Z.X. Zhang r Energy Economics 22 2000 587]614 596
Table 7
a
Ž .
Changes in CO emissions from fossil fuel among selected countries and regions
2
1990]1996 1995]1996
b
OECD q
7.8 q
2.6 European Union
q 0.9
q 2.3
Denmark q
41.0 q
20.6 Germany
y 7.8
q 2.1
Netherlands q
10.0 q
2.6 United Kingdom
y 1.0
q 2.9
United States q
8.4 q
3.3 Canada
q 5.5
q 1.6
Japan q
14.3 q
1.8 Australia
q 9.5
q 2.2
New Zealand q
10.7 q
4.0 Norway
q 14.5
q 7.3
CIS and C E Europe y
31.0 y
2.6 Developing countries
q 32.0
q 5.1
World q
6.4 q
2.7
a
Ž .
Source: Jefferson 1997 . A positive sign indicates an increase; A negative sign indicates a decline.
b
Excluding Mexico, Korea, Hungary and Poland.
contribution has been too little appreciated. While China is making such an impressive achievement, we might ask how the OECD countries perform in this
regard. They accounted for 50.3 of global CO emissions in 1996 compared with
2
Ž .
49.6 in 1990 Jefferson, 1997 and promised at the Earth Summit in June 1992 to individually or jointly stabilise emissions of CO and other greenhouse gases at
2
their 1990 levels by 2000. As shown in Table 7, the total CO emissions in the
2
OECD countries rose by 7.8 between 1990 and 1996. On their current trends, CO emissions in the US and EU would be 18 and 8 above the promised targets
2
Ž .
in 2000, respectively Jefferson, 1997; Energy Information Administration, 1999 . Therefore, it is fair to say that, with few exceptions, most of the OECD countries
are unlikely to meet their voluntary commitments to stabilising CO emissions at
2
their 1990 levels by 2000.
3. Economic effects of future carbon limits for China
As shown in Fig. 1, China whose contribution to global CO emissions is already
2
high will surpass the US to become the world’s leading CO emitter by 2020. Thus,
2
advocates of controlling CO emissions call for substantial efforts from China.
2
However, the Chinese authorities have claimed that China cannot be expected to make a significant contribution to solving the carbon emission problem, by arguing
that ignoring the industrialised countries’ responsibility for the majority of global CO emissions and simply asking for special action on China’s part would seriously
2
harm China’s economic development and improvement of living standards. This contrasts sharply with the wishes of proponents of controlling CO emissions. This
2
Z.X. Zhang r Energy Economics 22 2000 587]614 597
section is devoted to explaining this difference in opinion, by analysing the economic effects of possible future carbon limits for China.
3.1. Main features of the CGE model of the Chinese economy Ž
. To this end, a dynamic computable general equilibrium CGE model of the
Chinese economy has been developed.
8
The main features of the CGE model for China are as follows.
This CGE model of the Chinese economy operates by simulating the operation of markets for factors, products and foreign exchange. It is highly non-linear, with
equations specifying supply and demand behaviour across all markets. Moreover, with focus being placed on addressing such energy and environmental issues as
quantifying the economy-wide effects of policies aimed at limiting CO emissions,
2
the model pays particular attention to modelling the energy sector and its linkages to the rest of the economy, because the CO emissions from fossil fuel combustion
2
are the main source of man-made CO emissions, which in turn are the major
2
cause of the greenhouse effect. This makes our CGE model different from other CGE models for China in several aspects. Our model includes ten producing
sectors and is made up of the following blocks: production and factors, prices, income, expenditures, investment and capital accumulation, foreign trade, energy
and environment, welfare, and market clearing conditions and macroeconomic balances. In our CGE model, energy use is disaggregated into coal, oil, natural gas
and electricity. Along with capital, labour and intermediate inputs, the four energy inputs are regarded as the basic inputs into the nested constant elasticity of
substitution-Leontief production function. Moreover, our model incorporates an explicit time dimension, and has a transparent representation of the rate of
Ž .
autonomous energy efficiency improvement AEEI unrelated to energy price increases if dynamic linkages proceed. So, the effect of the AEEI parameter
9
can easily be assessed. Thus, our CGE model, which is also rich in treatment of foreign
trade and is appropriate for modelling the household consumption, allows en- dogenous substitution among energy inputs and alternative allocation of resources
as well as endogenous determination of foreign trade and household consumption in the Chinese economy in order to cope with environmental restrictions, at both
8
Ž .
