Energy Policy 73 (2014) 269–277

Contents lists available at ScienceDirect

Energy Policy
journal homepage: www.elsevier.com/locate/enpol

Distributional effects of a carbon tax on Chinese households:

A case of Shanghai
Zhujun Jiang a,b,n,1, Shuai Shao c,nn,1
a
Institute for Advanced Research, Shanghai University of Finance and Economics, Shanghai 200433, China
b
Key Laboratory of Mathematical Economics, Shanghai University of Finance and Economics, Ministry of Education, Shanghai 200433, China
c
Institute of Finance and Economics Research, School of Urban and Regional Science, Shanghai University of Finance and Economics, Shanghai 200434, China

H I G H L I G H T S

The direct distributional effect of carbon tax presents a weak progressivity.

The indirect distributional effect of carbon tax is significantly regressive.

The comprehensive distributional effect of carbon tax is regressive.

The Suits index of carbon tax is  0.078.

Imposing carbon tax on fossil fuels can intensify income inequality.

art ic l e i nf o a b s t r a c t

Article history: As an effective policy instrument to reduce CO2 emissions, the effects of a carbon tax on distribution
Received 8 February 2014 have been the critical factor in determining whether a carbon tax will be acceptable in China. Taking
Received in revised form Shanghai as an example, which is the economic center and front-runner of China, this paper estimates
2 May 2014
the distributional effect of a carbon tax on households in various income groups by using the input–
Accepted 4 June 2014
Available online 7 July 2014
output model and the Suits index. The results indicate that the comprehensive distributional effect of the
carbon tax is regressive. The expenditure of the low-income group caused by the carbon tax accounts for
Keywords: 0.853% of the total expenditure, while that of the high-income group 0.712%. The direct distributional
Carbon tax effect presents a weak progressivity, while the indirect one is significantly regressive, and the latter is
Distributional effect
much larger than the former. Moreover, the Suits index of the carbon tax is  0.078, implying that the
Suits index
carbon tax burden on the low-income group is the highest and thus that a carbon tax can intensify
income inequality. Therefore, when introducing a carbon tax, some rational associated redistribution or
compensation measures, such as purposive transfer payments, should be implemented to restrict or
even eliminate the regressivity of the carbon tax.
& 2014 Elsevier Ltd. All rights reserved.

1. Introduction emission-reduction measure (Baranzini et al., 2000). As an envir-

As a major issue of human survival and development, global
climate warming has paid extensive attention all over the world.
Among the various climate-change mitigation policies, the carbon
tax is extensively advocated since it is regarded as the lowest cost
n
Corresponding author at: Institute for Advanced Research, Shanghai University
of Finance and Economics, Shanghai 200433, China. Tel.: þ 86 2165902562.
nn
Corresponding author. Tel.: þ 86 2165902562.
E-mail addresses: jiang.zhujun@mail.shufe.edu.cn (Z. Jiang),
shao.shuai@mail.shufe.edu.cn (S. Shao).
1
Authors contributed equally to this work and are listed in alphabetical order
by surname.
onmental tax imposed on CO2 emissions, the carbon tax can
reduce CO2 emissions and thus mitigate climate warming by
increasing tax burden. Theoretically, the essential role of a carbon
tax is to correct distorted price signals and optimize the resource
allocation by internalizing the environmental externality caused
by anthropogenic CO2 emissions. Therefore, the carbon tax is
considered as one of the most market-efficient economic mea-
sures in reducing CO2 emissions. Supporters of the carbon tax
argue that it has three main advantages. First, a carbon tax can
enhance the competitive power of renewable energy in cost and
thus promote the utilization of renewable energy. Second, the
revenue from a carbon tax can be used to subsidize environmental
protection project and energy-saving and emission-reducing

