can improve environmental quality, but the 44 nature of this relationship can vary by level of development (low evidence, medium agreement) (Knight 45 et al., 2017; Chen et al., 2020; Hailemariam et al., 2020; Huang and Duan, 2020; Liobikienė and 46 Rimkuvienė, 2020; Rojas-Vallejos and Lastuka, 2020; Uddin et al., 2020). Differences in the energy 47 and carbon intensities of consumption and the composition of consumption baskets across populations Do Not Cite, Quote or Distribute 2-53 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 and nations matter for emissions. (Jorgenson et al., 2016; Grunewald et al., 2017). There is evidence to 2 suggest that more equal societies place a higher value on environmental public goods (Baumgärtner et 3 al., 2017; Drupp et al., 2018). Additional research shows that reducing top income inequality in OECD 4 countries can reduce carbon emissions and improve environmental quality (Hailemariam et al., 2020) 5 and that the effect of wealth inequality, measured as the wealth share of the top decile, on per capita 6 emissions in high-income countries is positive (Knight et al., 2017). Evidence from 40 sub-Saharan 7 African countries suggests that a rise in income inequality contributed to increasing CO2 emissions 8 between 2010 and 2016, controlling for other drivers like economic growth, population size, and 9 inflation (Baloch et al., 2020). 10 The key development objective of eradicating extreme poverty (Hubacek et al., 2017a; Chakravarty and 11 Tavoni, 2013; Malerba, 2020) and providing universal access to modern energy services (Pachauri et 12 al., 2018, 2013; Singh et al., 2017; Pachauri, 2014) only marginally effect carbon emissions (robust 13 evidence, medium agreement). Shifts from biomass to more efficient energy sources and collective 14 provisioning systems for safe water, health, and education are associated with reduced energy demand 15 (Baltruszewicz et al., 2021). Efforts to alleviate multi-dimensional poverty by providing decent living 16 standards universally, however, may require more energy and resources. Recent estimates of the 17 additional energy needed are still within bounds of projections of energy demand under climate 18 stabilisation scenarios (Rao et al., 2019; Pascale et al., 2020; Hubacek et al., 2017b; a); Kikstra et al., 19 2021). Bottom-up estimates suggest that achieving decent living standards requires 13–40 GJ per capita 20 annually, much less than the current world average energy consumption of 80 GJ per capita in 2020 21 (Millward-Hopkins et al., 2020) (medium evidence, high agreement). Aggregate top-down estimates 22 suggest that achieving a high Human Development Index (HDI) score above 0.8 requires energy 23 consumption between 30–100 GJ per capita yr-1 (Lamb and Rao, 2015). There is some evidence, 24 however, of a decoupling between energy consumption and HDI over time (Akizu-Gardoki et al., 2018). 25 The emissions consequences of poverty alleviation and decent living also depend on whether 26 improvements in well-being occur via energy- and carbon-intensive industrialisation or low-carbon 27 development (Semieniuk and Yakovenko, 2020; Fu et al., 2021; Huang and Tian, 2021). 28 2.4.4 Rapid and Large-scale Urbanisation as a Driver of GHG Emissions 29 Economic growth and urbanisation go hand in hand and are both influencing GHG emissions. However, 30 the exact role of urban development in driving emissions is multi-faceted and heterogeneous, depending 31 on development status and other regional factors (medium evidence, high agreement) (Jorgenson et al., 32 2014; Lamb et al., 2014; Liddle and Lung, 2014; Creutzig et al., 2015; Pincetl, 2017; Azizalrahman and 33 Hasyimi, 2019; Muñoz et al., 2020). This calls for a differentiated assessment. This section assesses the 34 process of rapid urban growth in developing countries and how emissions change over time when cities 35 grow rapidly, that is, when urban populations and infrastructure expand at fast speed and at a massive 36 scale (Seto et al., 2017; Elmqvist et al., 2021). To distinguish, Section 2.6 includes the carbon footprint 37 of urban lifestyles and the difference in emissions profiles between already urbanised and less urbanised 38 areas. Chapter 8 deals with urban strategies for climate change mitigation. 39 Urban development is most significant and rapid in developing and transition countries, accompanied 40 by a substantial migration of rural populations to urban areas (Apergis and Li, 2016; Azizalrahman and 41 Hasyimi, 2019; Wang et al., 2019c) and associated impacts on land use (Richardson et al., 2015). If the 42 trend of developing countries following infrastructure stock patterns in industrialised nations continues 43 until 2050, this could cause approximately 350 GtCO2 from the production of materials (Müller et al., 44 2013). This would be equivalent to 70% of the 500 GtCO2 estimated remaining carbon budget from the 45 beginning of 2020 to limit global warming to 1.5°C with a likelihood of 50% (IPCC, 2021b). 46 In many developing countries across the world, the process of urban expansion leads to higher per capita 47 consumption-based GHG emissions (medium evidence, high agreement) (Jorgenson et al., 2014; Yao 48 et al., 2015; Zhang et al., 2016; Wood et al., 2018a; Muñoz et al., 2020). The high disparity between Do Not Cite, Quote or Distribute 2-54 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 rural and urban personal carbon footprints in these countries (Wiedenhofer et al., 2017) (see Section 2 2.6) means that migration to urban areas increases overall emissions as levels of income and expenditure 3 rise, leading to further economic growth and infrastructure development in urban areas (Müller et al., 4 2013; Li et al., 2015; Wang and Yang, 2016; Zhang et al., 2016; Wiedenhofer et al., 2017; Cetin and 5 Bakirtas, 2019; Fan et al. 2019; Li and Zhou, 2019; Xia et al., 2019; Sarkodie et al., 2020). 6 For total production-based emissions in general, urbanisation is thought to have a smaller effect than 7 changes in population, GDP per capita, and energy and emissions intensities, which are all more 8 influential (Lin et al., 2017). Another driver of urban emissions is rising ambient air temperature caused 9 by urban land expansion, which will likely drive a substantive increase in air conditioning use and cold 10 storage for food (Huang et al., 2019). Specific emission drivers, however, depend on city- and place- 11 specific circumstances such as income, household size, density, or local climate (Baiocchi et al., 2015; 12 Wang et al., 2019a). Geographical factors, urban form, and transport/fuel costs are dependent on each 13 other, and, together with economic activity, have been found to explain 37% of urban direct energy use 14 and 88% of urban transport energy use in a global sample of 274 cities (Creutzig et al., 2015). 15 16 17 2.5 Technological Change is Key to Reducing Emissions 18 Technological change for climate change mitigation involves improvement in and adoption of 19 technologies, primarily those associated with energy production and use. Technological change has had 20 a mitigating effect on emissions over the long term and is central to efforts to achieving climate goals 21 (high confidence). Progress since AR5 shows multiple low-carbon technologies are improving and 22 falling in cost (high confidence); technology adoption is reaching substantial shares, and small-scale 23 technologies are particularly promising on both (medium confidence). Faster adoption and continued 24 technological progress can play a crucial role in accelerating the energy transition. However, the 25 historical pace of technological change is still insufficient to catalyse a complete and timely transition 26 to a low-carbon energy system; technological change needs to accelerate (high confidence). This section 27 assesses the role of technological change in driving emissions reductions and the factors that drive 28 technological change, with an emphasis on the speed of transitions. Incentives and support for 29 technological change affect technology outcomes (Sivaram et al., 2018; Wilson et al., 2020a). Work 30 since AR5 has focused on evaluating the effectiveness of policies, both those that accelerate 31 technological change by enhancing knowledge – technology push – and those that increase market 32 opportunities for successful technologies – demand pull – (Nemet, 2013), as well as the importance of 33 tailoring support to country contexts (Rosenbloom et al., 2020; Barido et al., 2020), including the limits 34 of policies to date that price carbon (Lilliestam et al., 2020). Section 2.8 and Chapter 13 describe how 35 these polices affect emissions, Cross-Chapter box 12 in Chapter 16 and Chapter 14 discuss transition 36 dynamics, and Chapter 16 provides a more detailed assessment of the evolution and mitigation impacts 37 of technology development, innovation, and transfer. 38 2.5.1 Technological Change Has Reduced Emissions 39 Technological change that facilitates efficient energy utilisation from production to its final conversion 40 into end-use services is a critical driver of carbon emissions reductions (high confidence). Technological 41 change can facilitate stringent mitigation, but it also can reduce these effects by changing consumer 42 behaviour such as through rebound effects (see Section 2.6 and Chapter 16). AR6 includes an entire 43 chapter on innovation, technology development, and transfer (Chapter 16). A focus gained in this 44 section is the extent to which aligned, credible, and durable policies can accelerate technological change 45 factors to put emissions reductions on a trajectory compatible with reaching UNFCCC goals. Do Not Cite, Quote or Distribute 2-55 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 Technological change has facilitated the provision of more diverse and efficient energy services 2 (heating, cooling, lighting, and mobility) while generating fewer emissions per unit of service. As seen 3 in Section 2.4, in Kaya identity terms (Lima et al., 2016) (see Glossary): population and economic 4 growth are factors that have increased emissions, while technological change has reduced emissions 5 (Peters et al., 2017). These Kaya statistics show that while technological change can facilitate the 6 transition to a low-carbon economy, it needs to proceed at a much faster pace than historical trends 7 (Peters et al., 2017). 8 Multiple challenges exist in accelerating the past rate of technological change. First, an array of physical 9 assets in the energy system are long-lived and thus involve substantial committed carbon (see Section 10 2.7) (Knapp, 1999; Cui et al., 2019). A process of “exnovation,” accelerating the phase-out of incumbent 11 technology through intentional policy (e.g., by pricing carbon), provides a means to address long 12 lifetimes (Davidson, 2019; Rosenbloom and Rinscheid, 2020). Second, countries may not have the 13 capacity to absorb the flows of ideas and research results from international knowledge spillovers due 14 to weak infrastructure, limited research capacity, lack of credit facilities (see Chapter 15, Section 15.5), 15 and other barriers to technology transfer (Adenle et al., 2015). In a developing country context, 16 processes of innovation and diffusion need to include competence-building systems (Lema et al., 2015; 17 Perrot and Sanni, 2018; Stender et al., 2020). Third, public policy is central to stimulating technological 18 change to reduce emissions; policy depends on creating credible expectations of future market 19 opportunities (Alkemade and Suurs, 2012), but the historical evidence shows that, despite recent 20 progress, policies related to energy and climate over the long term have been inconsistent (Taylor, 2012; 21 Nemet et al., 2013; Koch et al., 2016). Bolstering the credibility and durability of policies related to 22 low-carbon technology are crucial to accelerating technological change and inducing the private sector 23 investment required (Helm et al., 2003; Habermacher et al., 2020). 24 2.5.2 A Low-Carbon Energy Transition Needs to Occur Faster Than Previous 25 Transitions 26 An illuminating debate on the possibility of faster transitions has emerged since AR5 – with diverging 27 assumptions about future technological change at the core of the discourse (Bazilian et al., 2020; Lu 28 and Nemet, 2020). Table 2.5 summarises these arguments. 29 2.5.2.1 Energy transitions can occur faster than in the past 30 Recent studies have identified examples supporting fast energy transitions (Sovacool, 2016; Bond et 31 al., 2019; Reed et al., 2019). One describes five rapid national-scale transitions in end-use technologies, 32 including lighting in Sweden, cook-stoves in China, liquefied petroleum gas stoves in Indonesia, ethanol 33 vehicles in Brazil, and air conditioning in the USA (Sovacool, 2016). Adoption of electric vehicles in 34 Norway and in cities in China have also been rapid (Rietmann and Lieven, 2019; Li et al., 2020; 35 Fridstrøm, 2021). Examples in energy supply, include electrification in Kuwait, natural gas in the 36 Netherlands, nuclear electricity in France and Sweden, combined heat and power in Denmark, 37 renewable energy in Uruguay, and coal retirements in Ontario, Canada (Qvist and Brook, 2015). 38 Reasons that these exemplars could be applied more broadly in the future include: growing urgency on 39 climate change, shifting motivation from price response to proactive resource scarcity, and an increase 40 in the likelihood of technological breakthroughs (medium confidence) (Sovacool, 2016; Bazilian et al., 41 2020). The emergence of smaller unit scale, “granular” technologies described below also creates the 42 potential for faster system change (Trancik, 2006; Grubler et al., 2018; Wilson et al., 2020a). Prices of 43 energy services and government actions that affect demand are critical to the speed and extent of energy 44 transitions (Kramer and Haigh, 2009). Reasons scholars consider for expecting a fast transition include: 45 intentional policy and alignment with goals; globalisation which diversifies sources and integrates 46 supply chains; collective action via the Paris Agreement; as well as bottom-up grassroots movements 47 and private sector initiatives (Kern and Rogge, 2016). Political support for change can also speed Do Not Cite, Quote or Distribute 2-56 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 transitions (Burke and Stephens, 2017; Stokes and Breetz, 2018), as can the credibility of transition- 2 related targets (Li and Pye, 2018; Rogge and Dütschke, 2018). 