Working Paper No 2013-34 – October Border Carbon Ajustment in Europe and Trade Retaliation: What would be the Cost for European Union? Jean Fouré, Houssein Guimbard & Stéphanie Monjon Highlights A border carbon adjustment in the European Union would imply export losses in the first year of implementation, which would mainly affect the USA (2 billion dollars), Russia (1.8 billion), and China (1 billion). Although small at the macroeconomic scale, these export losses would certainly lead to disputes at the World Trade Organization (WTO). The possible ensuing trade retaliation would entail export losses for the EU focused on agricultural goods, of an order of magnitude of 6 billion dollars, as soon as retaliation begins. Energy intensive and trade exposed sectors in the EU would increase production due to the BCA, compensating for around 20% of the drop due to carbon pricing, at the expense of production by partners. This effect would be magnified in the case of trade retaliation. Other European sectors would be penalized by both the BCA and trade retaliation; whereas consumers' real income would not be significantly impacted, making a BCA more dependent on the political weight of the sectors concerned. CEPII Working Paper Border carbon adjustment in Europe and trade retaliation Abstract Unilateral climate policy, such as carbon pricing, represents an additional cost to the economy, especially to energy- intensive industrial sectors, as well as those exposed to international competition. A border carbon adjustment (BCA) is often presented as an attractive policy option for countries that want to go ahead without waiting for a global climate agreement. We used the computable general equilibrium model MIRAGE-e to simulate the impact of the introduction of a BCA on imports of energy intensive products in EU and EFTA countries and to evaluate the export losses their main trade partners would suffer. Given that a BCA is a trade measure, it would certainly lead to disputes at the World Trade Organization (WTO). If the BCA is considered illegal, the losses suffered by some partners may justify retaliation, as authorized by a WTO dispute settlement. The overall aggregated impacts of these measures would be negative but marginal, meaning that neither the BCA nor trade retaliation would have a marked impact on consumers' real income or GDP, while prohibitive retaliatory tariffs are more likely to target sensitive products in the EU. A BCA would ultimately be a signal of the EU's willingness to maintain an ambitious climate policy. Keywords Emission trading scheme, border carbon adjustment, trade retaliation. JEL D58, F18, Q56. CEPII Working Paper Border carbon adjustment in Europe and trade retaliation Border carbon adjustment in Europe and trade retaliation: What would be the cost for European Union? 1 Jean Fouré∗ Houssein Guimbard† Stéphanie Monjon‡ 1. Introduction Effective and efficient climate protection ultimately requires cooperative action across all major greenhouse gas (GHG) emitting countries. But the agreement reached in Copenhagen in 2009 has led to a fragmented world in which each country has unilaterally decided (or not) to reduce its GHG emissions by 2020, as well as the intensity of the reduction. The current climate regime groupssomecountrieswhoaretryingtomoveforward, otherswhoseektoreducetheiremissions but without really departing from “business-as-usual” and others who have no commitment to emission reduction. In this fragmented climate regime, a border carbon adjustment (BCA) is sometimes seen as an appealing policy option for countries who intend to implement more stringent unilateral policies and worry about possible carbon leakage, which would jeopardize the effectiveness of their action (Krugman, 2009; Helm et al., 2012).2 The aim of a BCA is basically to level the playing field by imposing a similar constraint on the GHG emissions of the imported and domestically produced goods. Applied as a complementary instrument to binding domestic emission pricing, a BCA can reduce carbon leakage and increase global cost effectiveness. Focusing on the Kyoto protocol and assuming countervailing carbon levies for all sectors in countries who are reducing their GHG emissions, Babiker and Rutherford (2005) found that BCA can substantially reduce the real income losses for the coalition mem- bers by shifting a great part of the carbon policy burden to non-coalition