JID: FMRE

ARTICLE IN PRESS [m5GeSdc;April 13, 2023;18:55]

Fundamental Research xxx (xxxx) xxx

Contents lists available at ScienceDirect

Fundamental Research
journal homepage: http://www.keaipublishing.com/en/journals/fundamental-research/

Article

The carbon border adjustment mechanism is inefficient in addressing

carbon leakage and results in unfair welfare losses
Xinlu Sun a, Zhifu Mi a,∗, Lu Cheng a, D’Maris Coffman a, Yu Liu b
a
The Bartlett School of Sustainable Construction, University College London, London WC1E 7HB, United Kingdom
b
College of Urban and Environmental Sciences, Peking University, Beijing 100871, China

a r t i c l e i n f o a b s t r a c t

Article history: The European Commission has proposed a Carbon Border Adjustment Mechanism (CBAM) to reduce carbon leak-
Received 12 September 2022 age and create a level playing field for its domestic products and imported goods. Nevertheless, the effectiveness
Received in revised form 13 January 2023 of the proposal remains unclear, especially when it triggers threats of retaliation from trading partners of the
Accepted 21 February 2023
European Union (EU). We apply a Computable General Equilibrium (CGE) model - Global Trade Analysis Project
Available online xxx
(GTAP) - to assess the economic and environmental impacts of different CBAM schemes. Here we show that the
effectiveness of the CBAM to address carbon leakage risks is rather limited, and the CBAM raises concerns over
Keywords:
global welfare costs, GDP losses, and violation of equality principles. Trade retaliation leads to multiplied welfare
Carbon border adjustment mechanism
Carbon leakage losses, which would mostly be borne by poor countries. Our results question the carbon leakage reduction effect
Climate change mitigation of a unilateral trade policy and suggest that climate change mitigation still needs to be performed within the
Floor carbon price framework of international cooperation.
Trade retaliation

1. Introduction countries will undermine EU and global decarbonization efforts. To deal

with carbon leakage risks, the European Commission intends to impose
The aim of a Carbon Border Adjustment Mechanism (CBAM) is to a carbon levy on imports from non-EU regions or to include imports
equalise the price of carbon between domestic products and imports in the EU ETS. Such a measure would see imported goods be hit with
with the objective of addressing the longstanding problem of carbon an environmental cost equivalent to that imposed when produced in
leakage. As global society is urgently demanding a green recovery from the EU and thus protect the competitiveness of the EU’s products. The
the COVID-19 pandemic, the European Commission published its pro- CBAM is expected not only to help protect the abatement achievements
posal to introduce the CBAM, which is part of the “Fit for 55 package” of the EU’s efforts but also to precipitate the climate actions of non-EU
to cut emissions by at least 55% by 2030. In Dec 2022, the European countries [4,5].
Council and Parliament reached a provisional deal on the CBAM. The However, the potential of the CBAM to reduce global carbon leakage
CBAM is designed to function in parallel with the EU Emission Trading remains unclear, as its effectiveness is affected by various internal and
System (EU ETS) and to prevent carbon leakage. These goals are in ac- external factors. The internal factors are related to policy design, includ-
cord with the ambition of the European Union (EU) as one of the main ing the scope of emissions, carbon price level, carbon intensity measure-
leaders in climate change mitigation in the world. ment and revenue usage. Firstly, the scope of emissions accounted in the
The CBAM helps to create a level playing field for the EU’s CBAM is an important factor [6,7]. Researchers believe that for a better
domestically-produced and imported goods [1]. As the climate rules in indication of the carbon contents of import goods and for more effective
the EU have become increasingly ambitious and strict, the carbon price measurement, both direct carbon emissions during production and in-
in the EU ETS continues to increase due to the tightening emission quota. direct carbon emissions from electricity usage and material production
This increase leads to increased production costs for EU goods and gives should be accounted for [4,8]. However, from a practical perspective,
a competitive advantage to products outside the EU that use carbon- this inevitably causes immense challenges for data collection and lo-
intensive production processes [2,3]. This situation not only erodes the gistical management. Secondly, the level of the carbon price directly
competitiveness of EU products but may also lead to a production relo- influences the price changes of imported products and thus affects the
cation of carbon-intensive industries to regions without strong climate effectiveness of the CBAM [9]. The future development of the EU ETS
mitigation measures. The ensuing emission increase in non-abatement and the improvement of relevant policies will introduce uncertainties

∗
Corresponding author.
E-mail address: z.mi@ucl.ac.uk (Z. Mi).

https://doi.org/10.1016/j.fmre.2023.02.026
2667-3258/© 2023 The Authors. Publishing Services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. This is an open access article under the CC BY
license (http://creativecommons.org/licenses/by/4.0/)

Please cite this article as: X. Sun, Z. Mi, L. Cheng et al., The carbon border adjustment mechanism is inefficient in addressing carbon leakage and
results in unfair welfare losses, Fundamental Research, https://doi.org/10.1016/j.fmre.2023.02.026
JID: FMRE
ARTICLE IN PRESS [m5GeSdc;April 13, 2023;18:55]