3 The important role of leader countries is often missed when looking only at global aggregates (Meckling 4 and Hughes, 2018); leaders accumulate important knowledge, provide scaled market, and set positive 5 examples for followers (medium confidence) (Schwerhoff, 2016; Buchholz et al., 2019). In recent years, 6 the conception of where leadership, climate-relevant innovation, and technology transfer originate has 7 shifted to considering more meaningfully direct South-South and South-North forms of technology 8 transfer, flows of capital, drivers for market access, origins of innovation, and other forms of 9 cooperation (Urban, 2018; Köhler et al., 2019). Recent evidence shows South-South trade is enabling 10 clean technology transfer (Gosens, 2020). Leaders can initiate a process of “catalytic cooperation” in 11 which they overcome collective action problems and stimulate rapid change (Hale, 2018). Similarly, 12 “sensitive intervention points” – targeted support of social movements, technologies, or policies 13 themselves – can lead to rapid and self-sustaining change (Farmer et al., 2019), such as support for 14 photovoltaics in Germany in the 2000s and student climate activism in Europe in 2019. The focus on 15 leadership, catalysts, and intervention points reflects a systemic view of transitions that emphasises 16 interactions and interdependence (Geels, 2018; Meckling and Hughes, 2018). Technological change 17 has been at the core of transitions, but is best understood as part of a system in which social aspects are 18 crucial (medium confidence) (Cherp et al., 2018; Köhler et al., 2019; Overland and Sovacool, 2020). 19 20 Table 2.5 Summary reasons to expect a fast energy transition and reasons to expect a slow transition. Fast transition Slow transition Evidentiary basis Technology and country cases over 50 Historical global system over 200 years years Systems Complementary technologies enable Difficult integration with existing integration infrastructure Economics Falling costs of nascent technology Mature incumbent technologies Up-front costs and capital constraints Technology Digitalisation and global supply chains Long lifetimes of capital stock More abundant innovation Difficult to decarbonise sectors Granular technology Actors Proactive efforts for transition Risk-averse adopters Bottom-up public concern Attributes do not appeal to consumers Mobilised low-carbon interest groups Rent-seeking by powerful incumbents Governance Leaders catalyse faster change Collective action problems 21 22 2.5.2.2 Reasons that transitions will occur at historical rates of change 23 Recent work has also reasserted previous claims that the speed of a low-carbon transition will follow 24 historical patterns (low confidence). Broad transitions involve technological complexity, time- 25 consuming technological development, risk-averse adopters, high up-front costs, and low immediate 26 individual adoption benefits, attributes which are not all present in the examples of rapid change 27 described above (Grubler et al., 2016). Additional factors that slow transitions include: the need for the 28 transition to occur globally, thus requiring nations with unequal economic resources and development 29 circumstances to engage in near-universal participation; slow progress in recent decades; intermittence 30 of renewables, and the time involved in building supporting infrastructure (Smil, 2016); difficulty in Do Not Cite, Quote or Distribute 2-57 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 decarbonising transportation and industry (Rissman et al., 2020); and material resource constraints 2 (Davidsson et al., 2014). 3 2.5.3 Improvements in Technologies Enable Faster Adoption 4 Since AR5, multiple low-carbon technologies have shown dramatic improvement, particularly solar 5 PV, wind, and batteries (high confidence). The observed pace of these changes and the likelihood of 6 their continuation support the arguments in the previous section that future energy transitions are likely 7 to occur more quickly than in the past (medium confidence). 8 2.5.3.1 Technological change has produced dramatic cost reductions 9 A wide array of technologies shows long-term improvements in performance, efficiency, and cost. 10 Among the most notable are solar photovoltaics, wind power, and batteries (high confidence) (see 11 Chapters 6 and 16). PV’s dynamics are the most impressive, having fallen in cost by a factor of 10,000 12 from the first commercial application on a satellite in 1958 (Maycock and Wakefield, 1975) to power 13 purchase agreements signed in 2019 (IRENA, 2020). Wind has been on a nearly as steep trajectory 14 (Wiser and Bolinger, 2019) as are lithium-ion battery packs for electric vehicles (Nykvist and Nilsson, 15 2015; Service, 2019). The future potential for PV and batteries seems especially promising given that 16 neither industry has yet begun to adopt alternative materials with attractive properties as the cost 17 reductions and performance improvements associated with the current generation of each technology 18 continue (medium confidence) (Kwade et al., 2018). A key challenge is improving access to finance, 19 especially in developing country contexts, where the costs of financing are of crucial importance 20 (Creutzig et al., 2017; Schmidt, 2019). 21 2.5.3.2 Technological change has accelerated since AR5 22 Figure 2.22 shows changes in the costs of four dynamic energy technologies. One can see rapid changes 23 since AR5, cost data for which ended in 2010. Solar PV is by far the most dynamic technology, and its 24 cost since AR5 has continued on its steep decline at about the same rate of change as before AR5, but 25 now costs are well within the range of fossil fuels (high confidence) (see Chapter 6). Very few 26 concentrating solar power (CSP) plants had been built between the 1980s and 2012. Since AR5, 4 GW 27 have been built and costs have fallen by half. Onshore wind has continued its pace of costs reductions 28 such that it is well within the range of fossil fuels. Offshore wind has changed the most since AR5. 29 Whereas costs were increasing before AR5, they have decreased by 50% since. None of these 30 technologies shows indications of reaching a limit in their cost reductions. Crucial to their impact will 31 be extending these gains in the electricity and transportation sectors to the industrial sector (Davis et 32 al., 2018). 33 34 Figure 2.22: Learning curves for renewable energy technologies 2000–2019 Do Not Cite, Quote or Distribute 2-58 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 Range of fossil fuel levelised cost of electricity indicated as horizontal dashed lines spanning the range of 2 USD50–177 MWh-1. Dashed lines are power functions fit to data for AR4–AR5 in black and for post-AR5 3 (2012) in blue. Blue areas show ranges between the 10th and 90th percentile in each year. 4 Source: Data from Nemet (2019), IRENA (2020). 5 6 2.5.3.3 Granular technologies improve faster 7 The array of evidence of technology learning that has accumulated both before and since AR5 8 (Thomassen et al., 2020) has prompted investigations about the factors that enable rapid technology 9 learning. From the wide variety of factors considered, unit size has generated the strongest and most 10 robust results. Smaller unit sizes, sometimes referred to as ‘granularity’, tend to be associated with faster 11 learning rates (medium confidence) (Sweerts et al., 2020; Wilson et al., 2020). Examples include solar 12 PV, batteries, heat pumps, and to some extent wind power. The explanatory mechanisms for these 13 observations are manifold and well established: more iterations are available with which to make 14 improvements (Trancik, 2006); mass production can be more powerful than economies of scale 15 (Dahlgren et al., 2013); project management is simpler and less risky (Wilson et al., 2020); the ease of 16 early retirement can enable risk-taking for innovative designs (Sweerts et al., 2020); and they tend to 17 be less complicated (Malhotra and Schmidt, 2020; Wilson et al., 2020). Small technologies often 18 involve iterative production processes with many opportunities for learning by doing and have much of 19 the most advanced technology in the production equipment than in the product itself. In contrast, large 20 unit scale technologies – such as full-scale nuclear power, CCS, low-carbon steel making, and negative 21 emissions technologies such as bioenergy with carbon capture and sequestration (BECCS) – are often 22 primarily built on site and include thousands to millions of parts such that complexity and system 23 integration issues are paramount (Nemet, 2019). Despite the accumulating evidence of the benefits of 24 granularity, these studies are careful to acknowledge the role of other factors in explaining learning. In 25 a study of 41 energy technologies (Figure 2.23), unit size explained 22% of the variation in learning 26 rates (Sweerts et al., 2020) and a study of 31 low-carbon technologies showed unit size explained 33% 27 (Wilson et al., 2020). Attributing that amount of variation to a single factor is rare in studies of 28 technological change. The large residual has motivated studies, which find that small-scale technologies 29 provide opportunities for rapid change, but they do not make rapid change inevitable; a supportive 30 context, including supportive policy and complementary technologies, can stimulate more favourable 31 technology outcomes (high confidence). Do Not Cite, Quote or Distribute 2-59 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 Figure 2.23 Learning rates for 41 energy demand, supply, and storage technologies 3 Source: Sweerts et al. (2020). 4 There is also evidence that small technologies not only learn but get adopted faster than large 5 technologies (medium confidence) (Wilson et al., 2020b). Some of the mechanisms related to the 6 adoption rate difference are related to those for cost reductions; for example, smaller, less lumpy 7 investments involve lower risk for adopters (Dahlgren et al., 2013; Wilson et al., 2020b). The shorter 8 lifetimes of small technologies allow users to take advantage of new performance improvements 9 (Knapp, 1999) and access a large set of small adopters (Finger et al., 2019). Other mechanisms for faster 10 adoption are distinctly related to markets: modular technologies can address a wide variety of niche 11 markets (Geels, 2018) with different willingness to pay (Nemet, 2019) and strategically find protected 12 niches while technology is maturing (Coles et al., 2018). 13 2.5.4 Rapid Adoption Accelerates Energy Transitions 14 The transition to a more sustainable energy system depends not just on improvement in technologies, 15 but also on their widespread adoption. Work since AR5 has also substantiated the bidirectional causal 16 link between technology improvement and adoption. Cost reductions facilitate adoption, which 17 generates opportunities for further cost reductions through a process of learning by doing (medium 18 confidence). The rate of adoption is thus closely related to the speed at which an energy transition is 19 possible. 20 Results of integrated assessment models (IAMs) show that scale-up needs are massive for 2°C 21 scenarios. Using logistic growth rates of energy shares as in previous work (Wilson, 2012; Cherp et al., 22 2021), most of these technologies include annual adoption growth rates of 20% in the 2020s and 2030s, 23 and are in line with recent adoption of wind and solar. However, it is important to realise that IAMs 24 include faster adoption rates for some mitigation technologies than for others (Peters et al., 2017). 25 Growth rates in IAMs for large-scale CCS – biomass, coal, and gas – are between 15 and 30% (25th and 26 75th percentiles) (Figure 2.24). So few plants have been built that there is little historical data to which 27 to compare this expected growth; with only two full scale CCS power plants built and a 7% growth rate 28 if including industrial CCS. In contrast, IAMs indicate that they expect much lower rates of growth in Do Not Cite, Quote or Distribute 2-60 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 future years for the set of technologies that has been growing fastest in recent years (wind and solar), 2 without strong evidence for why this should occur. 3 The overall pattern shows that IAMs expect growth in small-scale renewables to fall to less than half 4 of their recent pace and large-scale CCS to more than double from the limited deployment assessed 5 (high confidence). The emerging work since AR5 showing the rapid adoption and faster learning in 6 small-scale technologies should prompt a keener focus on what technologies the world can depend on 7 to scale up quickly (Grubb et al., 2021). In any case, the scenario results make quite clear that climate 8 stabilisation depends on rapid adoption of low-carbon technologies throughout the 2020–2040 period. 9 10 Figure 2.24 Growth of key technologies (2020–2040) in Paris-consistent mitigation scenarios compared to 11 historical growth 12 Comparisons of historical growth (grey bars) to growth in 2020–2040 mitigation scenarios (dots). Values 13 on vertical axis are logistic annual growth rates for share of each technology in electricity supply. 14 Horizontal arrangement of dots within technology categories indicates count of scenarios at each growth 15 rate. 16 Source: Data on scenarios from Chapter 3, historical data from BP (2021). 