Regional Effects of an Emissions-Reduction Policy in China: Taxes, Subsidies and the

Method of Financing

Anping Chen,

School of Economics,

Jinan University,

Guangzhou, China,

anping.chen@yahoo.com.cn

and

Nicolaas Groenewold,*

Economics,

University of Western Australia,

Perth, Australia

nic.groenewold@uwa.edu.au

*Corresponding author. Acknowledgements: none yet but we hope we’ll get some

acknowledgeable comments soon!

1

Regional Effects of an Emissions-Reduction Policy in China: Taxes, Subsidies and the

Method of Financing

Abstract

The issue of the possible adverse effects of a reduction in pollution on the economy is a very

real one for China, given its public commitment to substantial cuts in CO2 emissions by 2020.

An important part of this issue is the regional dimension – the pollution reduction is likely to

have significantly different effects across the regions and so possibly exacerbate the already

large and persistent inter-regional disparities in China. Policy choices will therefore be

complicated and will need to be carried out with a clear understanding of the impact of

alternative policies at national and regional levels. One important policy choice which has

received little, if any, attention in the regional context is the one between a tax on pollution

and a subsidy on abatement activity. We help elucidate this policy choice by exploring the

tax-subsidy issue in a small theoretical model which captures some of the salient features of

the Chinese regions and the tax/expenditure system and which we solve numerically based on

a parameterisation achieved using data for the Chinese economy. We find that (i) The

adverse economic effects of a reduction in emissions are greater in the interior than in the

coast but (ii) the coast is worse-off than the interior due to relative price changes and the

effects of the government budget constraint; (iii) the effects of a tax on macroeconomic

variables such as wages, profits and output are greater than they are in the case of a subsidy,

particularly in the short run, but (iv) there is little to choose between the two instruments as

far as welfare is concerned, particularly in the long run; (v) the method of financing a tax or

subsidy also matters more for standard economic variables than it does for welfare and that

the difference also dissipates over time.

1

1. Introduction

China’s high growth rate in the past 30-plus years has brought great benefits to the

country on one hand, but on the other has also resulted in serious problems such as

environmental deterioration and widening regional disparities. The rapid growth of carbon

emissions is claimed to be one important factor which has contributed to national

environmental degradation and which has also spilled over from China to the global

environment. China’s carbon dioxide emissions have jumped from 1422 million tonnes (Mt)

in 1978 to 8979 Mt in 2011, representing an average growth rate of 5.8 per cent during this

period.1 Given the size and growth of the Chinese economy, it is not surprising that China

has become the largest emitter of carbon in the world, accounting 26.4 per cent of the total

world emissions in 2011.

Partially because of the realisation of the problems which high emissions impose on

sustainable economic development and partially because of international pressure to reduce

emissions, the Chinese central government announced its carbon emission mitigation target in

1999. It promised to reduce CO2 emissions per unit of GDP by 40-50 per cent below 2005

levels by 2020. To implement this commitment, China has set a national target of reducing

CO2 intensity (CO2 emission per unit of GDP) by 17 per cent over the 12th

Five-Year Plan

(2011-2015). A new development concept called “Beautiful China” has been advocated in

the 18th

National Congress of the Communist Party of China in November 2012, reflecting

the greater environmental concerns in China.

These environmental concerns have simulated a number of potential policy responses,

including a proposal to cap CO2 emission for each province, establishing industrial energy

efficiency audits, setting targets for the deployment of renewable electricity generation,

introducing a carbon tax, developing markets for trading carbon emissions permits, providing

1 The data on emissions come from The BP Statistical Review of World Energy 2012 (BP, 2012).

2

financial subsidies for carbon reduction and so on. Among these policies, a carbon tax has

received much attention, and is believed to be one of the most likely mitigation instruments in

the near future (Liang and Wei, 2012).

The main rationale for a carbon tax is to internalise the externalities associated with

emissions. It is expected that a carbon tax will lead to an increase in the price of goods which

are pollution-intensive relative to other goods and a shift in the economic structure from high-

emission-intensity to low-emission-intensity production. Thus, the imposition of such a

policy can be expected to result in widespread reallocations within the economy.