http://dx.doi.org/10.1016/j.enpol.2014.06.005
0301-4215/& 2014 Elsevier Ltd. All rights reserved.
270 Z. Jiang, S. Shao / Energy Policy 73 (2014) 269–277
technology. Finally, compared with carbon emissions trading, a
carbon tax has lower administrative cost, more convenient imple-
mentation process, and more predictable effects.
A carbon tax has been carried out in some countries and has
achieved some desired results. However, until now, there are only
a few developed countries actually imposing carbon taxes, such as
Finland, Sweden, Denmark, Norway, Netherlands, and Italy. Most
countries are cautious about the carbon tax considering that it
may produce some negative impacts on national economy by
affecting energy price and supply–demand relationship, increasing
enterprises' costs, and generating the regressive effect of taxation.
Discussions on mechanisms and channels of a carbon tax affecting
national economy have been academic focus issue as well as the
crucial basis for governments to determine whether to levy a
carbon tax. Among various potential impacts of a carbon tax on
economy, the influential direction and degree of a carbon tax on
distribution are the critical factor in determining whether a carbon
tax will be accepted by the public. Hence, it is also a key issue
focused on by both governments and academic circles.
Related studies mainly focus on whether a carbon tax will
become a regressive tax or not. Whereas, the consensus on this
issue is not reached since different conclusions are drawn by using
different samples. Empirical evidences from most developed
countries indicate that a carbon tax has a regressive effect on
distribution (Poterba, 1991; Jacobsen et al., 2003; Kerkhof et al.,
2008; Shammin and Bullard, 2009), while some researchers argue
that the distributional effect of a carbon tax is proportional or
progressive (Symons et al., 2000; Tiezzi, 2005; Dissou and
Siddiqui, 2014). Bureau (2011) found that a carbon tax was
regressive before revenue recycling. However, Gonzalez (2012)
argued that carbon taxes were not necessarily regressive and
whether the way revenue was recycled became a major determi-
nant of how the carbon tax costs were distributed. Most existing
studies focus on developed countries and there are fewer studies
on developing countries. In developing countries, many factors,
such as market forces, price regulation, and import restriction, are
likely to generate the distributional effect of environmental policy.
The results of Brenner et al. (2007) and Yusuf and Resosudarmo
(2007) on China and Indonesia, respectively, indicate that a carbon
tax has the progressive impact on distribution.
In 2009, the Chinese government announced its emission–
reduction target that CO2 emissions per unit of GDP would be
reduced by 40–45% in 2020 than 2005 level. As a part of this
target, China's 12th “Five-Year Plan” explicitly proposed an obli-
gatory indicator that CO2 emissions per unit of GDP should decline
by 17% compared with 2010 level. As an effective policy instru-
ment to reduce CO2 emissions, the pleas to implement a carbon
tax are becoming louder in China, but the understanding of the
Chinese government in the potential effects of a carbon tax is not
clear, especially in the distributional effect of the carbon tax. As a
developing great power characterized by obvious imbalance
among regions in natural resource endowment and economic
development, the distributional effect of a carbon tax is particu-
larly complex in China. The issue has become one of the vital
references in determining whether to implement a carbon tax in
China. Hence, the distributional effect of the carbon tax should be
paid more attention to.
In recent years, although studies on China's economic effect of
a carbon tax have become plentiful, most existing literature focus
on the impact of a carbon tax on the macro-economy or the
competitive power of various industries, while the potential effect
of a carbon tax on distribution is paid little attention to. Based on
the CGE model, Cao (2009) carried out a comparative analysis on
transfer payment effects of different allocation modes of the
carbon tax revenue in various proportional taxation scenarios,
but did not directly discuss the distribution effect of a carbon tax.
Su et al. (2009) analyzed the effects of a carbon tax on China's
macro-economy, industrial development, and income distribution
in various scenarios of tax rate. However, Su et al. (2009) focused
on the effect of a carbon tax on the income inequality of urban and
rural dwellers and did not analyze the influential mechanism of
the carbon tax on distribution.
In view of limitations above, taking Shanghai as an example,
which is an international metropolis and China's economic center
and whose economic growth and energy consumption are all on
the top in China, this paper estimates the distributional effect of a
carbon tax on different income groups and discusses the influen-
tial mechanism and channels by using the input–output model to
provide some theoretical basis and decision-making reference for
the implementation of a carbon tax in China.
As the economic center and front-runner of China, Shanghai is
expected to play a leading role in climate change mitigation. With
the rapid development, the dependence of Shanghai's economy on
energy consumption is increasing. According to Shanghai Statistical
Yearbook, the total energy consumption in Shanghai has increased
from 394.67 million tons of coal equivalent (tce) in 1993 to
1292.61 million tce in 2011, with a average annual growth rate
of over 10%. Moreover, with the deepening of energy dependence,
Shanghai's CO2 emissions also present an increasing trend. The
total CO2 emissions from production sectors in Shanghai have
increased from 525.02 million tons in 1993 to 1292.61 in 2011,
with an average annual growth rate of above 8% (Yang and Shao,
2013). Obviously, Shanghai is encountering the great pressure to
realize the target of energy saving and emission reduction (Shao et
al., 2011). Therefore, Shanghai is a representative sample in China,
and the study on Shanghai can attain some necessary decision-
making references to the policy design of a carbon tax in China.
The remainder of the paper is structured as follows. Section 2
addresses the fossil energy consumption of households in various
income groups in Shanghai. Section 3 estimates direct and indirect
effects of a carbon tax on various income groups. In Section 4, the
progressivity of a carbon tax is measured and discussed. Conclu-
sions and policy implications are provided in Section 5.
2. Energy consumption in various income groups
As reported in Shanghai Statistical Yearbook, in 2010, the
primary energy consumption in Shanghai reached 111.61 million
tce, in which coal, oil, and natural gas consumptions were 58.76
million tons, 32.98 million tons, and 4.50 billion cubic meters,
respectively. As the main source of CO2 emissions, the share of
fossil fuels in the total primary energy consumption is about 85%
in Shanghai. According to the Word Bank Report, above 70% of the
total CO2 emissions are from fossil fuels in the world. Moreover, it
is a consensus that imposing the carbon tax upstream is more
effective than downstream. On one hand, it is more practicable to
impose on fossil fuels than on CO2 emissions per se. On the other
hand, the administrative costs of a carbon tax could be lower
through its upstream implementation (Metcalf, 2009). Hence, like
some countries where a carbon tax has been imposed (e.g.,
Sweden, Norway, Denmark), we treat fossil fuels as the carbon
tax base.
In addition, it is necessary to classify the households to discuss
the distributional effect of a carbon tax. Similar to most related
studies, we employ the classification by income levels of house-
holds. According to the recent classification standard of the
National Bureau of Statistics of China, the Tabulation on the 2010
Population Census of Shanghai Municipality shows that in 2010, the
urban population of Shanghai is 20.56 million, accounting for
89.3% of the total population of Shanghai, while the rural popula-
tion only accounts for 10.7%. Furthermore, due to the limitation of
 Z. Jiang, S. Shao / Energy Policy 73 (2014) 269–277 271

data availability, we cannot attain the detailed data about the Table 1
energy consumption of various income levels of rural households. Per capita energy consumptions of various income groups in 2010 (Unit: kgtce).
Source: Authors calculate by using related statistical data.
Therefore, we focus on the urban households which contain
almost 90% of the total population of Shanghai. According to Group Transport fuela Residential fuelb
Shanghai Statistical Yearbook, the urban households can be divided
into five groups: low, mid-low, mid, mid-high, and high income Low 20.67 56.37
groups. Mid-low 20.87 67.44
Mid 67.27 69.86
The energy consumption types of households include coal, Mid-high 105.93 69.86
petrol, diesel, liquefied petroleum gas (LPG), fuel gas, and elec- High 233.47 85.08
tricity.2 We estimate various types of energy consumptions of
a
different income groups in Shanghai by using the related data Transport fuels contain petrol and diesel.
b
Residential fuels contain coal, natural gas, pipeline gas, and LPG.
because it is unavailable directly from the statistical data. The
estimation process is as follows.
With respect to the consumption of petrol and diesel, we can 350
first obtain the per capita traffic expenditure of each group from Transport fuels Residential fuels
Shanghai Statistical Yearbook, which contains four aspects: family
300
vehicle, transport fuels, the services of vehicle, and traffic fees
according to China Urban Life and Price Yearbook. Compared the