X. Sun, Z. Mi, L. Cheng et al. Fundamental Research xxx (xxxx) xxx

to the carbon price within the EU [10]. The average carbon price in the limited solution to address the carbon leakage problems caused by the
EU ETS was valued at $90.33/t in the last three months in 2021, while EU ETS and brings scant carbon reduction effects to global carbon abate-
price level in the literature was apparently lower [11]. Thirdly, two ment. In addition, the CBAM will cause welfare losses and inequality
ways of calculating carbon contents of the imported products are usu- in developing countries. Retaliation will weaken the carbon reduction
ally considered in modelling the effects of a CBAM. Practically, requiring effect and further lead to more than seven-fold global welfare losses,
the importers or foreign producers to report the facility-level emissions which will be borne by poor countries. However, cooperation between
and activity data is preferable. But when modelling, one way is to use countries to promote floor carbon price brings much more carbon re-
the average emission intensity in the export country to reflect the spe- duction gains and greatly mitigate carbon leakage risks. Clearly, the
cific carbon content of the products from different origins. The defect implementation of the CBAM must be based on effectively solving prob-
of this approach is the complicated data collection and monitoring for lems of justice and efficiency. This study indicates the insufficiency of
the policy implementation. Another approach, using the EU production the unilateral trade measurement in tackling carbon leakage risks and
technology as the reference emissions, eliminates the overloaded admin- points out the necessity of international cooperation in climate change
istrative cost. Also, this will treat the imports from different countries mitigation.
identically and avoid discrimination on the imports [12]. Fourthly, the
revenue usage of the CBAM influences the macroeconomic impact of 2. Material and methods
the CBAM. A simulation comparing the revenue recycle and no recy-
cling indicates that the macroeconomic gains by the CBAM revenue is 2.1. The global trade analysis project model
much more than the direct trade interactions in the long run [13]. How
the CBAM revenue will be used has not been clarified in detail, but most A CGE model is typically utilized to address the impacts of trade poli-
of them will be made into the EU budget as general income to support cies. In particular, the GTAP model is a well-established and commonly
investments in green and digital transition [14,15]. used CGE model for evaluating the economic and environmental impacts
The external factor that impacts the implementation of the CBAM in- of global trade policies [29]. It is a comparative static model that shows
cludes the potential legal disputes and other countries’ responses. As the the differences between different possible states of the global economy
CBAM has the potential to enhance the climate ambition and facilitate [30]. The economic simulation of a policy shock is based on the optimal
the climate action outside the EU, it provides an opportunity for all na- behaviour of several agents, namely, private households, governments,
tions to abate carbon emissions together. If other countries are willing to and companies. Consumers in the model aim to maximize their utility,
cooperate and implement more ambitious climate policy in response to while producers aim to maximize their profits and minimize their costs.
the CBAM, for example, a global carbon tax or removing fossil subsidies Products are categorized into two groups: energy commodities (i.e.,
[11], this would bring more carbon reduction benefits. However, coop- coal, gas, crude oil, petroleum products, and electricity) and other com-
eration is not necessarily guaranteed. Non-EU countries would interpret modities. All input sets are aggregated as two composite bundles of pri-
the carbon levy as a source of income for the EU’s own fiscal budget mary factors and intermediate inputs. Substitution of the input sets is
and recognize the CBAM as protectionist [16–18]. For a more resilient structured by multi-level nested constant elasticity of substitution func-
transition to Net Zero, the EU passed the NextGenerationEU (NGEU) Re- tions. The production of each sector is based on the assumption of con-
covery Plan in 2020 for economic recovery. In total, a stimulus package stant returns to scale and a completely competitive market. In terms of
of over €2 trillion has been assigned by NGEU to rebuild a post-COVID- international trade, commodities from domestic production and imports
19 Europe [14] and the CBAM was proposed as a source of revenue to are treated as imperfect substitutes based on the Armington assumption
support NGEU investment [13]. Many countries have expressed their [31]. Regional and national economies are connected by trade in com-
concerns about the CBAM. The United States warned the EU that the modities.
CBAM should be a “last resort” [19], while Australia criticized the CBAM GTAP-E is the energy-environmental extension of the standard GTAP
for running “the risk of enhancing protectionism” and being “detrimen- model, the most frequently employed extension of the standard GTAP
tal to global growth” [20]. Major developing countries protest against model for research on carbon mitigation policies (i.e., carbon tax and
the CBAM because it shifts the economic burden of developed-world cli- emission trading) [32,33]. It focuses on a more precise structure of en-
mate policies to the developing world through terms-of-trade effects and ergy production and consumption, as well as the related carbon emis-
aims to protect of EU domestic production [21–24]. The CBAM compels sions (from fossil fuel combustion). The core idea is to consider substi-
other countries to make the same efforts to decarbonize their economic tution within and between fossil fuels, other types of energy sources,
structure and may be regarded as contravening the common but differ- capital and labour. To do this, the model takes two steps to take energy
entiated responsibility principle of the United Nations Framework Con- commodities out of the intermediate input nest and incorporate them
vention on Climate Change (UNFCCC) [25]. In addition, the CBAM may into the value-added nest [32]. First, energy commodities are separated
be charged with violating the General Agreement on Tariffs and Trade into electricity and non-electricity groups including fossil fuels, where
(GATT) because it favours domestic companies [26]. Other countries substitutions are allowed within and between the two groups. Then, the
may therefore treat the CBAM as a carbon-levy trade barrier. Thus, the energy composite is combined with capital and other primary factors
use of counter-tariffs by other countries may be inevitable and further in a value-added-energy nest through a constant elasticity of substitu-
cause uncertainty regarding the impacts of the CBAM [4]. tion structure. The elasticity of substitution between capital and energy
Previous studies on the EU CBAM are insufficient to indicate the ef- within the capital-energy composite nest (𝜎 KE-inner ) is usually smaller
fectiveness of the CBAM because they did not consider all the factors compared with the elasticity of substitution between the capital-energy
discussed above [6,27,28]. Here, we quantitatively evaluate the eco- composite and other primary factors (𝜎 VAE ), producing an overall neg-
nomic and environmental impacts of the EU CBAM by comprehensively ative substitution elasticity between capital and energy (𝜎 KE-outer ). The
considering the internal and external factors, and the costs burdened relationships are as follows:
by countries differing in income levels. Based on seven counterfactual
𝜎𝐾𝐸−𝑜𝑢𝑡𝑒𝑟 = (𝜎𝐾𝐸−𝑖𝑛𝑛𝑒𝑟 − 𝜎𝑉 𝐴𝐸 )∕𝑆𝐾𝐸 +𝜎𝑉 𝐴𝐸 ∕𝑆𝑉 𝐴𝐸 (1)
scenarios, we simulate alternative CBAM schemes with different carbon
prices, emission scopes and carbon intensity measurement, revenue us- Where SKE is the cost share of the capital-energy composite in the
age and in particular, cooperation and retaliation by other countries. value-added-energy nest, and SVAE is the cost share of the value-added-
The simulation is conducted in a Global Trade Analysis Project (GTAP) energy composite in the output nest (see supplementary Fig. S1 for the
Computable General Equilibrium (CGE) model with the latest data in production structure of GTAP-E).
2014. In this study, the carbon intensity and technology of production Carbon emissions are calculated based on the energy consumption by
of the initial equilibrium is used. The results show that the CBAM is a firms, government, and private household. It is assumed that emissions

2
JID: FMRE
ARTICLE IN PRESS [m5GeSdc;April 13, 2023;18:55]

X. Sun, Z. Mi, L. Cheng et al. Fundamental Research xxx (xxxx) xxx

are proportional to energy consumption [33], for example: evaluate the impact of the CBAM, the complete list of sectors in the EU
ETS is considered in the simulations (see carbon-intensive sectors plus
𝑔 𝑐𝑜2𝑓 𝑑 (𝑒, 𝑖, 𝑟) = 𝑞 𝑓 𝑑(𝑒, 𝑖, 𝑟) (2)
oil products and electricity in the energy sectors in supplementary Table
Where gco2fd is changes of emissions from firms’ domestic energy S1).
consumption; qfd is the changes of firms’ domestic energy consumption; Both the direct emissions and embodied emissions (plus indirect
e represents energy type, including coal, oil, gas, and oil products; i emissions) of import products are considered to determine the scope
represents industrial sector; r represents country. of emission accounting. According to the proposal of the provisional
The latest GTAP model data, version 10, is employed in this study, deal of the European Council and Parliament, the CBAM will apply to
and the reference year in GTAP 10 is 2014, which is the latest year direct emissions for some products and both types of emissions for the
available. Although it might be better to update the data to the actual other. Indirect emissions from electricity utilization during production
year of CBAM implementation, the economic effect of the COVID-19 of cement, fertilizer and power will also be covered, which will improve
pandemic would cause profound uncertainty in updating and project- policy efficiency and mirror the scope of the EU ETS. The literature on
ing the data. Therefore, this study conducted a counterfactual analysis the impacts of the CBAM agrees that both the direct and indirect car-
based on analysing the 2014 data. In GTAP 10, 141 countries and 65 sec- bon emissions of import products should be considered to ensure the
tors are included. In this study, the 141 countries are aggregated into effectiveness of the CBAM [4]. However, they also admit that account-
26 countries/regions, and the 65 sectors are aggregated into 17 sectors ing for direct carbon emissions in the CBAM would be more practical
(see supplementary Table S1). Based on data aggregation, the five types due to a lack of technical and data support to measure embodied car-
of energy products are separated into five individual sectors to better bon emissions [25]. To deal with data availability, some scholars have
simulate the trade flows in the energy market and the carbon emission suggested that a benchmark carbon intensity representing the average
changes of different types of energy goods. In addition, the five primary performance of a sector may be feasible [7], for example, using the EU
factors in production are considered in GTAP-E, namely, land, capital, carbon intensity to calculate carbon content in the EU’s imports. How-
natural resources, skilled labour, and unskilled labour. Limitations exist ever, this approach fails to measure the emissions of an individual region
in this study due to data availability (for example, failure to reflect the [25]. Others have proposed that indirect emissions mainly come from
impact of energy prices change in 2022 and the war in Ukraine), and electricity usage, and they have measured embodied emissions based on
the assumptions of the CGE model (for example perfect information, ra- the proportion of electricity usage and the total emissions of the elec-
tional behaviour, etc.), the study enables the assessment of long-term tricity sector [34]. In this study, the benchmark scenario (D90) accounts
impact of the CBAM. The results help to compare the effectiveness of for only direct emissions and scenario E90 employs the embodied car-
various CBAM schemes to reduce carbon leakage caused by the EU ETS. bon emissions of imports to explore the impact of the emission scope
To test the robustness of the results, we conduct a sensitivity analy- on CBAM effectiveness. Both scenarios measure carbon content accord-
sis with respect to the Armington elasticity and substitution elasticity ing to the local production process. And based on D90, the scenario
of capital-energy composite to show how changes in parameters in the D90_EUintst using the EU’s production carbon intensity is designed.
CGE model will affect the robustness of our results (see supplemental We calculate the direct carbon emissions of import products based
materials for details). on the technologies in exporting countries and the indirect carbon emis-
sions due to electricity usage in the production process, which can better
2.2. Scenarios of different policy schemes reflect the actual carbon content in trade goods. The calculation is as the
following Eq. 3 and Eq. 4 [35]:
As a pre-simulation scenario, the current carbon prices are imple-
𝑬 dirt,r = 𝑪 r .∕ 𝑽 𝑶𝑴 r (3)
mented in the existing carbon markets, for example, a carbon price at
$90/t is implemented in the EU (supplementary Table S2 for carbon ( )
prices of other countries). Carbon leakage rate is calculated as the pro- 𝑬 emb,r = 𝑪 𝑟 + 𝑽 𝑫 𝑭 𝑴 power,r ⋅ 𝑐power,r ∕ vo𝑚power,r .∕𝑽 𝑶𝑴 r (4)
portion of increased carbon emissions outside the regions with the cor-
where Edirt,r is the vector of the direct emission intensity in country r,
responding carbon price to the reduced carbon emissions in the regions.
Cr is the vector of the sectoral carbon emissions from fossil fuel, and
In designing different scenarios regarding possible CBAM schemes,
the elements of vector VOMr are the values of the sectoral total out-
it is necessary to consider the following issues: the sectors and countries
put in country r. The embodied emission intensity Eemb,r in country r
to which the CBAM applies, carbon accounting methods, carbon price
is the direct emission intensity plus indirect emission intensity, where
and the referred technology of the production process. Five scenarios
VDFMpower,r is the consumption of electricity by each sector in country
simulating different CBAM schemes are set in this study: a benchmark
r, cpower,r and vompower,r are the carbon emission and total output of the
scenario, D90, with direct (scope 1) emissions accounted for and an av-
electricity sector in country r, and the two variables are elements of the
erage carbon price level in line with the EU ETS; an indirect emission
vector Cr and VOMr , respectively.
(scope 1 and 2 emissions) scenario, E90; a higher carbon price scenario,
The carbon price may greatly affect the effectiveness of the CBAM
D200; a scenario, D90_EUintst, with EU’s production carbon intensity
policy. According to the CBAM proposal of the European Commission,
rather than the local production carbon intensity applied for tariff calcu-
the carbon price of the CBAM will not exceed that of the EU ETS to
lation; and scenario D90_otax simulates the impact of the CBAM revenue
avoid any violation of GATT principles [14]. Thus, the price in the GTAP
usage, where output tax is reduced in the EU by recycling the CBAM rev-
model, US$90.33/t, is set as the carbon price under basic scenario D90,
enue. Before simulating the impacts of the CBAM, a baseline scenario
which is the average price in the EU ETS in the last month in 2021.
is set by implementing the carbon prices of the current carbon markets
As the carbon cap will be further constrained in the future, the carbon
in the EU and other ETS. The carbon prices in these countries are set
price will increase due to the decrease in permits. Thus, scenario D200
as the average level in 2021 or in the last month in 2021 when carbon
simulates the impact of a carbon price increase from $90/t to $200/t.
price increased dramatically (supplementary Table S2). Carbon prices in
The carbon pricing in the CBAM finally produce an increase in the tariff
non-EU countries are adjusted according to the emission coverage ratio
rate of the imported products to the EU. Tariff shocks (change power)
compared with the EU ETS [11]. The CBAM schemes under all scenar-
of the EU’s imports are calculated via Eqs. 5 - 7:
ios apply to all sectors covered in the EU ETS. Currently, the European ( )
council and Parliament agree on several carbon-intensive products at a VXCO2i,r,EU = 𝑝CBAM − 𝑝r × 𝑒i,r × VXM𝐷i,r,EU (5)
high level of carbon leakage risk that are covered by the CBAM: iron
and steel, cement, fertilizer, aluminium, hydrogen, and electricity gen- VIM𝑆 i,r,EU
eration, with the goal to include all EU ETS sectors by 2030. To better tm𝑠i,𝑟 = (6)
VIW𝑆 i,r,EU