17 18 2.6 Behavioural Choices and Lifestyles 19 2.6.1 Introduction 20 This section synthesises how behavioural choices, lifestyles, and consumption preferences affect energy 21 use and emissions. Household consumption is the largest component of a country’s gross domestic 22 product (GDP) and the main contributor to greenhouse gas emissions through direct energy 23 consumption for heating and cooling or private transportation and indirectly through carbon emitted 24 during production of final consumption items. There is great variation in individual, groups and 25 household behavior and consumption patterns within and between countries and over time. A number 26 of factors affect people’s consumption patterns and associated carbon emissions, such as socio- 27 demographics, socio-economic status, infrastructure and access to public services; the regulatory frame; 28 availability, affordability and accessibility of more or less sustainable choices on markets; individual 29 values and preferences (Dietz et al., 2009). Do Not Cite, Quote or Distribute 2-61 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 Do Not Cite, Quote or Distribute 2-62 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 Figure 2.25 Carbon footprints per capita income and expenditure category for 109 countries ranked by 2 per capita income (consumption-based emissions) 3 Notes: Countries and income categories are dependent on data availability. Blue dots represent income 4 quintiles (lowest, low, middle, higher, and highest) of EU countries and the United States. Orange dots are 5 for the developing country group provided by the World Bank for 4 expenditure categories: lowest, low, 6 middle and higher (Hubacek et al., 2017b). Red diamonds represent average per capita carbon footprints. 7 Countries are ranked from the lowest per capita income (bottom) to the highest income (top) within each 8 country group. Countries are grouped using the IPCC's six categories high-level classification. Footprint 9 values for higher income groups in the World Bank data are less reliable. 10 11 Carbon footprints vary between and within countries and show an uneven distribution because of 12 differences in development levels, economic structure, economic cycle, available public infrastructure, 13 climate and residential lifestyles (Bruckner et al., 2021). Similar emission characteristics can also be 14 found within a country, see, for example for China (Feng et al., 2013), for the US (Pizer et al., 2010; 15 Wang et al., 2018; Miehe et al., 2016; Feng et al., 2013; Hubacek et al., 2017b) for Brazil (Sanches- 16 Pereira et al., 2016), for Latin American countries (Zhong et al., 2020). 17 In western countries, the largest contribution to the household carbon footprint is from transportation, 18 housing, and consumption of food (Druckman and Jackson, 2015). These three items’ joint contribution 19 varies in different countries depending on consumption patterns and account for 58.5%, on average, in 20 EU 25 countries. (Tukker and Jansen, 2006). However, different countries and even regions within 21 countries may have different emission patterns due to differences in income, lifestyle, geography, 22 infrastructure, political and economic situation. For example, the main contributors to the average US 23 household is private transport (19.6%), followed by electricity (14.8%) and meat (5.2%) (Jones and 24 Kammen, 2011), while the UK households have 24.6% emissions on energy and housing, 13.7% 25 emissions on food, and 12.2% emissions on consumables (Gough et al., 2011). A study of 49 Japanese 26 cities found that energy (31%), food (27%), and accommodation (15%) were the largest sources of 27 household emissions (Long et al., 2017). An overview investigation of Japan’s household emissions 28 found that energy, food, and utility are the three main emissions sources, but their shares are dependent 29 on age (Shigetomi et al., 2014). See section 12.4 in chapter 12 and Box 5.4 in chapter 5 for more in- 30 depth discussion on food systems and dietary shifts towards lower emission food. 31 In terms of rapidly growing economies, China is the most extensively researched country. China’s 32 household emissions were primarily derived from electricity and coal consumption, as well as residents’ 33 consumption of emission-intensive products, such as housing (33.4%), food (23.6%), private 34 transportation and communications (14.8%) (Wang et al., 2018). Space heating was the largest 35 contributor among various daily energy uses in northern cities (Yang and Liu, 2017). In comparison, 36 Indonesian rural households have a larger emission share on foods and a much smaller share on services 37 and recreation than urban households (Irfany and Klasen, 2017). Urban Indonesian households have a 38 much larger share of transport related emissions (Irfany and Klasen, 2017). Analysis from the 39 Philippines shows that on average households in urban areas emit twice as much as rural ones because 40 of much lower direct energy use in homes and for transport in rural areas (Serino, 2017). In other 41 emerging economies, such as India, Brazil, Turkey and South Africa, a high share of transport related 42 carbon emissions among urban middle- and high-income households is evident (Huang and Tian, 2021). 43 44 2.6.2 Factors affecting household consumption patterns and behavioural choices 45 Households’ carbon emissions are closely linked to activities and consumption patterns of individuals 46 and as a group in households. Individual and group behaviour, in turn, is shaped by economic, 47 technological, and psychological factors, social contexts (such as family ties, friends and peer-pressure) 48 and cultural contexts (social identity, status, and norms) as well as the natural environment (number of Do Not Cite, Quote or Distribute 2-63 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 heating and cooling days) and physical infrastructure, or geography (Jorgenson et al., 2019). For 2 example, a city with an excellent bicycle infrastructure will make it safer and easier for citizens to 3 become highly mobile by using their bikes; a city that has less density and is dominated by automobile 4 infrastructure induces more people to travel by car (see Chapter 8 and 10). As a consequence, many 5 climate relevant consumption acts are not consciously decided upon or deliberately made part of a 6 lifestyle but are strongly influenced by the factors listed above. Chapter 5 provides more in-depth 7 discussion on behavioural drivers and examples of behavioural interventions and policies that can be 8 used to reduce emissions. 9 Demographic characteristics such as age, sex, and education constitute an important set of determinants 10 influencing emissions patterns. People of different genders have different consumption patterns. For 11 example, men tend to consume more food (especially meat) than women, leading to higher food-related 12 emissions. Also, men spend more money on vehicles and driving (Wang et al., 2018). Similar evidence 13 has been found in Germany, Greece, Norway, and Sweden, where men’s energy use is 8%, 39%, 6%, 14 and 22% higher than women’s, respectively (Räty and Carlsson-Kanyama, 2010). 15 Income. Due to the differences that shape individuals’ consumption patterns there are enormous 16 differences in carbon footprints associated with income being one of the most important predictors. 17 Globally, households with income in the top 10% (income higher than USD23.03 PPP per capita per 18 day) are responsible for 36% to 45% of GHG emissions, while those in the bottom 50% (income less 19 than USD2.97 PPP per capita per day) are responsible for only 13-15% of emissions depending on the 20 study (Chancel and Piketty, 2015; Semieniuk and Yakovenko, 2020; Hubacek et al., 2017b) (Figure 21 2.25). The average carbon footprint of the high household incomes is more than an order of magnitudes 22 larger than that of the lowest expenditure group (Feng et al. 2021). For example, Zhang et al. (2016) 23 analysed the impact of household consumption across different income households on the whole CO 2 24 emissions in China and concluded that the impact on CO2 emissions generated by urban households’ 25 consumption are 1.8 times as much as that of rural ones. High-income households have higher emission 26 related to transport and entertainment, such as recreational expenditure, travel, and eating out, than low- 27 income households. Low-income households tend to have a larger share on necessities such as fuel for 28 heating and cooking (Kerkhof et al., 2009). Figure 2.25, shown above, depicts the carbon footprint per 29 capita ranked by per capita income. 30 Age. The effect of population ageing on emissions is contested in literature. Ageing when accompanied 31 by shrinking household size and more energy-intensive consumption and activity patterns results in 32 increased emissions. However, an ageing labour force can also dampen economic growth and result in 33 less of energy-intensive activity like driving, which decreases emissions (Liddle, 2011; Liddle and 34 Lung, 2010). An ageing of the population characterises the demographic transition in both developed 35 and developing countries. The implications of ageing for emissions depend on labour force 36 participation of the elderly and difference in the consumption and investment patterns of different age 37 groups (O’Neill et al., 2012). Analysis using panel macro data from OECD countries suggests that shifts 38 in age and cohort composition have contributed to rising GHG emissions since the 1960s (Menz and 39 Welsch, 2012; Nassen, 2014). Household-level data over time for the US provides evidence that 40 residential energy consumption increases over the lifetime of household members, largely also due to 41 accompanying changes in household size (Estiri and Zagheni, 2019). Similar insights emerge from 42 Japan, where analysis shows that those in their 70s or older, a group that is growing in size in Japan, 43 have higher emissions than other age groups (Shigetomi et al., 2014, 2019, 2018). Recent analysis from 44 China suggests that the shift to smaller and ageing households is resulting in higher carbon emissions 45 because of the accompanying time-use and consumption shifts (Yu et al., 2018; Li and Zhou, 2019). 46 An increase in the dependency ratio, i.e. the proportion of children under 15 and people over 65 relative 47 to the working-age population, in other analysis, has been shown to lead to reduced CO2 emissions in 48 China (Wei et al., 2018; Li and Zhou, 2019). Implications of the nature of this relationship are important Do Not Cite, Quote or Distribute 2-64 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 to policy discussions of working hours and retirement age that are likely to have an influence on 2 emissions. For example, children and youth tend to emit more education related emissions than adults 3 (Han et al., 2015). Older people tend to have higher emissions related to heating and cooling being more 4 sensitive to temperature (Meier and Rehdanz, 2010). 5 Household size. Per capita emissions tend to decrease with family size as living together becomes more 6 energy efficient (Qu et al., 2013). The household size in most countries is decreasing (Liu et al., 2011), 7 but the decrease rate differs across countries and show, for example, higher decrease rate in China than 8 Canada and UK (Maraseni et al., 2015). The evidence shows that shifts to smaller households are 9 associated with larger per-capita footprints (Liddle and Lung, 2014; Underwood and Zahran, 2015; 10 Wiedenhofer et al., 2018; Ivanova et al., 2017), at least in developed countries (Meangbua et al., 2019). 11 Urban Living. The carbon footprint of individuals and households is also significantly influenced by 12 urban-rural differences (Ivanova et al., 2018; Wiedenhofer et al., 2018). In some part, the difference 13 can be explained by the effect of locational and spatial configuration characteristics such as levels of 14 compactness/density, centrality, proximity and ease of access to services. In all these parameters, urban 15 areas score higher as compared with rural or peri-urban (outlying and suburban) areas, thus influencing 16 household emissions in different ways. Urban households tend to have higher emissions than rural 17 households (O’Neill et al., 2010; Liu et al., 2011), but with different energy and consumption structure. 18 For example, rural households have more diverse energy inputs, such as biomass, biogas, solar, wind, 19 small hydro and geothermal in addition to coal (Maraseni et al., 2016). 20 In terms of indirect emissions, urban households have more service related emissions, such as from 21 education and entertainment than rural households, while rural households tend to have higher 22 emissions related to food consumption or transportation (Büchs and Schnepf, 2013; Maraseni et al., 23 2016) but this is strongly dependent on the specific situation of the respective country as in poorer 24 regions rural transport might be mainly based on public transport with lower carbon emissions per 25 capita. Centrality and location also place a role on the level of urban household emissions. Studies on 26 US households found that residents in the urban core have 20% lower household emissions than 27 residents in outlying suburbs, which show a large range of household emissions (from -50% to +60%) 28 (Kahn, 2000; Jones and Kammen, 2014). From a global average perspective, higher population density 29 is associated with lower per capita emissions (Liddle and Lung, 2014; Liu et al., 2017). 30 Location choices are a significant contributor to household emissions. Suburbanites generally purchase 31 large, spacious homes with larger heating and cooling requirements. Commuting distance and access to 32 public transportation, recreation areas, city centres, public services, and shops are other important 33 neighbourhood-specific determinants of carbon emissions (Baiocchi et al., 2010) (see more on this in 34 urban and the transport chapters 8 and 10). 