members. A recent 1We thank Nicolas Lagarde for his participation in early developments of this paper. We also acknowledge helpful advicefromLionelFontagné,SébastienJean,theFrenchMinistryofFinanceandtheparticipantstothe15thAnnual Conference on Global Economic Analysis and to the 3rd Association of Environmental and Resource Enconomists Summer Conference. All errors remain ours. ∗CEPII ([email protected]) †CEPII ([email protected]) ‡PSL*, University Paris-Dauphine (LEDa-CGEMP), CEPII and CIRED ([email protected]) 2A unilateral climate policy in a country can increase emissions (i.e. carbon leakage) in non-abating (or less acting) countries. Indeed, the climate policy reduces the consumption of some fossil fuels in the abating country, and hence their price. This encourages the rest of the world to consume more fossil fuels as their price has gone down. Another source of carbon leakage is possible loss of market shares by domestic firms in favor of foreign firms,duetotheimposedcarbonconstraint,orthemigrationofindustrialsitestowardsregionswithalessstringent climatepolicy. ABCAcanonlylimitthecarbonleakageofthelattertype. Carbonleakageimpairstheeffectiveness of a climate policy. 3 CEPII Working Paper Border carbon adjustment in Europe and trade retaliation initiative by the Energy Modelling Forum, an international expert platform for the discussion of energy and environmental problems, investigated the efficiency and distributional impacts of BCA as a complementary instrument to domestic climate policy and reached the same conclu- sions (Böhringer et al., 2012b).3 Thus, while border carbon adjustments do have some appeal based on their implications for carbon leakage and global cost effectiveness, they also shift a larger part of the economic abatement from regions who implement the BCA to other regions, including developing countries. This loss of agents’ real income might create an incentive for non-acting (or less acting) countries to price carbon,4 but also to implement trade retaliation. As a trade measure, a BCA may be contested by a World Trade Organization (WTO) member under its dispute settlement mechanism. Even if several recent analyses, including a report from WTO-UNEP (2009), conclude that theoretically it is possible to design a WTO-compatible border adjustment (Ismer and Neuhoff, 2007; Pauwelyn, 2007; Eichenberg, 2010), the risk of a WTO dispute arising over climate related trade measures is high (Werksman et al., 2009; Holmes et al., 2011). Somedevelopingcountrieshavealreadyexpressedtheirconcernsabouttheuseoftrademeasures for carbon motives and affirmed their determination to fight all initiatives of this kind. In 2010, the Indian Environment Minister Jairam Rames stated that “India [would] bring a WTO challenge against any ’carbon taxes’ that rich countries impose on Indian imports” (ICTSD, 2010). Before the international conference on the climate in Doha in December 2012, with China’ssupport, Indiapre-emptivelyaskedtheUnitedNationstotableaproposaltobanclimate- related protectionist measures, including border taxes (TWN, 2012). The conflict opposing the EU and a broad coalition of countries, including Russia, India, the United States, and China, on the inclusion in the European Emission Trading Scheme (EU ETS)5 of all the airlines operating in the EU perfectly illustrates the risk of trade war (Reuters, 2011). While the economic impacts of border measures have already been addressed, to our knowledge no study has assessed the implications of trade retaliation for a region that implements a BCA for climate reasons. A country that wishes to use this instrument needs to assess this risk. Our objective is thus to compare possible benefits of implementing a BCA for a coalition and the costs it would have to bear if its main trade partners decided to apply retaliatory trade sanctions. This question is interesting for the more general issue of the use of trade measures as complementary policies to climate policies. We focus on the EU because this region is among the most ambitious in terms of emission reduction, and we limit the implementation of the BCA 3Forinstance,Böhringeretal.(2012a)conclude: “RegardingleakagereductionandthealleviationofadverseEITE output effects, we find that border carbon adjustments are by far the most effective instrument since they directly level the playing field between regulated domestic EITE production and unregulated EITE production abroad. Border carbon adjustments can provide non-negligible global cost savings.” 