3
JID: FMRE
ARTICLE IN PRESS [m5GeSdc;April 13, 2023;18:55]

X. Sun, Z. Mi, L. Cheng et al. Fundamental Research xxx (xxxx) xxx

’ VIMSi,r,EU +VXCO2i,r,EU VIMSi,r,EU 1990 to 2019 in the World Bank Temporary Trade Barriers Database
tmsi,r − tmsi,r VIWSi,r,EU
− VIWSi,r,EU and identified the sectors most involved in trade reactions to EU im-
Δtmsi,r = =
tmsi,r VIMSi,r,EU ports. In designing the trade retaliation under scenario Retaliation, we
VIWSi,r,EU referred to these major trade partners and corresponding sectors to de-
( ) VXMDi,r,EU termine the countries initiating countermeasures and the targeted sec-
= 𝑝CBAM − 𝑝r × ei,r × (7)
VIMSi,r,EU tors (see supplementary Table S4). The tariff changes were determined
based on export losses, which means that trade retaliation aims to cre-
where Eq. 5 calculates the value of carbon tariff paid for the export i ate the same export losses for the EU as the retaliating country expe-
from country r to the EU (VXCO2i,r,EU ), Eq. 6 is the equation measuring rienced. Specifically, tariff changes are first determined by the value
tariff tmsi, r in the GTAP, Eq. 7 measures the tariff change power Δtmsi, r of tariff burden experienced by the raiser country, calculated by Eq. 8.
of the export i from country r to the EU induced by an ad valorem carbon Then, an extra tariff shock beyond scenario D90 is determined by the
tax. pCBAM is the carbon price in the CBAM scheme, the unit of which GTAP model, and the export losses for the EU are estimated. The tariff
is $ per ton carbon dioxide emissions, and pr is the domestic carbon is then adjusted until the export losses of the EU and the raiser country
price in country r (supplementary Table S2); ei,r is the carbon intensity are equal.
of product i made in country r, i.e., the emissions contained per unit of ∑
exported product i; VXWDi,r,EU is the value of export i from country r 𝑖 VXCO2i,r,EU
Δtmsj,EU,r = (8)
to the EU at the world price; VIMSi,r,EU and VIWSi,r,EU is the value of VIMSj,EU,r
import i from country r to the EU at EU’s market price and world price,
respectively. Additionally, ei,r is an element of vector Edirt,r or Eemb,r . where Δtmsj,EU,r is the initial tariff change power of the imports j from
As the proposed scheme of the CBAM by the Europe Commission, the the EU to the raiser country r; VIMSi,EU,r, is the value of import j from
European Free Trade Association (EFTA) countries will be exempted for the EU to country r at the market price. In this way, the change in the
the CBAM in this study. The amendments adopted by the European Par- retaliatory tariff is determined to compensate for export loss due to the
liament emphasized the technical and financial support for decarboniza- CBAM.
tion in the Less Developed Countries (LDC) countries. In this study, LDC
countries are also exempted. An aggregated region named LDC contains 2.4. Gini coefficient
the less developed countries defined by the United Nations [36] (sup-
plementary Table S3). Note that some of the less developed countries The Gini coefficient, proposed by the Italian statistician and socialist
on the UN’s list are aggregated in the “rest of the world (ROW)”. There- Corrado Gini, is a widely-used statistic to measure inequality of income
fore, exemptions are offered to EFTA, LDC, and the ROW under each and wealth [39]. A Gini coefficient of zero represents the perfect equal-
scenario. ity, indicating people all have the same income or wealth. In contrast, a
Gini coefficient of one delineates absolute inequality, meaning one per-
son has all the income or wealth whereas other people have none. We
2.3. Scenarios of abatement cooperation and trade retaliation
calculated Gini coefficients to evaluate the impact of CBAM on inequal-
ity. The following Eq. 9 is the calculation of the Gini coefficient in this
Furthermore, in consideration of the efficiency, justice, and legal is-
study:
sues of CBAM design, we set two scenarios to simulate the impact of
different responses of non-EU countries, including international cooper- ∑
26 ∑
26

ation and trade retaliation by other countries. The scenario Cooperation 𝐺= Popi Incmi +2 Popi (1 − T_incmi ) − 1 (9)
𝑖=1 𝑖=1
assumes that global cooperation is agreed on and the floor carbon prices
is implemented where carbon prices are determined according to the de- Where G is the Gini coefficient; 𝑃 𝑜𝑝i is the ratio of population in
velopment level of the countries. Based on the IMF’s proposal for a floor country i to total population in the world, 𝐼𝑛𝑐𝑚i is the ratio of income
carbon price scheme, the developed countries agree on a carbon price in country i to the total income in the world; 𝑇 _𝑖𝑛𝑐𝑚i is the cumulative
at $75/t, while high-income and low-income developing countries have proportion of income in country i, and i denotes the number of coun-
less carbon price at $50/t and $25/t [37]. As the carbon price in the EU try/region (i = 1, 2, 3, …, 26).
ETS already exceeds the price in this scheme, here, we assume that the Similarly, by replacing the income-related statistics to carbon emis-
developed countries implemented a carbon price the same as in the EU sions, we also calculated the carbon emission Gini coefficient (C-Gini)
($90/t), while higher- and lower-income developing countries (HID and [40]. The Gini coefficient and C-Gini coefficients are 0.60 and 0.45 when
LID, respectively) have carbon prices at $60/t and $30/t, respectively. current carbon prices are implemented, indicating carbon inequality is
The impact of both the higher carbon prices in these countries and the relatively moderate compared with income inequaility [40].
CBAM charged for imports from the developing countries are simulated
in the scenario Cooperation. 2.5. Data sources
By contrast to possible cooperation, many other countries, especially
developing countries, may claim that the CBAM is trade protectionism The trade, economic, population and carbon emission data are ac-
based on environmental trade barriers. They have expressed their con- cessed via the GTAP database (https://www.gtap.agecon.purdue.edu)
cerns about and strong opposition to any type of trade measure that for the basic CGE modelling. The trade barrier data, used to determine
would impose carbon costs on their exports [16,19,20,22,23]. Previous the sensitive sectors in the historical trade countermeasures, are from
conflicts on the EU’s proposal to include international aviation into the the World Bank Temporary Trade Barriers Database (https://www.
EU ETS confirm the potential for trade disputes induced by the CBAM worldbank.org/en/data/interactive/2021/03/02/temporary-trade-
[38]. Non-EU countries may retaliate against the CBAM by imposing a barriers-database). The latest carbon price in the EU ETS is provided
tariff on EU products. by International Carbon Action Partnership Allowance Price Ex-
The scenario Retaliation is designed where the EU’s major trade part- plorer (https://icapcarbonaction.com/en/ets-prices). Carbon prices
ners retaliate on the CBAM. As the trade dispute resolution process of the of other countries are from the World Bank carbon dashboard
World Trade Organization (WTO) incurs great costs, it is suggested that (https://carbonpricingdashboard.worldbank.org/). Developing and de-
major economies will tend to choose trade retaliation, while small coun- veloped countries are classified according to the World Economic Out-
tries will not. Thus, under scenario Retaliation, we investigated the ma- look database by International Monetary Fund [41] (https://www.imf.
jor trade partners of the EU and the sensitive sectors in the trade dispute org/external/pubs/ft/weo/2015/02/weodata/groups.htm#cc). Cate-
between these countries. We combed through trade dispute records from gories of the developing countries into high- and low-income groups are