Per capita energy consumption/kgtce

expenditure of each group of Shanghai with the corresponding
data of China, we can estimate the share of transport fuel
expenditure of each group in the total traffic expenditure and
then get the per capita expenditure of each group on transport
fuels in Shanghai. Considering that the majority of transport fuels
in China are petrol and diesel, the expenditure on transport fuels
can be treated as that on petrol and diesel. Multiplying the per
capita expenditure of each group on transport fuels by the
corresponding population of each group, the share of the expen-
diture of each group on petrol and diesel in the total expenditure
of five groups on them can be calculated. The share is equal to the
share of real petrol and diesel consumption of each group in the
total petrol and diesel consumption since all the households in
Shanghai face the unified oil prices. Furthermore, China Energy
Statistical Yearbook reports the various energy consumption of
urban households in Shanghai, including the petrol and diesel
consumption. At last, multiplying the total petrol and diesel
consumption in Shanghai by the share above and then divided
by the population of each group, we can attain the per capita
petrol and diesel consumption of each group.
With respect to residential fuels like coal and gas, the related
statistical data do not report their detailed categorical data.
However, we can obtain the per capita expenditure of each group
on residential fuels from Shanghai Residents' Life and Price Year-
book, which can be regarded as a whole of all kinds of residential
fuels. Multiplying the per capita expenditure of each group on
residential fuels by the corresponding population of each group,
the share of the expenditure of each group on them in the total
expenditure of five groups can be estimated. Since all the house-
holds in Shanghai face the unified fuel prices, the share can also be
treated as the share of real residential fuels consumption of each
group. Based on the related data from China Energy Statistical
Yearbook, summing up the consumptions of coal, natural gas,
pipeline gas, and LPG in the unit of coal equivalent, we can obtain
the total residential fuels consumption of urban households in
Shanghai. Furthermore, multiplying the total residential fuels
consumption by the share above and then divided by the popula-
tion of each group, the per capita residential fuels consumption of
each group in Shanghai can be calculated.3
Estimated per capita energy consumptions of various groups
in 2010 are reported in Table 1 and Fig. 1. According to “energy
2
Since electricity is a secondary energy source and does not emit CO2 in its
utilization process, it is not been taken into account in this paper.
3
The evidence from the micro-level data, such as the survey data of China's
urban residents, may be more accurate and credible, but the energy data in China,
especially the micro-level energy data, are very limited. Hence, we mainly use the
data released by the related statistical yearbook.
250
200
150
100
50
0
Low Mid-low Mid Mid-high High
Fig. 1. Per capita energy consumption of various income groups in 2010.
ladder” theory, the households with a low income usually rely on
more traditional fuels like coal, while those with a high income
mainly use more modern fuels, such as natural gas and transport
fuels (petrol and diesel) (Barnes et al., 1997; Masera et al., 2000).
Our results are consistent with this theory (see Fig. 1). The
transport fuels consumption raises rapidly with the rise of house-
holds' income. The per capita transport fuels consumption of the
low-income group is only 20.67 kgtce, while that of the high-
income group reaches 233.47 kgtce, which is ten times more than
the former, resulting from the more possession of private vehicles
of the high-income group. In the low-income group, the con-
sumption of transport fuels is obviously less than that of residen-
tial fuels, while in the high-income group, the corresponding
situation is reverse. The results indicate that the greater the
income gap, the more obvious the energy ladder phenomenon.
3. Distributional effects of carbon tax
Since the households in different income groups present
different lifestyles and consumption patterns, they are likely to
substantially differ from the carbon tax payments. A carbon tax
can affect prices of fossil fuels and consumers' expenditure in two
channels. On one hand, due to the rise of fuel prices caused by a
carbon tax, households' expenditure on fuels as basic means of
subsistence (e.g., coal, petrol, and gas) will directly increase.
On the other hand, since a carbon tax will push up the prices
of products and services, especially energy-intensive ones, by
increasing the cost of production factor and raw material, house-
holds' expenditure on those products and services will indirectly
increase. Such a channel can be regarded as the indirect effect of
272 Z. Jiang, S. Shao / Energy Policy 73 (2014) 269–277

a carbon tax on households' expenditure (Symons et al., 1994; Table 2

Rapanos, 1995; Cornwell and Creedy, 1996; Labandeira and Direct effect of a carbon tax.
Labeaga, 1999). Therefore, the distributional effect of a carbon
Group Per capita tax Share of the tax payment
tax can be divided into the direct effect and the indirect effect. payment (RMB) in expenditure (%)
Wier et al. (2005) pointed out that using the total households'
expenditure may be more preferable than using the income to Low 3.062 0.024
analyze the regressivity of the carbon tax. First, the tax base is Mid-low 3.508 0.022
Mid 5.472 0.025
expenditure, making it relevant to measure the regressivity Mid-high 7.029 0.026
relative to expenditure. Second, since households seek to smooth High 12.767 0.031
consumption over their life cycle according to permanent income
hypothesis, actual income may be inappropriate to measure the
regressivity. Thus, we adopt the impact of a carbon tax on the
households' expenditure to estimate its distributional effect,
which is measured by the households' tax payments relative to 15 0.035
Per capita tax payment
expenditure for the deciles. Proportion of carbon tax

3.1. Direct effect of carbon tax 12

0.030

Proportion of tax payment/%

Per capita tax payment/CNY
The direct effect of a carbon tax refers to affecting the house-
holds' expenditure through the direct fuel consumption. Assuming 9
that there is no energy substitution and households' energy
demand keeps constant when introducing a carbon tax,4 the direct 0.025

effect can be expressed as follows: 6

TAX direct
i ¼ ∑ TAX direct
ij ¼ ∑ Q j CEF j T ð1Þ
j j
0.020
where subscript i denotes various groups of households, subscript j 3