4Barrett(2011)considersthatthetwoprimarymotivesforusingtraderestrictionsarelimitingcarbonleakageand free riding. 5Since 2005, an ETS caps the GHG emissions generated by heavy and energy industries. This instrument covers more than 10,000 industrial production units in the EU, about 40% of Member States’ emissions. 4 CEPII Working Paper Border carbon adjustment in Europe and trade retaliation to the energy-intensive and trade-exposed (EITE) sectors covered by the EU ETS. The analysis was based on the use of the MIRAGE-e model, the energy-oriented version of MIRAGE, CEPII’s global computable general equilibrium (CGE) model. The first scenario rep- resents commitments concerning reductions in GHG emissions. It models an increasing con- straint on GHG emissions by highly emissive industrial sectors in the EU up to 2020 based on the decision taken by the European Parliament and Council in 2009, including the integration of the aviation sector from 2012 (European Commission, 2009). The emission reductions in the non-ETS sectors and in the rest of the world are consistent with the pledges taken during the 15th Conference of Parties of the UNFCCC held in Copenhagen in 2009 and confirmed in the Durban Agreement the following year (Dellink et al., 2011). The second scenario encom- passes the implementation of a BCA to imports of EITE goods covered by the EU ETS. Like EU producers, foreign producers must surrender allowances equal to the emissions linked to the production (but not to transport) of the imported products. The third scenario assumes that the EU’s main trade partners bring a WTO challenge against the BCA and apply trade retaliation compatible with the WTO framework, here modeled as an increase in custom duties. We found that even if a BCA were applied only at the level of EITE sectors, it could lead to export losses for the EU’s trade partners, to an extent that is comparable with the amounts already identified by the WTO dispute settlement as a justification for retaliation. The overall impact on European exports of both border adjustment and trade retaliation would then be slightly negative in the majority of the sectors examined. These trade measures would limit the drop in production in the EITE sectors, but at the expense of the other sectors. Nevertheless, neither the BCA nor retaliation would have sizeable impacts on real income or GDP in the EU or on the retaliators, while leading to a small decrease in global emissions. Section 2 explains the methodology used and the policy simulations. Results are discussed in Section 3. Section 4 concludes. 2. Methodology 2.1. The MIRAGE-e model MIRAGE-e (Fontagné et al., 2013) is adapted from MIRAGE, the multi-sectoral and multi- regional CGE model (Bchir et al., 2002; Decreux and Valin, 2007), which was developed for, and has been extensively used to assess trade liberalization and agricultural policy scenarios. Based on a sequential dynamic recursive set up, MIRAGE is used to evaluate a future for the world economy, using exogenous projections for macroeconomic variables like GDP or labor force. MIRAGE-e offers improved energy modeling whose dynamics rely on a new baseline built following the MaGE model (Macroeconometrics of the Global Economy, Fouré et al., 2013). Each sector is modeled as a representative firm, exhibiting constant returns to scale, whose production function combines value added and intermediate consumption in fixed shares. Value 5 CEPII Working Paper Border carbon adjustment in Europe and trade retaliation added comprises imperfectly substitutable production factors: capital, skilled labor, unskilled labor, land, natural resources, and energy. Allprimaryfactorendowments(labor, land, naturalresourcesandcapital)areassumedtobefully employed and their regional variations are exogenously taken from MaGE projections. Installed capital stock is assumed to be immobile (sector specific), while investment, which represents the longtermpossibilityofacapitalmarketadjustment, isallocatedacrosssectors(perfectmobility) according to their rate of return to capital. Skilled labor is perfectly mobile across sectors, while unskilled labor is imperfectly mobile between agricultural and other sectors. Land is assumed to be imperfectly mobile between agricultural sectors, and natural resources are sector-specific. Consumption of the five energy goods by firms is aggregated in a single bundle which mainly substitutes with capital. This energy aggregate is subject to specific improvements in produc- tivity resulting from the growth model MaGE. Among energy producing sectors, substitutions between energies are allowed only for the generation of electricity. Moreover, energy productivity does not improve in other energy sectors. 