4
JID: FMRE
ARTICLE IN PRESS [m5GeSdc;April 13, 2023;18:55]

X. Sun, Z. Mi, L. Cheng et al. Fundamental Research xxx (xxxx) xxx

Table 1
Economic and environmental impacts of domestic carbon prices in EU ETS and other countries.

Carbon reduction Emission leakage Leakage rate Welfare Real income loss Real GDP loss
Mt Mt % Billion$ % %

ETS
At current carbon prices −1609.2 86.8 5.4 −108.7 0.16 0.14
EU ETS −588.9 119.1 20.2 −86.2 0.13 0.11
ETS of other developed countries −401.8 25.4 6.3 −15.9 0.02 0.02
ETS of developing countries −685.4 9.2 1.3 −6.5 0.01 0.01
D200
EU ETS - 200$/t −961.8 228.0 23.7 −196.4 0.29 0.25
Price increase from 90 to 200$/t −372.9 108.9 29.2 −110.2 0.16 0.14
Cooperation
Total effects of floor carbon prices −7960.1 44.0 0.6 −318.8 0.47 0.41
Developed - 90$/t −3372.1 295.0 0.9 −125.1 0.18 0.16
Higher-income developing - 60$/t −4187.2 158.7 3.8 −95.8 0.14 0.13
Lower-income developing - 30$/t −850.4 40.0 4.7 −7.2 0.01 0.01

Notes: The baseline is the initial equilibrium where no carbon prices and CBAM exist. Scenario ETS simulates the effects of the carbon prices in the current emission
trading systems in different regions. The implemented carbon prices and emission trading systems can be found in supplementary Table S2. Scenario D200 simulates
a domestic carbon price at $200/t only in EU and the EFTA countries. Scenario Cooperation simulates the effects of floor carbon prices in three country groups,
where developed countries, excluding the EU and EFTA, have a carbon price at $90/t, the same as the EU ETS; higher-income developing countries have a moderate
carbon price at $60/t; lower-income developing countries have a low carbon price at $30/t. The column carbon reduction means reduced carbon emission in the
regions where the corresponding carbon price is implemented, and emission leakage is the increased carbon emissions outside the regions with carbon price. Leakage
rate is the ratio of emission leakage to carbon reduction.

from the World Bank country classification [42]. National consumption- this study is the latest level and much higher than the history level.
based carbon emissions are available from Global Carbon Budget This carbon leakage rate is consistent with results in the literature [45].
[43] (https://www.globalcarbonproject.org/carbonbudget/21/data. Carbon prices in other developed countries like the UK, US, Canada etc.,
htm). lead to reduced carbon emissions by 401.8 Mt and emission leakage by
25.4 Mt. Carbon leakage rate of current carbon prices in these developed
3. Results countries is 6.3%. Carbon prices in developing countries, such as China,
Mexico, and South Africa, can reduce carbon emissions by 658.4 Mt
3.1. Limited carbon leakage reduction effects are found accompanied with a carbon leakage rate of 1.3%.
Although the cost-benefit efficiency of the CBAM should be affirmed,
The carbon leakage rate caused by the current carbon price ($90/t) the overall environmental benefits of the CBAM are insufficient to ad-
EU ETS, measured by the ratio of increased emissions in non-EU regions dress the carbon leakage caused by the EU ETS. In the benchmark sce-
to the reduced emissions in the EU and EFTA regions, is about 20.2% nario D90, the emission reduction in non-EU countries is 22.5 Mt, ac-
(Table 1). Scenario ETS, where the current carbon prices in the EU and counting for only 18.9% of the carbon emission leakage caused by the
other ETS are implemented to the corresponding carbon markets, shows EU ETS (Table 2). Such emission reduction gains are offset by 4.9 Mt
that the EU ETS leads to about 588.9 Mt carbon emission reduction in emission increase in the EU and exemption areas. Consequently, to-
the EU and EFTA countries, while carbon emissions increase by 119.1 tal carbon emission reduction globally is 17.4 Mt. Regarding the eco-
Mt in the non-EU countries. The reduced carbon emissions are higher nomic impacts, it is affirmable that the CBAM is effective. The global
than in the literature [44], because the carbon price implemented in welfare losses (measured by equivalent variation) are approximately

Table 2
Economic and environmental impacts of different policy schemes and responses at the corresponding carbon prices.

CO2 in non-EU Emission leakage CO2 in EU and Welfare Real income loss Real GDP loss C-Gini change
reduction exempted areas
Mt % Mt Billion$ % % %

ETS. At current carbon prices

D90 −22.5 18.9 4.9 −0.67 0.001 0.001 0.019
E90 −29.7 24.9 6.7 −0.78 0.001 0.001 0.019
D90_Euinten −10.5 8.8 1.6 −0.59 0.001 0.001 0.002
D90_otax −22.3 18.7 5.5 −0.30 0.000 0.000 0.019
Retaliation −19.6 16.5 1.9 −4.58 0.006 0.005 0.022
D200. EU ETS at 200$/t
D200 −36.4 16.0 8.1 −2.3 0.003 0.003 0.022
Cooperation. Floor carbon prices
Cooperation −10.3 8.6 2.9 −0.05 0.000 0.000 0.012

Notes: the emission and economic impacts of the CBAM scenarios in this table are the variation compared with equilibrium at the corresponding carbon prices in
Table 1. Emission leakage reduction is the ratio of carbon reduction in non-EU countries (excluding EFTA, LDC, and ROW) to emission leakage due to carbon prices
in the EU and EFTA. D90 is the scenario in which the CBAM accounts for direct carbon emissions and sets a carbon price of $90/t. E90 is the scenario with indirect
emissions (scope 1 and 2 emissions) accounted for and a carbon price at $90/t. D90_EUintst is the scenario with the policy scheme of D90 but use carbon intensity of
the EU production processes to calculate CBAM tariff changes. D90_otax simulates the impact of the CBAM revenue usage by reducing the output tax to compensate
for the price change in the EU. Scenario Retaliation simulates the non-cooperation responses of countries faced with a CBAM, in which the main trade partners of
EU raise trade retaliation and impose tariffs on anti-dumping products based on historical trade disputes. D200 is the scenario with direct emissions accounted for
and a carbon price at $200/t. Scenario Cooperation simulates a global cooperation scheme where countries agree on global floor carbon prices, and EU implements
the CBAM to the countries with lower carbon prices than the price in EU ETS.