35 Time Use. A study on the emissions implications of time use (Wiedenhofer et al., 2018) found that the 36 most carbon intensive activities are personal care, eating and drinking and commuting. Indirect 37 emissions are also high for repairs and gardening. In contrast, home-based activities such as sleep and 38 resting, cleaning and socializing at home have low carbon intensities per hour of time use. The same 39 study also found that households in cities and with higher incomes tend to substitute personal activities 40 for contracted services, thus shifting away from households to the service sector (Wiedenhofer et al., 41 2018). Improvements in the efficiency of time or resource use are diminished by rebound effects which 42 have been shown to reduce emissions savings by 20-40% on average (Gillingham et al., 2015), while 43 other authors argue that potentially the size of the rebound effect could be larger (Saunders, 2015) (see 44 more coverage of the rebound effect in Chapters 9 and 16). Lifestyle shifts brought about by using 45 information technologies and socio-technological changes are inducing alterations in people’s daily 46 activities and time-use patterns. Do Not Cite, Quote or Distribute 2-65 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 The reduction of working hours is increasingly discussed as an approach to improve well-being and 2 reduce emissions (Wiedenhofer et al., 2018; Fitzgerald et al., 2015, 2018; Melo et al., 2018; Smetschka 3 et al., 2019). For instance, analysis of differences in working hours across US states for the period 2007- 4 2013 shows that there is a strong positive relationship between carbon emissions and working hours, 5 which holds even after controlling for other differences in political, demographic and economic drivers 6 of emissions (Fitzgerald et al., 2018). In other analysis, this relationship is seen to hold in both 7 developed and developing countries (Fitzgerald et al., 2015). One recent study, however, finds evidence 8 of nonlinear relationships between working time and environmental pressure in EU-15 countries 9 between 1970 and 2010, in cases where non-work time is spent instead in carbon-intensive leisure 10 activities (Shao and Shen, 2017). 11 Social Norms. Evidence from experiments in the US shows that social norms can not only help in 12 reducing a household’s absolute level of electricity use but also shift the time of use to periods when 13 more renewable electricity is in the system (Horne and Kennedy, 2017). Analysis from Sweden shows 14 that adoption of sustainable innovations like solar panels is influenced by perceived behaviour and 15 expectations of others (Palm, 2017). Similar conclusions emerge from analysis in the Netherlands on 16 the adoption of electric vehicles and smart energy systems (Noppers et al., 2019). 17 Broader contextual factors and cultural trends towards consumerism, individualization and defining 18 self-worth through conspicuous consumption can drive emissions up (Chancel and Piketty, 2015). 19 However, cohort and generational shifts can drive emissions down. For instance, evidence, from 20 millennials in the OECD shows that fewer younger people have driving licenses compared to older 21 generations (Kuhnimhof et al., 2012). Similar, findings are evident from analysis for the US, where 22 changing attitudes, decreased employment and rising virtual mobility explain decreased travel by 23 Millennials (McDonald, 2015). Analysis for France shows that baby boomers are higher emitters than 24 other generations (Chancel, 2014). A change in social norms is taking place with the spread of the 25 sharing economy by which consumers share or borrow goods from other consumers. Sharing 26 opportunities are more advanced within the mobility sector (Greenblatt and Shaheen, 2015). Successful 27 car and bike sharing have rapidly expanded in countries such as China, Indonesia, Mexico, Brazil and 28 Turkey. Technology and data advances are currently barriers to spreading of sharing in low- and lower 29 middle-income cities but the potential offered by these technologies to allow poor countries to leapfrog 30 to more integrated, efficient, multimodal transport systems is important (Yanocha et al., 2020). Despite 31 this potential it is unclear how much shared mobility contributes to transport decarbonization or to make 32 it worse as it takes away riders from public transit (ITF, 2019). The evidence so far shows that the 33 potential positive impacts of shared mobility with pooled rides in lowering travel costs, abating 34 congestion, and reduced GHG emissions have not materialized to date (Merlin, 2019) (See Chapter 5). 35 Education & Environmental Knowledge. A positive relationship was found between general and 36 carbon-specific knowledge and the attitude towards carbon-specific behaviours in US consumers 37 (Polonsky et al., 2012). One example, pertaining to students, found that the gain of environmental 38 knowledge resulted in more environmentally favourable attitude among these high school students 39 (Bradley et al., 1999). A comparison across states in the USA, for example, shows that environmental 40 awareness can be a mitigating factor of territorial GHG emissions (Dietz et al., 2015). A 1% increase 41 in 'environmentalism' – defined as the "environmental voting record of the state's Congressional 42 delegation" (Dietz et al., 2015) – leads to a 0.45% decrease in emissions. 43 Environmental knowledge is not always directly translating into decreased ecological footprint 44 (Csutora, 2012). While pro-environmental action is lagging behind, research shows that this is not 45 caused by people undervaluing the environment but rather by people structurally underestimating how 46 much others care (Bouman and Steg, 2019). Other evidence shows that there are multiple causal 47 pathways through which a more educated population can effect emissions, some of which may be 48 positive and others negative (Lutz et al., 2019). A more educated population is more productive and can Do Not Cite, Quote or Distribute 2-66 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 drive higher economic growth and therefore emissions (Lenzen and Cummins, 2013). Moreover, 2 education that is designed to specifically inform decision makers of the impacts of their decisions and 3 provide behavioural nudges can be a way to reduce emissions (Duarte et al., 2016). 4 Status Competition. As part of a larger consumer society and consumer culture, based on consumer- 5 oriented lifestyles, products frequently provide a source for identity and fulfilment (Stearns, 2001; 6 Baudrillard, 2017; Jorgenson et al., 2019). People pursue cultural constructs such as status, comfort, 7 convenience, hygiene, nutrition, and necessity. Consumption is, by and large, not an end in itself but a 8 means to achieve some other end, and those ends are diverse and not necessarily connected to one 9 another (Wilk, 2010). This shows that consumption patterns cannot be sufficiently understood without 10 also considering the context, for example the cultural and social contexts leading to status competition 11 and status-related consumption (Veblen, 2009; Schor and J.B., 2015; Wilk, 2017). Status seeking can 12 work to reduce emissions when ‘green products’ such as an electric car or photovoltaics on the roof 13 become a sign for high-status (Griskevicus Tybur, and Van Den Bergh, 2010). It also can work to 14 increase emissions through visible and high-carbon intensive consumption items such a larger homes, 15 fuel-inefficient SUVs cars, and long-distance vacations (Schor, 1998), driven by a notion of having ‘to 16 keep up with the Joneses’(Hamilton, 2011). This can lead to formation of new habits and needs, where 17 products and services become normalized and are quickly perceived as needed, reinforced through 18 social networks and advertisement, making it psychologically easy to convert a luxury item to a 19 perceived necessity (Assadour, 2012). For example, the share of adults who consider a microwave a 20 necessity was about one third in 1996 but had increased to more than two thirds in 2006, but retreated 21 in importance during the recession years 2008-2009 (Morin and Taylor, 2009). Similar ups and downs 22 have been observed for television sets, air conditioning, dishwasher or the clothes dryer. (Druckman 23 and Jackson, 2009). What is considered a basic need and what is a luxury is subject to change over 24 one’s lifetime and in relation to others (Horowitz, 1988). This shows that the boundaries of public’s 25 luxury-versus-necessity perceptions are malleable (Morin and Taylor, 2009). 26 Inequality. Global inequality within and between countries has shifted over the last decades expanding 27 consumption and consumer culture (Castilhos and Fonseca, 2016; Alvaredo et al., 2018; Short and 28 Martínez, 2020). The rise of middle class income countries, mostly in Asia, eg. China, India, Indonesia 29 and Vietnam, and the stagnating incomes of the middle classes in developed economies reduced 30 between countries income differences; meanwhile the population under extreme poverty (threshold of 31 1.9 USD per person/day) is now concentrated in Sub-Saharan Africa and South Asia (Milanović, 2016). 32 A major pulling apart between top and bottom incomes occurred in parallel within countries. Since 33 1980, the top 1% richest individuals in the world captured twice as much growth as the bottom 50% 34 individuals (Friedman and Savage, 2017; Alvaredo et al., 2018). The influence of these dual inequality 35 trends on lifestyles, new consumption patterns and carbon emissions at regional, local and global scale 36 are large and have led to the fastest growth of global carbon emissions, in particular, for fast emerging 37 economies (see section 2.2. and 2.3). Emissions remain highly concentrated, with the top 10% per capita 38 emitters contributing to between 35-45% of global emissions, while bottom 50% emitters contribute to 39 13-15% of global emissions (Hubacek et al., 2017a). Furthermore, the top 1% of income earners by 40 some estimates could have an average carbon footprint 175 times that of an average person in the bottom 41 10% (Otto et al., 2020). The top 10% high emitters live in all continents, and one third of them live in 42 emerging countries (Chancel and Piketty, 2015; Semieniuk and Yakovenko, 2020; Hubacek et al., 43 2017a). Mitigation pathways need to consider how to minimize the impacts of inequality on climate 44 change and the different mechanisms and effects coming into play between inequality of income and 45 emissions (see 2.4.3) (Baek and Gweisah, 2013; Berthe and Elie, 2015; Hao et al., 2016; Grunewald et 46 al., 2017). 47 Inequality trends catalyses impact at a demand level, mobilizing rapid lifestyles changes, symbolic 48 consumption and ideals of material improvements and upward mobility (Castilhos et al., 2017) and Do Not Cite, Quote or Distribute 2-67 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 emulation of high-carbon emissions intensive lifestyle of the wealthy (Gough, 2017). Decoupling 2 energy use and emissions from income growth and, the decarbonisation of energy services have not 3 counteracted these trends (see 2.4.1). Alternative options to deal with carbon inequality like sharing 4 global carbon emissions among high emitters (Chakravarty et al., 2009; Chakravarty and Tavoni, 2013) 5 or addressing the discourse of income distribution and the carbon intensity of high emitters lifestyles 6 (Gössling, 2019; Otto et al., 2019; Hubacek et al., 2017b). are met with caution that such alternatives 7 may necessitate difficult and hard to implement institutional changes (Semieniuk and Yakovenko, 8 2020). Growing inequality within countries may make re-composition of emission intensive 9 consumption more difficult and, it may also exacerbate redistribution and social cohesion dilemmas 10 (Gough, 2017; Römpke et al., 2019). Climate mitigation action has different motivational departures in 11 unequal context. An emerging global `middle class' strengthens consumption at the margin as evidence 12 by first-time purchases of white goods with likely impacts on energy demand (Wolfram et al., 2012), 13 and with a warming climate, the increased use of air conditioning (Davis and Gertler, 2015). Inequality 14 may affect the willingness of rich and poor to pay for environmental goods or accept policies to protect 15 the environment (Baumgärtner et al., 2017). Unequal departure for action is strongly manifested in cities 16 of all sizes in developing countries with low-income urban residents hardest hit in lock-in situations 17 such as lack of access to transportation and jobs (Altshuler, 2013; Mattioli, 2017), lack of green spaces 18 (Joassart-Marcelli et al., 2011), poor access to waste collection (King and Gutberlet, 2013) and to energy 19 and clean water provision. The exacerbation of these conditions constraint the feasibility for achieving 20 emissions reductions through lifestyle or behavioural changes alone (Oxfam, 2015; Baiocchi et al., 21 2010). High inequality limits mitigation efforts, and conversely advancing mitigation should not 22 contribute to deepen existing inequalities (Rao and Min, 2018; Saheb et al., 2019). It is critically 23 important to account for varying demands and affordability across heterogeneous household groups in 24 access to quality energy, education, health, decent jobs and services, while recomposing consumption 25 and balancing societal trade-offs via policies to boost the inclusion of low income and energy poor 26 population groups (Pachauri et al., 2013). Further, there is a need to reduce inequalities and improve 27 the capabilities people have to live the lives they value (Sen, 1999; Gough et al., 2011; Gough, 2017; 28 Aranoff et al., 2019) 29 30 2.7 Emissions associated with existing and planned long-lived 31 infrastructure 32 2.7.1 Introduction: clarification of concepts 33 Carbon lock-in can be understood as inertia in a system that limits the rate of transformation by a path- 34 dependent process (Seto et al., 2016). For example, long lifetimes of infrastructures such as power 35 plants, roads, buildings or industrial plants may influence the rate of transformation substantially and 36 lock societies into carbon-intensive lifestyles and practices for many decades (Unruh, 2000, 2002; 37 Unruh and Carrillo-Hermosilla, 2006; Grubler, 2012; Seto et al., 2016; Sovacool, 2016). Infrastructure 38 stock evolution depends not only on technological and economic factors, but also on institutional and 39 behavioural ones that are often mutually reinforcing. That is, physical infrastructure such as the built 40 environment of urban areas can shape behaviour and practices of daily life, which in turn change the 41 demand for such infrastructure and lock-in energy demand patterns (Creutzig et al., 2016; Makido et 42 al., 2012; Banister et al., 1997; Shove and Trentmann, 2018; Seto et al., 2016). 