The consequences of non-participation in the Paris Agreement

Larch, Mario; Wanner, Joschka Article — Published Version The consequences of non-participation in the Paris Agreement European Economic Review Provided in Cooperation with: Kiel Institute for the World Economy – Leibniz Center for Research on Global Economic Challenges Suggested Citation: Larch, Mario; Wanner, Joschka (2024) : The consequences of non-participation in the Paris Agreement, European Economic Review, Elsevier BV, Amsterdam, NL, Vol. 163, pp. 1-18, https://doi.org/10.1016/j.euroecorev.2024.104699 This Version is available at: https://hdl.handle.net/10419/307097 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ (XURSHDQ (FRQRPLF 5HYLHZ    $YDLODEOH RQOLQH  )HEUXDU\    7KH $XWKRUV 3XEOLVKHG E\ (OVHYLHU %9 7KLV LV DQ RSHQ DFFHVV DUWLFOH XQGHU WKH && %< OLFHQVH KWWSFUHDWLYHFRPPRQVRUJOLFHQVHVE\ Contents lists available at ScienceDirect European Economic Review journal homepage: www.elsevier.com/locate/eer The consequences of non-participation in the Paris Agreement$ Mario Larcha,b,c,d,e, Joschka Wanner f,g,d,∗ aDepartment of Law, Business & Eonomics, University of Bayreuth, Germany bCEPII, France cifo, Germany dCESifo, Germany eGEP, United Kingdom fJulius-Maximilians-Universität Würzburg (JMU), Germany gKiel Institute for the World Economy, Germany ARTICLE INFO JEL classification: F14 F18 Q56 Keywords: Climate change International trade Carbon leakage Fossil fuel supply ABSTRACT International cooperation is at the core of multilateral climate policy. How is its effectiveness harmed by individual countries not participating in the global mitigation effort? We use a multisector structural trade model with carbon emissions from production and a constant elasticity of fossil fuel supply function to simulate the consequences of unilateral non-participation in the Paris Agreement. Taking into account both direct and leakage effects, we find that nonparticipation of the US would eliminate more than a third of the world emissions reduction (31.8% direct effect and 6.4% leakage effect), while a potential non-participation of China lowers the world emission reduction by 24.1% (11.9% direct effect and 12.2% leakage effect). The substantial leakage is primarily driven by technique effects induced by falling international fossil fuel prices. In terms of welfare, the overwhelming majority of countries gain from the implementation of the Paris Agreement and most countries have only very little to gain from unilaterally deciding not to participate. 1. Introduction The coming into force of [the] Paris Agreement has ushered in a new dawn for global cooperation on climate change. (Then UN Secretary General Ban Ki-Moon, November 15th, 2016) [I]n order to fulfill my solemn duty to protect America and its citizens, the United States will withdraw from the Paris Climate Accord. (Then US President Donald Trump, June 1st, 2017) $Previous versions of this paper were circulated under the title ‘‘The Consequence of Unilateral Withdrawals from the Paris Agreement’’. We thank Franz Mildner and Fabian Schmid for excellent research assistance, Franziska Piontek and Leonie Wenz for sharing data on climate damages, as well as two anonymous referees and participants at the TRISTAN workshop 2018 in Bayreuth, ETSG 2018 in Warsaw, FIW Research Conference ‘‘International Economics’’ 2018 in Vienna, Midwest International Economics Group Meeting 2019 in Bloomington, European Association of Environmental and Resource Economists Conference 2020, Workshop ‘‘Shaping Globalization’’ 2021, German Economic Association Conference 2021, Goettingen Workshop ‘‘International Economics’’ 2022, the ‘‘Gravity at Sixty’’ Workshop 2022 in Vienna, as well as at research seminars in Innsbruck, Nottingham, St. Catharines (Brock University), Penn State, Philadelphia (Drexel University), Paris (CEPII), the Kiel Institute of the World Economy, Freiburg, the Mercator Institute on Global Commons and Climate Change, and the Humboldt University Berlin for helpful comments. All errors are our own. ∗Corresponding author at: Julius-Maximilians-Universität Würzburg (JMU), Germany. E-mail addresses: [email protected] (M. Larch), [email protected] (J. Wanner). https://doi.org/10.1016/j.euroecorev.2024.104699 Received 29 December 2022; Received in revised form 8 February 2024; Accepted 11 February 2024 (XURSHDQ (FRQRPLF 5HYLHZ     M. Larch and J. Wanner In December 2015, the parties to the United Nations Framework Convention on Climate Change (UNFCCC) reached a joint agreement to combat climate change. With its 195 signing countries, the Paris Agreement constitutes a truly global consensus to take appropriate measures to keep global warming well below two degrees Celsius. One centerpiece of the agreement are the Nationally Determined Contributions (NDCs) in which every country specifies an individual greenhouse gas (GHG) emission reduction target. While the reduction targets stated in the NDCs are very heterogeneous across countries, what is crucial and most likely explains at least part of the enthusiasm expressed in the first opening quote by former UN Secretary General Ban Ki-Moon is the fact that every country has a target. The sub-global coverage of the Paris Agreement’s most prominent predecessor, the Kyoto Protocol, severely harmed its effectiveness due to leakage effects (see e.g. Aichele and Felbermayr,2012,2015). Carbon leakage refers to the phenomenon that climate policies undertaken in some countries can lead to increased emissions in other places where no such policies are undertaken due to (i) production shifts of emission-intensive goods towards the un-(or less) regulated countries and (ii) falling fossil fuel prices on the world market that incentivize a more fossil fuel-intensive production (see e.g. Felder and Rutherford,1993). As the second opening quote by former US President Donald Trump clearly shows, the hope of achieving the world emission reduction that would result from adding up all national targets may be overly optimistic. Following through on the announcement, the United States officially left the agreement in November 2020.1Even though the United States has rejoined under Trump’s successor Joe Biden, the episode clearly demonstrates the fragility of the global consensus. Countries that decide not to commit to their emission targets harm the effectiveness of the Paris Agreement in two ways. First, and most obviously, the sum of the national targets is lowered if some countries drop their target (we call this the ‘‘direct effects’’). Second, and potentially just as importantly, non-participation