5
JID: FMRE
ARTICLE IN PRESS [m5GeSdc;April 13, 2023;18:55]

X. Sun, Z. Mi, L. Cheng et al. Fundamental Research xxx (xxxx) xxx

Carbon emission reduction is mainly attributed to reduced produc-

tion in carbon-intensive industries in developing countries with China,
India, Russia, Ukraine, and South Africa seeing a major impact in this
respect. The production of energy-intensive industries experiences a re-
location from developing countries to developed countries. Due to the
increased import price of energy-intensive goods in the EU, the boosted
domestic production of such products leads to a carbon emission rise in
the electricity, metal, chemical, and mineral sectors. Although carbon
emissions in most sectors in the EU increase as the CBAM protect the
comparativeness of domestic products, there is a significant emission
reduction in the transport sector because of a shrinkage in the usage of
oil products (Fig. 2 and supplementary data). Such reduced oil products
usage is resulted from both decreased demand of transport products and
price increase of oil products in the EU, while some reduction is substi-
tuted by increased gas usage due to the lower prices of gas goods. Out-
side of the EU, the production of ferrous metals and chemical products
increases in developed countries such as the EFTA and the UK, and de-
creases in developing countries such as China, India, and Russia. Such
production shrinkages cause emission reductions mainly in the electric-
Fig. 1. Changes in GDP, carbon emissions and welfare due to a carbon ity, metal, and mineral sectors in developing countries. The trends of
border adjustment mechanism. This scenario (D90) refers to a CBAM with production relocation put pressure on developing countries to upgrade
direct carbon emissions accounted for and a carbon price set to $90/t. HID de- industrial structure for green and sustainable development.
notes high-income developing countries and LID denotes low-income developing The carbon tariff on the carbon-intensive products shows a signifi-
countries. cant impact on other sectors, especially manufacturing. In many devel-
oped countries, like the EU, EFTA, the UK, Japan, and Korea, production
in manufacturing drops off, while in other countries, manufacturing pro-
$0.67 billion, accounting for 1% of the welfare loss of the EU ETS. This duction is boosted. In developing countries, such as India, the boosted
indicates that the CBAM causes an extra 1% economic loss but addresses manufacturing production is mainly driven from enlarged exports (sup-
18.9% of the carbon leakage of the EU ETS. plementary data). This is because the stricken carbon-intensive produc-
A clear impact on equality is seen in the modelling results. Car- tion in developing countries, suffering as a result of the CBAM, weakens
bon emission reductions created by the CBAM mainly occur in devel- the competitiveness of related sectors, and therefore the declined pro-
oping countries, namely, Russia, India, Turkey, South Africa, Ukraine, duction releases primary factor demand and decreases primary factor
etc., (Fig. 1), resulting from a shrinkage in production and exports in prices. Furthermore, the manufacturing sector in developing countries
these developing countries. Regarding economic impacts, GDP increases benefits from this and subsequently gains a comparative advantage in
in most developed countries. The EU enjoys the largest GDP increase global markets. The increased price of the imported raw material and
(0.1%) due to the CBAM. In contrast, developing countries, e.g., India, therefore increased production cost are the main reasons of the con-
Russia, Ukraine, and South Africa, experience a GDP decline due to the tracted downstream manufacturing in the EU. In other developed coun-
CBAM. On the other hand, the EU is the major beneficiary, enjoying a tries, the declined manufacturing production is because of the drop in
welfare increase of $2.6 billion, accounting for 73% of the total wel- the export of manufacturing products.
fare gains due to the CBAM. Ukraine experiences the most economic Overall, the CBAM leads to a reshaped international trade structure
loss in this analysis because of its export loss in ferrous metal indus- and deteriorated global trade conditions, which does a disservice to real
try. While the value of production in ferrous metal makes it one of the income [46] and creates increased inequality. Exports shrink in most
most important sectors in Ukraine amongst the secondary industry, it countries (Fig. 3). India, Russia, Ukraine, South Africa, and Saudi Ara-
is also amongst the most carbon-intensive sectors and suffers greatly bia are the major regions enduring hits to their exports, with exports
from the CBAM (Fig. 2). With price increase and demand decline in declining from 0.17% to 0.36%. In contrast, the EU, EFTA, and UK en-
the EU market after the implementation of the CBAM, exports of fer- joy the greatest increase by 0.02–0.1% in exports. The export increase
rous metal products from Ukraine to the EU are decreased by about for the EU is due to the increased intra-EU trade rather than exports to
34%. Although increase in ferrous metal exports to other countries off- the non-EU countries. A 0.02% increase in the Gini coefficient indicates
sets some of the negative impacts, the CBAM finally leads to ferrous that there is increased inequality through the widening gap of income
metal production in Ukraine declining by almost 6%. At the same time, per capita across countries (Table 2 and Supplementary Table S5). There
the declined household income in Ukraine leads to depressed consump- is a trend toward stronger international trade amongst the developing
tion by private households, especially in the service sector (accounting countries (developing-developing trade). The CBAM improve the com-
for more than half of the private consumption decrease), which is sen- parativeness of the EU’s domestically produced goods to some extent
sitive to income levels. Malaysia and Brazil are the only two develop- and imports from non-EU countries dramatically decrease while trade
ing countries where the carbon emissions increase. These two countries within the EU is boosted significantly. The CBAM not only gives the
benefit from the relatively low carbon intensity of the domestic produc- EU products a comparative advantage in the EU market, but also weak-
tion and therefore experience lower tariff impacts (Supplementary Fig. ens the comparativeness of the products in developing countries in the
S2). Chemistry in Malaysia is the most affected sector because of the world market while strengthening the comparativeness of the products
CBAM (Fig. 2). The shrunken demand in carbon-intensive products re- in the developed countries. This further leads to increased exports from
leases the demand of primary factors such as labour and capital in the the developed countries to all non-EU countries. As imports from devel-
developing countries, and therefore decreases the price of primary fac- oping countries to developed countries are decreased, the developing
tors. Such changes enhance the competitiveness of manufacturing goods countries seek stronger trade links between the developing countries to
and low-carbon manufacturing production rises. Export of manufactur- compensate for the export loss. Therefore, the CBAM witnesses the trend
ing goods from Malaysia to the developed countries increases and leads of stronger trade between developing countries.
to increased demand in transportation, and therefore increased carbon Four scenarios regarding different policy schemes are simulated
emissions. based on D90: 1) the embodied emission accounting method (indirect

6
JID: FMRE
ARTICLE IN PRESS [m5GeSdc;April 13, 2023;18:55]

X. Sun, Z. Mi, L. Cheng et al. Fundamental Research xxx (xxxx) xxx

Fig. 2. Impacts on the carbon emissions and production of 17 sectors. This scenario (D90) refers to a CBAM with direct carbon emissions accounted for and a
carbon price set to $90/t. (a) Sectoral emission changes of the major impacted countries. (b) Sectoral emission changes of developed countries. (c) Sectoral emission
changes of developing countries. (d) Sectoral production changes of the major impacted countries. (e) Sectoral production changes of developed countries. (f) Sectoral
production changes of developing countries.

emissions in the production process, i.e., scope 1 and 2 emissions) is countries. In scenario E90, the carbon emission reduction in non-EU re-
adopted in scenario E90; 2) a higher carbon price of $200/t is applied gions is 29.7 Mt, increased by 32% compared with the direct emission
in scenario D200; 3) carbon intensity based on the EU domestic pro- accounting in scenario D90 (Table 2). In addition, the negative impacts
duction technology is applied in scenario D90_EUints; and 4) the CBAM of the welfare loss increased by 16%, indicating enhanced efficiency of
revenue is used to reduce output tax to compensate the price increase the policy scheme by including indirect emissions in the CBAM. China is
due to the CBAM in the EU in scenario D90_otax. the country most negatively impacted by the E90 scheme. The welfare
Firstly, consideration of both direct emissions and indirect emissions loss rises by almost sevenfold from $1.1 billion to $7.5 billion. Metal
due to electricity usage contained in import products enhances the effec- products suffer most in China due to the E90 scheme, with the produc-
tiveness of the CBAM scheme while adding to the pressure on developing tion shrinkage increased from 0.002% to 0.12%. Scope 2 emissions of