43 There is a broad literature on carbon lock-in related to infrastructure that has analysed different 44 geographical scales and sectors, with a strong focus on the power sector (Fisch-Romito et al., 2020). 45 Available quantifications differ in the time frames of analysis that can be classified as backward- 46 looking, static for a given year, or forward-looking using scenarios (Fisch-Romito et al., 2020). 47 Quantifications also differ in the indicators used to describe carbon-lock in. Literature has assessed how Do Not Cite, Quote or Distribute 2-68 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 delays in climate policy affect the evolution of fossil-fuel infrastructure stock in the short term (Bertram 2 et al., 2015; McGlade et al., 2018; Kefford et al., 2018), overall mitigation costs (Luderer et al., 2016; 3 Riahi et al., 2015), or the transition risks from premature retirements or underutilisation of existing 4 assets (Iyer et al., 2015; van Soest et al., 2017; Lane et al., 2016; Farfan and Breyer, 2017; Cui et al., 5 2019; Malik et al., 2020; Wang et al., 2020a; Johnson et al., 2015; Luderer et al., 2016; Kefford et al., 6 2018; Fofrich et al., 2020; Pradhan et al., 2021). Only a few authors have relied on indicators related to 7 institutional factors such as technology scale or employment (Erickson et al., 2015; Spencer et al., 8 2018). A complementary literature has explored how the sheer size of the world’s fossil fuel reserves 9 (and resources) and financial interest of owners of these could contribute to supply-side dynamics that 10 sustain the use of fossil fuels (McGlade and Ekins, 2015; Heede and Oreskes, 2016; Jewell et al., 2013; 11 Bauer et al., 2016; Jakob and Hilaire, 2015; Welsby et al., 2021). 12 One way of quantifying potential carbon lock-in is to estimate the future CO2 emissions from existing 13 and planned infrastructure (Davis et al., 2010; Davis and Socolow, 2014) based on historic patterns of 14 use and decommissioning. Such estimates focus on CO2 emissions from operating infrastructure and do 15 not comprise any upstream or downstream emissions across the lifecycle, which are provided elsewhere 16 in the literature (Müller et al., 2013; Fisch-Romito, 2021; Krausmann et al., 2020; Creutzig et al., 2016). 17 Moreover, estimates tend to focus on energy, while, for example, the agricultural sector is usually not 18 covered. Another strand of literature quantifies lock-in by estimating fossil-fuel related CO2 emissions 19 that are hard-to-avoid in future scenarios using integrated assessment models (Kriegler et al., 2018b; 20 Luderer et al., 2018). The remainder of this chapter will assess potential carbon lock-in through those 21 two related strands of literature. 22 2.7.2 Estimates of future CO2 emissions from long-lived infrastructures 23 Table 2.6 summarizes studies that apply an accounting approach based on plant-level data to quantify 24 future CO2 emissions from long-lived fossil fuel infrastructure (Davis and Socolow, 2014; Smith et al., 25 2019; Rozenberg et al., 2015; Davis et al., 2010; Tong et al., 2019; Cui et al., 2019; Pfeiffer et al., 2018; 26 Pradhan et al., 2021; Edenhofer et al., 2018). Differences between studies arise in the scope of the 27 infrastructure covered (incl. resolution), the inclusion of new infrastructure proposals, the exact 28 estimation methodology applied as well as their assessments of uncertainties. Other studies provide 29 analysis with a sectoral focus (Vogl et al., 2021; Bullock et al., 2020) or with a regional focus on the 30 power sector (Shearer et al., 2017, 2020; Tao et al., 2020; González-Mahecha et al., 2019; Grubert, 31 2020) . 32 Assuming variations in historic patterns of use and decommissioning, comprehensive estimates of 33 cumulative future CO2 emissions from current fossil fuel infrastructures are 720 (550-910) GtCO2 34 (Smith et al., 2019) and 660 (460-890) (Tong et al., 2019) (Table 2.6, Figure 2.26) (high confidence). 35 This is about the same size than the overall cumulative net CO2 emissions until reaching net zero CO2 36 of 510 (330-710) Gt in pathways that limit warming to 1.5°C with no or limited overshoot (Chapter 3). 37 About 50% of cumulative future CO2 emissions from current fossil fuel infrastructures come from the 38 power sector and 70% of these (or about 40% of the total) are from coal plants only. Like global annual 39 CO2 emissions (Friedlingstein et al., 2020; Peters et al., 2020), future CO2 emissions from fossil-fuel 40 infrastructures have increased over time, i.e. future CO2 emissions from fossil fuel infrastructure 41 additions in a given year are still outgrowing “savings” from infrastructure retirements (Davis and 42 Socolow, 2014; Tong et al., 2019). This could add further inertia to the system as it may require more 43 and faster retirement of fossil-based infrastructures later and leads to higher costs for meeting the 44 climate goals (e.g. Johnson et al., 2015; Bertram et al., 2015). 45 Estimates of total cumulative future CO2 commitments from proposed infrastructure focus only on the 46 power sector due to data availability (Table 2.6, Figure 2.26). Infrastructure proposals can be at various 47 stages of development involving very different probabilities of implementation. About one third of the 48 currently proposed projects are more probable as they are already under construction (Cui et al., 2019). Do Not Cite, Quote or Distribute 2-69 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 Pfeiffer et al. (2018) and Tong et al. (2019) assess the cumulated CO2 emissions from proposed 2 infrastructure in the entire power sector at 270 GtCO2 and 190 GtCO2 respectively. Estimates of CO2 3 emissions implications for new coal power infrastructure plans are more frequent (Pfeiffer et al., 2018; 4 Edenhofer et al., 2018; Cui et al., 2019; Tong et al., 2019) ranging between 100 and 210 GtCO2. 5 Differences across estimates of future CO2 emissions from proposed power infrastructure mostly reflect 6 substantial cancellations of coal infrastructure proposals in 2017 and 2018 (Tong et al., 2019). 7 Do Not Cite, Quote or Distribute 2-70 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 8 Table 2.6 - Comparing cumulative future CO2 emissions estimates from existing and proposed long-lived infrastructures by sector. 9 Future CO2 emissions estimates are reported from the “year of dataset”. Note that in some cases, the totals may not correspond to the sum of underlying sectors 10 due to rounding (based on Tong et al., 2019). Initial estimates of future CO2 emissions from fossil fuel infrastructures by Davis et al. (2010) are considerably lower 11 than more recent estimates by Smith et al. (2019) and Tong et al. (2019) due to substantial growth in fossil energy infrastructure as represented by more recent 12 data. Estimates presented here are rounded to two significant digits. Davis et al. Davis and Socolow Rozenberg et al Edenhofer et al. Pfeiffer et al. Smith et al. Cui et al. Tong et al. (2019) (2010) (2014) (2015) (2018) (2018) (2019) (2019) Year of Year of Gt Year of Year of Year of Year of Year of Gt Year of GtCO2 GtCO2 GtCO2 GtCO2 GtCO2 GtCO2 dataset dataset CO2 dataset dataset dataset dataset dataset CO2 dataset 350 360 Electricity 220 2009 310 2012 - - - - 310 2016 2009* 2018 - - (260-450) (240-490) 260 Coal 2009 210 2012 - - 190 2016 220 2016 - - 2018 340 2017 (180-360) Gas, oil, 98 and other 2009 100 2012 - - - - 88 2016 - - 2018 - - (65-140) fuels 150 160 Industry 100 2009 - - - - - - 2009 2017 - - (120-190) (110-220) Existing 92 64 Transport 120 2009 - - - - - - 2017 2017 - - (73-110) (53-75) Residential, commercial, 120 74 53 2009 - - - - - - 2009* 2018 - - and other (91-160) (52-110) energy 660 2013 500 720 660 All Sectors (370- - - - - - - - - (280-700) (550-910) (460-890) 890) 190 Electricity - - - - 270 2016 - - 2018 - - (140-230) 97 Coal - - 150 2016 210 2016 - - 2018 180 2017 Proposed (74-120) Gas, oil, 91 and other - - - - 60 2016 - - 2018 - - (68-110) fuels All Sectors + Proposed 850 Electricity (600-1,100) 13 Do Not Cite, Quote or Distribute 2-71 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 The global estimate of future CO2 emissions from current and planned fossil-fuel infrastructures is 850 2 (600-1100) GtCO2 (Tong et al., 2019). This already exceeds total cumulative net CO2 emissions in 3 pathways that limit warming to 1.5°C with no or limited overshoot (see above). It is about the same size 4 than the total cumulative net CO2 emissions of 890 (640-1160) GtCO2 from pathways that limit likely 5 warming to 2°C (Chapter 3). Hence, cumulative net CO2 emissions to limit likely warming to 2°C or 6 lower could already be exhausted by current and planned fossil fuel infrastructure (medium confidence) 7 even though this estimate only covers a fraction of all infrastructure developments over the 21st century 8 as present in mitigation pathways, does not cover all sectors (e.g. AFOLU) and does not include 9 currently infrastructure development plans in transport, buildings, and industry due to a lack of data. 10 Hence, the Paris climate goals could move out of reach unless there are dedicated efforts to early 11 decommissioning, and reduced utilization of existing fossil fuel infrastructures, cancellation of plans 12 for new fossil fuel infrastructures, or compensation efforts by removing some of the CO 2 emissions 13 from the atmosphere (Smith et al., 2019; Tong et al., 2019; Cui et al., 2019; Pradhan et al., 2021). For 14 example, Fofrich et al. (2020) suggest in a multi-model study that coal and gas power infrastructure 15 would need to be retired 30 (19-34) and 24 (21-26) years earlier than the historical averages of 39 and 16 36 years when following 1.5°C pathways and 23 (11-33) and 19 (11-16) years earlier when following 17 2°C pathways. Cui et al. (2019) arrive at more conservative estimates for coal power plants, but only 18 consider the existing and currently proposed capacity. Premature retirement of power plants pledged by 19 members of the Powering Past Coal Alliance would cut emissions by 1.6 GtCO2, which is 150 times 20 less than future CO2 emissions from existing coal power plants (Jewell et al., 2019). 21 Few quantifications of carbon lock-in from urban infrastructure, in particular urban form, have been 22 attempted, in part because they also relate to behaviours that are closely tied to routines and norms that 23 co-evolve with “hard infrastructures” and technologies, as well as “soft infrastructure” such as social 24 networks and markets (Seto et al., 2016). There are some notable exceptions providing early attempts 25 (Guivarch and Hallegatte, 2011; Lucon et al., 2014; Erickson and Tempest, 2015; Driscoll, 2014; IPCC, 26 2014b; Creutzig et al., 2016). Creutzig et al. (2016) attempt a synthesis of this literature and estimate 27 the total cumulative future CO2 emissions from existing urban infrastructure at 210 Gt and from new 28 infrastructures at 495 Gt for the period 2010-2030. Do Not Cite, Quote or Distribute 2-72 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 3 Figure 2.26 Future CO2 emissions from existing and currently planned fossil fuel infrastructure in the 4 context of Paris carbon budgets in GtCO2 based on historic patterns of infrastructure lifetimes and 5 capacity utilization. Future CO2 emissions estimates of existing infrastructure for the electricity sector as 6 well as all other sectors (industry, transport, buildings, other fossil fuel infrastructures) and of proposed 7 infrastructures for coal power as well as gas and oil power. Grey bars on the right depict the range (5th – 8 95th percentile) in overall cumulative net CO2 emissions until reaching net zero CO2 in pathways that 9 limit warming to 1.5°C with no or limited overshoot (1.5°C scenarios), and in pathways that limit likely 10 warming to 2°C (2°C scenarios). 11 Source: Based on (Tong et al., 2019) and (Edenhofer et al., 2018). 12 13 2.7.3 Synthesis – Comparison with estimates of residual fossil fuel CO2 emissions 14 A complementary strand of literature uses Integrated Assessment Models (IAMs) to assess the 15 cumulative gross amount of unabated CO2 emissions from fossil fuels across decarbonisation pathways 16 that are not removed from the system even under strong (short- and long-term) climate policy ambition. 17 Lower bound estimates for such a minimum amount of unabated residual CO2 emissions across the 21st 18 century that is not removed from the system even under very ambitious climate policy assumptions may 19 be around 600-700 GtCO2 (Kriegler et al., 2018b). This range increases to 650-1800 GtCO2 (Table 2.7) 20 as soon as a broader set of policy assumptions are considered including delayed action in scenarios that 21 limit warming to 1.5°C and 2°C respectively (Luderer et al., 2018). 