can induce carbon leakage that lowers the achieved world reduction below the remaining sum of national targets. Different from the direct effects, leakage effects (and hence the total effects) of unilateral non-participation cannot be simply calculated, but have to be solved using a multi-country general equilibrium framework. The most common approach to investigate the global effects of different trade and climate policies is the use of computable general equilibrium (CGE) models (see e.g. Böhringer et al.,2012, for an overview of various prominent CGE models). A recent strand of literature (Egger and Nigai,2015;Shapiro,2016; Larch and Wanner,2017;Larch et al.,2018;Shapiro and Walker,2018;Farrokhi and Lashkaripour,2021;Shapiro,2021;Caron and Fally,2022) incorporates environmental components into structural gravity models as an alternative approach.2Gravity models are the workhorse models in the empirical international trade literature. Just as CGE models, they can be used to conduct ex-ante analyses of different policy scenarios. Compared to typical CGE models, they tend to sacrifice some detail in the model structure in favor of higher analytical tractability and direct estimation of key model parameters. Given gravity’s great success in predicting trade flows (see e.g. Head and Mayer,2014;Costinot and Rodríguez-Clare,2014, for surveys on gravity models and their performance), it is likely to capture well leakage that occurs via production shifts and international trade. The main model of Larch and Wanner (2017), as well as the models by Shapiro (2016), Shapiro and Walker (2018), and Farrokhi and Lashkaripour (2021) exclusively focus on this leakage channel. In this paper, we extend the model of Larch and Wanner (2017) by considering fossil fuel resources that are internationally traded and supplied according to a constant elasticity of fossil fuel supply function, as proposed in the CGE context by Boeters and Bollen (2012). The resulting extended gravity model will capture leakage effects via international trade and via the international fossil fuel market and hence allow a quantification of the total emission reduction losses associated with unilateral non-participation in the Paris Agreement. At the same time, the model structure remains tractable enough to allow an analytical and quantitative decomposition of the national emission changes into scale, composition, and technique effects as is often done in the theoretical and empirical literature on trade and the environment (see e.g. Grossman and Krueger,1993;Copeland and Taylor,1994,2003). This decomposition can generate important insights into the channels through which international climate policies are effective. Our analysis of the effects of non-participation complements other studies that investigate the Paris Agreement and its implications. For example, Glanemann et al. (2020) investigate whether the Paris goal of keeping global warming well below two degrees is economically sensible: it is because avoided damages outweigh mitigation costs. Rogelj et al. (2016) analyze whether individual national goals are sufficient to jointly achieve the two (or even 1.5) degree Celsius target: they are not. Aldy and Pizer (2016), Aldy et al. (2017), and Iyer et al. (2018) aim to make the different NDCs comparable in their implied required mitigation efforts of the different countries. Rose et al. (2018) investigate one particular way for efficiently achieving the reduction pledges, namely by linking different emissions trading schemes. Nong and Siriwardana (2018) analyze the consequences of a US withdrawal on the US economy, finding, among others, a significant drop in energy prices. Böhringer and Rutherford (2017) and Winchester (2018) show that the introduction of carbon tariffs is not a credible threat to the US to try to keep them in the agreement. Kemp (2017) considers measures that can be taken to reduce the damage to the effectiveness of the agreement due to a US withdrawal, e.g. by incorporating cooperation with US states. We contribute to the literature by quantifying the harm done by countries not participating in the Paris Agreement taking into account both direct effects and emission shifts (leakage) resulting from general equilibrium adjustments of supply and demand of goods and fossil fuels. The rest of this paper proceeds as follows. Section 2presents our extended structural gravity model, shows how counterfactual analyses can be performed in this framework, and derives the emission change decomposition. In Section 3, the data sources 1Additionally, a small number of other signing countries of the agreement (Iran being the largest among them in terms of carbon emissions) have not yet moved on to ratification. 2Pothen and Hübler (2018) develop a hybrid model, combining an Eaton and Kortum (2002)-type gravity trade structure with a CGE model production structure. (XURSHDQ (FRQRPLF 5HYLHZ     M. Larch and J. Wanner and descriptive statistics are presented, as well as the gravity estimation procedure. We discuss the results of simulating the nonparticipation for each country in Section 4. In Section 5, we derive a model extension with multiple fossil fuels of varying carbon intensities, leading to a fourth, substitution, effect on emissions, and rerun the simulations using the extended model. Section 6 concludes. 2. Model In this section, we present an extended structural gravity model that includes multiple sectors, a multi-factor production function including an energy input, energy production including an internationally tradable fossil fuel resource, a constant elasticity of fossil fuel supply (CEFS) function following Boeters and Bollen (2012), as well as emissions associated with fossil fuel usage. The model builds on the framework by Larch and Wanner (2017), but deviates by (i) modeling the energy market leakage channel using a CEFS function,3(ii) linking emissions directly to fossil fuel use rather than to general energy use, and (iii) explicitly including a carbon tax that countries can use to achieve emission reduction targets. 2.1. Supply 2.1.1. Goods production There is a set of countries and a set of sectors . Each country 𝑗∈produces a differentiated variety in each of the 𝑙∈ sectors according to the following Cobb–Douglas production function: 𝑞𝑖 𝑙=𝐴𝑖 𝑙(𝐸𝑖 𝑙)𝛼𝑖 𝑙𝐸 ∏ 𝑓∈ (𝑉𝑖 𝑙𝑓 )𝛼𝑖 𝑙𝑓 , where 𝐴𝑖 𝑙is a sectorand country-specific productivity parameter, 𝛼𝑖 𝑙𝐸, and 𝛼𝑖 𝑙𝑓 denote production cost shares, and 𝑉𝑖 𝑙𝑓 the usages of a production factor 𝑓∈. Countries are endowed with a fixed factor supply 𝑉𝑖 𝑓and factors are mobile across sectors, but internationally immobile. 