Ž .
Zhang 1997a and Zhang and Folmer 1998 have argued that in analyzing the economic impacts of limiting CO emissions, a CGE approach is generally considered an appropriate tool. For a detailed
2
Ž .
description of the CGE model for China and its application, see Zhang 1997a, 1998a,b .
9
The AEEI parameter accounts for all but energy price-induced energy conservation. Energy conservation of this type is available at zero or negative net cost. In cost-benefit analysis of greenhouse
gas control, this implies, ceteris paribus, a higher optimal level of emission reduction than the case where abatement costs are always positive. Energy conservation of this type is taking place regardless of
the development of energy prices. It may be brought about by regulations. It may also occur as a result of ‘good housekeeping’ or of a shift in the economic structure away from energy-intensive heavy
manufacturing towards services. In the case where the parameter lowers the growth rate of CO
2
emissions over time, and therefore decreases the amount by which CO emissions need to be
2
constrained, the economic impacts of a given carbon constraint will also be lower.
Z.X. Zhang r Energy Economics 22 2000 587]614 598
sectoral and macroeconomic levels. The equilibrium solution to the model for a given year produces a wealth of detailed information, including market clearing
prices, GNP, productivity levels by industry, investment by industry, final consump- tion levels by commodity, employment by industry, imports and exports by com-
modity, fuel-specific production in physical terms, energy consumption patterns,
Ž .
and CO emissions. Moreover, the Hicksian equivalent variation EV is calcu-
2
lated to measure the welfare impacts of, say, emission abatement policies. EV takes the pre-policy equilibrium income and consumer prices as given and mea-
sures the maximum amount of income that a consumer would be willing to pay to avoid the price change. Because EV measures income change at pre-policy prices,
this makes it more suitable for comparisons among a variety of policy changes compared with the compensating variation. At each point in time, if EV is positive,
post-policy welfare is improving; if negative, post-policy welfare is worsening. Furthermore, the CGE model incorporates an explicit tax system. This makes it
suitable for estimating the ‘double dividend’ from the imposition of a carbon tax that is incorporated as a cost-effective means of limiting CO emissions.
10 2
3.2. The baseline scenario for the Chinese economy until the year 2010 Using this CGE model, a baseline scenario for the Chinese economy has been
developed under a set of assumptions about the exogenous variables. The baseline scenario is characterised by a rapid economic growth, with gross national product
Ž .
GNP being expected to grow at an average annual rate of 7.95 for the period from 1990 to 2010. Although the calculated rates of GNP growth are lower than
those achieved in the early 1980s and 1990s, they are well in line with the government targets of GNP growth rate, which are set at 8]9 per annum for the
period 1990]2000 and at 7.2 thereafter to 2010.
11
Rapid economic growth will lead to increased energy consumption and hence CO emissions, despite substan-
2
tial energy efficiency improvement. As shown in Table 8, total energy consumption is expected to rise from 987.0 Mtce in 1990 to 2560.4 Mtce in 2010. Consequently,
the baseline CO emissions are expected to grow from 586.9 MtC in 1990 to 1441.3
2
MtC in 2010, at an average annual rate of 4.59 for the period to 2010. While the absolute amounts of CO emissions in China are increasing in line with its rapid
2
economic development, its carbon emissions per unit of output are expected to be cut approximately in half during the period 1990]2010.
On a per capita basis, China’s energy consumption of 0.86 tce in 1990 is expected to rise to 1.80 tce in 2010, while the corresponding CO emissions of 0.51 tC in
2
10
A carbon tax is more cost-effective in terms of target achievement than an energy tax. Moreover, compared with an energy tax, a carbon tax is less burdensome in that it raises a smaller amount of
Ž .
government revenues for a given reduction of CO emissions see Zhang, 1997a .
2 11
Converted to the period 1990]2010, the government target of GNP growth rate ranges from 7.6 to 8.1 per annum.
Z.X. Zhang r Energy Economics 22 2000 587]614 599
1990 are expected to rise to 1.01 tC in 2010. Although the amounts are expected to double over twenty years, they are still well below the corresponding current world
average levels, which were equal to 2.12 tce and 1.14 tC, respectively in 1990 Ž
. Zhang, 1997a .