Given the growing concern about the adverse global effects of carbon emissions, it is

not surprising that the effects of policies designed to reduce such emissions have been

extensively analysed. Analysis has been under taken both within small theoretical models as

well as in large Computable General Equilibrium (CGE) models. The first category includes

papers such as Hoel (2006) which derives the optimal carbon taxes for cooperating countries,

Fischer (2008) which considers the socially optimal level of R&D in abatement technology

when optimal carbon tax or carbon pricing is not possible and Galinato and Yoder (2010)

which solves for sector-specific pollution taxes to maximise utility of the representative

household.

In the CGE class, there are models for many countries; e.g., Bohriger and Hutherford

(1997) for Germany, Gilbert and Netcalf (2009) for the U.S., Meng, Siriwardana and McNeill

(2013) for Australia, Callan et al. (2009) for Ireland, Bye and Jacobsen (2011) for Norway

and Devarajan et al. (2011) for South Africa.

There are relatively few papers which develop models to explore the effects of carbon

taxes on emission reduction in China. Garbaccio, Ho and Jorgenson (1999) builds a CGE

model of Chinese economy which includes carbon emissions besides many other elements of

the economy. They find the potential for a “double dividend” in China, i.e., a decrease in

3

emissions of CO2 and a long-run increase in GDP and consumption if the carbon tax revenue

is spent on investment. In a new generation of the Garbaccio, Ho and Jorgenson model,

Fisher-Vanden and Ho (2007) finds that the effects of carbon taxes on the economy are

affected by the reform of the capital market in China.

Lu, Tong and Liu (2012) constructs a dynamic recursive CGE model and estimates

the effects of carbon taxes and complementary policies on the Chinese economy. They find

that carbon taxes can reduce emissions substantially with little negative impact on economic

growth. In addition, the use of carbon-tax revenue to replace indirect taxes on firms and to

provide lump-sum subsidies to consumers has large effects on production and consumption.

Liang, Fan and Wei (2007) develops a CGE model with 16 sectors and simulates the

effects of a carbon tax policy in China. They find that the negative macroeconomic impact of

a carbon tax on the economy can be alleviated by properly relieving or subsidising production

sectors. Using a similar model, Liang and Wei (2012) explores the impact of a carbon tax on

household disposable income. They find that a carbon tax will not only widen the urban–

rural gap, but also reduce the living standards of households in both urban and rural areas.

All the papers on China above focus their analysis on the effects of a tax on carbon

emission on the macro economy. None of them evaluates the regional economic effects of

the emissions control. We argue that this is a serious shortcoming since there exists

significant heterogeneity across regions in China. In general, the coastal region has a higher

per-capita GDP as well as higher carbon emissions but a lower emission intensity compared

with the interior region (Li and He, 2010). Since emissions-control policy will almost

certainly have regionally-differentiated effects, it is possible that it will exacerbate existing

inter-regional differences at a time when policy-makers at all levels agree they are a serious

problem which need to be addressed in their own right.2

2 For recent discussion of regional disparities, see Chen (2010), Chen and Groenewold (2012) and Lin, Lin and Ho (2013).

4

This is not to say that the regional effects of carbon taxes have been completely

ignored; several papers have extended the CGE analysis to include regions. 3

Li and He

(2010) builds a regional CGE model for China with 30 regions and 23 sectors and analyses

the effects of a uniform carbon tax on the regions. They find that the welfare losses of the

provinces in the central and western regions are bigger than those in the coastal region. The

carbon tax will enlarge the regional income gap if other support measures are not

implemented.

Zhang et al. (2012) builds a regional CGE model for China with 30 provinces, three

regions in the world and 26 commodity groups and assesses the impact of alternative

approaches to achieving the emissions reduction target in the 12th

Five-Year Plan with an

endogenous tax on CO2 embodied in energy used. They find that a regionally-differentiated

target and a single national uniform target have differing effects on the economy across

provinces.

Finally, Pu and Hayashiyama (2012) builds a multi-regional CGE model for China

with 8 regions and 30 commodity sectors to evaluate the effects of an energy resource tax on

China’s regional economy. They find that such a tax can reduce emissions with minor

adverse effects on macroeconomic variables and the effects are differential across regions.