refers to various fuels, TAX direct

i denotes the direct effect of a
carbon tax of the ith group, TAX direct
ij is the direct effect of a carbon
tax of the ith group based on the jth fuel, Q j and CEF j are the
consumption and CO2 emissions coefficient of the jth fuel, respec-
tively, and T is the carbon tax per unit of CO2 emissions, i.e., the
carbon tax rate.
We estimate CEF j by using the reference method and para-
meters in 2006 IPCC Guidelines for National Greenhouse Gas
Inventories together with China's released relevant parameters
(see the detailed below).
With respect to the CO2 emissions coefficient of transport fuels,
according to China Energy Statistical Yearbook, we can first calcu-
late the shares of petrol and diesel in the total transport fuels
consumption of urban residents in Shanghai, respectively. Then,
multiplying their shares by the CO2 emissions coefficients of petrol
and diesel, respectively, and summing them up, we can obtain the
weighted CO2 emissions coefficient of transport fuels. With respect
to the CO2 emissions coefficient of residential fuels, because their
consumptions are measured in the standard unit of coal equiva-
lent, we first convert the CO2 emissions coefficient measured in
the physical quantity into that in the standard unit of coal
equivalent. Then, multiplying their CO2 emissions coefficients by
their shares in the total residential fuels consumption, respec-
tively, and summing them up, we can get the weighted CO2
emissions coefficient of residential fuels.
The Research Group of Ministry of Finance of China suggested
that China's carbon tax rate should be 10 RMB per ton of CO2 at the
beginning of its implementation, while the Planning Institute of
Ministry of Environmental Protection of China proposed 20 RMB
per ton of CO2. Chen (2011) argued that the rational range of the
4 simulate the indirect distributional effect of a carbon tax on
The fact may not be so. However, we focus on households’ consumption of
conventional fuels, which are difficult to be substituted by new fuels in the short
term. Moreover, considering that the change of fuels demand depends on their
price elasticity per se and cross-price elasticity, levying a carbon tax will push up
the prices of fuels with high carbon emissions and thus lead to more demand for
fuels with low carbon emissions. Such an increase in demand may in turn cause the
rise of prices of low-carbon fuels. Undoubtedly, it is very difficult to precisely
forecast the change of various fuels consumption of various groups. Hence, we
make the simplifying assumptions above.
0 0.015
Group Low Low-mid Mid Mid-high High
Fig. 2. Direct effect of a carbon tax.
carbon tax rate should be 19–46 RMB per ton of CO2 for China's
industry in 2009–2020. Referring to the studies above, the middle
level of their purposed carbon tax rates, i.e., 20 RMB per ton of CO2
is chosen in this paper. The estimated results of the direct effect of
a carbon tax on various income groups are presented in Table 2
and Fig. 2.
The direct distributional effect of a carbon tax mainly relies on
households' fuels consumption and its structure, which are closely
related to households' income level. It is easily seen that the direct
carbon tax payment increases with the income level, and its level
is relatively low. Even for the high-income group who bears the
largest direct impact, the carbon tax payment is only 12.767 RMB,
accounting for about 0.031% of the total household expenditure.
Hence, the direct effect of a carbon tax presents a weak progres-
sivity and the tax burdens on various income groups are all
relatively small.
3.2. Indirect effect of carbon tax
3.2.1. Methods
As mentioned above, the indirect effect of a carbon tax roots
in the increasing production cost of commodity induced by the
carbon tax. Since it involves industrial linkage, the input–output
model reflecting industrial relationship is a prevalent method to
households (Symons et al., 1994, 2000; Cornwell and Creedy,
1996; Labandeira and Labeaga, 1999; Wier et al., 2005; Ahmad
and Stern, 2009; Grainger and Kolstad, 2010; Datta, 2010; Sun and
Ueta, 2011; Fan and Zhang, 2013; Mathur and Morris, 2014). The
existing research samples cover developed and developing coun-
tries, e.g., US, UK, France, Italy, Spain, Danish, Australia, India,
Pakistan, and China.
 Z. Jiang, S. Shao / Energy Policy 73 (2014) 269–277
This approach generally assumes that no energy substitution
takes place in the production process when introducing a carbon
tax, and the tax burdens on producers are fully transmitted into
final commodity prices. Thus, those households that demand a
commodity will eventually pay the carbon tax paid by the
producer of this commodity (Labandeira and Labeaga, 1999;
Wier et al., 2005; Ahmad and Stern, 2009; Sun and Ueta, 2011).
We will employ this approach, although the degree of transmis-
sion will depend on technological development and substitution
possibilities in industries as well as in households. Moreover,
because the estimate of indirect carbon tax payments may suffer
from some uncertainty, the tax payments are assumed to be linear
in the production of each sector (Labandeira and Labeaga, 1999;
Wier et al., 2005; Grainger and Kolstad, 2010; Fan and Zhang,
2013). Thus, the carbon tax payments from the indirect energy
demand of households can be expressed as follows:
TAX indirect
i ¼ TEðI  AÞ  1 Cei
where subscript i denotes various groups of households, TAX indirect
is the total indirect carbon tax payments of households as a
consequence of the consumption of goods and services, T is the
carbon tax per unit of CO2 emissions, E is a vector of the CO2
emission intensity (i.e., CO2 emissions per unit of economic out-
put) of each sector in the input–output table, ðI  AÞ  1 is the
Leontief inverse matrix, I is a unit matrix and A is a coefficient
matrix describing inter-sector commodity flows, C is a matrix of
the composition of consumption commodity aggregates, i.e.,
the private consumption commodity aggregates apportioned by
production sectors, and e is a vector including the households'
expenditure of each commodity consumption.
3.2.2. Data
To get the CO2 emission intensity of each industry, the CO2
emissions should be estimated. Since we assume the carbon tax
levying upon fossil fuels, we consider the CO2 emissions of each
industry from the fossil fuels combustion rather than those from
the production process. We employ the method and parameters
Table 3
CO2 emission intensity of 42 sectors in 2007 (Unit: kg/RMB).
Sector CO2 emission
intensity
Production and supply of electric power and heat power 0.7369
Processing of petroleum, coking, and nuclear fuel 0.1882
Smelting and pressing of ferrous metals 0.1641
Transport, storage, and post 0.1292
Manufacture of raw chemical materials and chemical products 0.0558
Agriculture, forestry, animal husbandry, and fishery 0.0511
Manufacture of non-metallic mineral products 0.0503
Manufacture of paper and paper products 0.0349
Manufacture of rubber 0.0291
Manufacture of textile 0.0252
Wholesale, retail trades, hotels, and catering services 0.0202
Manufacture of medicines 0.0141
Processing of food from agricultural products 0.0129
Manufacture of chemical fibers 0.0127
Manufacture of beverage 0.0120
Extraction of petroleum and natural gas 0.0116
Manufacture of foods 0.0109
Construction 0.0106
Processing of timber and manufacture of wood, bamboo, rattan, 0.0099
palm, and straw products
Smelting and pressing of non-ferrous metals 0.0085
Manufacture of plastics 0.0082
a
Expect the extraction of petroleum and natural gas, mining and processing industries in Shanghai have no actual output and fuel consumption in most years due to the
poor natural resources endowment in Shanghai.
273
used in our previous study (Shao et al., 2011) to estimate the CO2
emissions of each industry, i.e., the reference method and para-
meters in the 2nd Volume (Energy Volume) of 2006 IPCC Guide-
lines for National Greenhouse Gas Inventories together with China's
released relevant parameters. To obtain more accurate results, we
consider all types of final fossil fuels consumption reported in the
statistical data except electricity and heat since CO2 is not directly
emitted in the utilization of electricity and heat. The 15 kinds of
involved energy source include raw coal, cleaned coal, coke, coke
oven gas, other gases, crude oil, petrol, kerosene, diesel, fuel oil,
liquefied petroleum gas, refinery gas, natural gas, other petroleum
products, and other coking products. As we know, this paper takes
into account the most types of energy source among the related
studies on Shanghai. The estimation formula is as follows:
C i ¼ ∑ C ij ¼ ∑ Eij  N j  CC j  Oj  M ð3Þ
j j
ð2Þ where subscript i denotes the industrial sector, subscript j refers to
various fuels, C i stands for the gross energy-related CO2 emissions
of the ith sector, C ij is the CO2 emissions of the ith sector based on
the jth fuel, Eij denotes the total energy consumption of the ith
sector based on the jth fuel, N j , CC j , and Oj are the net calorific
value, carbon content, and the carbon oxidation factor of the jth
fuel, respectively, and M is the molecular weight ratio of CO2 to
carbon (44/12).
The use of special parameters of each country is encouraged by
the IPCC (2006) based on its methods. Therefore, we adopt the
principle of priority to select the related parameters announced
officially in China and the second choice of the defaults provided
by the IPCC (2006) to assure the accuracy of the results. The
involved data are derived from Shanghai Statistical Yearbooks on
Industry, Energy, and Transport, China Energy Statistical Yearbook,
and The People's Republic of China National Greenhouse Gas
Inventory.
According to the classification of industrial sectors in the China
Energy Statistical Yearbook, we re-classify and integrate 133 sectors
in the 2007 Input–Output Tables of Shanghai into 42 sectors, whose
output data can be derived from Shanghai Statistical Yearbooks on
Sector CO2 emission
intensity
Manufacture of textile wearing apparel, footwear and caps 0.0071
Other sectors 0.0065
Manufacture of metal products 0.0062
Manufacture of articles for culture, education, and sport activities 0.0053
Manufacture of general purpose machinery 0.0050
Printing, reproduction of recording media 0.0049
Manufacture of artwork and the recycling and disposal of waste 0.0042
Manufacture of leather, fur, feather and related products 0.0032
Manufacture of transport equipment 0.0026
Manufacture of furniture 0.0021
Manufacture of special purpose machinery 0.0021
Manufacture of electrical machinery and equipment 0.0019
Production and supply of fuel gas 0.0013
Production and supply of water 0.0011
Manufacture of measuring instruments and machinery for 0.0006
cultural activity and office work
Manufacture of communication equipment, computers and other 0.0005
electronic equipment
Manufacture of tobacco 0.0003
Mining and washing of coal N.A.a
Mining and processing of ferrous metal ores N.A.a
Mining and processing of non-ferrous metal ores N.A.a
Mining and processing of non-metal ores N.A.a
274 Z. Jiang, S. Shao / Energy Policy 73 (2014) 269–277