3.3. Carbon abatement: counterfactual policy simulations Using the CGE model, we have then analysed the implications of two scenarios
under which China’s CO emissions in 2010 will be cut by 20 and 30, respectively
2
Ž .
relative to the baseline see Fig. 3 . The two emission targets are less restrictive in that they are not compared with the level of emissions in a single base year, but
with the baseline CO emissions in 2010, the latter being 2.46 times that in 1990.
2
3.3.1. Macroeconomic effects
12
The carbon tax required to achieve a 20 cut in CO emissions in 2010 relative
2
to the baseline is estimated to be US18 at 1987 prices, while the corresponding figure necessary to achieve a 30 cut in CO emissions in 2010 is estimated to be
2
US35 at 1987 prices.
13
Table 9 converts the carbon taxes into fuel-specific ad valorem tax rates.
14
Two important observations can be made. First, the carbon taxes and the fuel-specific tax rates differ significantly among the two scenarios. As can be seen
in Table 9, a larger absolute cut in CO emissions will require a higher carbon tax.
2
A higher tax also implies higher fuel-specific tax rates because a carbon tax becomes higher relative to the baseline prices of fossil fuels. Moreover, carbon
taxes and the fuel-specific tax rates rise at an increasing rate as the target of CO
2
emissions becomes more stringent, indicating that large reductions in carbon emissions can only be achieved by ever-larger increases in carbon taxes. Comparing
Scenarios 1 and 2, for example, shows that the carbon tax goes up to 95 while the carbon reduction increases by only 50. Second, although the same carbon tax is
imposed in each scenario, tax rates differ considerably among different types of fossil fuels, depending on both the carbon content and the price of fuel in the
absence of carbon taxes. Given that coal is the least expensive and gives rise to the highest CO emissions per unit of energy content of all fossil fuels, it is not
2
surprising that coal has the highest tax rates. A surprising result is that natural gas
12
Ž .
See Zhang 1997a, 1998a for the detailed results about the effects on sectoral production and employment as well as the contribution of changes in both level and structure of economic activity, a
change in energy input coefficients, and a change in direct energy consumption by households to energy consumption reduction as a result of the imposition of carbon taxes.
13
Although a carbon tax is incorporated as a means in this modelling exercise, it is not first order in considering China’s response strategies for climate change. The main concern is whether China can
afford or were determined to commit to an emissions cap. If there were an emissions cap for China, using a carbon tax would be not impossible, given that a carbon tax is a cost-effective means of limiting
carbon emissions and that China has used emissions charges to control emissions of a number of pollutants, including SO emissions in the acid rain control area.
2 14
Ž .
See Zhang 1997a for converting a given carbon tax into fuel-specific ad valorem tax rates.
Z.X. Zhang r Energy Economics 22 2000 587]614 600
Fig. 3. CO emissions in China under alternative scenarios.
2
has a higher tax rate than oil,
15
although the former has fewer CO emissions per
2
unit of energy content. This is mainly because prices of natural gas in the absence Ž
. of carbon taxes or the pre-tax price in the baseline are not rising faster than oil
prices. As a result, tax rates for gas become higher relative to its prices,
16
although absolute levels of the tax per unit of energy content are higher for oil than for gas.
As far as electricity is concerned, it is indirectly rather than directly affected by carbon taxes via the taxation of inputs used to generate electricity. This results in
15
The results are broadly consistent with the OECD’s GREEN modelling results for North America, Ž
. the Pacific and China Burniaux et al., 1991 . Similar findings are also presented in the study of Ingham
Ž .
and Ulph 1991 , who analysed the effect of carbon taxes on the UK manufacturing sector.
16
This implies the increase in the post-tax price of natural gas relative to that of oil as shown in Table 10. This can be shown as follows. Denoting the pre-tax prices of natural gas and oil by P and P , the
g o
carbon tax per unit of carbon emitted by T
c
, and the amount of carbon emitted per heat unit of natural gas and oil by e and e , the fractional change in the price of natural gas relative to that of oil as a
g o
result of the imposition of a carbon tax is given by Ž
c
. 1 q T e rP
g g
c
Ž .