7
JID: FMRE
ARTICLE IN PRESS [m5GeSdc;April 13, 2023;18:55]

X. Sun, Z. Mi, L. Cheng et al. Fundamental Research xxx (xxxx) xxx

are enhanced by higher carbon prices, the CBAM shows lessened capac-
ity to address carbon leakage problems. The reduced carbon emissions
outside the EU accounts for 16% of the total emission rise due to leak-
age, which is lower than the leakage reduction rate in D90.
Thirdly, the scenario D90_EUintst uses the carbon intensity of the
EU production process to calculate tariff changes. This scheme helps to
simplify the policy implementation but reduces the effect of the CBAM.
The carbon emission reduction by the CBAM in non-EU regions only
accounts for 8.8% of the EU ETS carbon leakage, while the economic
impact is slightly changed. The carbon emissions per unit production
in the EU, especially of the carbon-intensive products, are amongst the
lowest globally due to the ambitions of the EU in decarbonizing and
mitigating climate change. Therefore, using the EU technology for mea-
suring carbon content of the imports significantly relieves the pressure
that the non-EU countries experienced, but also reduce the effectiveness
of the CBAM and weakens the decarbonization motivation of producers
outside the EU.
Fourthly, revenue recycling plays an important role in protecting
the EU regarding the macroeconomic impacts of the CBAM. With the
CBAM revenue used to reduce output tax in the EU, the environmental
benefits are only slightly affected but the economic cost is considerably
Fig. 3. Impacts on exports and imports of countries and regions. This sce- reduced. However, it is worth noting that such welfare loss reduction is
nario (D90) refers to a CBAM with direct carbon emissions accounted for and a not because of the overall benefits of the global nations but exclusively
carbon price set to $90/t. enjoyed by the EU. By recycling the CBAM revenue, the welfare gains
of the EU increased by 40%, while other regions experience more wel-
fare loss. Detailed simulation results in these scenarios can be found in
metal products account for 70% of total scope 1 and scope 2 emissions. supplementary Figs. S3–8.
China is at a disadvantage because the carbon intensity of power gener-
ation in China is amongst the highest in the world, at more than twice 3.2. Trade retaliation multiplies economic costs and increases inequality
that of the EU. The embodied emission scheme could hopefully improve
the motivation to decarbonize power generation in the world. The scale of trade conflicts includes the countries taking counter-
Secondly, higher carbon prices could lift carbon reduction globally measures, the products targeted by the trade conflicts, and the range of
and put more pressure on carbon-intensive products outside the EU, but tariff changes. Here, we designed a scenario to simulate the impacts of
the effectiveness of the CBAM is weakened. Assuming the carbon price trade retaliation based on scenario D90. Under scenario Retaliation, the
will rise to $200/t, the price changes in the EU ETS lead to an extra EU’s major trade partners impose an extra tariff on EU products that are
carbon reduction by 372.9 Mt CO2 emissions. And this further leads to often the objects of anti-dumping measures, and the retaliatory tariff is
108.9 Mt carbon leakage, causing an enlarged carbon leakage rate of the determined by export losses the raiser experiences.
EU ETS from 20.2% to 23.7%. At the same time, carbon emissions re- The results imply that trade retaliation against the CBAM has lim-
duced outside the EU by the CBAM are 36.4 Mt, with 8.1 Mt rebounded ited impacts on the carbon reduction achievement of the CBAM but
in the EU and exempted areas. Although the reduced carbon emissions leaves the world more vulnerable to the economic loss and inequality

Fig. 4. Economic costs for different income groups. (a) Welfare costs for different income groups under scenario D90 without trade retaliation. (b) Welfare costs
for different income groups under scenario Retaliation with trade retaliation. Countries are classified as developed or developing according to the World Economic
Outlook database by International Monetary Fund [41]. Carbon footprint per capita is calculated based on national consumption-based carbon emission from the
Global Carbon Budget [43] and population from the GTAP database and the United Nations [47].

8
JID: FMRE
ARTICLE IN PRESS [m5GeSdc;April 13, 2023;18:55]

X. Sun, Z. Mi, L. Cheng et al. Fundamental Research xxx (xxxx) xxx

in climate change mitigation. The carbon reduction in non-EU regions production in these countries will lead to far larger environmental ben-
amounts to 19.6 Mt under the scenario Retaliation, which is less than the efits to the world’s climate change mitigation.
emission reduction when there is no trade retaliation. Retaliatory tar-
iffs on EU productions increase trade within the non-EU countries and 4. Discussion
thus offset the carbon emissions reduction via the CBAM. On the other
hand, economic losses are multiplied by trade retaliation. Under the Simulations of different internal factors of the CBAM indicate that
scenario Retaliation, the EU, Ukraine, and Saudi Arabia experience the accounting for embodied emissions and using the carbon intensity of
most GDP losses, ranging from 0.1% to 1.5%. The welfare loss amounts the local production technologies lead to more environmental benefits,
to 4.6 billion dollars, which is nearly seven times the welfare loss un- and that revenue used to reduce output tax further brings the EU more
der scenario D90. The EU suffers the most welfare losses of $4.3 billion. economic gains. Firstly, embodied emissions accounting opens the op-
International trade conditions are unambiguously deteriorated, with an portunity for the CBAM to facilitate the global energy transition and
expansion of export losses. Most countries experience risen prices of do- technology upgrading. The indirect emissions of products vary greatly
mestic production and therefore exports. For example, chemistry and across countries, and in this respect, electricity usage is the most impor-
plastic are amongst the most impacted sectors in South Africa, with tant source [6,34]. Both electricity usage in production, and the carbon
prices edged up by 0.5% and 0.3% domestically. Consequently, exports intensity of electricity generation, are different due to the technology
in India, South Africa, and Russia decrease by 0.6%, 0.5% and 0.4%, factors in different countries [48]. Therefore, countries that rely more on
respectively. high carbon-intensive electricity, such as China and India, suffer more
The welfare losses of retaliating against the CBAM and the corre- from the CBAM applying embodied emission accounting. In contrast,
sponding trade barriers between countries leave poor countries more countries with cleaner energy generation will gain comparative advan-
vulnerable. As trade retaliation causes welfare losses or gains for differ- tages from such a scheme. For instance, Brazil enjoys a GDP increase
ent countries, the wealthy countries as a group enjoy net benefits while because of its energy structure. In 2019, non-fossil energy in Brazil ac-
the poor countries as a group suffer net losses (Fig. 4). The EFTA coun- counted for 45.7% of its total energy consumption, which is amongst
tries are the richest and receive welfare gains of $18.8 per capita under the highest in the world [49]. Secondly, measuring the carbon intensity
scenario Retaliation. On the other hand, people in Ukraine face welfare of the production process in the original export country improves the
losses of $33.2 per capita. Clearly, the economic costs of the CBAM and effectiveness of the CBAM. As it is designed, the CBAM needs to com-
trade retaliation are borne by the poorest group, thereby leaving eco- pensate for the carbon abatement cost of the EU’s domestic production
nomically vulnerable people worse off. and improve the comparative advantages of the EU’s products. Measur-
ing the carbon content of imported goods according to the average level
3.3. The CBAM is less effective than international cooperation in the origins reflects the technology level of the production process and
therefore improves the motivation of the producers to decarbonize their
A Cooperation scenario sees significantly enhanced carbon reduction production process. The CBAM, if based on the average carbon inten-
benefits with slightly increased economic costs. In the scenario Cooper- sity of EU producers rather than the actual carbon content of imports,
ation, the floor carbon price is implemented according to the develop- could of course reduce the complication and cost of policy implemen-
ment of the countries. The developed countries impose a carbon price tation. But it is inadequate to compensate the carbon abatement cost of
equal to the carbon price in the EU, and high-income emerging coun- the EU’s domestic producers and thus shows less efficiency in address-
tries have a carbon price lower than the EU ETS at $60/t, while the ing carbon leakage. Thirdly, the CBAM revenue, recycled to reduce the
low-income emerging countries have the lowest carbon price at $30/t. output tax and compensate for the increased price in the EU’s domes-
The carbon reduction effects of the three types of carbon prices are tic market, further deteriorates the inequality conditions caused by the
3372.1 Mt in developed countries, 4187.2 Mt in high-income developing CBAM. With reduced output tax, the EU has a significant increase in
countries, and 850.4 Mt in low-income developing countries. The car- welfare gains, but other countries bear an extra welfare loss. Consid-
bon leakage rates of the floor carbon prices are 0.9%, 3.8% and 4.7% ering the macroeconomic impact of the revenue recycling, there is an
respectively. As the floor carbon prices are implemented, the global car- increased legal risk that the CBAM may be recognized as domestic pro-
bon leakage is greatly reduced to 0.6%. Comparing the economic loss tectionism and a resource for the EU’s own fiscal budget.
of carbon prices in different countries, the price in high-income devel- Although the environmental benefits of the CBAM can be enlarged
oping countries leads to higher abatement benefits at a lower economic with appropriate scheme settings, the overall carbon leakage reduction
cost. The global welfare and GDP loss caused by the carbon prices in the effect is rather limited, and the higher carbon price brings with it even
developed countries is 0.18% and 0.16% respectively. These economic less effectiveness of the CBAM to address carbon leakage problems. The
losses are relatively equal with the EU ETS and carbon prices in the CBAM scheme at the current carbon price in the EU ETS can reduce
developing countries. But the carbon reduction in the high-income de- emissions outside the EU by only 10.5–29.7 Mt, accounting for a small
veloping countries is apparently larger, about 14% of global total carbon proportion of the emission leakage caused by the EU ETS (8.8%−24.9%).
emissions. Worse still, the effectiveness of the CBAM to address the carbon leakage
The CBAM in the scenario Cooperation imposes a tariff on the devel- risks of the EU ETS is weakened under a higher carbon price, leading
oping countries to compensate for the carbon price gap between the EU to greater economic burden as well as risks of shrinking the market size
and the developing countries. The results show that the carbon reduc- of the EU’s products [50]. To improve compatibility with GATT actions,
tion by the CBAM is 10.3 Mt in the non-EU regions, which accounts for it is necessary to constrain the carbon price in the CBAM so as not to
8.6% of the carbon leakage caused by the EU ETS and 46% of the bench- exceed the price level in the EU ETS. As carbon quotas tighten and car-
mark CBAM scheme in scenario D90. Considering the carbon prices in bon prices rise, the carbon leakage ratio is enlarged in the EU ETS in
the developing countries in scenario Cooperation, the effectiveness of the future, but the CBAM shows less capability to address such leakage.
the CBAM pales because the carbon reduction only accounts for 0.2% of The carbon reduction by the CBAM accounts for a lower proportion of
the carbon reduction by carbon prices in these countries. Overall, floor the total carbon leakage (from 18.9% to 16%). Therefore, the increased
carbon price can impose carbon abatement measures globally and essen- ambition of mitigating climate change in the EU needs to be accompa-
tially reduce carbon leakage by acknowledging the differentiated levels nied with further measurements to boost climate ambition outside the
of carbon reduction costs according to the degree of economic develop- EU as the unilateral competitive policy scheme shows less capacity in
ment. Therefore, compared with the unilateral CBAM scheme, assisting dealing with such imbalanced climate efforts.
the developing countries to promote their climate change mitigation Regarding the weak carbon leakage reduction effect, one reason is
ambition and address the carbon emission embedded in the domestic that the CBAM cannot fully compensate for the competitiveness losses