Thus, there are relatively few papers using models which permit an analysis of the

regional dimension of the effects of a tax to reduce carbon emissions. Moreover, there

appears to be considerable disagreement on the nature and extent of regional effects.

Our paper builds on the existing literature as follows. Like Li and He (2010), Zhang

et al. (2010), Pu and Hayashiyama (2012), we investigate the regional economic effects of a

carbon tax in China. But, while they use CGE models, we use a small two-region theoretical

model. Our regions are based on the common distinction in China between the coastal and

3 The regional economic effects of other carbon control policies, such as emission cap has received attention in recent work,

e.g., Aunan et al. (2007), Vennemo et al. (2009) , Chen and Groenewold (2013).

5

interior regions. We specify out model to include various aspects of the Chinese economy,

including the household registration or hukou system and elements of its tax and expenditure

system. Our model has only two goods, one produced in each region, identical households in

each region and identical firms in each region. It abstracts from open-economy

considerations. It is, therefore, several orders of magnitude smaller than a normal CGE

model, and we argue that it is, by comparison, quite transparent so that we can more easily

trace the effects of the carbon tax through the model structure. In addition, our model

requires far less disaggregated data, an attractive feature given the paucity of China’s data

which creates difficulties in calibrating the parameters of large numerical models (Pu and

Hayashiyama, 2012).

Moreover, we explore the effects not only of carbon taxes but also of subsidies. The

subsidies we model are not simply lump-sum recycling of pollution-tax revenues as in

Garbaccio, Ho and Jorgenson (1999), Liang and Wei (2012) and Lu, Tong and Liu (2012).

Rather, they are emissions-reduction instruments in their own right and, following Bye and

Jacobsen (2011), we argue that they need to be financed and will therefore have further

economic consequences through the relevant government budget constraint. This leads us to

explore and compare various methods of financing.

We find that whether a pollution tax or an abatement subsidy is used to reduce

emissions matters more for economic variables such as wages, output, income and

consumption that it does for welfare. Moreover, the similarity in welfare effects is stronger

in the long run than in the short run. The same may be said about the method of financing:

whether a subsidy is financed by reducing government consumption expenditure, government

infrastructure expenditure, output taxes or the rate of VAT often matters for economic

outcomes but relatively little for welfare. It is important therefore for governments to decide

whether they want to target more visible economic outcomes such as output, wages and

6

incomes or more basic welfare. Moreover, a government focus on the short run may lead

them to overreact to offset effects which will dissipate over time.

The structure of the paper is as follows. In section 2 we develop the model, we set out

the simulations in section 3 and report the results of these simulations in section 4.

Conclusions are presented in section 5.

2. The model

We use a simple two-region model with some features reflecting Chinese economic

characteristics. The simplest regional division of China is into coastal and inland (or interior)

regions (denoted C and I respectively). These two regions have been the basis for the

discussion of regional policy until the mid-1980s. It has also been the scheme used in much

empirical work on regional issues in China.4 We use this two-region division in our model.

5

Each region has households, firms and regional governments. There is also a central

government. Households supply labour to firms which produce output. Households receive

income from wages, profits and capital rental and they use this income to purchase some of

each region’s output; in addition, they receive a government-provided consumption good

which is private in the rival sense.

Firms within a region produce a homogeneous output which differs across regions.

We therefore talk of a single industry consisting of identical firms within each region. Firms

use three conventional factors – labour, land and capital, as well as a government-provided

public good which we call infrastructure. Capital is inter-regionally mobile in the short run

4 Recent papers using this classification include Whalley and Zhang (2007), He et al. (2008), Fleisher et al. (2010) and Su

and Jefferson (2012). 5 The coastal region consists of Beijing, Tianjin, Hebei, Guangdong, Hainan, Shandong, Fujian, Zhejiang, Jiangsu, Shanghai,

Liaoning and Guangxi with the remaining provinces being allocated to the interior region. The interior therefore consist of:

Shanxi, Inner Mongolia, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan, Sichuan, Chongqing, Guizhou, Yunnan,

Shaanxi, Gansu, Qinghai, Ningxia, Tibet, Xinjiang.

7

while labour, land and infrastructure are not. In the long run labour is also mobile between

regions.