1 q T e rP
o o
Rearranging, we have the change in the price of natural gas relative to that of oil Ž
c
. P q T e
P
g g
o c
Ž .
P q T e P
o o
g
Given that the pre-tax price of natural gas is not rising faster than that of oil and that natural gas emits Ž
. approximately quarter less carbon per heat unit than oil Zhang, 1997a , the post-tax price of natural
gas thus increases relative to that of oil.
Z.X. Zhang r Energy Economics 22 2000 587]614 601
Table 8
a
Energy-related results for the baseline scenario of the Chinese economy 1990
1997 2010
Ž .
Energy consumption million tce 987.0
1440.0 2560.4
Coal’s share in total energy 76.2
73.5 67.5
Ž .
consumption Ž
. Energy consumption per capita tce
0.86 1.16
1.80 Ž
. CO emissions million tons of carbon
586.9 847.3
1441.3
2 b
Ž .
Carbon intensity of GDP GNP 0.802
0.551 0.427
CO emissions per capita 0.51
0.69 1.01
2
Ž .
tons of carbon
a
Ž .
Sources: Zhang 1997a, 1998a . Own calculations.
b
Measured in tC per thousand US at 1980 prices.
tax rates for electricity of approximately 20 in Scenario 1 and 38 in Scenario 2. They are considerably lower than those for coal, since coal accounts for only
approximately 18 of total electric utility costs. Imposing fuel-specific tax rates contributes in turn to increases in prices of coal,
oil, natural gas and electricity. Comparing the increases in Table 10 with the corresponding fuel-specific tax rates in Table 9, we can see that, as would be
expected, they are almost the same. As shown in Table 10, even under the two less restrictive carbon emission
scenarios, China’s GNP drops by 1.5 and 2.8, respectively and its welfare measured in Hicksian equivalent variation drops by 1.1 and 1.8, respectively in
2010 relative to the baseline. This indicates that the associated GNP and welfare losses tend to rise more sharply as the degree of the emission reduction increases.
Put another way, the economic costs of incremental environmental policy actions increase with the level of emission reduction. This is reflected by, for instance, the
price elasticity of carbon abatement, which rises from y0.40 in Scenario 1 to y
0.32 in Scenario 2. This increasing marginal cost of emission reduction implies that further reductions in CO emissions are becoming increasingly more difficult.
17 2
Given the fact that most studies surveyed by the Intergovernmental Panel on Climate Change second assessment report estimate that the economic losses under
Ž very restrictive carbon limits e.g. stabilisation or even 20 below 1990 levels in
. Ž
2010 are reported not to exceed 2 of GNP for the OECD countries IPCC, .
1996 , our results also support the general finding from global studies that China would be one of the regions hardest hit by carbon limits.
18
This, combined with the
17
Ž .
This finding also corresponds to other CGE studies. See, for example, Conrad and Schrder 1991 Ž
. Ž
. for Germany; Jorgenson and Wilcoxen 1993a,b for the United States; Beausjour et al. 1995 for
Ž .
Canada; and Martins et al. 1993 for the global study.
18
There are many ways in measuring the fairness of sacrifice. Here we interpret the results according to the equal sacrifice criterion that cost incurred as a fraction of GNP should be equal for each country.
Z.X. Zhang r Energy Economics 22 2000 587]614 602
Table 9
a
Carbon taxes and fuel-specific tax rates in 2010 Scenario 1
Scenario 2 Ž
. Carbon tax 1987 USrtC
18 35
Ž .
Coal 64.0
122.0 Ž
. Oil
14.4 28.2
Ž .
Natural gas 46.3
90.9 Ž
. Electricity
19.8 38.1
a
Source: Own calculations.
industrialised countries being responsible for the majority of global CO emissions,
2
explains the Chinese government stance on carbon abatement. Table 10 also shows that the reduction in total CO emissions is larger than the
2
reduction in total energy consumption. This is due to a shift in fuel consumption away from coal towards oil as shown in Table 11, the latter being less carbon-pol-
luting than coal. Moreover, the larger the reduction in CO emissions, the larger
2
the extent to which such fuel switching takes place. 3.3.2. Comparison with GLOBAL 2100 and GREEN
Because of the global characteristics of climate change and China’s potential importance as a source of CO emissions, there exist, though relatively few, global
2
models that cover various political-economic regions and that treat China as a separate region. Global models in this tradition include the well-known GLOBAL
Ž .