22 Notably, the lower end of residual fossil fuel emissions in IAM scenarios (Luderer et al., 2018) is 23 remarkably similar to global estimates from the accounting studies of the previous section as shown in 24 Table 2.6. Yet, there are important conceptual and interpretative differences that are also reflected in 25 the very different distribution of reported future CO2 emissions attached to current and future fossil fuel 26 infrastructures (Table 2.7). Accounting studies start from granular, plant-based data for existing fossil 27 fuel infrastructure and make statements about their future CO2 emission assuming variations of historic Do Not Cite, Quote or Distribute 2-73 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 patterns of use and decommissioning. Expansions to the future are limited to proposals for new 2 infrastructures that we know of today. Scenario studies quantifying residual fossil fuel emissions start 3 from aggregate infrastructure descriptions, but dynamically update those through new investment 4 decisions in each time step across the 21st century based on the development of energy and energy 5 service demands as well as technology availability, and guided by defined climate policy goals (or their 6 absence). 7 In accounting studies, estimated of future CO2 emissions from current fossil fuel infrastructures are 8 dominated by the power sector with its large fossil fuel capacities today. In contrast, scenario studies 9 highlight residual emissions from non-electric energy – particularly in the transport and industry sectors. 10 Fossil-fuel infrastructure in the power sector can be much more easily retired than in those sectors, 11 where there are fewer and more costly alternatives. IAMs therefore account for continued investments 12 into fossil-based energy technologies in areas with limited decarbonisation potential, such as some areas 13 of transportation (in particular aviation, shipping and road-based freight) or some industrial processes 14 (such as cement production or feedstocks for chemicals). This explains the key discrepancies observable 15 in Table 2.7. Overall, our assessment of these available lines of evidence therefore strongly emphasises 16 the importance of decommissioning, as well as reduced utilization of existing power sector 17 infrastructure as well as continued cancellation of new power sector infrastructures in order to limit 18 warming to well below 2°C (high confidence) (Luderer et al., 2018; Kriegler et al., 2018b; Fofrich et 19 al., 2020; Cui et al., 2019; Chen et al., 2019). This is important as the power sector is comparatively 20 easy to decarbonise (IPCC, 2014a; Krey et al., 2014; Méjean et al., 2019; Davis et al., 2018) and it is 21 crucial to make space for residual emissions from non-electric energy end-uses that are more difficult 22 to mitigate (high confidence). Any further delay in climate policy substantially increases carbon lock- 23 in and mitigation challenges as well as a dependence on carbon dioxide removal technologies for 24 meeting the Paris climate goals (Kriegler et al., 2018b; Luderer et al., 2018). Do Not Cite, Quote or Distribute 2-74 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 Table 2.7 Residual (gross) fossil fuel emissions (GtCO2) in climate change mitigation scenarios strengthening mitigation action after 2020 (“early strengthening”), 2 compared to scenarios that keep NDC ambition level until 2030 and only strengthen thereafter. 3 Cumulative gross CO2 emissions from fossil fuel and industry until reaching net zero CO2 emissions are given in terms of the mean as well as minimum and 4 maximum (in parentheses) across seven participating models: AIM/CGE, GCAM, IMAGE, MESSAGES, POLES, REMIND, WITCH. Scenario design prescribes a 5 harmonised, global carbon price in line with long-term carbon budget. Delay scenarios follow the same price trajectory, but 10 years later. Carbon dioxide 6 removal requirements represent ex-post calculations that subtract gross fossil fuel emissions from the carbon budget associated with the respective long-term 7 warming limit. We take the carbon budget for limiting warming to 1.5°C with a 50% probability and to 2°C with a 67% probability (Canadell et al., 2021). Hence, 8 carbon dioxide removal (CDR) requirements reflect a minimum amount of CDR for a given mitigation trajectory. Results are reported at 2 significant digits. 9 Sources: (Luderer et al., 2018; Tong et al., 2019) Future CO2 emissions from existing and planned Residual fossil fuel emissions - cumulative gross CO2 emissions from fossil fuel and industry until reaching net zero CO2 fossil fuel infrastructure (accounting studies) emissions (in GtCO2) Tong et al. (2019) Early strengthening from (2020) Delayed strengthening from 2030 GtCO2 Year Well below 2°C Below 1.5°C in Well below 2°C Below 1.5°C in 2100 2100 Electricity 550 Electricity 180 130 250 200 proposed Existing (380-730) 2018 (140 - 310) (90 - 160) (220 - 340) (190 - 230) AND Non-electric supply Non-electric supply 100 59 120 75 (42 - 130) (27 - 83) (55 - 150) (40 - 100) Existing AND future instalments Industry 160 Industry 260 140 290 200 (110-220) 2017 (160 - 330) (86 - 180) (200 - 370) (130 - 250) Transportation 64 Transportation 310 170 310 200 (53-75) 2017 (190 - 370) (110 - 220) (250 - 400) (140 - 260) Buildings 74 Buildings 110 58 120 73 (52-110) 2018 (75 - 110) (35 - 77) (80 - 150) (51 - 93) All sectors and proposed 850 All sectors (2021 – net zero CO2) 960 570 1100 770 Existing electricity (730 - 1100) (400 - 640) (900 - 1200) (590 - 860) All sectors (2021-2100) 1300 850 1400 1000 (600-1,100) (970 - 1500) (650 - 1100) (1200 - 1600) (860 - 1300) Implied minimum requirement 150 350 250 500 for carbon dioxide removal until (0 – 350) (150 – 600) (50 – 450) (360 – 800) 2100 10 Do Not Cite, Quote or Distribute 2-75 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 2 2.8 Climate and Non-Climate Policies and Measures and their Impacts on 3 Emissions 4 2.8.1 Introduction 5 The key to achieving climate change mitigation targets includes crafting environmentally effective, 6 economically efficient and socially equitable policies. For the purposes of this section, policies are 7 defined broadly as actions to guide decisions to reach explicit goals and, accordingly, climate 8 (mitigation) policies are the ones whose primary objective is to reduce GHG emissions. They include a 9 range of domains from economic and institutional to R&D and social policies and are implemented by 10 various instruments (e.g., market-based and regulatory in the economic domain) and measures (e.g., 11 legal provisions and governance arrangements in the institutional domain) (see Chapter 13 and the 12 Glossary about mitigation policies). Yet GHG emissions are also affected by policies enacted in various 13 social, economic, and environmental areas to pursue primarily non-climatic objectives. This section 14 presents succinct assessments of the outcomes and effectiveness of a few selected policy instruments 15 applied in the last two decades targeting climate protection (Sections 2.8.2 and 2.8.3) and GHG 16 emissions impacts of selected other policies primarily aiming improvements in environmental quality 17 and natural resource management (Section 2.8.4).12 18 It is rather difficult, though not impossible, to discern the genuine impacts of climate and non-climate 19 policies on GHG emissions. Most of current and past policies target only a small part of global emissions 20 in a limited geographical area and/or from a small number of economic sectors. However, in addition to 21 the targeted region or sector, policies and measures tend to affect GHG emissions in other parts of the 22 world. Emissions leakage is the key channel by which such phenomena and complex interactions 23 occur. 13 Uncertainties in impacts, synergies, and trade-offs between policies and measures also 24 complicate the evaluation of emissions impacts. These make it challenging to identify the impacts of 25 any specific policy or measure on emissions of any specific region or sector. Rigorous statistical analyses 26 are necessary for building strong empirical evidence, but the experience with climate-related policy 27 experiments to date is limited. 28 2.8.2 Comprehensive Multinational Assessments 29 Comprehensive multinational evaluations with wider regional and sectoral coverage enable the 30 assessment of emissions impacts without distortions from emissions leakage. Among the wide range of 31 climate policy instruments, pricing carbon such as a carbon tax or an emissions trading system has been 32 one of the most widely used and effective options to reduce GHG emissions (robust evidence, high 33 agreement). In a comparison of 142 countries with and without carbon pricing, countries with a carbon 34 price show an annual CO2 emission growth rates of 2 percentage points lower than countries without 35 such policies (Best et al., 2020). A more comprehensive evaluation of carbon prices shows that countries 36 with a lower carbon pricing gap (a higher carbon price) tend to be more carbon-efficient, that is, they 37 have a lower carbon intensity of GDP (OECD, 2018). 14 An empirical analysis of the effects of FOOTNOTE12 This section only reviews emission impacts of selected policy instruments. Other important aspects such as equity and cost-effectiveness are assessed in Chapter 13, presenting comprehensive evaluations of policies and measures. FOOTNOTE13 Refer to Chapter 13 on policies and institutions for detailed discussion of emissions leakages and complex interactions from policy mixes. FOOTNOTE14 The OECD (2018) measures carbon prices using the effective carbon rate (ECR), which is the sum of three components: specific taxes on fossil fuels, carbon taxes, and prices of tradable emissions permits. The Do Not Cite, Quote or Distribute 2-76 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 environmental regulation and innovation on the carbon emissions of OECD countries during the period 2 1999–2014 indicates that a 1% increase in environmentally friendly patents reduced carbon emissions 3 by 0.017%, and a 1% increase in environmental tax revenue per capita reduced carbon emissions by 4 0.03% (Hashmi and Alam, 2019). 5 Domestic and international climate legislation have also contributed to the reduction of GHG emissions. 6 An empirical analysis of legislative activity in 133 countries over the period 1999–2016 based on panel 7 data indicates that each new law reduced annual CO2 emissions per unit of GDP by 0.78% nationally in 8 the first three years and by 1.79% beyond three years. Additionally, climate laws as of 2016 were 9 associated with an annual reduction in global CO2 emissions of 5.9 GtCO2 and 38 GtCO2 cumulatively 10 since 1999 (Eskander and Fankhauser, 2020). It is notable that 36 countries that accepted legally binding 11 targets under the Kyoto Protocol all complied (Shishlov et al., 2016). It is impossible to disentangle 12 precisely the contribution of individual mitigation policies, but it is clear that the participating countries, 13 especially those in the OECD, did make substantial policy efforts with material impact (Grubb, 2016). 14 An ex-post evaluation shows a significant impact of the Protocol on emissions reductions (Maamoun, 15 2019). 16 Renewable energy policies, such as Renewable Portfolio Standards and Feed-in-Tariff, have played an 17 essential role in the massive expansion of renewable energy capacities, another key driver of GHG 18 emissions reductions (robust evidence, high agreement). The drivers of decreasing CO2 emissions in a 19 group of 18 developed economies that have decarbonised over the period 2005–2015 has been shown 20 to be the displacement of fossil fuels by renewable energy and decreases in energy use (Le Quere et al., 21 2019). Renewable energy policies both at the EU and Member States level have played an essential role 22 in abating GHG emissions (ICF International, 2016). 23 2.8.3 National, Sectoral, and Cross-Sectoral Policies 24 2.8.3.1 National and regional carbon pricing 25 Carbon prices (e.g., carbon taxes and GHG emissions trading schemes), are among the widely used 26 climate policy instruments across the globe, together with technology support instruments (see IRENA 27 (2018)). As of May 2020, there were 61 carbon pricing schemes in place or scheduled for 28 implementation, consisting of 31 emissions trading schemes (ETSs) and 30 carbon tax regimes, covering 29 12 GtCO2-eq or about 22% of annual global GHG emissions (World Bank, 2020). The performance of 30 carbon pricing in practice varies by countries and sectors, and depends on the policy environment (robust 31 evidence, high agreement). 32 The European Union Emissions Trading Scheme (EU ETS), the longest-standing regional climate policy 33 instrument to date, has reduced emissions, though the estimates of the amount vary by study, by country, 34 and by sector; ranging from 3 to 28% (McGuinness and Ellerman, 2008; Ellerman et al., 2010; Abrell 35 et al., 2011; Anderson and Di Maria, 2011; Egenhofer et al., 2011; Petrick and Wagner, 2014; 36 Arlinghaus, 2015; Martin et al., 2016). The EU ETS avoided emitting about 1.2 GtCO2 between 2008 37 and 2016 (3.8%), almost half of what EU governments promised to reduce under their Kyoto Protocol 38 commitments (Bayer and Aklin, 2020). 39 China’s emission trading pilots have resulted in a decline in carbon intensity in the pilot provinces by 40 adjusting the industrial structure (Zhou et al., 2019). The Regional Greenhouse Gas Initiative (RGGI) 41 in the USA has induced leakage in emissions through increases in electricity generation in surrounding 42 non-RGGI areas, but it has led to the reduction of emissions by way of changes in the fuel mix from 43 coal to gas (Fell and Maniloff, 2018). Actual emissions declined in six of the ten ETSs for which data carbon pricing gap measures the difference between actual ECRs and benchmark rates. The carbon pricing gap indicates the extent to which polluters do not pay for the damage from carbon emissions. Do Not Cite, Quote or Distribute 2-77 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 is available, although other factors such as the 2009 recession, have had significant impacts on those 2 emissions as well (Haites et al., 2018). 