We model emissions in keeping with much of the recent literature in the area by

assuming that pollution is proportional to output; see, e.g., Fischer (2008), Galinato and

Yoder (2010) and Bye and Jacobsen (2011).6 The emissions-intensity of output is allowed to

vary across the regions. Firms can reduce actual emissions at each level of output by

incurring abatement costs. We ignore the disutility of pollution since we are not interested in

deriving the optimal level of pollution for which this would be necessary. While we analyse

the welfare effects of various model shocks, we do not include the direct effects on utility of

reductions in pollution and in this we follow Bye and Jacobsen (2011). Like them, we focus

on emissions which cause global warming which is related to world-wide pollution levels.

Since global pollution is not greatly affected at the margin by current regional emissions

levels, even for a country as large as China, we ignore the direct effect of regional emissions

on welfare. We assume that firms must pay a pollution tax to the central government. They

may also receive government subsidies to defray abatement costs.

We distinguish between central and regional governments, with the latter including all

sub-national government levels although we recognise that, in practice, the latter level

includes several layers (provincial, prefecture, county and township). In our model, both

levels of government provide households with a consumption good. In addition to this, the

regional governments also provide infrastructure which is a public factor of production.

Moreover, both levels of government may pay firms a subsidy to encourage pollution-

abatement activity. We also include a transfer from the central to the regional governments.

On the taxation side, in addition to the pollution tax we assume three taxes in the

model in a way which broadly reflects the stylised facts of the Chinese taxation system. The

6 An alternative to our treatment of emissions is to assume that production uses “environmental capital” (Hosoe and Naito,

2006) so that permits to pollute are treated as a factor of production; see, e.g., Beladi and Rapp (1993), Beladi and Frasca

(1996), Rosendahl (2008), Hadjiyianniset al. (2009) and Boucekkine and Germain (2009).

8

first is a national VAT, the rate for which is set by the central government at the same level

for both regions and the proceeds from which are shared between the central government and

the regions. The other two taxes are levied by each regional government on the output

produced in its own region.

We assume that households supply labour inelastically to firms in their own region

(each household supplying one unit) and choose consumption to maximise utility. Firms

choose factor inputs, output and abatement activity to maximise profits, taking the factor

prices, tax rates and the abatement subsidy level as parameters. We assume that governments

are exogenous but each must satisfy a budget constraint.

We consider the behaviour of households, firms and governments in more detail in

turn.7

2.1 Households

Households derive utility from the consumption of the two privately-produced goods

(one produced by the firms in each region) as well as from a good supplied by governments.

We assume a representative household in each region i (i = I, C) and that the utility function

for this household is of the constant-elasticity-of-substitution (CES) form:

(1)

1

( )i i Ii Ii Ci Ci i iV C C GH

, i = I, C

where Vi = utility, Cji = real private consumption of good j (j = I, C), GHi = real government-

provided consumption, in region i and βi, γji, δi and ρ, are parameters with 0i , 0<ji <1,

0<i<1, Ii + Ci + i = 1, and ρ> -1.

7 A list of variables is given in Appendix 1.

9

Households receive income from wages and profits paid by firms in the region in

which they live.8 Household income in each region is measured in terms of its own output.

Thus in the interior region the household budget constraint is:

CII +P-1

CCI = JI,

and in the coast it is:

PCIC + CCC = JC,

where Ji (i = I, C) is income in region i in terms of its own output, P denotes the relative price

PI/PC and income is measured net of the VAT which we account for when we define income

below.9

Utility maximisation subject to the household budget constraint gives the

consumption demand functions:

(2a) 1

11 1

III

CI

II

JC

P P

,

(2b) 1

1

CIC

CC

IC

JC

P P

(2c) 1

11

ICI

CI

II

JC

P P

,

(2d) 1

1

1

CCC

CC

IC

JC

P P

Household income is derived from wages, profits and capital rental income.

Households own a unit of labour each which they supply to firms in their own region, they

8 We therefore abstract from inter-regional commuting and from inter-regional firm ownership. Each assumption simplifies

the extent of inter-regional links without sacrificing any essential relationships. 9 The simple structure of the model implies that the VAT is equivalent to a tax on consumption and, given that households

spend all their income, it is also equivalent to an income tax.