Ž .
2100 Manne and Richels, 1992 and GREEN Burniaux et al., 1992 . We have compared our results with those obtained by GLOBAL 2100 and GREEN, in
terms of both the baseline scenarios and the carbon constraint ones. Here we only briefly show the results of a comparison of the carbon constraint scenarios with
GLOBAL 2100 and GREEN.
19
From Table 12, it can be seen that our estimates of a reduction in GNP growth are higher than those by GLOBAL 2100 and GREEN. Although it is difficult to
provide a completely rigorous explanation for the differences between these results, which goes beyond the scope of this article, there are possibilities of
identifying the sources of the differences, if not to quantify their significance. This difference might be related to three factors. First, our baseline of carbon emissions
is higher than that in GLOBAL 2100 and GREEN, indicating that they are the larger size of the gap between uncontrolled emissions and a particular target in our
study and hence the higher costs incurred for compliance with the target. Second, our model is relatively disaggregated compared with both GLOBAL 2100 and
GREEN. This implies less substitutability in our model, leading to higher economic costs. Third, model types matter. While in our single-country model one branch of
19
Ž .
See Zhang 1997a for a detailed comparison with other studies for China in terms of both the baseline scenarios and carbon constraint ones.
Z.X. Zhang r Energy Economics 22 2000 587]614 603
Table 10
a
Ž .
Main macroeconomic effects for China in 2010 percentage deviations relative to the baseline Scenario 1
Scenario 2 GNP
y 1.521
y 2.763
Welfare y
1.078 y
1.753 Private consumption
y 1.165
y 2.972
Investment y
0.686 y
1.832 Exports
y 5.382
y 7.447
Imports y
1.159 y
2.128 Energy consumption
y 19.468
y 29.322
CO emissions y
20.135 y
30.112
2
Price elasticity of carbon abatement y
0.396 y
0.317 Price of coal
q 64.954
q 123.095
Price of oil q
15.296 q
29.144 Price of natural gas
q 46.813
q 90.564
Average price of fossil fuels q
50.888 q
94.895 Price of electricity
q 22.785
q 43.256
Terms-of-trade q
3.636 q
3.822 Nominal wage rate
y 1.807
y 3.043
Real exchange rate y
0.004 y
0.021 User price of capital
y 1.777
y 4.228
Prices of exports q
3.633 q
3.801 Prices of imports
y 0.004
y 0.021
a
Source: Own calculations. A positive sign indicates an increase; a negative sign indicates a decline.
industry is estimated to be negatively affected under the carbon constraints, this would not always be the case in a global model such as GREEN because of the
relative improvement in Chinese branch goods’ competitiveness via trade realloca- tion. The differing effects brought about by the imposition of unilateral carbon
taxes or regional carbon taxes could be part of the explanation for the higher GNP losses in our model.
With respect to the carbon taxes required to achieve the same percentage of carbon reductions in 2010 relative to the baseline, our estimates are on the one
hand higher than those by GREEN. This is because GREEN has a smaller size of
Table 11
a
Ž .
Breakdown of China’s fossil fuel use in 2010 Baseline
Scenario 1 Scenario 2
Coal 74.0
69.2 67.5
Oil 22.1
26.9 28.7
Natural gas 3.9
3.9 3.8
Total fossil fuel 100.0
100.0 100.0
a
Source: Own calculations.
Z.X. Zhang r Energy Economics 22 2000 587]614 604
the gap between the uncontrolled emissions and the emission target, and because Ž
GREEN has lower baseline prices of fossil fuels in GREEN, although the year 1985 is chosen as the base year, China’s input]output table for 1981 is used, while
in our model China’s 1987 input]output table is used. Given that fossil fuels, Ž
. particularly coal, were more heavily subsidised in 1981 than in 1987 Zhang, 1997a ,
. it is not surprising that GREEN requires lower carbon taxes than our model . On
the other hand, our estimates are lower than those by GLOBAL 2100. This is because GLOBAL 2100 assumes lower rates of the autonomous energy efficiency
improvement, and because GLOBAL 2100 considers limited options for reducing CO emissions and overstates the costs of some important alternative low carbon-
2
Ž .