Industry, Energy, and Transport. The calculated CO2 emission Table 4

intensity of each sector in 2007 is shown in Table 3. Indirect effect of a carbon tax.
Table 3 indicates that the production and supply sector of
Group Per capita tax Share of the tax payment
electric power and heat power in Shanghai presents the highest payment (RMB) in expenditure (%)
CO2 emission intensity as a result of 0.737 kg/RMB. This is mainly
because that the coal-fired power plants are dominant in Shang- Low 104.08 0.829
hai, which consume a large number of coal as the highest emitting Mid-low 130.46 0.817
Mid 160.43 0.742
fossil fuel. As expected, some energy-intensive sectors, such as Mid-high 195.90 0.732
processing of petroleum, coking, nuclear fuel, smelting and press- High 277.29 0.681
ing of ferrous metals, and transport, storage, and post also show a
relatively high CO2 emission intensity, while the manufactures of
tobacco, communication equipment, computers and other electro-
nic equipment, and measuring instruments and machinery for
cultural activity and office work have a remarkably low CO2 300 Direct effect Indirect effect 0.87
emission intensity. Proportion of tax payment

250 0.84
3.2.3. Results and discussion
According to Shanghai Statistical Yearbook, the expenditure of

Proportion of tax payment/%

Per capita tax payment/CNY
households in Shanghai includes eight categories: food, clothing, 200
0.81
residence, household facilities and articles, health care and med-
ical services, transport and communications, education, cultural
150
and recreation, other miscellaneous goods and services. As shown 0.78

in Fig. 3, there is an obvious difference in both scale and structure

of expenditure among various groups in Shanghai. The low-
income group pays more for food, with the highest Engel coeffi-
cient of 42%. With the rise of income level, the share of food in the
total expenditure decreases, while the transport and communica-
tions expenditure noticeably increases. Such results indicate that
the expenditure of the low-income and high-income groups is
mainly used for basic requirements and the high quality of life,
respectively.
Based on the input–output table released by the Shanghai
Statistics Bureau, we merge the input–output table into 42 sectors.
And then, we match the residential consumption categories to the
corresponding sectors in the input–output table. According to
Eq. (2), the indirect effects of a carbon tax on various income
groups are estimated and presented in Table 4.
The estimated results indicate that although the indirect
carbon tax payment increases with the income level, its share in
the total expenditure presents a downward trend from low-
income group to high-income group. Hence, the indirect effect of
a carbon tax shows a significant regressivity. Compared with the
high-income group, the low-income group bears heavier tax. The
share of the carbon tax payment in the total expenditure of the
low-income group is 0.829%, while this share of the high-income
45000 Food Clothing Residence
Health Care and Medical Services
Education, Cultural and Recreation Services
40000 Miscellaneous Goods and Services
35000
30000
Expenditure/CNY
Household Facilities and Services
Transport and Communications
100
0.75
50
0.72
0
Group Low Low-high Mid Mid-high High
Fig. 4. Total distributional effect of a carbon tax.
group is 0.681%. Summing up the direct effect in Table 2 and the
indirect effect in Table 4, we can get the total distributional effect
of a carbon tax (see Fig. 4).
Similar to the results of the indirect effect, Fig. 4 suggests that
the total distributional effect of a carbon tax presents an evident
regressivity. As income rises, the total carbon tax payment
increases, but it constitutes a smaller and smaller share in the
total expenditure, decreasing from 0.853% of the low-income
group to 0.712% of the high-income group. Hence, the low-
income group bears more tax burden than other income groups.
This finding is consistent with the conclusions of most current
studies (Cornwell and Creedy, 1996; Brännlund and Nordström,
2004; Kerkhof et al., 2008; Shammin and Bullard, 2009; Mathur
and Morris, 2014), i.e., the distributional effect of a carbon tax is
regressive. In addition, we find that the indirect effect is much
larger than the direct. This means that the indirect distributional
effect of a carbon tax on households induced by pushing up the
prices of products and services, especially energy-intensive ones,
is far greater than its direct distributional effect caused by
consuming fuels.