10

own the capital in equal shares in the economy as a whole and they own the firms in equal

shares in their region.10

Wages, profits, and capital income are all measured in terms of

output of the region in which they originate. Recalling that household income is measured in

terms of units of output of their own region, we have:

(3a) (1+TV)JI = ΠHI +WI + RKIKI/L + P-1

RKCKC/L,

(3b) (1+TV)JC = ΠHC +WC + RKCKC/L + PRKIKI/L,

where TV = the VAT rate, Wj = the real wage in industry j, ΠHi = real profit distribution per

household in region i, RKj= capital rental rate in industry j, Kj = capital stock in industry j, and

L= national population.

Inter-regional migration has been an important spatial equilibrating mechanism in

regional models. In our model we allow for migration from one region to another although,

given the slow reaction of migration to economic incentives, we allow for it only in the long

run. Since the household registration system (hukou) is a prominent feature of Chinese

internal migration, we include it in our model by assuming it increases the costs of migration

where the cost of migrating from the interior to the coast increases with the population

density of the coast relative to the interior, reflecting a greater resistance to further migrants

from coastal residents, the more crowded the coastal cities become.11

To simplify the

analysis, we assume that if hukou costs were removed, migration would occur from the poor

to the rich region.12

In the models with free migration it is customary to assume that migration occurs until

utility is equalised across regions. But under the hukou system, people will be worse-off in

10 These assumption for capital ownership departs from that for the other two factors to allow for inter-regional capital

mobility. 11 See Liu (2005) for a general description and history of the hukou system. 12 This avoids the discontinuities which result from two-way costly migration; see Mansoorian and Myers (1993) for an

analysis of a model with such discontinuities and Woodland and Yashida (2006) for an approach similar to ours but applied

to international immigration. Other authors such as Groenewold and Hagger (2007) have avoided the discontinuity by

assuming migration to be costless but this is not consistent with the presence of hukou restrictions.

11

the (poorer) interior since they will have to incur costs to obtain hukou for the coastal region.

We therefore model the migration equilibrium condition as:

(4) /

, 0/

C CC I

I I

L AV V

L A

where Li is the population and Ai the area of region i so that Li/Ai is the population density in

region i; μ can be thought of as the hukou parameter – the larger is μ the greater will be the

difference in utilities across the two regions (since the coastal population density exceeds that

in the interior so that the term in brackets exceeds one).

2.2 Firms

We assume that there is a given number of firms in each region which, without loss of

generality, we set equal to 1. Two goods are produced in the economy and it is assumed that

firms in each region are completely specialised. We call the two goods interior and coastal

goods according to the region in which they are produced. In each region, firms use their

fixed endowment of land, hire labour from households in their own region and capital from

households across the country and combine them with infrastructure provided by the regional

government to produce output. Production technology is assumed to be Cobb-Douglas with

constant returns to scale:

)(1( ) ( ) ( ) ,LjLj Kj Gj Kj Gj

j j j j j jY LAND L K GRF

where Lj = employment, Kj = capital used, and GRFj = infrastructure provided by the regional

government, all for industry j. We can simplify by writing:

)(1( ) Lj Kj Gj

j j jD LAND

so that:

(5) ( ) ( )Lj Kj Gj

j j j j jY D L K GRF

, 0 , , ,(1 ) 1, ,Lj Kj Gj Lj Kj Gj j I C

12

Firms produce emissions in proportion to their output and reduce it by abatement

activity:

(6) Ej = εjYj - Bj ,j = I, C

where Ej represents emissions, εj is the emissions intensity of output (at zero abatement levels)

and Bj denotes abatement. Abatement activity is costly and we follow Fischer (2008) in

specifying a simple quadratic cost-of-abatement function. In particular, we assume that

abatement costs are given by:

COSTj(Bj) = ωjBj2 - SjBj

where ωj is a constant and Sj denotes a subsidy per unit of abatement. The firm pays a wage

rate Wj, a capital rental rate of RKj, an output tax at the rate Tj as well as an emissions tax of

TEj per unit of emissions. Profits for the representative firm in region j are then:

(7) ΠFj = (1-Tj)Yj –WjLj – RKjKj – TEj(εjYj-Bj) – (ωjBj2-SjBj), j = I,C.