9
JID: FMRE
ARTICLE IN PRESS [m5GeSdc;April 13, 2023;18:55]

X. Sun, Z. Mi, L. Cheng et al. Fundamental Research xxx (xxxx) xxx

that domestic firms face because the carbon tariff on the imports is rela- the establishment and improvement of carbon pricing in non-EU coun-
tively small. Therefore, the changes in the price are slight. For example, tries and therefore promote global climate action. There is no denying
the ad valorem tariff rates by the CBAM on chemical products and min- that the intention of the CBAM to address carbon leakage is in accor-
eral products from China are 4.3% and 6.6%, according to indirect emis- dance with the EU’s climate ambition and its domestic efforts. However,
sion accounting (supplementary data and Fig. S2), respectively. Other evasion measures may also be taken by EU trade partners, for example,
manufacturing sectors and regions usually impose even lower carbon inaccurate emissions report or symbolic policy. The intensified global
tariffs. Consequently, the affected trade volume is narrow because the competition resulting from the CBAM may lead to a detrimental envi-
change in demand is small. In other words, the competitive advantages ronment for international trust, making sustainable development with
of EU and non-EU carbon-intensive products change slightly, and thus, equality and climate change mitigation far more costly. Overall, climate
the effect of global industrial restructuring is limited. change mitigation and global carbon reduction should be based on a
Another reason is related to the low level of price elasticities for thorough transition of the economic structure and energy mix. Tech-
intermediate inputs. In the EU, 80.7% of total imports are purchased nology upgrading and green development are critical in this process
by firms as intermediate inputs for domestic production, while the [55,56]. Developing countries have an urgent need for access to green
other 19.3% of total imports are consumed by government and private technologies and the necessary financial support. However, the CBAM
households. In addition, the import proportion of the carbon-intensive scheme is a solution based on competitiveness protection, and it neglects
products purchased by the EU firms is higher than other products. technology transfer. We suggest that major economies should pursue the
Specifically, in 2014, 18.5% of their carbon-intensive intermediate chance to promote international climate cooperation and enable devel-
input products were imported, while only 11.1% of the non-carbon- oping countries to access green technologies. For example, the floor car-
intensive inputs were imported. As intermediate input purchases are bon prices in this study are effective in addressing the carbon leakage of
less sensitive to price than final consumption, there are innate defects the abatement countries. In addition, improving the motivation of more
in the CBAM mechanism regardless of its design. The increasing carbon ambitious carbon pricing in developing countries is efficient with less
prices in the future could expand the environmental benefits, but economic costs but more emission reduction gains.
they will also cause adverse impacts on EU firms by increasing their
production costs and contracting their market size [48]. Declaration of competing interest
Last but not least, the essential deficiency of the CBAM is that it
fails to address carbon leakage happening in the global energy market. The authors declare that they have no conflicts of interest in this
Due to the very nature of the mechanism of the CBAM, it mainly im- work.
proves competitiveness of domestic carbon-intensive products by rais-
ing the price of import products, which is one of three patterns of carbon Acknowledgements
leakage. The second is the international relocation of carbon-intensive
industries. Firms tend to transfer their carbon-intensive production to This study was supported by the Engineering and Physical Sciences
non-abatement countries to avoid the environmental costs in abatement Research Council (EP/V041665/1), the British Council (2019-RLWK11–
countries. Finally, carbon leakage happening in the global energy mar- 10577). The authors would like to thank Dr Fergus Green for his insight-
ket is also very important [51], especially for large emissions abatement ful suggestions to help improve the manuscript.
groups [3]. Carbon reduction measures cause decreases in the demand
for fossil fuel in abatement countries and further lead to decreases in the
Supplementary materials
energy price in the global market. These decreases adversely stimulate
the energy demand of non-abatement countries and increases their car-
Supplementary material associated with this article can be found, in
bon emissions. The CBAM does not limit this last type of carbon leakage
the online version, at doi:10.1016/j.fmre.2023.02.026.
because it has little impact on reducing non-EU countries’ domestic de-
mand for fossil fuel energy [50,52,53]. Additional measures to restrict
References
fossil fuel supply may assist in tackling this problem [54].
If other countries initiate trade retaliation in response to the CBAM, [1] M.G. Morgan, Managing carbon from the bottom up, Science 289 (2000) 2285.
concerns of inequality will be enlarged. Through the CBAM, the EU en- [2] A. Kander, M. Jiborn, D.D. Moran, et al., National greenhouse-gas accounting for
joys the most economic gains. In addition to the EU, other major de- effective climate policy on international trade, Nat. Clim. Change. 5 (2015) 431–435.
[3] K.W. Steininger, C. Lininger, L.H. Meyer, et al., Multiple carbon accounting to sup-
veloped countries gain competitive advantages by carbon-based tariffs port just and effective climate policies, Nat. Clim. Change. 6 (2016) 35–41.
and obtain increased demand for carbon-intensive products. In contrast, [4] C. Izard, C. Weber, S. Matthews, Scrap the carbon tariff, Nat. Clim. Change. 1 (2010)
developing countries usually experience a reduction in total production 10–11.
[5] S. Tagliapietra, G.B. Wolff, Form a climate club: united States, European Union and
due to declined market share, and have the most welfare losses. Conse- China, Nature 591 (2021) 526–528.
quently, the gap between developing countries and developed countries [6] J. Fouré, H. Guimbard, S. Monjon, Border carbon adjustment and trade retaliation:
is further widened and global Gini coefficient is enlarged. Trade retal- what would be the cost for the European Union? Energy Econ. 54 (2016) 349–362.
[7] S. Monjon, P. Quirion, Addressing leakage in the EU ETS: border adjustment or
iation, justified by the export and production losses, leads to an offset output-based allocation? Ecol. Econ. 70 (2011) 1957–1971.
of the CBAM, with reduced environmental benefits but multiplied eco- [8] C. Böhringer, B. Bye, T. Fæhn, et al., Alternative designs for tariffs on embodied
nomic losses. As a result, the EU would be faced with a significant loss carbon: a global cost-effectiveness analysis, Energy Econ. 34 (2012) S143–S153.
[9] United Nations Conference on Trade and Development, A European Union Carbon
of real income and domestic production because it is a net exporter of
Border Adjustment Mechanism: implications for developing countries, (2021) 31.
carbon-intensive goods and would be sensitive to trade barriers in these [10] G. Perino, New EU ETS Phase 4 rules temporarily puncture waterbed, Nat. Clim.
sectors. However, retaliation still fails to compensate for the losses of Change 8 (2018) 262–264.
[11] United Nations Economic and Social Commission for Asia and The Pacific, Asia-
developing countries. Even worse, the inequality is magnified as poor-
Pacific Trade and Investment Report 2021: accelerating Climate-smart Trade and
est countries bear the most economic costs while the richest enjoy the Investment for Sustainable Development, United Nations, New York, 2021.
most benefits. [12] M. Chepeliev, Possible implications of the European carbon border adjustment mech-
In the light of the ineffectiveness of the CBAM as a unilateral mea- anism for Ukraine and Other EU trading partners, Energy Res. Lett. 2 (2021),
doi:10.46557/001c.21527.
sure, in reducing carbon leakage and the aggravated inequality, we sug- [13] Sanna Markkanen, Jorge Viñuales, Hector Pollitt, Hosuk Lee-Makiyama, Bence, Kiss-
gest that international cooperation on technological innovation is vital Dobronyi, Arushi Vaishnav, Kevin Le Merle, Lauren Gomez Cullen, On the Border-
for global climate change mitigation. As is officially stated in the pro- line: the EU CBAM and its place in the world of trade, Cambridge Institute for Sus-
tainability Leadership, University of Cambridge, Cambridge, UK, 2021.
posal of the CBAM, strong international cooperation will strengthen the [14] European Commission, Proposal for a Regulation of the European Parliament and of
joint climate action [14]. The signal effect of the CBAM may precipitate the Council Establishing a Carbon Border Adjustment Mechanism, Brussels (2021).