25000

4. Measurement of carbon tax progressivity

20000

The tax progressivity is the most commonly used concept to

15000
study the tax distribution in different income groups. The Suits
10000
index proposed by Suits (1977) is widely adopted to examine the
progressivity and regressivity of taxes, which is based on the
5000 principle of the concentration curve and the Gini coefficient. Fig. 5
depicts an example of a concentration curve used to calculate the
0 Suits index, where the accumulated percent of the tax burden is
Group Low Low-mid Mid Mid-high High
plotted vertically while the accumulated percent of the income on
Fig. 3. Expenditure structure of various income groups in Shanghai in 2010. the horizontal axis.
 Z. Jiang, S. Shao / Energy Policy 73 (2014) 269–277 275

100 B Table 5
Suits index of a carbon tax.

Effect Suits index

Direct 0.022
Accumulated percent of tax burden

Indirect  0.082
Total  0.078

50 100

C
80

Accumulated percent of tax burden (direct)

L

60

0 A
0 50 100
Accumulated percent of income 40
Fig. 5. An example of the concentration curve used to calculate the Suits index.

Following Suits (1977), the Suits index is specified as follows:

20
S ¼ ðK  LÞ=K ¼ 1  ðL=KÞ ð4Þ
where S denotes the Suits index (i.e., the index of tax progressiv-
ity), K refers to the area of the triangle OAB in Fig. 5, and L stands 0
0 20 40 60 80 100
for the area OABC between the curve and the horizontal axis OA.
The Suits index varies from  1 to 1. In the case of regressivity, Accumulated percent of income