Profits are maximised by choosing employment, capital usage, output and abatement activity

with the wage, the output tax rate, the capital rental rate, the emission tax rate and the supply

of government infrastructure taken as given. Profit-maximisation therefore implies the usual

marginal conditions:

(8a) (1 )Lj j j Ej j j jY T T W L , j = I, C,

(8b) (1 )Kj j j Ej j Kj jY T T R K , j = I, C,

(8c) Bj = (TEj + Sj)/2ωj, j = I, C.

All factor markets clear. On the labour supply side, each household is assumed to

provide one unit of labour to the firms in its own region. We assume that wages adjust to

clear labour markets so that labour force, labour supply, employment, the number of

households and population are all equal. Nationally, capital is in fixed supply but it is inter-

regionally mobile so that the given national supply is allocated across regions so as to

equalise rental rates:

13

(9) RKI = RKC

2.3 Governments

The central government derives revenue from the emissions tax and the VAT, the

latter of which is shared with the regional governments. In addition to its VAT share, each

regional government levies tax on output in its own region. All tax revenue is assumed to be

measured in terms of the output of the region in which the tax is paid. Thus the central

government receives VAT revenue as well as pollution-tax revenue in terms of output I and C,

depending on the origin of the tax. In addition, it may make transfers to the government of

region i and pay a subsidy to the firms in region i both in terms of region i’s output. Thus the

net amount of region I’s output it receives is θTVLIJI +TEIEI – TRI – SCIBI and the net amount

of region C’s output it has for disposal is θTVLCJC + TECEC– TRC - SCCBC, where θ is the

central government’s share of the VAT revenue, TRi is the transfer to regional government i

and SCi is the central government’s subsidy to abatement activities by firms in region i. Each

of the net revenues are costlessly converted into a government good with units of the goods

chosen so that one unit of region I’s good converts to one unit of the government good and

one unit of region C’s good converts to η units of the government good. Thus the central

government’s budget constraint may be written as:

(10) LIGCI + LCGCC =[θTVLIJI +TEIEI – TRI – SCIBI]+η [θTVLCJC + TECEC– TRC –SCCBC ]

where GCi is the amount of the government consumption good provided by the central

government per household in region i.

Regional governments receive a share (1-θ) of VAT revenue collected from

households in their region, levy an output tax on firms in their region and receive lump-sum

transfers from the central government, all measured in terms of their own region’s output.

They use some of the revenue for subsidising the abatement activities of firms in their own

14

region and convert the remainder to a government good with the same “technology” as the

central government’s. Each regional government provides some of the government good to

households as a consumption good (in equal per capita amounts) within its own region as

well as providing some to firms as infrastructure, a public good. The regional governments’

budget constraints have the form:

(11a) LIGRHI + GRFI = TIYI+ (1–θ)TVLIJI +TRI –SRIBI

(11b) LCGRHC + GRFC =η[TCYC+ (1–θ)TVLCJC+TRC –SRCBC],

where GRHi is the provision of the government good per household by the regional

government in region i.

2.4 Definitions and Closure

The relationship between GHi and its components is given by:

(12) GHi = GRHi + GCi, i = I, C,

and between Si and its components is given by:

(13) Si = SCi + SRi, i = I, C.

Goods markets clear in each region:

(14) Yi = LICiI + LCCiC + TiYi+TVLiJi+TEiEi, i = I, C

Firms distribute profits to households in their own region in equal per capita amounts:

(15) ΠFi = LiΠHi, i = I, C

The trade between regions must balance:

(16) LCPCIC = LICCI

There is a given national labour force ( = population), L:

(17) LI + LC = L,

and a given national capital stock, K:

(18) KI + KC = K.

15

The national level of emissions is given by:

(19) EI + EC = E.