𝐸𝑖 𝑙denotes the energy input. Markets are assumed to be perfectly competitive and goods are hence sold at marginal costs: 𝑝𝑖 𝑙=𝛤𝑖 𝑙 𝐴𝑖 𝑙 (𝑒𝑖)𝛼𝑖 𝑙𝐸 ∏ 𝑓∈ (𝑤𝑖 𝑓)𝛼𝑖 𝑙𝑓 ,(1) where 𝛤𝑖 𝑙=(𝛼𝑖 𝑙𝐸)−𝛼𝑖 𝑙𝐸 ∏𝑓∈(𝛼𝑖 𝑙𝑓 )−𝛼𝑖 𝑙𝑓 ,𝑒𝑖is the energy price in country 𝑖, and 𝑤𝑖 𝑓are the factor prices. 2.1.2. Energy production Different from the other production factors, countries are not endowed with a fixed energy supply, but the energy input has to be produced itself according to the following Cobb–Douglas production function: 𝐸𝑖=𝐴𝑖 𝐸(𝑅𝑖)𝜉𝑖 𝑅∏ 𝑓∈ (𝑉𝑖 𝐸𝑓)𝜉𝑖 𝑓, where 𝜉𝑖 𝑅and 𝜉𝑖 𝑓denote the input cost shares and 𝑅𝑖is the usage of a fossil fuel resource. We abstract from trade costs in fossil fuels and assume that they are freely internationally tradable, implying a perfectly integrated world fossil fuel market.4A country’s carbon emissions are modeled as proportional to its fossil fuel use.5 The energy price depends on the factor prices and technological parameters, as well as on the global fossil fuel price 𝑟. Additionally, countries can charge a carbon tax 𝜆𝑖on the fossil fuel use: 𝑒𝑖=𝛤𝑖 𝐸 𝐴𝑖 𝐸((1 + 𝜆𝑖)𝑟)𝜉𝑖 𝑅∏ 𝑓∈ (𝑣𝑖 𝑓)𝜉𝑖 𝑓,(2) where 𝛤𝑖 𝐸=(𝜉𝑖 𝑅)−𝜉𝑖 𝑅∏𝑓∈(𝜉𝑖 𝑓)−𝜉𝑖 𝑓. 3The base model of Larch and Wanner (2017) only features the trade leakage channel, while the small model extension presented in their work relies on an energy resource in fixed supply. 4A very insightful paper that allows for a role of geography in one specific fossil fuel market (crude oil), is Farrokhi (2020). There, a gravity-type pattern arises due to a combination of fixed costs and unobserved refiner-supplier-pair frictions. Farrokhi (2020) finds the extent to which the oil market deviates from an integrated global market to be ‘‘modest’’, encouraging us in our simplifying assumption at this point, in particular as our counterfactual scenarios leave bilateral trade costs unaffected. 5Note that this implies two simplifications: the only type of greenhouse gas we account for in the model is CO2and in terms of CO2, we account only for combustion emissions and abstract from process emissions (e.g. in cement production). (XURSHDQ (FRQRPLF 5HYLHZ     M. Larch and J. Wanner 2.1.3. Fossil fuel supply In modeling the global supply of the fossil resource 𝑅𝑊, we use a constant elasticity of fossil fuel supply function as proposed by Boeters and Bollen (2012): 𝑅𝑊=𝜁(𝑟 𝑃)𝜂,(3) where 𝜁is a supply shifter, 𝑃a global price index, and 𝜂denotes the supply elasticity. The total fossil fuel supply 𝑅𝑊stems from the different countries according to their varying fossil fuel endowment shares 𝜔𝑖(with ∑𝑖∈𝜔𝑖=1 ). These fossil endowment shares are also used to aggregate national price indices to the global level: 𝑃≡∏𝑖∈(𝑃𝑖∕𝜔𝑖)𝜔𝑖, with 𝑃𝑖≡∏𝑙∈(𝑃𝑖 𝑙∕𝛾𝑖 𝑙)𝛾𝑖 𝑙, where 𝛾𝑗 𝑙 represents country 𝑗’s expenditure share for sector 𝑙. As the name suggests, the chosen supply function ensures that the fossil fuel supply reacts with a constant elasticity to changes in the real fossil fuel price. As pointed out by Boeters and Bollen (2012), this is a difference (and advantage) in comparison to the more standard procedure of a nested production structure with a natural resource in fixed supply entering the uppermost nest.6 Avoiding the assumption of a resource in fixed supply further allows us to link emissions directly to the quantity of the resource employed in production, rather than e.g. indirectly linking it proportionately to the energy use. Note the key role of 𝜂for the energy market leakage channel. The more elastic the supply, the less a negative fossil fuel demand shock will change the fossil fuel price and hence the smaller the incentive for a country without its own climate policy to rely more heavily on fossil fuels and thus the smaller the energy market leakage effect. 2.1.4. Income Countries generate income from (i) the expenditure on their national production factors, (ii) their share of the global supply of fossil fuels, and (iii) the carbon tax charged on its fossil fuel use: 𝑌𝑖=∑ 𝑓∈ 𝐼𝑖 𝑓+𝐼𝑖 𝑅+(𝜆𝑖 1+𝜆𝑖)𝜉𝑖 𝑅∑ 𝑙∈ 𝛼𝑖 𝑙𝐸𝑌𝑖 𝑙,(4) where 𝐼𝑖 𝑓≡𝑤𝑖 𝑓[𝑉𝑖 𝐸𝑓 +∑𝑙∈𝑉𝑖 𝑙𝑓 ]denotes the factor incomes, 𝐼𝑖 𝑅≡𝜔𝑖𝑅𝑊𝑟the fossil resource income, and 𝑌𝑖 𝑙≡𝑞𝑖 𝑙𝑝𝑖 𝑙are the sectoral values of production. 2.2. Demand 2.2.1. Utility Consumers in country 𝑗obtain utility according to the following utility function: 𝑈𝑗=⎡⎢⎢⎢⎣∏ 𝑙∈⎛⎜⎜⎝[∑ 𝑖∈ (𝛽𝑖 𝑙) 1−𝜎𝑙 𝜎𝑙(𝑞𝑖𝑗 𝑙) 𝜎𝑙−1 𝜎𝑙]𝜎𝑙 𝜎𝑙−1 ⎞⎟⎟⎠ 𝛾𝑗 𝑙⎤⎥⎥⎥⎦⎡⎢⎢⎢⎣ 1 1+(1 𝜇𝑗∑𝑖∈𝑅𝑖)2⎤⎥⎥⎥⎦ , where 𝛽𝑖 𝑙represents a preference parameter for goods from different origins, 𝑞𝑖𝑗 𝑙is the amount of good 𝑙from country 𝑖consumed in country 𝑗,𝜎𝑙stands for the sectoral elasticity of substitution, 𝜇𝑗is a parameter that captures 𝑗’s disutility from global carbon emissions, and 𝑅𝑖is country 𝑖’s fossil fuel use which is proportional to its emissions. The utility function hence combines sectoral CES utility from consumption of goods from different origins in an upper-tier Cobb–Douglas utility function (implying constant sectoral expenditure shares), as well as disutility from global emissions in the functional form chosen by Shapiro (2016) to ensure almost constant social costs of carbon around the baseline emission level. Carbon emissions are treated as a pure externality and are therefore not taken into account in consumption decisions. 2.2.2. Gravity Introducing iceberg trade costs 𝑇𝑖𝑗 𝑙(with 𝑇𝑖𝑗 𝑙=𝑇𝑗𝑖 𝑙≥1and 𝑇𝑖𝑖 𝑙=1), we can express sectoral bilateral trade shares as an Eaton and Kortum (2002)-type gravity expression that contrasts country 𝑖’s cost of serving market 𝑗(in terms of technology, input costs, and trade costs) to all other suppliers: 𝜋𝑖𝑗 𝑙=(𝛽𝑖 𝑙𝑝𝑖 𝑙𝑇𝑖𝑗 𝑙)1−𝜎𝑙 ∑𝑘∈(𝛽𝑘 𝑙𝑝𝑘 𝑙𝑇𝑘𝑗 𝑙)1−𝜎𝑙=(𝛽𝑖 𝑙𝑝𝑖 𝑙𝑇𝑖𝑗 𝑙 𝑃𝑗 𝑙)1−𝜎𝑙 .(5) Note that our calibration of the model will also include one non-tradable sector. This can simply be achieved in the model with infinite trade costs in the respective sector, implying fully domestic sourcing (𝜋𝑖𝑖 𝑙=1). The climate policies considered in this paper and discussed in more detail in the next section affect the production costs and hence the prices of producers in different countries and sectors differently and hence alter international trade patterns. This will capture the production relocation leakage channel, as low/no carbon price countries gain competitiveness and market shares in emission-intensive industries and hence specialize in these products. 