3 The evidence of environmental effectiveness of carbon taxes in Western European countries is varied 4 depending on country and study (robust evidence, high agreement). A significant impact is found in 5 Finland but insignificant impacts are found in Denmark and the Netherlands, and there are mixed results 6 for Sweden (Lin and Li, 2011; Brännlund et al., 2014). Only six of the 17 taxes, where data are available, 7 have reduced actual emissions subject to the tax. Tax rates tend to be too low in many cases and the 8 scale and frequency of the rate changes has not been sufficient to stimulate further emissions reductions 9 (Haites et al., 2018). 10 2.8.3.2 Selected sectoral climate policy instruments 11 Many governments have implemented sector-specific policies, in addition to nationwide measures, to 12 reduce GHG emissions (high confidence). Examples of sectoral climate policies include carbon taxes 13 on transportation fuels, low-carbon fuel standards, and regulation of coal power generation. 14 The implementation of a carbon tax and value-added tax on gasoline and diesel in Sweden resulted in 15 significant reductions of CO2 emissions in the transportation sector (Shmelev and Speck, 2018; 16 Andersson, 2019). An assessment of a variety of carbon tax schemas across various sectors in the 17 European Union shows a negative relationship between CO2 emissions and a CO2 tax (Hájek et al., 18 2019). In British Columbia (Canada), the carbon tax resulted in a decrease in demand for gasoline and 19 a reduction in total GHG emissions (not exclusive to the transportation sector) estimated to be between 20 5 and 15% (Murray and Rivers, 2015; Rivers and Schaufele, 2015). Low Carbon Fuel Standards in 21 California have contributed to reducing carbon emissions in the transportation sector by approximately 22 9.85–13.28% during 1997–2014 (Huseynov and Palma, 2018). 23 The power sector typically accounts for a large portion of countries’ CO2 emissions. Market-based 24 regulation and government subsidies in China contributed to improving operational efficiency and 25 reducing emissions (Zhao et al., 2015). In addition, the implementation of ultra-low emission standards 26 also has resulted in a significant reduction in emissions from China's power plants (Tang et al., 2019). 27 Mandatory climate and energy policies, including the California Global Warming Solutions Act, 28 reduced CO2 emissions by 2.7–25% of the average state-level annual emissions from the power sector 29 over the period 1990–2014 in the USA. Mandatory GHG registry/reporting, electric decoupling and 30 public benefit fund have been effective in further decreasing power sector emissions in the USA (Martin 31 and Saikawa, 2017). In the UK electricity sector, a carbon price floor, combined with electricity market 32 reform (competitive auctions for both firm capacity and renewable energy), displaced coal, whose share 33 fell from 46% in 1995 to 7% in 2017, halving CO2 emissions, while renewables grew from under 4% in 34 2008 to 22% by 2017 (Grubb and Newbery, 2018). See Chapter 13 for more. 35 An alternative approach to a carbon tax is an indirect emissions tax on fuels such as an excise tax, or on 36 vehicles, based on the expected CO2 intensity of new passenger vehicles. Vehicle purchase taxes can 37 result in a reduction in GHG emissions through reducing the CO2 emissions intensity of vehicles, while 38 also discouraging new vehicle purchases (Aydin and Esen, 2018). For example, a vehicle tax policy in 39 Norway resulted in a reduction of average CO2 intensity per kilometre of 7.5 gCO2 km-1 (Ciccone, 2018; 40 Steinsland et al., 2018). Despite such evidence, studies of carbon pricing find that additional policies are 41 often needed to stimulate sufficient emissions reductions in transportation (medium confidence) 42 (Tvinnereim and Mehling, 2018). 43 Electric vehicles (EVs) powered by clean electricity can reduce GHG emissions and such policies are 44 important for spurring adoption of such vehicles and GHG emission reductions (Kumar and Alok, 2020; 45 Thiel et al., 2020). The extent to which EV deployment can decrease emissions by replacing internal 46 combustion engine-based vehicles depends on the generation mix of the electric grid (Abdul-Manan, 47 2015; Nichols et al., 2015; Canals Casals et al., 2016; Hofmann et al., 2016; Choi et al., 2018; Teixeira Do Not Cite, Quote or Distribute 2-78 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 and Sodré, 2018), although even with current grids EVs reduce emissions in almost all cases (Knobloch 2 et al., 2020). Policy incentives for EV adoption can be an effective mechanism to increase EV sales 3 (Langbroek et al., 2016) and may include charging discounts, purchase subsidies, regulations, and 4 government leadership (medium confidence) (Bakker and Jacob Trip, 2013; Silvia and Krause, 2016; 5 Teixeira and Sodré, 2018; Qiu et al., 2019; Santos and Davies, 2020). The presence of charging 6 infrastructure and publicly available charging increases the adoption rate of EVs (Vergis and Chen, 7 2015; Javid et al., 2019). A comparison of EV adoption rates across 30 countries shows a positive 8 correlation between charging stations and EV market share (Sierzchula et al., 2014). A rollout of 80,000 9 DC fast chargers across the United States is estimated to have resulted in a 4% reduction in emissions 10 compared to a baseline of no additional fast chargers (Levinson and West, 2018). More recently, bans 11 on internal combustion engine vehicles have provided a much more direct approach to stimulating the 12 adoption of EVs and its supporting infrastructure; however, the efficacy of such measures depends on 13 enforcement (Plötz et al., 2019). 14 Public transit can reduce vehicle travel and reduce GHG emissions by reducing the number of trips taken 15 by private vehicles and the length of those trips (medium confidence). Changes to the operation of public 16 transportation systems (such as density of bus stops, distance from stops to households, duration and 17 frequency of trip times, and lowering ridership costs) can result in a mode shift from private car trips to 18 public transit trips (Cats et al., 2017; Choi, 2018; Carroll et al., 2019). These changes in the public transit 19 system operation and network optimisation have been shown to have reduced GHG emissions in cases 20 such as San Francisco, in which the cost optimisation of the transit network was estimated to decrease 21 emissions by a factor of three (Cheng et al., 2018) and Barcelona, where the optimisation of the urban 22 bus system was estimated to reduce GHG emissions by 50% (Griswold et al., 2017). For every 1% 23 increase in investment in transit services and transit-oriented design, there is an estimated 0.16% 24 reduction in private vehicle kilometres travelled per capita (McIntosh et al., 2014). 25 Bike and car sharing programmes can reduce GHG emissions (medium confidence). Albeit a study of 26 eight cities in the United States with larger bike share systems and higher ridership found that their 27 potential to reduce total emissions is limited to <0.1% of total GHG emissions from the transportation 28 sectors of these cities (Kou et al., 2020). The emissions reductions effects of car-sharing programmes 29 depends on the specifics of programmes: the mode shift from public transit to car-sharing services can 30 outweigh the decreases in GHG emissions associated with decreased cars on the road (Jung and Koo, 31 2018), whereas car-sharing programmes with electric vehicle fleets may reduce GHG emissions (Luna 32 et al., 2020). 33 2.8.4 Emission Impacts of Other Related Policies 34 Policies other than those intended directly to mitigate GHG emissions can also influence these 35 emissions. Policies to protect the stratospheric ozone layer is a case in point. Implementing the Montreal 36 Protocol and its amendments, emissions of controlled ozone-depleting substances (ODSs) (those 37 covered by the protocol) declined to a very low level of about 1.4 GtCO2-eq yr-1 by 2010, avoiding GHG 38 emissions of an estimated 13.3–16.7 GtCO2-eq yr-1 (9.7–12.5 GtCO2-eq yr-1 when accounting for the 39 ozone depletion and hydrofluorocarbons (HFCs) offsets) (Velders et al., 2007). Yet fluorinated gases 40 (F-gases), the substances introduced to substitute ODSs are also potent GHGs. See Section 2.2 for 41 emissions data and Chapter 13 on current policies to mitigate HFCs and other F-gases. GHG 42 implications of two other categories of non-climate policies are briefly assessed in this section. 43 2.8.4.1 Co-impacts of air quality, sector-specific and energy policies on climate mitigation 44 Co-impacts of local or regional air pollution abatement policies for climate mitigation are widely studied 45 in the literature. Cross-border externalities of air pollution have also made these a focus of several 46 international agreements (Mitchell et al., 2020). Evaluating the effectiveness of such treaties and policies 47 is difficult because deriving causal inferences and accurate attribution requires accounting for several Do Not Cite, Quote or Distribute 2-79 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 confounding factors, and direct and indirect spillovers (Isaksen, 2020). Nevertheless, several studies 2 assess the effectiveness of such treaties and regulations (De Foy et al., 2016; Li et al., 2017a, 2017b; 3 Morgenstern, 2018; Mardones and Cornejo, 2020). However, there is little ex-post empirical analysis 4 and a greater focus on ex-ante studies in the literature. 5 At a local scale, air pollutants are often co-emitted with GHGs in combustion processes. Many air quality 6 policies and regulations focus on local pollution from specific sources that can potentially either 7 substitute or complement global GHG emissions in production and generation processes. Also, policies 8 that reduce certain air pollutants, such as SO2, have a positive radiative forcing effect (Navarro et al., 9 2016). The evidence on individual air pollution control regulation and policies for GHG emissions is 10 therefore mixed (medium evidence, medium agreement). Evidence from the USA suggests that increased 11 stringency of local pollution regulation had no statistically detectable co-benefits or costs on GHG 12 emissions (Brunel and Johnson, 2019). Evidence from China suggests that the effectiveness of policies 13 addressing local point sources differed from those of non-point sources and the co-benefits for climate 14 are mixed, though policies addressing large industrial point sources have been easier to implement and 15 have had significant impact (Huang and Wang, 2016; Xu et al., 2016; van der A et al., 2017; Dang and 16 Liao, 2019; Fang et al., 2019; Yu et al., 2019). Legislation to reduce emissions of air pollutants in Europe 17 have significantly improved air quality and health but have had an unintended warming effect on the 18 climate (Turnock et al., 2016). 19 Often, the realisation of potential co-benefits depends on the type of pollutant addressed by the specific 20 policy, and whether complementarities between local pollution and global GHG emissions are 21 considered in policy design (Rafaj et al., 2014; Li et al., 2017a) (medium evidence, high agreement). 22 Effective environmental regulations that also deliver co-benefits for climate mitigation require 23 integrated policies (Schmale et al., 2014; Haines et al., 2017). Uncoordinated policies can have 24 unintended consequences and even increase emissions (Holland et al., 2015). Many studies suggest that 25 policies that target both local and global environmental benefits simultaneously may be more effective 26 (Klemun et al., 2020) (medium evidence, medium agreement). Furthermore, air pollution policies aimed 27 at inducing structural changes, for example closure of polluting coal power plants or reducing motorised 28 miles travelled, are more likely to have potential positive spillover effects for climate mitigation, as 29 compared to policies incentivising end-of-pipe controls (Wang, 2021). 30 Other policies that typically have potential co-benefits for climate mitigation include those specific to 31 certain sectors and are discussed in Chapters 5–11. Examples of such policies include those that 32 encourage active travel modes, which have been found to have ancillary benefits for local air quality, 33 human health, and GHG emissions (Fujii et al., 2018). Policies to reduce energy use through greater 34 efficiency have also been found to have benefits for air quality and the climate (Tzeiranaki et al., 2019; 35 Bertoldi and Mosconi, 2020) (robust evidence, medium agreement). Important air quality and climate 36 co-benefits of renewable or nuclear energy policies have also been found (Lee et al., 2017; Apergis et 37 al., 2018; Sovacool and Monyei, 2021) (medium evidence, medium agreement). 38 Policies specific to other sectors such as encouraging green building design can also reduce GHG 39 emissions (Eisenstein et al., 2017). Evidence from several countries also show that replacing polluting 40 solid biomass cooking with cleaner gas-burning or electric alternatives have strong co-benefits for 41 health, air quality, and climate change (Anenberg et al., 2017; Singh et al., 2017; Tao et al., 2018) (robust 42 evidence, high agreement). 43 2.8.4.2 Climate impacts of agricultural, forestry, land use, and AFOLU-related policies 44 Policies on agriculture, forestry, and other land use (AFOLU), and AFOLU sector-related policies have 45 had a long history in many developing and developed countries. Co-impacts of these policies on the 46 climate have been only marginally studied, although their impacts might be quite important because the Do Not Cite, Quote or Distribute 2-80 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 AFOLU sector is responsible for 24% of total GHG emissions (robust evidence, high agreement). The 2 results of afforestation policies around the world and the contribution to CCS are also important. 