20
polluting energy technologies see Williams, 1990 . Despite numerical differences across models in the carbon tax rates and their
associated costs, the following consensuses emerge:
21
First, a larger absolute cut in CO emissions will require a higher carbon tax. Moreover, carbon tax rises at an
2
increasing rate as the target of CO emissions becomes more stringent, indicating
2
that large reductions in carbon emissions can only be achieved by ever-larger increases in carbon taxes. Second, the associated GNP losses rise as the carbon
emission targets become more stringent. Moreover, they tend to rise more sharply as the degree of the emission reduction increases. Third, China would be one of
the regions hardest hit by carbon limits. This is reflected by the fact that China’s GNP losses under less restrictive carbon limits are in the same range as the often
reported estimates for industrialised countries under very restrictive carbon limits.
Table 13 shows the carbon tax levels across the countries and regions considered. It can be seen that there are significant differences in the carbon taxes required in
order to achieve the same percentage of emission reductions relative to the baseline. As shown in Table 13, the carbon taxes would be much higher in the
industrialised countries than in the developing countries, because the industrialised countries already have relatively energy-efficient economies, have limited possibili-
ties for substituting less polluting energy sources, and already have high pre-carbon
Ž .
tax energy prices as a result of existing energy taxes Hoeller and Coppel, 1992 . Moreover, Table 13 clearly indicates that the carbon taxes required in China in
order to achieve the same percentage of emission reductions relative to the baseline are much lower than those of the industrialised countries and the world
average. This provides the economic rationale for the development of carbon credit investment projects in China. Through the so-called Clean Development Mecha-
Ž .
nism CDM under the Kyoto Protocol, developing countries will be encouraged to combat global climate change. Now the US even wants to go beyond this by
20
This clearly indicates that model structure does matter in comparing model outcomes. Our model and GREEN are similar in the sense that both models are of CGE type and thus allow for the
straightforward comparisons. By contrast, GLOBAL 2100 is an optimisation model with a detailed treatment of the energy sector but a highly aggregated description of the economy.
21
The first two consensuses are also in line with general findings from other CGE studies; see, e.g. Ž
. Ž
. Conrad and Schrder 1991 for Germany; Jorgenson and Wilcoxen 1993a,b for the United States;
Ž .
Ž .
Beausjour et al. 1995 for Canada; and Martins et al. 1993 for the global study.
Z.X. Zhang r Energy Economics 22 2000 587]614 605
Table 12
a
A comparison of CO emission reductions, carbon taxes and growth effect across models in 2010
2 b
c b
Ž .
CO emissions Carbon tax
GNP GDP
2 d
GLOBAL 2100 Scenario 1
y 18.036
57.999 y
0.783 Scenario 2
y 32.657
165.837 y
2.127 Ž
. Ž
. Ž
. Scenario 1
y 20.135
73.480 y
0.976 Ž
. Ž
. Ž
. Scenario 2
y 30.112
147.066 y
1.893
c
GREEN Scenario 1
y 17.535
8.000 y
0.200 Scenario 2
y 32.135
20.000 y
0.500 Ž
. Ž
. Ž
. Scenario 1
y 20.135
10.137 y
0.253 Ž
. Ž
. Ž
. Scenario 2
y 30.112
18.337 y
0.458 Our CGE model
Scenario 1 y
20.135 17.929
y 1.521
Scenario 2 y
30.112 34.983
y 2.763
a
Ž .
Ž .
Sources: Manne 1992 , Martins et al. 1993 , Own calculations.
b
Ž .
Percentage deviations relative to the corresponding baseline negative values indicate declines .
c
Measured in US dollars per ton of carbon. In GLOBAL, carbon taxes are measured at 1990 prices, in GREEN at 1985 prices, and in our model at 1987 prices.
d
The figures in parentheses result from interpolating the carbon taxes required and the associated GDP losses that have originally been estimated by GLOBAL 2100 and GREEN in order to achieve the
same percentage of carbon reductions as those in our study.
demanding major developing countries like China to commit themselves to some kind of limitation on greenhouse gas emissions, and threats that its ratification of
the Kyoto Protocol is conditional on this. How should China respond to this challenge? This leads to another large subject.
4. China’s strategies at the climate change negotiations subsequent to Buenos Aires