L 4K and 1 rS o0, while a proportional tax implies L ¼K and
S ¼0, and a progressive tax yields LoK and 0 oS r1. Suggested in
Suits (1977):
1 n
S ¼ 1 ∑ ½Tðyi Þ þTðyi  1 Þðyi  yi  1 Þ
10000 i ¼ 1
where yi denotes the accumulated percent of the income, mea-
sured on the horizontal axis, which ranges from 0 to 100, Tðyi Þ is
the corresponding accumulated percent of the tax burden from a
given tax, and n stands for the number of households' income
groups.
Since the 1980s, many studies employed the Suits index to
measure the distribution effect of taxation or public expenditure.
For instance, Kienzle (1981) empirically demonstrated that the
Suits measurement methodology applied in a broader and more
useful sense in public taxation and expenditure analysis than the
previous methods. Sarte (1997) combined a dynamic equilibrium
model with the Suits index to analyze the effects of progressive
taxation on the degree of income inequality in a competitive
economy and asserted that the Suits index is a helpful indication of
the tax burden borne by each agent type. Using the Suits index,
Metcalf (1999) assessed the distributional impact of a shift toward
greater reliance on various environmental tax policies. Further-
more, Price and Novak (1999) and Hansen et al. (2000) employed
the Suits index to investigate the issue of the tax incidence of
lotteries. The studies for China were conducted by Liu and Nie
(2004, 2009), who used the Suits index to estimate the distribu-
tional impact of three major indirect taxes, including value added
tax, consumption tax, and business tax. Recently, Sterner (2012)
also adopted the Suits index to examine the distributional effects
of transport fuels taxation in seven European countries and found
some weak evidence of regressivity.
However, the study using the Suits index to measure the
distributional effect of a carbon tax in China is still absent.
According to Eq. (5), the Suits index of a carbon tax in Shanghai
is estimated and reported in Table 5. Once levying a carbon tax on
Fig. 6. Concentration curve of the direct effect.
fossil fuels, the Suits index of the direct effect is 0.022, indicating
ð5Þ that the direct effect of the carbon tax is progressive, but the
progressivity is small, approximately proportional. The Suits
indices of the indirect effect and the total effect are  0.082 and
 0.078, respectively. Therefore, on the whole, levying a carbon tax
on fossil fuels appears regressive in Shanghai since the tax burden
on low-income households is greater than that of high-
income ones.
Furthermore, we present the concentration curves of distribu-
tional effects of a carbon tax in Figs. 6–8, which also consolidate
the results in Table 5. Fig. 6 shows that the concentration curve
is below the diagonal line but very close to it. This suggests that
the progressivity of the direct effect of a carbon tax is weak in
Shanghai. Figs. 7 and 8 show that the concentration curves are
above the diagonal line, further indicating that the indirect effect
and the total effect of a carbon tax in Shanghai are all regressive.
5. Conclusions and policy implications
China has become the largest CO2 emissions country in the
world, accounting for one-quarter of global CO2 emissions in 2011
and 80% of the world's rise in CO2 emissions since 2008 (Peters et
al., 2012; Liu et al., 2013). With the deepening of industrialization
and urbanization, China's economic development has to encounter
the rigid energy demands and severe environmental constrains.
Energy saving and emission reduction have become the necessary
way to resolve the conflicts among economy, energy, and the
environment, and to realize the sustainable development. As the
economic center of China, Shanghai should play a leading role in
energy saving and emission reduction. Although a carbon tax is
generally regarded as an effective policy instrument to reduce
CO2 emissions, it will inevitably generate the redistribution of
276
100
Accumulated percent of tax burden (indirect)
80
60
40
20
0
0 20
Accumulated percent of income
Fig. 7. Concentration curve of the indirect effect.
100
80
Accumulated percent of tax burden(total)
60
40
20
0 the damage to the social welfare caused by a carbon tax and thus
0 20
Accumulated percent of income
Fig. 8. Concentration curve of the total effect.
households' income. Thus, this paper estimates and discusses the
impacts of a carbon tax on the expenditure of households in
different income groups in Shanghai. The main conclusions are
summarized as follows.
First, households' energy consumption accords with the
“energy ladder” theory. The distributional effect of a carbon tax
can be divided into two parts: the direct and indirect effects. The
expenditure resulting from a carbon tax increases with the rise of
households' income level. The expenditure from the indirect effect
of a carbon tax is more than that from the direct effect. The direct
distributional effect presents a weak progressivity, while the
indirect distributional effect is significantly regressive. Thus, the
total distributional effect is also regressive, indicating that low-
income households bear more tax burden. The direct effect largely
depends on the quantity and structure of households' fuels
consumption, while the decisive factors of the indirect effect are
the CO2 emission intensity of various economic sectors and
households' expenditure structure.
Second, both the Suits index and the concentration curve of the
carbon tax indicate that imposing a carbon tax on fossil fuels is
Z. Jiang, S. Shao / Energy Policy 73 (2014) 269–277
regressive and thus can intensify income inequality. The progres-
sivity of the direct distributional effect is approximately propor-
tional. However, the indirect and the total distributional effects are
all significantly regressive. Therefore, the tax burden on the low-
income group is the highest.
The distributional effect of a carbon tax is a key factor in
determining whether it can be accepted by the public and be
successfully implemented, especially under the special back-
ground of the widening income gap in China. This study indicates
that the high-income group consumes more transport fuels than
other groups. Barker and Köhler (1998) also argued that imposing
a carbon tax on transport fuels will result in a progressive effect.
Hence, at the beginning of implementing a carbon tax, it can be
considered to impose it first on transport fuels and then to
gradually expand the scope of taxation. Moreover, as pointed out
by Hammar and Jagers (2007), different carbon tax rates should be
applied to different energy products. For instance, petrol and
diesel should be imposed on by higher tax rates while those
essential energy sources for heating, which is generally used by
40 60 80 100 low-income households, should be imposed on by lower tax rates.
Alternatively, households whose income level is below the stan-
dard line can be exempted from the carbon tax.
Undoubtedly, compared with other mitigation measures of
climate change, it is one of important features of a carbon tax to
increase tax revenue. Some researchers believe that a carbon tax
can generate a result of “double dividend” (Pearce, 1991; Goulder,
1995). If such a “double dividend” hypothesis is tenable, the
redistribution of the carbon tax revenue will play a key role in
its distributional effect. This means that the regressivity of a
carbon tax can be restricted or even eliminated by rationally
designing the redistribution mode of the tax revenue. Barker and
Köhler (1998) testified that different redistribution modes can lead
to different distributional effects of a carbon tax. When the carbon
tax revenue is used to reduce the payroll tax, a carbon tax will be
weakly regressive. When the revenue is used for transfer pay-
ments, it will be strongly progressive. Therefore, when introducing
a carbon tax, some associated redistribution or compensation