10
JID: FMRE
ARTICLE IN PRESS [m5GeSdc;April 13, 2023;18:55]

X. Sun, Z. Mi, L. Cheng et al. Fundamental Research xxx (xxxx) xxx

[15] European Parliament, Text Aopted - Carbon border adjustment mechanism, Brussels [37] S.B. Blomberg, G.D. Hess, How Much Does Violence Tax Trade? Rev. Econ. Stat. 88
(2022). (2006) 599–612.
[16] K. Appunn, Emission reduction panacea or recipe for trade war? The EU’s carbon [38] S. van Renssen, Round two for EU climate policy, Nat. Clim. Change. 3 (2013) 13–14.
border tax debate, J. Energy Transit. (2021) https://www.cleanenergywire.org/ [39] C. Gini, Measurement of Inequality of Incomes, Econ. J. 31 (1921) 124.
factsheets/emission-reduction-panacea-or-recipe-trade-war-eus-carbon-border-tax- [40] Z. Mi, J. Zheng, J. Meng, et al., Economic development and converging household
debate. (accessed October 21, 2021). carbon footprints in China, Nat. Sustain. 3 (2020) 529–537.
[17] European Commission, Carbon Border Adjustment Mechanism: questions and [41] International Monetary Fund, World Economic Outlook, 2015. https://www.imf.
Answers, (2021). https://ec.europa.eu/commission/presscorner/detail/en/qanda_ org/external/pubs/ft/weo/2015/02/weodata/groups.htm#cc.
21_3661. [42] World Bank, World Bank Country and Lending Groups, (2021). https://
[18] European Commission, The Commission proposes the next generation of datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-
EU own resources, (2021). https://ec.europa.eu/commission/presscorner/detail/ and-lending-groups (accessed September 1, 2022).
en/ip_21_7025. [43] G.P. Peters, S.J. Davis, R. Andrew, A synthesis of carbon in international trade, Bio-
[19] Leslie Hook, John Kerry warns EU against carbon border tax, Financ. Times. (2021) geosciences 9 (2012) 3247–3276.
https://www.ft.com/content/3d00d3c8-202d-4765-b0ae-e2b212bbca98. (accessed [44] P. Bayer, M. Aklin, The European Union emissions trading system reduced CO2 emis-
October 23, 2021). sions despite low prices, Proc. Natl. Acad. Sci. 117 (2020) 8804–8812.
[20] Sarah Anne Aarup, Barbara Moens, Australia attacks EU carbon border [45] C. Böhringer, E.J. Balistreri, T.F. Rutherford, The role of border carbon adjustment
levy plans, Politico (2021) https://www.politico.eu/article/cbam-detrimental-to- in unilateral climate policy: overview of an energy modeling forum study (EMF 29),
growth-australian-trade-minister-says/. Energy Econ. 34 (2012) S97–S110.
[21] E.J. Balistreri, C. Böhringer, T.F. Rutherford, Carbon policy and the structure of [46] U. Kohli, Real GDP, real domestic income, and terms-of-trade changes, J. Int. Econ.
global trade, World Econ. 41 (2018) 194–221. 62 (2004) 83–106.
[22] Muyu Xu, David Stanway, China says EU’s planned carbon border tax violates [47] U. DESA, World Population Prospects 2019. United Nations. Department of Eco-
trade principles, Reuters (2021) https://www.reuters.com/business/sustainable- nomic and Social Affairs, World Popul. Prospects 2019 (2019).
business/china-says-ecs-carbon-border-tax-is-expanding-climate-issues-trade- [48] J.-M. Burniaux, J. Chateau, R. Duval, Is there a case for carbon-based border
2021-07-26/. tax adjustment? An applied general equilibrium analysis, Appl. Econ. 45 (2013)
[23] Asit Ranjan Mishra, BRICS summit is likely to strongly oppose carbon tax pro- 2231–2240.
posed by EU, Mint (2021) https://www.livemint.com/news/india/brics-summit-is- [49] Bernard Looney, BP statistical review of world energy, London, UK, 2021.
likely-to-strongly-oppose-carbon-tax-proposed-by-eu-11631044909479.html. [50] A. Antimiani, V. Costantini, C. Martini, et al., Assessing alternative solutions to car-
[24] M. Springmann, Integrating emissions transfers into policy-making, Nat. Clim. bon leakage, Energy Econ. 36 (2013) 299–311.
Change. 4 (2014) 177–181. [51] M.B. Budolfson, D. Anthoff, F. Dennig, et al., Utilitarian benchmarks for emissions
[25] T. Kuusi, M. Björklund, V. Kaitila, et al., Carbon border adjustment mechanisms and pledges promote equity, climate and development, Nat. Clim. Change. 11 (2021)
and their economic impact on Finland and the EU, Prime Minister’s Office, 827–833.
Helsinki, 2020. https://www.etla.fi/en/publications/carbon-border-adjustment- [52] D. Helm, C. Hepburn, G. Ruta, Trade, climate change, and the political game theory
mechanisms-and-their-economic-impact-on-finland-and-the-eu/. of border carbon adjustments, Oxf. Rev. Econ. Policy. 28 (2012) 368–394.
[26] C. Böhringer, C. Fischer, K.E. Rosendahl, et al., Potential impacts and challenges of [53] O. Kuik, M. Hofkes, Border adjustment for European emissions trading: competitive-
border carbon adjustments, Nat. Clim. Change. 12 (2022) 22–29. ness and carbon leakage, Energy Policy 38 (2010) 1741–1748.
[27] D. Gros, C. Egenhofer, The case for taxing carbon at the border, Clim. Policy. 11 [54] F. Green, R. Denniss, Cutting with both arms of the scissors: the economic and politi-
(2011) 1262–1268. cal case for restrictive supply-side climate policies, Clim. Change. 150 (2018) 73–87.
[28] B. Bednar-Friedl, T. Schinko, K.W. Steininger, The relevance of process emissions for [55] B. Probst, S. Touboul, M. Glachant, et al., Global trends in the invention and diffusion
carbon leakage: a comparison of unilateral climate policy options with and without of climate change mitigation technologies, Nat. Energy. 6 (2021) 1077–1086.
border carbon adjustment, Energy Econ. 34 (2012) S168–S180. [56] N. Nakicenovic, P.D. Lund, Could Europe become the first climate-neutral continent?
[29] J. Lin, M. Du, L. Chen, et al., Carbon and health implications of trade restrictions, Nature 596 (2021) 486.
Nat. Commun. 10 (2019) 4947.
[30] E.L. Corong, M.E. Tsigas, The standard GTAP model, Version 7, J. Glob. Econ. Anal. Zhifu Mi is a professor of Climate Change Economics and
2 (2017) 119. the director of Research of The Bartlett School of Sustain-
[31] P.S. Armington, A theory of demand for products distinguished by place of produc- able Construction, University College London (UCL). He has
tion, Staff Pap 16 (1969) 159–178. published papers in leading journals, such as Lancet, Science
[32] J.-M. Burniaux, T.P. Truong, GTAP-E: an energy-environmental version of the GTAP Advances, Nature Energy, Nature Food, and Nature Sustainabil-
model, GTAP Tech. Pap. (2002) 18. ity. He was named in the Clarivate Highly Cited Researchers
[33] R. McDougall, A. Golub, GTAP-E release 6: a revised energy-environmental version list and was awarded the World Sustainability Award. In ad-
of the GTAP model (GTAP Technical Paper No. 15), Cent. Glob. Trade Anal. Purdue dition, his research was awarded the Best Paper Award in
Univ. West Lafayette USA (2007). Energy Economics, Top 50 Earth and Planetary Sciences Ar-
[34] Y. Dong, M. Ishikawa, T. Hagiwara, Economic and environmental impact analysis ticles in Nature Communications, Best Early Career Article in
of carbon tariffs on Chinese exports, Energy Econ. 50 (2015) 80–95. Environmental Research Letters, and Highly Cited Original Pa-
[35] J. Meng, Z. Mi, D. Guan, et al., The rise of South–South trade and its effect on global per in Applied Energy. He is an Editor-in-Chief of Structural
CO2 emissions, Nat. Commun. 9 (2018) 1871. Change and Economic Dynamics and an Executive Editor of
[36] United Nations, UN List of Least Developed Countries, (2021). https://unctad.org/ Journal of Cleaner Production.
topic/least-developed-countries/list.

11