2.2. Data and aggregation of regions and sectors For this paper, MIRAGE-e was calibrated with data from the GTAP-7 database for the year 2004 (Narayanan and Walmsley, 2008). The database contains integrated data on bilateral trade flows and input-output matrixes for 57 sectors and 113 countries and regions, covering the whole world. Two other GTAP databases that are consistent with the previous one were also used: energy consumption and CO emissions. 2 Table 1 shows how GTAP sectors and regions are aggregated in this paper. The geographical aggregation used is precise enough to represent the emission reduction commitments made in Copenhagen in 2009. Designing an aggregation that perfectly isolates the sectors concerned by the EU ETS is not an easy task. The data sources (43 goods sectors in GTAP, corresponding e.g. to around 15,000 products in the Integrated Tariff of the European Communities – TARIC – classification) make it difficult to achieve perfect correspondence. For instance, the sector “Paper products, publishing” includes some activities whose emissions due to the fossil fuel combustion are not subject to the EU ETS, publishing, for example. Being aware of this limitation, we identified seven GTAP sectors that could correspond to the EU ETS. Except for electricity, all these EU ETS sectors are considered as EITE. MIRAGE-e also distinguishes three additional energy sectors (coal, oil and gas). The other sectors are aggregated in six large sectors: an extraction sector, two agricultural sectors, the rest of industry, and two service sectors. 6 CEPII Working Paper Border carbon adjustment in Europe and trade retaliation Table 1 – Overview of regions and sectors Regions Sectors European Union (EU27) EU ETS sectors European Free Trade Association (EFTA) Paper products, publishing (ppp) United States of America (USA) Petroleum, coal products (p_c) Canada Chemical, rubber, plastic products (crp) Japan Mineral products nce (nmm) Australia and New Zealand (ANZ) Ferrous metals (i_s) Russian Federation Metals nce (nfm) Rest of Europe Electricity (ely) Brazil Energy sectors China and Hong-Kong (ChinaHK) Coal (coa) South Africa Oil (oil) India Gas (gas), gas distribution (gdt) Oil producing countriesa Other sectors Other Latin America Cropsc Other South-east Asia Livestockd Least Developed Countries (LDC)b Minerals nce (omn) Rest of the World Rest of Industry Transporte Servicesf Note: ’nce’ means ’not classified elsewhere’ aVenezuela,RestofWesternAsia,IslamicrepublicofIran,RestofNorthAfrica,Nigeria bCambodia, Lao People’s Democratic Republic, Indonesia,Rest of Southeast Asia, Bangladesh, Rest of South Asia, Senegal, Rest of Western Africa, Central Africa, South-Central Africa, Ethiopia, Madagascar, Malawi,Mozambique,Tanzania,Uganda,Zambia,RestofEasternAfrica. cPaddy rice (pdr), Wheat (wht), Cereal grains nce (gro), Vegetables, fruit, nuts (v_f), Oil seeds (osd), Sugarcane,sugarbeet(c_b),Plant-basedfibers(pfb),Cropsnce(ocr). dCattle,sheep,goats,horses(ctl),Animalproductsnce(oap),Rawmilk(rmk),Wool,silk-wormcocoons (wol),Forestry(frs),Fishing(fsh). eTransportnce(otp),Seatransport(wtp),Airtransport(atp). fWater(wtr),Construction(cns),Trade(trd),Communication(cmn),Financialservicesnce(ofi),Insur- ance(isr),Businessservicesnce(obs),Recreationandotherservices(ros),PublicAdministration,Defence, Health,Education(osg),Dwellings(dwe). CEPII Working Paper Border carbon adjustment in Europe and trade retaliation 2.3. The scenarios All counterfactual analyses depart from a scenario with no climate policy called a “Business-As- Usual” (BAU)scenario. Theaimofthisreferencescenarioistobuildaframeworkforcomparison with the simulations of climate policies. The scenario called “ETS” is a stylized version of the emission reduction commitments taken in Copenhagen in 2009 as well as the decisions taken by the EU. The “BCA” scenario differs from the ETS scenario by assuming the implementation of a BCA to imports of goods whose production in the EU is covered by the EU ETS. The “TR” scenario