6In this approach (taken e.g. in a paper on commodity trade by Fally and Sayre,2018), the fossil fuel supply elasticity changes endogenously with the stringency of climate policy measures taken. (XURSHDQ (FRQRPLF 5HYLHZ     M. Larch and J. Wanner 2.3. Climate policy We will implement climate policy via carbon taxes in the model. Countries can charge a national carbon tax 𝜆𝑖on the use of fossil fuels to fulfill specific emission targets 𝑅𝑖. We will run different scenarios in all of which all countries around the world will fulfill the emission reduction targets specified in their NDCs, except for one country that decides not to participate in the agreement. We can use the scenario to pin down the chosen level of the carbon tax 𝜆𝑖in the model. Denoting the set of committed (or cooperating) countries by 𝑐𝑜𝑝, the country that is not part of the agreement chooses a zero carbon tax, while all other countries choose their carbon tax exactly at the required level to ensure that their realized emissions are equal to their targeted emission level7: 𝜆𝑖=⎧ ⎪ ⎨ ⎪ ⎩ 0if 𝑖∉𝑐𝑜𝑝, 𝜉𝑖 𝑅∑𝑙∈𝛼𝑖 𝑙𝐸𝑌𝑖′ 𝑙 𝑅𝑖′𝑟′−1 if 𝑖∈𝑐𝑜𝑝. (6) 2.4. Trade balance, market clearing and equilibrium Trade is assumed to be balanced and the national energy and factor markets, as well as the international goods and fossil fuel markets, are all assumed to clear: ∑ 𝑖∈∑ 𝑙∈ 𝜋𝑖𝑗 𝑙𝛾𝑗 𝑙𝑌𝑗=∑ 𝑗∈∑ 𝑙∈ 𝜋𝑗𝑖 𝑙𝛾𝑖 𝑙𝑌𝑖(7) 𝐸𝑖=∑ 𝑙 𝐸𝑖 𝑙(8) 𝑉𝑖 𝑓=∑ 𝑙∈ 𝑉𝑖 𝑙𝑓 +𝑉𝑖 𝐸𝑓,(9) 𝑌𝑖 𝑙=∑ 𝑗∈ 𝜋𝑖𝑗 𝑙𝛾𝑗 𝑙𝑌𝑗,(10) 𝑅𝑊𝑟=∑ 𝑖∈(1 1+𝜆𝑖)𝜉𝑖 𝑅∑ 𝑙∈ 𝛼𝑖 𝑙𝐸𝑌𝑖 𝑙.(11) Definition 1. For given factor endowments 𝑉𝑖 𝑓, productivities 𝐴𝑖 𝑙and 𝐴𝑖 𝐸, preference shifters 𝛽𝑖 𝑙, and trade costs 𝑇𝑖𝑗 𝑙, an equilibrium under climate policy structure {𝑐𝑜𝑝, 𝑅𝑖}is a set of factor prices 𝑤𝑖 𝑓, energy prices 𝑒𝑖, carbon taxes 𝜆𝑖, a world fossil fuel price 𝑟, and global fossil fuel supply 𝑅𝑊that satisfy equilibrium conditions (1)–(11). Note that the equilibrium could also be expressed for a given set of carbon taxes rather than for a given coalition with a set of emission targets. Then, 𝜆𝑖becomes exogenous and we can drop Eq. (6) from the equilibrium conditions. 2.5. Equilibrium in changes Following Dekle et al. (2007,2008), we can re-express the equilibrium of our model in changes, as this allows us to perform counterfactual analyses without the need to identify the level of the factor endowments 𝑉𝑖 𝑓, productivities 𝐴𝑖 𝑙and 𝐴𝑖 𝐸, and preference shifters 𝛽𝑖 𝑙.8We follow their ‘‘hat notation’’ which indicates the change of the respective variables, i.e. 𝑥 =𝑥′ 𝑥, where the prime indicates a counterfactual value in response to a policy shock and values without a prime correspond to the baseline equilibrium. Definition 2. Let {𝑣𝑖 𝑓,𝑒 𝑖,𝑟,𝑅 𝑊}be a baseline equilibrium under climate policy structure {𝜆𝑖}and {𝑣𝑖′ 𝑓,𝑒 𝑖′,𝜆 𝑖′,𝑟 ′,𝑅 𝑊′}be a counterfactual equilibrium under climate policy structure {𝑐𝑜𝑝, 𝑅𝑖′ }. Then, {𝑣𝑖 𝑓,𝑒𝑖, 1+𝜆𝑖,𝑟,  𝑅𝑊}satisfy the following equilibrium conditions (12)–(19): Carbon tax change:  1+𝜆𝑖=⎧ ⎪ ⎨ ⎪ ⎩ 1if 𝑖∉𝑐𝑜𝑝, 𝜉𝑖 𝑅∑𝑙𝛼𝑖 𝐸,𝑙 ∑𝑗𝜋𝑖𝑗 𝑙𝜋𝑖𝑗 𝑙𝛾𝑗 𝑙𝑌𝑗′ 𝜉𝑖 𝑅∑𝑙𝛼𝑖 𝐸,𝑙 ∑𝑗𝜋𝑖𝑗 𝑙𝛾𝑗 𝑙𝑌𝑗(𝑅𝑖′ 𝑅𝑖𝑟)−1 if 𝑖∈𝑐𝑜𝑝. (12) 7Note that we treat the targeted emission level 𝑅𝑖′as exogenously given. This is in contrast to two important recent contributions in the trade and environment literature by Farrokhi and Lashkaripour (2021) and Kortum and Weisbach (2021) that both consider optimal climate policies in an international setting. Kortum and Weisbach (2021), however, consider a two-country setting and Farrokhi and Lashkaripour (2021) abstract, as previously mentioned, from the energy market leakage channel, while our model brings together a multi-country setting and a consideration of both key leakage channels. 8In principle, it would also allow us to avoid identification of the iceberg trade costs. We nevertheless estimate these and use a fitted trade network for our baseline equilibrium. 𝜋𝑖𝑗 𝑙in the following hence refers to fitted rather than observed trade shares. Using fitted rather than observed trade shares avoids zero trade flows leading to the implicit assumption of infinite trade costs between some countries, as well as potential problems of overfitting (see Dingel and Tintelnot, 2021). Further, we can use the calculation of fitted trade shares to eliminate trade imbalances in the data that otherwise may lead to numeraire dependency or non-zero global imbalances in the counterfactual results (see Costinot and Rodríguez-Clare,2014;Ossa,2016). (XURSHDQ (FRQRPLF 5HYLHZ     M. Larch and J. Wanner Trade share change: 𝜋𝑖𝑗 𝑙=((𝑒𝑖)𝛼𝑖 𝐸,𝑙 ∏𝑓( 𝑤𝑓 𝑖)𝛼𝑖 𝑓,𝑙)1−𝜎𝑙 ∑𝑘𝜋𝑘𝑗 𝑙((𝑒𝑘)𝛼𝑘 𝐸,𝑙 ∏𝑓( 𝑤𝑓 𝑘)𝛼𝑘 𝑓,𝑙)1−𝜎𝑙.(13) Price index change:  𝑃𝑗 𝑙=⎛⎜⎜⎝∑ 𝑖 𝜋𝑖𝑗 𝑙((𝑒𝑖)𝛼𝑖 𝐸,𝑙 ∏ 𝑓( 𝑤𝑓 𝑖)𝛼𝑖 𝑓,𝑙)1−𝜎𝑙⎞⎟⎟⎠ 1∕(1−𝜎𝑙) .(14) Fossil fuel supply change:  𝑅𝑊=⎛⎜⎜⎜⎜⎝ 𝑟 ∏𝑖(∏𝑙( 𝑃𝑖 𝑙)𝛾𝑖 𝑙)𝜔𝑖⎞⎟⎟⎟⎟⎠ 𝜂 .(15) Counterfactual income: 𝑌𝑗′=∑ 𝑓( 𝑤𝑗 𝑓𝐼𝑗 𝑓)+ 𝑅𝑊𝑟𝐼𝑗 𝑅+(𝜆𝑗′ 1+𝜆𝑗′)𝜉𝑗 𝑅∑ 𝑙 𝛼𝑗 𝐸,𝑙 ∑ 𝑖 𝜋𝑗𝑖 𝑙𝜋𝑗𝑖 𝑙𝛾𝑖 𝑙𝑌𝑖′.(16) Factor price change:  𝑤𝑖 𝑓=1 𝐼𝑖 𝑓∑ 𝑙((𝛼𝑖 𝑓,𝑙 +𝜉𝑖 𝑓𝛼𝑖 𝐸,𝑙)∑ 𝑗 𝜋𝑖𝑗 𝑙𝜋𝑖𝑗 𝑙𝛾𝑗 𝑙𝑌𝑗′).(17) Energy price change: 𝑒𝑖=(( 1+𝜆𝑖)𝑟)𝜉𝑖 𝑅∏ 𝑓( 𝑤𝑖 𝑓)𝜉𝑖 𝑓.(18) Fossil fuel price change: 𝑟 =∑𝑖(1 1+𝜆𝑖′)𝜉𝑖 𝑅∑𝑙𝛼𝑖 𝐸,𝑙 ∑𝑗𝜋𝑖𝑗 𝑙𝜋𝑖𝑗 𝑙𝛾𝑗 𝑙𝑌𝑗′ ∑𝑖(1 1+𝜆𝑖)𝜉𝑖 𝑅∑𝑙𝛼𝑖 𝐸,𝑙 ∑𝑗𝜋𝑖𝑗 𝑙𝛾𝑗 𝑙𝑌𝑗( 𝑅𝑊)−1 .(19) 2.6. Decomposition of emission changes As emissions are proportional to a country’s fossil fuel use, emissions in country 𝑖can be written as: 𝑅𝑖=𝜉𝑖 𝑅(∑𝑙∈𝛼𝑖 𝑙𝐸𝑌𝑖 𝑙) (1 + 𝜆𝑖)𝑟=𝜉𝑖 𝑅𝛼𝑖 𝐸  𝑌𝑖 𝑃𝑖(𝑟𝑖 𝑃𝑖)−1 ,(20) where  𝑌𝑖≡∑𝑙∈𝑌𝑖 𝑙denotes total (nominal) production, 𝛼𝑖 𝐸≡∑𝑙∈𝛼𝑖 𝑙𝐸 𝑌𝑖 𝑙  𝑌𝑖is the production-share-weighted average energy cost share, and 𝑟𝑖≡(1+𝜆𝑖)𝑟is the national price for fossil fuels (including the carbon tax). Intuitively, the level of emissions in a country depends on (i) how much is spend for energy inputs in production, (ii) which share of the energy input expenditure is paid for fossil fuel inputs in energy production, and (iii) how expensive fossil fuels are (both in terms of the world market price and the national carbon tax). Following Grossman and Krueger (1993)andCopeland and Taylor (1994) (as well as Larch and Wanner,2017, in a structural gravity context), the change in emissions can then be decomposed into three parts9: 𝑑𝑅𝑖≈𝜕𝑅𝑖 𝜕( 𝑌𝑖∕𝑃𝑖)𝑑( 𝑌𝑖∕𝑃𝑖) ⏟⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏟⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏟ scale effect +𝜕𝑅𝑖 𝜕𝛼𝑖 𝐸 𝑑𝛼𝑖 𝐸 ⏟⏞⏟⏞⏟ composition effect +𝜕𝑅𝑖 𝜕(𝑟𝑖∕𝑃𝑖)𝑑(𝑟𝑖∕𝑃𝑖) ⏟⏞⏞⏞⏞⏞⏞⏞⏞⏞⏟⏞⏞⏞⏞⏞⏞⏞⏞⏞⏟ technique effect . 9Details on the three components are given in Appendix A (XURSHDQ (FRQRPLF 5HYLHZ     M. Larch and J. Wanner 2.7. Welfare effects Welfare changes are a combination of real income changes and changes in climate damages (i.e. in disutility from global emissions) and are given by:  𝑊𝑗= 𝑌𝑗  𝑃𝑗⎡⎢⎢⎢⎣ 1+(1 𝜇𝑗𝑅𝑊)2 1+(1 𝜇𝑗𝑅𝑊′)2⎤⎥⎥⎥⎦ . 