To summarise, the model consists of the 37 equations, (1) to (19) in 54 variables:

Vi, Cji, GHi, P, Ji, ΠHi, RKj, Kj, TEj, Ej, Bj, Sj, Dj, Yj, Lj, ΠFj, TV, Tj, Wj, TRi, SCi, SRi, GRHi,

GRFj, GCi, θ, L, K, and E, of which 19 are exogenous:. Dj, TEi, SCi, SRi, Tj, TRi, one of (GRHI,

GRFI), one of (GRHC, GRFC), one of (GCI, GCC), θ, TV, L, K, so that there are 35

endogenous variables: Vi, Cji, GHi, P, Ji, ΠHi, Yj, Lj, ΠFj, RKj, Kj, E, Ej, Bi, Si, Wj, one of

(GRHI, GRFI), one of (GRHC, GRFC), and one of (GCI, GCC).

Two equations, however, are redundant since (3), (5), (15), (16) and the household

budget constraint can be used to derive (14) so that the balance between the number of

equations and the number of endogenous variables is restored.

2.5 Short-run and long-run versions of the model

We distinguish between short-run and long-run versions of the model based, as in

Krugman (1991), on differences in closure assumptions. We define the short run as the

length of time before inter-regional migration begins to respond to the changes in VI and VC.

In terms of the model, this simply involves suspending equations (4) and (17) and making LI

and LC exogenous in the short-run simulations. The long run is used to refer to the simulation

results using the model as set out above.

2.6 The linearised, numerical version of the model

The model as it stands is too complicated to solve analytically so that we linearise it in

terms of proportional changes and calibrate the parameters using data for China’s regions.

The linearised version is given in Appendix 3. Calibration is discussed in Appendix 4.

16

3. The simulations

We ran a number of simulations differentiated by the pollution control instrument as

well as by the method used to finance the subsidy or dispose of the tax revenue. There are

three possible instruments ( a central government imposed emissions tax, a subsidy for

abatement paid by the central government and a subsidy paid by the regional governments)

and a variety of financing methods. We do not report the results of all of them in detail but

focus on the following five:

1. An emissions tax levied by the central government “financed” by a change in government

consumption expenditure.

2. A subsidy financed by government consumption expenditure. A comparison of the first

two simulations will allow us to address the tax v. subsidy question.

3. A tax “financed” by a change in the VAT tax rate levied by the central government. A

comparison of this simulation to simulation 1 will throw light on the importance of the

central government’s method of financing.

Regional governments also pay abatement subsidies to firms in their own region and

our final two simulations examine the effects of these. They are:

4. A regional government subsidy financed by a cut in infrastructure expenditure.

5. A regional government subsidy financed by a rise in the regional tax rate.

In addition to these five simulations we also ran a number of others. In particular, we

also examined the effects of a central government policy which sets out to achieve different

emission reduction in different regions. This reflects Chinese government policy discussion

in which it has been proposed that the interior region be less heavily taxed than the coast

since it is poorer as well as having a higher emissions intensity so that a uniform tax would be

expected to impose a proportionally greater burden on the interior provinces. We mention

17

the results of these simulations briefly in the next section and report detailed results in

Appendix 6.

4. The results

4.1 An emissions tax “financed” by central government consumption expenditure

In this simulation we impose a regionally uniform increase in the emissions tax TEi

and allow the central government’s budget constraint to be satisfied by changes in its level of

consumption expenditure (equi-proportionate changes in each regional). The level of

regional government consumption expenditure is assumed to adjust to satisfy the regional

government budget constraints. Using the notation for the linearised model, we set tEI = tEC

such that e = -1, make gci endogenous (with the restriction that gcI = gcC) to balance the

central government budget and grhi endogenous to clear the regional governments’ budgets.13

The results are reported in Table 1.

[Table 1 about here]

We begin with the short-run results. Recall that the distinction between short and

long runs is that in the former there is no inter-regional migration so that regional labour

forces are given while in the long run labour is free to migrate from one regional to another

so that regional labour forces are endogenous (although the national labour force is taken as

given).

In broad outline, the tax increase encourages firms to reduce emissions, both by

reducing output and by increasing abatement activity. Firms in the interior attempt to reduce

output by more than in the coast since they have a higher emissions intensity. They attempt

to reduce output by reducing factor demands. However, total national factor supplies are

fixed and, in the case of labour in the short run, they are fixed in each region in the absence of

13 The linearised model is reported in Appendix 3. The lower-case symbols denote the proportional change in their upper-

case counterparts. thus, e.g., gci is the proportional change in CGi.