is similar to the BCA scenario but assumes that some trade partners of the EU who are negatively impacted by the BCA will implement trade retaliation. In the BAU scenario, GDPs are set exogenously from the baseline growth trajectory produced with the MaGE model. Thus, total factor productivity adjusts such that GDP growth rates in MIRAGE-e match the baseline growth trajectory, under the additional assumptions on sector- specific differentials detailed in Fontagné et al. (2013).6 In this reference scenario, prices of fossil energies (coal, oil and gas) come from IEA (2011) projections. Local natural resources endogenously adjust to match these fossil fuel price targets, withalinearsmoothingbetween2005and2014toovercomepricevolatility, whichwouldprevent the model from solving. Contrary to the BAU scenario, GDPs become endogenous in the simulations with climate poli- cies, while the total factor productivities are equal to the values computed in the reference scenario. In addition, world primary energy prices become endogenous, while the corresponding reserves are assumed to remain at their reference level. 2.3.1. The ETS Scenario The EU sets its GHG emission target for 2020 at 20% below 1990 levels, which represents a 14% reduction in emissions between 2005 and 2020. This is divided into a reduction of 21% in the ETS sectors and a reduction of 10% in the non-ETS sectors and households.7 Norway, Liechtenstein and Iceland have implemented an ETS and have linked it to the EU ETS. In MIRAGE-e, EFTA is considered as a single region. We assume that this region links its ETS to the EU ETS from 2008 and applies the same reductions in emissions as in the EU both in the ETS and non-ETS sectors. The EU ETS gives some flexibility to firms covered by the scheme. Thus, the use of credits from Clean Development Mechanism (CDM) and Joint Implementation (JI) is allowed in the EU ETS, 6Namely, Crops and Livestock productivities are exogenous, while a 2 p.p. growth gap is maintained between the manufacturing and services sectors. 7Even if the model only covers CO2-energy emissions, we use the objectives in terms of GHG emissions. 8 CEPII Working Paper Border carbon adjustment in Europe and trade retaliation but only up to the limits proposed in the Directive 2009/29/EC (Article 11a).8 In MIRAGE-e, the use of international credits is modeled as a loosening of the constraint on emissions, through the availability of a greater number of allowances to the ETS sectors.9 After 2012, we assume that the aviation sector is included in the EU ETS.10 However, this sector is not represented explicitly as an ETS sector in MIRAGE-e. Firstly, data on aviation do not allow identification of routes that would be subject to ETS (i.e. departing from or arriving in the EU). Secondly, while the airlines can buy allowances created for the industry and energy sectors, the reverse is not possible. According to Boon et al. (2007), airlines will be at the origin of a high demand for industrial allowances. We account for this by reducing the number of allowances available in the scheme. The number of permits that are removed comes from Boon et al. (2007), corrected by the quantity of CDM and JI credits that can be used by airlines (11%). Since the beginning of the EU ETS, allowances have been allocated freely to industries on a lump-sum base and continue to be free for sectors exposed to the risk of carbon leakage during the third period of the scheme (2013-2020). We depart from this decision by assuming that from 2013 onwards, all the allowances are auctioned. In a CGE model with perfect competition and a single consumer-government agent, auctioned permits and free allocations are equivalent, as long as the number of permits received by firms is independent of their production.11 Technically speaking, the cap and trade system is implemented in the model by exogenously imposing the level of CO emissions in the EU and EFTA (one for ETS sectors, one for non- 2 ETS sectors plus households). As a result, two carbon prices emerge, matching the supply and demand for carbon allowances. This modeling assumption can be interpreted in two ways. First, for the ETS sectors, the corresponding price reflects the equilibrium price on the allowances market under perfect competition. Second, for non-ETS sectors and households, the price outcome corresponds to the level of carbon tax needed to achieve the reduction objective. For other countries, we assume that the emission reduction pledges made in Copenhagen are applied.12 Unlike under the Kyoto Protocol, the way countries report the emissions they plan to avoid is not standardized. Consequently, there is a variety of different pledges, in which designs (absolute/relative target, reference year etc.) differ considerably between countries. 