3. Data and estimation 3.1. Data sources Our main data source is the Global Trade Analysis Project (GTAP) 10 database (Aguiar et al.,2019). From GTAP, we take the data on carbon emissions, sectoral production, trade flows, factor expenditures, and expenditure for and income from fossil fuels.10 GTAP also provides estimates for the sectoral elasticities of substitution of which we make use.11 Unfortunately, no estimate is available for the fossil fuel supply elasticity. For our main model, we therefore choose the simple average of the values reported by Boeters and Bollen (2012) for the three different specific fossil fuels oil, gas, and coal, namely 𝜂=2 .12 The GTAP 10 data is given for the base year 2014. We hence construct our whole data set for this year. It captures 140 countries (some of which are in fact aggregates of several countries) covering the whole world. We aggregate the sectoral structure to one non-tradable and 14 tradable sectors.13 For the gravity estimation of bilateral trade costs, we rely on a set of standard gravity variables from the CEPII dataset by Head et al. (2010), namely bilateral distance (𝐷𝐼𝑆𝑇), an indicator variable for whether two countries share a common border (𝐶𝑂𝑁𝑇𝐼𝐺), and a second indicator variable for a common official language (𝐿𝐴𝑁𝐺). We complement these variables with an indicator variable for joint regional trade agreement (𝑅𝑇 𝐴) membership taken from Mario Larch’s RTA database (Egger and Larch,2008). We additionally construct a dummy variable that is equal to one for domestic trade flows and zero for all international trade (𝐼𝑁𝑇𝑅𝐴). The (I)NDCs of the signatory states of the Paris Agreement are collected and made available online at the United Nations NDC Registry.14 To translate the different emission targets into 2030 BAU reduction targets, we additionally use GDP and carbon emission projections by the US Energy Information Administration’s (EIA) International Energy Outlook 2016. For climate damages, we calibrate the disutility parameter 𝜇𝑗to the social cost of carbon estimate by Rennert et al. (2022). They estimate it to be 185 Dollars (in 2020 US dollars). Deflating their number to 2014 US dollars, we use a social cost of carbon of 168.53 Dollars. For the regional distribution of these damages, we rely on simulations by NGFS (2022), which in turn rely on the econometric climate damage estimates by Kalkuhl and Wenz (2020).15,16 The gravity, emission target, and climate damage data are all aggregated into the regional structure of the GTAP database. 3.2. National emissions and reduction targets We illustrate the data for two key country characteristics: the level of its emissions and the reduction target specified in its NDC. Fig. 1 displays the national levels of carbon emissions. China and the US stand out as the strongest emitters, followed by other large developed or emerging economies, such as India, Russia, Japan, Germany, and Canada. To make NDCs comparable, we standardize all reduction targets to percentage reductions of carbon emissions below the 2030 business-as-usual emission level.17 They hence relate to the counterfactual emission level enforced in the counterfactual scenarios by 𝑡𝑎𝑟𝑔𝑒𝑡𝑖=1−𝑅𝑖′∕𝑅𝑖.18 Details on the standardization are given in Appendix D. Fig. 2 reports the targets that result from this procedure and which are used in our counterfactual analyses.19 10 See Appendix B for details on the parametrization of the model. 11 See Table B.1 for the specific values across sectors. 12 In our model extension presented in Section 5we can directly use Boeters and Bollen (2012)’s values, specifically 𝜂𝑜𝑖𝑙 =𝜂𝑔𝑎𝑠 =1,𝜂𝑐𝑜𝑎𝑙 =4. 13 The 14 tradable sectors are agriculture, apparel, chemical, equipment, food, machinery, metal, mineral, mining, other, paper, service, textile, and wood. See Appendix C for the concordance to the 65 original GTAP sectors. 14 See https://www4.unfccc.int/sites/NDCStaging/Pages/All.aspx. Note that countries continuously update their NDCs. Our calculations incorporate all updates up until April 2022. 15 Specifically, we use the national median GDP changes from NGFS (2022)’s model runs using the integrated assessment model REMIND with the 50th percentile temperature projections and the median damages from Kalkuhl and Wenz (2020) in a scenario in which all countries implement their NDCs. 16 Finland, Mongolia, and the ‘‘Rest of European Free Trade Association’’ (comprising Iceland and Liechtenstein) are estimated to have positive effects of climate change according to these numbers. As our functional form does not allow for gains from climate change, we put their damages to zero. The ‘‘Rest of North America’’ (comprising Bermuda, Greenland, and Saint Pierre and Miquelon) is not covered by Kalkuhl and Wenz (2020) and we hence also put the corresponding damage to zero. 17 Note that strictly speaking the targets refer to CO2equivalents of all greenhouse gas emissions. Due to better data availability, we use carbon emission paths for the projections for 2030. 18 We calculate the reduction targets for the 2030 time frame, but refrain from projecting all model variables and parameters to 2030 and therefore implement all scenarios as changes from the 2014 baseline equilibrium (implying that 𝑅𝑖refers to national emissions in 2014). 19 The exact values are given in Table D.1 in Appendix D.. (XURSHDQ (FRQRPLF 5HYLHZ     M. Larch and J. Wanner Fig. 1. National Carbon Emissions in 2014. Fig. 2. Emission Reduction Targets in the Paris Agreement. Notes: This figure shows the emission reduction targets specified in the individual countries’ NDCs (or, where no NDCs are available, the Intended NDCs). To make the targets comparable, all are given as reductions below the business-as-usual emission path in 2030. National targets aggregate to a 25.4% global reduction compared to a BAU emission path. The large heterogeneity in the ambition of the targets becomes evident at first sight. While some Asian and African countries merely commit to not increase their emissions beyond the BAU path and some have rather mild targets (like the 11.3% of China), large parts of Europe and the Americas formulate strong targets that in some cases lower their emissions by more than half. Aggregating all national targets implies a 25.4% reduction of global emissions compared to a BAU emission path. 3.3. Gravity estimation Estimates of bilateral trade costs can be obtained based on the gravity Eq. (5) derived above. Approximating trade costs by a function of observable bilateral characteristics (captured by the vector 𝐳𝑖𝑗), collecting all (partly unobservable) importerand exporter-specific terms and introducing an error term yields the following regression equation: 𝑋𝑖𝑗 𝑙= exp(𝜋𝑖 𝑙+𝜒𝑗 𝑙+𝐳′ 𝑖𝑗𝜷𝑙)×𝜀𝑖𝑗 𝑙,(21) where 𝑋𝑖𝑗 𝑙=𝜋𝑖𝑗 𝑙𝛼𝑗 𝑙𝑌𝑗denotes trade flows from country 𝑖to country 𝑗in sector 𝑙. Following the suggestions by Feenstra (2004) and Santos Silva and Tenreyro (2006), respectively, we capture 𝜋𝑖 𝑙and 𝜒𝑗 𝑙by the inclusion of exporter and