measures should be implemented pertinently to the greatest
extent, such as purposive transfer payments, reducing or exempt-
ing some distorting taxes, and subsidizing low-income groups, so
as to restrain potential economic distortion, negative impacts, and
40 60 80 100
to assure optimal economic results of implementing a carbon tax.
Moreover, the implementation of a carbon tax is more likely to be
successful when wide economic reforms are carried out simulta-
neously, such as energy subsidies reform.
Acknowledgments
This paper is supported by the National Natural Science Founda-
tion of China (Nos. 71003068 and 71373153), the Ministry of
Education Foundation of China (No. 12YJC790081), the Program
for New Century Excellent Talents in University (No. NCET-13-0890),
Shanghai “Chen Guang” Program (No. 12CG45), the Shanghai Soft
Science Research Program (No. 14692103900), and the Key Project of
Zhejiang Statistics Research Program (No. 201416).
References
Ahmad, E., Stern, N., 2009. Effective Carbon Taxes and Public Policy Options:
Insights from India and Pakistan. Asia Research Centre Working Paper no. 28,
London School of Economics and Political Science, London, UK.
Baranzini, A., Goldemberg, J., Speck, S., 2000. A future for carbon taxes. Ecol. Econ.
32 (3), 395–412.
 Z. Jiang, S. Shao / Energy Policy 73 (2014) 269–277
Barker, T., Köhler, J., 1998. Equity and ecotax reform in the EU: achieving a 10
percent reduction in CO2 emissions using excise duties. Fisc. Stud. 19 (4),
375–402.
Barnes, D.F., Plas, R.V.D., Floor, W.M., 1997. Tackling the rural energy problem in
developing countries. Financ. Dev. 34 (2), 11–15.
Brännlund, R., Nordström, J., 2004. Carbon tax simulations using a household
demand model. Eur. Econ. Rev. 48 (1), 211–233.
Brenner, M.D., Riddle, M., Boyce, J., 2007. A Chinese sky trust? Distributional
impacts of carbon charges and revenue recycling in China. Energy Policy 35,
1771–1784.
Bureau, B., 2011. Distributional effects of a carbon tax on car fuels in France. Energy
Econ. 33 (1), 121–130.
Cao, J., 2009. Development of low carbon: the design of Chinese carbon tax policy
and the analysis of CGE model. J. Financ. Res. 12, 19–29.
Chen, S.Y., 2011. Marginal abatement cost and environmental tax reform in China.
Soc. Sci. China 3, 85–100.
Cornwell, A., Creedy, J., 1996. Carbon taxation, prices and inequality in Australia.
Fisc. Stud. 17 (3), 21–38.
Datta, A., 2010. The incidence of fuel taxation in India. Energy Econ. 32 (S1),
S26–S33.
Dissou, Y., Siddiqui, M.S., 2014. Can carbon taxes be progressive? Energy Econ. 42
(3), 88–100.
Fan, Y., Zhang, H., 2013. Income distribution impacts of carbon tax Chinese urban
residents and the design of carbon subsidy scheme. Econ. Theory Bus. Manag. 7,
81–91.
Gonzalez, F., 2012. Distributional effects of carbon taxes: the case of Mexico. Energy
Econ. 34 (6), 2102–2115.
Goulder, L.H., 1995. Effects of carbon taxes in an economy with prior tax
distortions: an intertemporal general equilibrium analysis. J. Environ. Econ.
Manag. 29 (3), 271–297.
Grainger, C.A., Kolstad, C.D., 2010. Who pays a price on carbon? Environ. Resour.
Econ. 46 (3), 359–376.
Hammar, H., Jagers, S.C., 2007. What is a fair CO2 tax increase? On fair emission
reductions in the transport sector. Ecol. Econ. 61 (2–3), 377–387.
Hansen, A., Miyazaki, A.D., Sprott, D.E., 2000. The tax incidence of lotteries:
evidence from five states. J. Consum. Aff. 34 (2), 182–203.
IPCC, 2006. 2006 IPCC Guidelines for National Greenhouse Gas Inventories, 〈http://
www.ipcc-nggip.iges.or.jp/public/2006gl/vol2.html〉.
Jacobsen, H.K., Birr-Pedersen, K., Wier, M., 2003. Distributional implications of
environmental taxation in Denmark. Fisc. Stud. 24 (4), 477–499.
Kienzle, E.C., 1981. Measurement of the progressivity of public expenditures and
net fiscal incidence. South. Econ. J. 48 (1), 197–203.
Kerkhof, C.A., Moll, C.H., Drissen, H., Wilting, C.H., 2008. Taxation of multiple
greenhouse gases and the effects on income distribution: a case study of the
Netherlands. Ecol. Econ. 67 (2), 318–326.
Liu, Z., Guan, D.B., Crawford-Brown, D., Zhang, Q., He, K., Liu, J., 2013. A low-carbon
road map for China. Nature 500, 143–145.
Liu, Y., Nie, H., 2004. Effect of indirect tax incidence on income distribution. Econ.
Res. J. 5, 22–30.
Liu, Y., Nie, H., 2009. Effects of value added tax and business tax on income
distribution. Financ. Trade Econ. 6, 63–68.
Labandeira, X., Labeaga, J., 1999. Combining input–output analysis and micro-
simulation to assess the effects of carbon taxation on Spanish households. Fisc.
Stud. 20 (3), 305–320.
277
Metcalf, G.E., 1999. A distributional analysis of green tax reforms. Natl. Tax J. 52 (4),
655–682.
Metcalf, G.E., 2009. Cost Containment in Climate Change Policy: Alternative
Approaches to Mitigation Price Volatility. NBER (National Bureau of Economic
Research) Working Paper no. 15125, 〈http://www.nber.org/papers/w15125〉.
Mathur, A., Morris, A.C., 2014. Distributional effects of a carbon tax in broader U.S.
fiscal reform. Energy Policy 66, 326–334.
Masera, O.R., Saatkamp, B.D., Kammen, D.M., 2000. From linear fuel switching to
multiple cooking strategies: a critique and alternative to the energy ladder
model. World Dev. 28, 2083–2103.
Pearce, D.W., 1991. The role of carbon taxes in adjusting to global warming. Econ. J.
101 (407), 938–948.
Poterba, J.M., 1991. Is the gasoline tax regressive? Tax Policy Econ. 5, 145–164.
Peters, G.P., Marland, G., Quéré, C.L., Boden, T., Canadell, J.G., Raupach, M.R., 2012.
Rapid growth in CO2 emissions after the 2008–2009 global financial crisis. Nat.
Clim. Chang. 2, 2–4.
Price, D.I., Novak, E.S., 1999. The tax incidence of the three Texas lottery games:
regressivity, race, and education. Natl. Tax J. 52 (4), 741–751.
Rapanos, V.T., 1995. The effects of environmental taxes on income distribution.
Eur. J. Polit. Econ. 11 (3), 487–501.
Sarte, P.G., 1997. Progressive taxation and income inequality in dynamic competi-
tive equilibrium. J. Public Econ. 66, 145–171.
Sterner, T., 2012. Distributional effects of taxing transport fuel. Energy Policy 41,
75–83.
Suits, D.B., 1977. Measurement of tax progressivity. Am. Econ. Rev. 67 (4), 747–752.
Shammin, R.M., Bullard, W.C., 2009. Impact of cap-and-trade policies for reducing
greenhouse gas emissions on U.S. households. Ecol. Econ. 68 (8–9), 2432–2438.
Su, M., Fu, Z.H., Xu, W., Wang, Z.G., Li, X., Liang, Q., 2009. The research of levying
carbon tax in China. Rev. Econ. Res. 72, 2–16.
Symons, E.J., Proops, J., Gay, P., 1994. Carbon taxes, consumer demand and carbon
dioxide emissions: a simulation analysis for the UK. Fisc. Stud. 15 (2), 19–43.
Symons, E.J., Speck, S., Proops, J.L.R., 2000. The Effects of Pollution and Energy Taxes
Across the European Income Distribution. Keele Department of Economics
Discussion Papers, Department of Economics, Keele University.
Sun, W., Ueta, K., 2011. The distributional effects of a China carbon tax: a rural–
urban assessment. Kyoto Econ. Rev. 80 (2), 188–206.
Shao, S., Yang, L.L., Yu, M.B., Yu, M.L., 2011. Estimation, characteristics, and
determinants of energy-related industrial CO2 emissions in Shanghai (China),
1994–2009. Energy Policy 39 (10), 6476–6494.
Tiezzi, S., 2005. The welfare effects and the distributive impact of carbon taxation
on Italian households. Energy Policy 33 (12), 1597–1612.
Wier, M., Birr-Pedersen, K., Jacobsen, H.K., Klok, J., 2005. Are CO2 taxes regressive?
Evidence from the Danish experience. Ecol. Econ. 52 (2), 239–251.
Yang, L.L., Shao, S., 2013. Techno-economic Drivers of Energy-related Industrial CO2
Emissions Growth in Shanghai (China) over 1994–2011. School of Urban and
Regional Science Working Paper, Shanghai University of Finance and
Economics.
Yusuf, A.A., Resosudarmo, B., 2007. On the Distributional Effect of Carbon Tax in
Developing Countries: The Case of Indonesia. Economics and Development
Studies Working Paper no. 200705, Department of Economics, Padjadjaran
University.