8“Allexistingoperatorsshallbeallowedtousecreditsduringtheperiodfrom2008to2020uptoeithertheamount allowed to them during the period from 2008 to 2012, or to an amount corresponding to a percentage, which shall not be set below 11%, of their allocation during the period from 2008 to 2012, whichever is the highest.” (Article 11a (8)) 9This implies that the price of the EU allowances and of the international credits is the same. 10In fact, at the end of 2012, the EU decided to delay the inclusion of foreign airlines by a year to allow time to negotiate with its partners in the framework of the International Civil Aviation Organization. 11With auctioning, the money that is collected increases the government’s revenues. On the other hand, a grand- fathered free allocation results in the redistribution of the collected revenue to the shareholders (under perfect competition), i.e. the consumer, who is the same agent as the government. 12For a detailed list of the pledges, see for instance Stern and Taylor (2010). 9 CEPII Working Paper Border carbon adjustment in Europe and trade retaliation Some pledges also correspond to an interval rather than to a single figure. For instance, China will endeavor to lower its CO emissions per unit of GDP by 40-45% by 2020 compared to the 2 2005 level, suggesting low and high pledges. We only consider low pledges. Unlike EU targets, some countries have made commitments in terms of CO intensity of GDP, or relative to the projected emissions in a BAU reference 2 (see Table A.2). We transform their pledges into absolute targets using our BAU scenario.13 Table A.2 lists countries’ commitments, translated into our geographical aggregation. 2.3.2. The BCA Scenario There are different ways of defining a BCA (Monjon and Quirion, 2010): it can target imports and exports or only imports; if imposed only on imports, a BCA can be a tax, or an obligation to surrender allowances; it can cover direct and indirect emissions, or only direct emissions, etc. In the model, we assume that the BCA targets European imports of EITE goods, whatever their origin, from 2013 onwards. The measure obliges importers to surrender a quantity of allowances corresponding to direct CO2 emissions generated during the production of the imported goods (thus excluding emissions underlain by electricity consumption). This differs from the more common assumption that importers have to pay a carbon tax (Böhringer et al., 2012b). With a “price-based” BCA, there is no direct impact on the allowance market, unlike an “allowance- based” BCA under which importers are committed to buy some allowances and therefore influ- ence their price.14 This modeling choice implies adjusting the supply of allowances: the emission reduction objective in 2020 was applied to the sum of emissions generated by the sectors ETS in the EU and by products that are imported and consumed in the EU. The BCA makes it possible to target consumption-based emissions rather than production-based emissions. The other elements of the BCA are defined in a more standard way (Manders and Veenendaal, 2008; Kuik and Hofkes, 2010; Böhringer et al., 2012b). The BCA applies only to imports of EITE goods. The evaluation of the number of allowances to surrender is based on the carbon content of imported products and varies with the country of origin. Finally, only direct emissions are taken into account.15 Considering a BCA applied irrespective of the origin of the imports and at the same time assuming that emission reduction pledges taken in Copenhagen are applied means some CO2 emissions have to be charged twice: once in the county that exports a CO2-emissive product to the EU or EFTA and once in the EU or EFTA. This double taxation may play against the BCA if a dispute related to it is raised at the WTO. 13In the BAU reference, India’s emissions are under the cap corresponding to its commitment. 14For a comparison of a “tax-based” BCA and an “allowance-based” BCA, see Monjon and Quirion (2011). 15Böhringer et al. (2012b) also include indirect emissions. 10