importer fixed effects and estimate the model in its multiplicative form (avoiding problems due to heteroskedasticity and zero trade flows) with the Poisson (XURSHDQ (FRQRPLF 5HYLHZ     M. Larch and J. Wanner even aggregating the non-participation gains of all countries with a positive effect leads to gains that are a very small fraction (2.4%) of the global gains of implementing the Paris Agreement. These welfare results seem to suggest that the free-riding problem in international climate cooperation (studied in detail by Nordhaus,2015) may be solvable with relatively minor transfer payments. At the same time, examples such as the US under Donald Trump demonstrate that aggregate national welfare losses from non-participation do not necessarily keep countries in the agreement. This could be due to decision-makers taking into account only the (more immediate) real income effects of mitigation policies and ignoring the (more long-run) gains from reduced climate damages. Additionally, one can contemplate a role for other political economy factors (e.g. related to distributional considerations, regionally concentrated job losses, and lobbying) in determining countries’ commitment to international climate change mitigation efforts (see e.g. Steckel and Jakob,2021, for an overview of the political economy findings in the coal context). 4.5. EU non-participation The European Union takes a special role in the Paris Agreement as all of its member countries are parties to the agreement individually, but at the same time, the EU is a party of its own to the treaty. Therefore, even though an EU non-participation decision would imply that a group of countries would drop out of the agreement, it can still be considered as a form of unilateral non-participation and we hence briefly consider its effects here.27 The total reduction loss of the EU leaving the Paris Agreement is 23.1% and hence very similar to the effect of Chinese non-participation (24.1%). However, this large harm to the agreement’s effectiveness stems primarily from a very large direct reduction loss of 18.5% from removing the ambitious EU reduction pledges. The endogenous component, on the other hand, is way smaller in the European than in the Chinese case with a leakage rate of only 5.6%, i.e. less than half of what we found for Chinese non-participation. While these numbers stress the importance of the EU as a large player in multilateral climate policy, they also indicate that its importance stems primarily from its potential to lead the way in terms of particularly ambitious reduction targets. 5. Model extension: Multiple fossil fuels The model developed in Section 2incorporated one single fossil fuel resource used in energy production and assumed emissions to be proportional to the fossil fuel usage. In this section, we allow for multiple fossil fuels with varying carbon intensities and potentially different supply elasticities. 5.1. Model We present the three model innovations in energy production, fossil fuel supply, and emission generation here and relegate details on the new model equilibrium to Appendix I. Fossil fuels used in energy production are now treated as a composite of different types of fossil fuels (specifically oil, gas, and coal): 𝐸𝑖=𝐴𝑖 𝐸(∏ 𝑣∈ (𝑅𝑖 𝑣)𝜌𝑖 𝑣)𝜉𝑖 𝑅∏ 𝑓∈ (𝑉𝑖 𝐸𝑓)𝜉𝑖 𝑓,(22) with ∑𝑣∈𝜌𝑖 𝑣=1. For each type of fossil fuel, supply is modeled with a separate CEFS function: 𝑅𝑊 𝑣=𝜁𝑣(𝑟𝑣 𝑃𝑣)𝜂𝑣 ,(23) with ∑𝑖∈𝑅𝑖 𝑣=𝑅𝑊 𝑣and 𝑃𝑣=∏𝑖∈(𝑃𝑖∕𝜔𝑖 𝑣)𝜔𝑖 𝑣. Fossil fuel types differ in their carbon intensity (𝜅𝑣). Hence, emissions are no longer simply proportional to 𝑅𝑖, but rather given by: 𝐸𝑀𝑖=∑ 𝑣∈ 𝜅𝑣𝑅𝑖 𝑣.(24) Emission taxes charged by Paris member countries to fulfill their reduction pledges take these carbon intensity differences into account and are e.g. hence higher in ad-valorem terms for coal than for gas, but equal across fuel types per ton of CO2. As in the base model, we can decompose the emission changes into scale, technique, and composition effects. Additionally, there is a substitution effect resulting from the change in the fossil fuel mix. See Appendices I.6 and I.8 for details on the decomposition and parametrization of the extended model, respectively. 27 Note that while all EU countries have the same reduction target of 55% below the 1990 emission level, this translates into different reductions compared to BAU. The standardized targets range from a mere commitment not to do worse than BAU in two Baltic countries (Estonia and Lithuania) to a very high 71% reduction target in Cyprus. (XURSHDQ (FRQRPLF 5HYLHZ     M. Larch and J. Wanner Fig. 9. Total Emission Reductions Lost (Model Extension) Notes: This figure shows the shares of the global emission reduction due to the Paris Agreement that is lost due to unilateral non-participation in the 140 different scenarios (in the extended model). On average, 1.2% of the global emission reductions are forgone. The loss shares range from 0.0% for a number of very small countries to 39.5% for the US. 5.2. Results Fig. 9 summarizes the most important results of the simulation of unilateral non-participation in the Paris Agreement in our extended model framework, namely the total percentage loss for the world emission reduction (i.e. it reproduces Fig. 7 from the main model results). Reassuringly, the overall pattern bears striking resemblance to our previous results. US non-participation still has by far the strongest effect (39.5%), followed by China (22.0%) and then a group of countries with effects between about 5 to 8% including e.g. Japan, Russia, Canada, Germany, and Brazil. On average, the incurred loss is slightly higher when additionally allowing for substitution between different fossil fuel sources (1.2 vs. 1.1%). The largest differences occur for Russia, whose non-participation is associated with a 1.7 percentage points higher reduction loss, and China, whose non-participation has a 2.2 percentage points weaker effect in the extended model. To gain a better insight into the differences in outcomes for the base and extended model, Fig. 10 displays the decomposition of the non-participating countries’ emission changes into scale, composition, technique, and substitution effect. As in the base model, the overall emission increases are primarily driven by the technique effects, i.e. generally more energy-intensive production. The new substitution effect in most cases additionally contributes to higher emissions in the non-committing countries. Hence, nonparticipating countries shift within their fossil fuel mix from relatively cleaner gas and oil to the most emission-intensive coal. This is because the