3 Both private and governmental policies can have a major impact on the climate. Experience indicates 4 that “climate proofing” a policy is likely to require some stimulus, resources, and expertise from 5 agencies or organisations from outside the country. Stimulus and support for adaptation and mitigation 6 can come from the UN system and from international development institutions (FAO, 2009). These 7 findings are also valid for small/organic farmers vis-à-vis large-scale agro-industry. For example, 8 small/medium and environmentally concerned farmers in Europe are often asking for more policies and 9 regulations, and see it as necessary both from a climate perspective and to maintain competitiveness 10 relative large agro-industrial complexes. Therefore, the need for governmental support for small 11 producers in regulations encompasses all AFOLU sectors. 12 Forestry case: zero deforestation 13 Forest is generally defined as land spanning more than 0.5 hectares with trees higher than 5 meters and 14 a canopy cover of more than 10%, or trees able to reach these thresholds in situ (FAO, 1998). Zero- 15 deforestation (i.e., both gross and net zero deforestation) initiatives generate results at multiple levels 16 (Meijer, 2014). Efforts to achieve zero-deforestation (and consequently emissions) are announced by 17 NGOs, companies, governments, and other stakeholder groups. NGOs engage through their 18 campaigning, but also propose tools and approaches for companies (Leijten et al., 2020). The extent to 19 which companies can actually monitor actions conducive to zero-deforestation pledges depends on their 20 position in the supply chain. Beyond the business practices of participating companies, achieving long- 21 term positive societal impacts requires upscaling from supply chains towards landscapes, with 22 engagement of all stakeholders, and in particular small producers. The various success indicators for 23 zero deforestation mirror the multiple levels at which such initiatives develop: progress towards 24 certification, improved traceability, and legality are apparent output measures, whereas direct-area 25 monitoring and site selection approaches target the business practices themselves. 26 Such efforts have led to the development of the High Carbon Stock (HCS) approach that combines 27 carbon stock values with the protection of HCS areas (including peatlands and riparian zones) and areas 28 important for the livelihoods of local communities (Rosoman et al., 2017). Long-term positive impacts, 29 however, will need to be assessed with hindsight and focus on national and global statistics. Successful 30 initiatives targeting zero deforestation at jurisdictional level would also need to improve the enforcement 31 of forest laws and regulations (EII, 2015; Meyer and Miller, 2015). 32 Large-scale agribusiness, banks, and consumer goods companies dominate supply chain-focused zero- 33 deforestation initiatives, but only the producers, including local communities and smallholders, can 34 change the production circumstances (TFD, 2014). Producers shoulder much of the burden for meeting 35 environmental requirements of pledges. And local communities and small producers are vulnerable to 36 being cut out when supply chains reorient. The zero-deforestation pledges do not always devise 37 programmes for introducing new sourcing strategies, and governments may have an important 38 contribution to make here, particularly in safeguarding the interests of small producers. 39 Other than in Brazil and Indonesia, beyond individual supply chains, there is still little evidence on 40 positive results of zero-deforestation commitments as information available for companies to judge their 41 progress is scarce. Moreover, many zero-deforestation pledges set targets to be achieved by 2020 or 42 2030, and, consequently, many companies have not yet reported publicly on their progress. Similarly, 43 only a few governments have yet shown progress in reducing deforestation, but the New York 44 Declaration on Forests, the SDGs and the Paris Agreement were adopted relatively recently. The 45 effectiveness of private-sector zero-deforestation pledges depends on the extent to which they can be 46 supported by governmental action and foster a cooperative environment with the engagement of all 47 stakeholders. Where the pledges are coordinated with regulation, multi-stakeholder dialogues, and 48 technical and financial support, a true paradigm shift becomes possible. Many governments are still Do Not Cite, Quote or Distribute 2-81 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 building the capacity to improve overall forest governance, but implementing ambitious international 2 targets is likely to depend on technical and major financial support that has not yet been mobilised. 3 4 2.9 Knowledge Gaps 5 • Global GHG emissions estimates are published less frequently and with greater reporting lags 6 than, for example, CO2 from fossil fuel and industry. Data quality and reporting frequency 7 remains an issue particularly in developing countries where the statistical infrastructure is not 8 well developed. Efforts to compile a global GHG emissions inventory by country, sector, and 9 across time that is annually updated based on the best-available inventory information, similar 10 to on-going activities for CO2, CH4 or N2O, could fill this gap. Uncertainties and their 11 methodological treatment in GHG emissions estimates are still not comprehensively 12 understood. 13 • There is a more fundamental data gap for F-gas emissions, where data quality in global 14 inventories is poor due to considerable gaps in the underlying activity data – particularly in 15 developing countries. Comprehensive tracking of F-gas emissions would also imply the 16 inclusion of other gases not covered under the Paris Agreement such as chlorofluorocarbons, 17 hydrochlorofluorocarbons and others. 18 • Currently, despite advances in terms of data availability, sectoral and spatial resolution, the 19 results in consumption-based emission estimates are dependent on the database used, the level 20 of sectoral aggregation and country resolution. More fine-grained data at spatial resolution as 21 well as the product level would support to explore the mitigation options at the sub-national 22 level, companies and households. 23 • Consumption-based emission accounts too suffer from lack of quantification of uncertainties at 24 the subnational level and especially in data-scarce environments such as for developing 25 countries. A better understanding of drivers that caused decoupling of emissions at the national 26 and especially sub-national level are important to explore. 27 • Understanding how social-economic drivers modulate emission mitigation is crucial. 28 Technological improvements (e.g. improved energy or land use intensity of the economy) have 29 shown persistent pattern over the last few decades but gains have been outpaced by increases in 30 affluence (GDP per capita) and population growth, leading to continued emissions growth. The 31 key gap in knowledge therefore is how these drivers of emissions can be mitigated by demand 32 management, alternative economic models, population control and rapid technological 33 transition to different extent and in different settings. More research on decoupling and 34 sustainability transformations would help to answer these questions. Key knowledge gaps also 35 remain in role of trade, in particular, how supporting low-carbon technologies in developing 36 and exporting countries can counteract the upward-driving effect of trade, and how to achieve 37 decoupling without outsourcing emissions to others and often to less developed regions. 38 • Understanding of how inequality affects emissions is in a nascent stage. Less is known about 39 the causal mechanisms by which different dimensions of inequality like income, socio- 40 economic, spatial, socio-cultural-gender and ethnicity affect emissions. In particular, limited 41 knowledge exists on the linkages between dimensions of inequality other than income or wealth 42 and emissions arising from different service demands. Research gaps are apparent on how 43 inequalities in living standards relate to emissions and how changes in inequalities between 44 genders, social groups, and other marginalised communities impact emissions trends. 45 • Digitalisation of the economy are often quoted as providing new mitigation opportunities, but 46 knowledge and evidences are yet limited- such as understanding of the role of smart apps and 47 the potentials and influence of disruptive technologies at the demand and supply side on GHG 48 emissions. 49 • Despite growing evidence of technological progress across a variety of mitigation areas and the 50 availability of increasingly precise data sets, knowledge gaps remain on technological change Do Not Cite, Quote or Distribute 2-82 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 and innovation and evidence on speed of transitions to clarify what would make them fast or 2 slow. Innovation is an inherently uncertain process and there will always be imperfect ex ante 3 knowledge on technological outcomes and their effects on mitigation. The extent to which a 4 low-carbon transition can proceed faster than historical examples is crucial to aid future 5 mitigation. That depends on a better understanding of the speed of building, updating and 6 replacing infrastructure. Additionally, how and whether financing for low-carbon technology 7 investment in low and middle income countries can be delivered at low-cost and sustained over 8 time are important questions. The emerging findings that small-scale technologies learn faster 9 and are adopted more quickly needs to be tested against a broader set of cases and in particular 10 against the large dispersion in data. 11 • Future CO2 emissions from existing and planned infrastructure is not well understood and 12 quantified outside the power sector. Further integration of bottom-up accounting and scenario 13 approaches from integrated assessment seems promising. Comprehensive assessments of hard- 14 to-abate residual fossil fuel emissions and their relationship to CO2 removal activities are 15 lacking, but will be important for informing net zero emissions strategies. 16 • Empirical evidence of emission impacts from climate policies, including carbon pricing, is not 17 sufficient for unambiguous attribution assessment, mainly due to the limited experience with 18 climate-related policy experiments to date. More attention to the methodology for 19 comprehensive evaluation of climate policies and measures, such as effective carbon rates is 20 apparent. Key knowledge gaps also exist on ex-post evaluations of climate and non-climate 21 policies and measures for their impact on emissions, particularly at the global scale, considering 22 national circumstances and priorities. 23 24 25 Frequently Asked Questions (FAQs) 26 FAQ 2.1 Are emissions still increasing or are they falling? 27 Global greenhouse gas (GHG) emissions continued to rise and reached 59±6.6 GtCO2-eq in 2019, 28 although the rate of growth has fallen compared to the previous decade. Still, emissions were higher 29 than at any point in human history before. Emissions were around 12% and 54% higher than in 2010 30 and 1990, respectively. Average annual GHG emissions for 2009–2019 were higher compared to the 31 periods 2000–2009 and 1990–1999, respectively. GHG emission growth slowed since 2010: while 32 average annual GHG emission growth was 2.1% between 2000 and 2010, it was only 1.3% for 2010– 33 2019. In order to stop the temperature increase, however, net emissions must be zero. 34 FAQ 2.2 Are there countries that have reduced emissions and grown economically at the same 35 time? 36 About 24 countries that have reduced territorial CO2 and GHG emissions for more than 10 years. 37 Uncertainties in emission levels and changes over time prevents a precise assessment in some country 38 cases. In the short observation period of 2010–2015, 43 out of 166 countries have achieved absolute 39 decoupling of consumption-based CO2 emissions from economic growth, which means that these 40 countries experienced GDP growth while their emissions have stabilised or declined. A group of 41 developed countries, such as some EU countries and the United States, and some developing countries, 42 such as Cuba, have successfully achieved an absolute decoupling of consumption-based CO2 emissions 43 and GDP growth. Decoupling has been achieved at various levels of per capita income and per capita 44 emissions. Overall, the absolute reduction in annual emissions achieved by some countries has been 45 outweighed by growth in emissions elsewhere in the world. 46 FAQ 2.3 How much time do we have to act to keep global warming below 1.5 degrees? 47 If global CO2 emissions continue at current rates, the remaining carbon budget for keeping warming to 48 1.5°C will likely be exhausted before 2030. Between 1850 and 2019, total cumulative CO2 emissions Do Not Cite, Quote or Distribute 2-83 Total pages: 127 Final Government Distribution Chapter 2 IPCC AR6 WGIII 1 from the fossil fuel industry (FFI) and agriculture, forestry, and other land use (AFOLU) were 2400 2 (±240 GtCO2). Of these, about 410±30 GtCO2 were added since 2010. This is about the same size as the 3 remaining carbon budget for keeping global warming to 1.5°C and between one third and half the 4 1150±220 (1350, 1700) GtCO2 for limiting global warming below 2°C with a 67% (50%, 33%) 5 probability, respectively (Canadell et al., 2021). At current (2019) rates of emissions, it would only take 6 8 (2-15) and 25 (18-35) years to emit the equivalent amount of CO2 for a 67th percentile 1.5°C and 2°C 7 remaining carbon budget, respectively. This highlights the dependence of 1.5°C pathways on the 8 availability of substantial CO2 removal capacities, as discussed in chapters 3, 4, and 12, but also Section 9 2.7 of this chapter. 10 11 12 13 Do Not Cite, Quote or Distribute 2-84 Total pages: 127