price decrease on the international coal market is particularly strong as coal is the most heavily taxed fossil fuel in the committed countries. However, there are a few notable exceptions, like China, India, Kazakhstan, and Poland, where the substitution effect counteracts the overall emission increase. This only occurs in countries with a high coal share in the initial fossil fuel mix. For example, if China does not participate in the Paris Agreement, there will be a smaller price decrease on fossil fuels compared to a scenario in which all countries fulfill their targets due to a smaller drop in fossil fuel demand. As China has a coal-intensive energy mix, this drop is the smallest for coal. Hence, China substitutes coal with oil and gas, leading to a negative substitution effect.28 6. Conclusions Despite potential problems of enforceability and an overall lack of ambition in the NDCs, the Paris Agreement has an important strength: its global coverage. This strength, however, stands on shaky ground, as illustrated by not all signatory states moving forward to ratification of the agreement and by the (temporary) withdrawal of one of its major parties, namely the United States. In this paper, we analyze the consequences of unilateral non-participation in the Paris Agreement on the achieved global emission reduction. To be able to account for both the direct effect of removing the non-participating country’s reduction target and the indirect effect of additional emission reductions due to carbon leakage, we use an extended multi-sector structural gravity model featuring emissions from fossil fuel use, carbon taxes, and a constant elasticity fossil fuel supply function. We find that single countries not participating in the Paris Agreement can severely hurt the effectiveness of the treaty, the worst case being US non-participation which would eliminate more than one-third of the overall emission reduction. Taking into account the endogenous emission adjustments beyond the mere absence of an emission target turns out to be of major importance, notably in the Chinese case, in which the reduction loss doubles if carbon leakage is added to the direct effect. Using a decomposition of 28 This relationship between the coal share and the substitution effect is illustrated in Figure I.1 in Appendix I. (XURSHDQ (FRQRPLF 5HYLHZ     M. Larch and J. Wanner Fig. 10. Decomposition of Emission Changes (Model Extension) Notes: This figure plots the decomposition of the emission changes into scale, composition, technique, and substitution effect for the 25 countries with the biggest reduction effect on world emissions and the rest of the world composite. emission changes into scale, composition, and technique effects, we find that emission increases in non-participating countries are mainly driven by a shift towards emission-intensive production techniques in response to a fall in the international fossil fuel price. Both the overall magnitude of the reduction losses and the relative importance of the different leakage channels have significant policy implications. Most importantly, our findings imply that global coverage is indeed crucial for the overall mitigation success of the agreement and therefore strong political efforts should be made to keep all large emitters on board. Further, if the global coverage breaks down, our findings on the strong energy market leakage channel suggest considering new climate policy instruments that specifically tackle the fossil fuel supply. Adding supply-side climate policies at the same time may have the potential to shift the incidence of climate rents and hence help avoid some countries losing from the implementation of the Paris Agreement, namely the major fossil fuel supply countries. Generally, however, the overall welfare effects of the agreement are overwhelmingly positive, and unilateral incentives for non-participation are weak. Further insights into the (political economy) drivers behind countries nevertheless questioning their mitigation efforts are needed in order to be able to design and evaluate policies that both effectively reduce carbon emissions and at the same time avoid feasibility pitfalls. Appendix A. Supplementary data Supplementary material related to this article can be found online at https://doi.org/10.1016/j.euroecorev.2024.104699. References Aguiar, Angel, Chepeliev, Maksym, Corong, Erwin, McDougall, Robert, Van der Mensbrugghe, Dominique, 2019. The GTAP data base: Version 10. J. Glob. Econ. Anal. 4 (1), 1–27. Aichele, Rahel, Felbermayr, Gabriel, 2012. Kyoto and the carbon footprint of nations. J. Environ. Econ. Manag. 63 (3), 336–354. Aichele, Rahel, Felbermayr, Gabriel, 2015. Kyoto and carbon leakage: An empirical analysis of the carbon content of bilateral trade. Rev. Econ. Stat. 97 (1), 104–115. Aldy, Joseph E., Pizer, William A., 2016. Alternative metrics for comparing domestic climate change mitigation efforts and the emerging international climate policy architecture. Rev. Environ. Econ. Policy 10 (1), 3–24. Aldy, Joseph E., Pizer, William A., Akimoto, Keigo, 2017. Comparing emissions mitigation efforts across countries. Climate Policy 17 (4), 501–515. Asheim, Geir B., Fæhn, Taran, Nyborg, Karine, Greaker, Mads, Hagem, Cathrine, Harstad, Bard, Hoel, Michael O., Lund, Diderik, Rosendahl, Knut E., 2019. The case for a supply-side climate treaty. Science 365 (6451), 325–327. Boeters, Stefan, Bollen, Johannes, 2012. Fossil fuel supply, leakage and the effectiveness of border measures in climate policy. Energy Econ. 34 (Supplement 2), S181–S189. Böhringer, Christoph, Balistreri, Edward J., Rutherford, Thomas, 2012. The role of border carbon adjustment in unilateral climate policy: Overview of an energy modeling forum study (EMF 29). Energy Econ. 34 (Supplement 2), S97–S110. Böhringer, Christoph, Fischer, Carolyn, Rosendahl, Knut Einar, Rutherford, Thomas, 2022. Potential impacts and challenges of border carbon adjustments. Nat. Climate Change 12 (1), 22–29. Böhringer, Christoph, Rutherford, Thomas, 2017. Paris after trump: An inconvenient insight. Oldenburg Discussion Papers in Economics, 400–17. Caron, Justin, Fally, Thibault, 2022. Per capita income, consumption patterns, and CO2 emissions. J. Assoc. Environ. Resour. Econ. 9 (2), 235–271. Copeland, Brian R., Taylor, M. Scott, 1994. North-south trade and the environment. Q. J. Econ. 109 (3), 755–787. Copeland, Brian R., Taylor, M. Scott, 2003. Trade and the Environment. Theory and Evidence. Princeton University Press, Princeton, NJ. (XURSHDQ (FRQRPLF 5HYLHZ     M. Larch and J. Wanner Costinot, Arnaud, Rodríguez-Clare, Andrés, 2014. Trade theory with numbers: Quantifying the consequences of globalization. In: Gopinath, Gita, Helpman, Elhanan, Rogoff, Kenneth (Eds.), Handbook of International Economics, fourth ed. vol. 4, North Holland, pp. 197–261. Dekle, Robert, Eaton, Jonathan, Kortum, Samuel, 2007. Unbalanced trade. Am. Econ. Rev.: Pap. Proc. 97 (2), 351–355. Dekle, Robert, Eaton, Jonathan, Kortum, Samuel, 2008. Global rebalancing with gravity: Measuring the burden of adjustment. IMF Staff Papers, 55, (3), pp. 511–540. Dingel, Jonathan I., Tintelnot, Felix, 2021. Spatial economics for granular settings. NBER Working Paper, 27287. Eaton, Jonathan, Kortum, Samuel, 2002. Technology, geography, and trade. Econometrica 70 (5), 1741–1779. Egger, Peter, Larch, Mario, 2008. Interdependent preferential trade agreement memberships: An empirical analysis. J. Int. Econ. 76 (2), 384–399. Egger, Peter, Nigai, Sergey, 2015. Energy demand and trade in general equilibrium. Environ. Resour. Econ. 60 (2), 191–213. Fally, Thibault, Sayre, James, 2018. Commodity trade matters. NBER Working Paper, 24965. Farrokhi, Farid, 2020. Global sourcing in oil markets. J. Int. Econ. 125, 103323. Farrokhi, Farid, Lashkaripour, Ahmad, 2021. Can trade policy mitigate climate change? Unpublished Working Paper. Feenstra, Robert C., 2004. Advanced International Trade: Theory and Evidence. Princeton University Press, Princeton, NJ. Felder, Stefan, Rutherford, Thomas, 1993. Unilateral CO2 reductions and carbon leakage: The consequences of international trade in oil and basic materials. J. Environ. Econ. Manag. 25 (2), 162–176. Glanemann, Nicole, Willner, Sven N., Levermann, Anders, 2020. Paris climate agreement passes the cost-benefit test. Nature Commun. 11 (1), 1–11. Grossman, Gene M., Krueger, Alan B., 1993. Environmental impacts of a North American free trade agreement. In: Garber, Peter M. (Ed.), The U.S.-Mexico Free Trade Agreement. MIT Press, Cambridge, MA, pp. 13–56. Harstad, Bard, 2012. Buy coal! A case for supply-side environmental policy. J. Polit. Econ. 120 (1), 77–116. Head, Keith, Mayer, Thierry, 2014. Gravity equations: Workhorse, toolkit, and cookbook. In: Gopinath, Gita, Helpman, Elhanan, Rogoff, Kenneth (Eds.), Handbook of International Economics, fourth ed. vol. 4, North Holland, pp. 131–195. Head, Keith, Mayer, Thierry, Ries, John, 2010. The erosion of colonial trade linkages after independence. J. Int. Econ. 81 (1), 1–14. Iyer, Gokul, Calvin, Katherine, Clarke, Leon, Edmonds, James, Hultman, Nathan, Hartin, Corinne, McJeon, Haewon, Aldy, Joseph, Pizer, William, 2018. Implications of sustainable development considerations for comparability across nationally determined contributions. Nature Clim. Change 8 (2), 124–129. Jensen, Svenn, Mohlin, Kristina, Pittel, Karen, Sterner, Thomas, 2015. An introduction to the green paradox: The unintended consequences of climate policies. Rev. Environ. Econ. Policy 9 (2), 246–265. Kalkuhl, Matthias, Brecha, Robert J., 2013. The carbon rent economics of climate policy. Energy Econ. 39, 89–99. Kalkuhl, Matthias, Wenz, Leonie, 2020. The impact of climate conditions on economic production. Evidence from a global panel of regions. J. Environ. Econ. Manag. 103, 102360. Kemp, Luke, 2017. US-proofing the Paris climate agreement. Climate Policy 17 (1), 86–101. Kortum, Samuel, Weisbach, David, 2021. Optimal unilateral carbon policy. Cowles Foundation Discussion Paper, 2659. Larch, Mario, Löning, Markus, Wanner, Joschka, 2018. Can degrowth overcome the leakage problem of unilateral climate policy? Ecol. Econom. 152, 118–130. Larch, Mario, Wanner, Joschka, 2017. Carbon tariffs: An analysis of the trade, welfare, and emission effects. J. Int. Econ. 109, 195–213. NGFS, 2022. Technical documentation to the NGFS Scenario. Tech. rep., Network for Greening the Financial System, Paris, France. Nong, Duy, Siriwardana, Mahinda, 2018. Effects on the U.S. economy of its proposed withdrawal from the Paris agreement: A quantitative assessment. Energy 159, 621–629. Nordhaus, William D., 2015. Climate clubs: Overcoming free-riding in international climate policy. Amer. Econ. Rev. 105 (4), 1339–1370. Ossa, Ralph, 2016. Quantitative models of commercial policy. In: Bagwell, K., Staiger, Robert W. (Eds.), Handbook of Commercial Policy. vol. 1, North-Holland, pp. 207–259. Pothen, Frank, Hübler, Michael, 2018. The interaction of climate and trade policy. Eur. Econ. Rev. 107, 1–26. Rennert, Kevin, Errickson, Frank, Prest, Brian C., Rennels, Lisa, Newell, Richard G., Pizer, William, Kingdon, Cora, Wingenroth, Jordan, Cooke, Roger, Parthum, Bryan, Smith, David, Cromar, Kevin, Diaz, Delavane, Moore, Frances C., Müller, Ulrich K., Plevin, Richard J., Raftery, Adrian E., Sevcikova, Hana, Sheets, Hannah, Stock, James H., Tan, Tammy, Watson, Mark, Wong, Tony E., Anthoff, David, 2022. Comprehensive evidence implies a higher social cost of CO2. Nature 610 (7933), 687–692. Richter, Philipp M., Mendelevitch, Roman, Jotzo, Frank, 2018. Coal taxes as supply-side climate policy: A rationale for major exporters? Clim. Change 150, 43–56. Rogelj, Joeri, den Elzen, Michel, Höhne, Niklas, Fransen, Taryn, Fekete, Hanna, Winkler, Harald, Schaeffer, Roberto, Sha, Fu, Riahi, Keywan, Meinshausen, Malte, 2016. Paris agreement climate proposals need a boost to keep warming well below 2 degrees celsius. Nature 534 (7609), 631–639. Rose, Adam, Wei, Dan, Miller, Noah, Vandyck, Toon, Flachsland, Christian, 2018. Policy brief - achieving Paris climate agreement pledges: Alternative designs for linking emissions trading systems. Rev. Environ. Econ. Policy 12 (1), 170–182. Santos Silva, João M.C., Tenreyro, Silvana, 2006. The log of gravity. Rev. Econ. Stat. 88 (4), 641–658. Shapiro, Joseph S., 2016. Trade costs, CO2, and the environment. Am. Econ. J.: Econ. Policy 8 (4), 220–254. Shapiro, Joseph S., 2021. The environmental bias of trade policy. Q. J. Econ. 136 (2), 831–886. Shapiro, Joseph S., Walker, Reed, 2018. Why is pollution from U.S. manufacturing declining? The roles of environmental regulation, productivity, and trade. Amer. Econ. Rev. 108 (12), 3814–3854. Sinn, Hans-Werner, 2008. Public policies against global warming: A supply side approach. Int. Tax Public Finance 15 (4), 360–394. Steckel, Jan C., Jakob, Michael, 2021. The political economy of coal: Lessons learnt from 15 country case studies. World Dev. Perspect. 24, 100368. Weisbach, David, Kortum, Samuel, Wang, Michael, Yao, Yujia, 2022. Trade, leakage, and the design of a carbon tax. NBER Working Paper, 30244. Winchester, Niven, 2018. Can tariffs be used to enforce Paris climate commitments? World Econ. 41 (10), 2650–2668.