18

inter-regional migration of labour. Thus the attempt to reduce employment has only wage

effects (each regional labour market clears) and the wage fall in the interior exceeds that in

the coast. The reduction in demand for capital is also larger in the interior with the result that

capital migrates from the interior to the coast and output falls in the interior and expands in

the coast. Profits fall, again by more in the interior than in the coast.

Because at the national level factors are in fixed supply and fully employed, there is

little overall reduction in emissions resulting from the output effects. Most of the reduction

in emissions results from changes in abatement since, with higher emissions taxes, abatement

is more cost-effective. The proportional changes in abatement are approximately the same

for the two regions but, since abatement is considerably larger relative to total emissions for

the coast, the reduction in emissions in the coast is greater.

The reduction in wages, capital returns and profits all lead to a fall in household

income, larger in the interior than in the coast. However, the re-allocation of output from the

interior to the coast shifts relative prices in favour of interior output and this relative-price

change reverses the difference in regional income changes so that in terms of purchasing

power, interior residents are better-off: their reduction of good I consumption is smaller than

that of the coastal residents and they increase consumption of good C while coastal residents

reduce their consumption of their own good.

On the government side, the central government does well out of the changes, not

surprisingly since it levies the increased pollution tax while regional governments suffer with

their tax revenues falling because of the fall in income. Thus the central government is able

to increase its provision of the consumption good to both regions while both regional

governments cut their expenditure on the consumption good. The central government effect

dominates, however, and overall government consumption increases in both regions.

19

Finally, the effect on welfare reflects both the change in private consumption and the

change in government consumption. For coastal residents, the rise in government

consumption is not enough to offset the effect on welfare of the fall in private consumption

and they are, consequently, worse-off as measured by utility. Interior residents, on the other

hand, find that their reduction in the private consumption of good I is more than offset by the

greater amount of good C and the government good they can consume and so are better-off

after the pollution tax increase.

In summary, the short-run results of the increase in the emissions tax are as follows.

Output is re-allocated in favour of C; wages, profits and incomes fall more in the interior than

they do in the coast. The extra government expenditure made possible by the extra tax

revenue also favours the coastal residents. Nevertheless, the relative-price change in favour

of the interior good favour interior residents sufficiently to offset these coastal advantages,

resulting in interior resident being better-off and coastal residents worse-off overall.

In the long run residents are able to migrate from one region to the other and they do

so from the coast to the interior in response to short-run utility differentials. This partially

reverses the short-run output effects: output now falls in both regions but by less in the

interior than it did in the short run and the relative price change is consequently smaller. The

income gap narrows and consumption possibilities change in favour of the coastal residents

and, while they are still worse-off relative to the initial equilibrium, they are better-off than in

the short run. Interior residents, on the other hand, are worse-off than they were in both the

initial equilibrium and the short run, although still not as badly-off as the coastal residents.

Thus, on the whole, the ability to migrate which distinguishes the short and long runs,

reduces but does not reverse the differences between the regions in output, income,

consumption and welfare.

20

4.2 An abatement subsidy financed by central government consumption expenditure

In this simulation the shock to central government abatement subsidies is constrained

to be equal across regions and designed to produce a fall in national emissions levels of 1 to

make the results comparable to those of the previous simulation and allow us to address the

tax v subsidy question. As in the tax case, the central government adjusts its consumption

expenditure to balance its budget with the proportional reduction constrained to be the same

for both regions.14

The results are reported in the second pair of columns in Table 1.

Consider the short-run effects first. As in the tax case, the required fall in emissions is

effectively achieved by an increase in abatement activities (a positive value of bi) with the

value of b being larger in the interior than in the coast. This reflects the combination of two

factors: first, a given increase in central government subsidies results in a greater increase in

total subsidies in the interior (because of a greater central share in the interior) and, second,

abatement elasticity with respect to subsidy differs across regions. Despite the greater

proportional increase in abatement in the interior, the reduction in emissions is actually

greater in the coast – this reflects a much higher initial level of abatement in the coast.

In contrast to the tax case, there are no effects on output, wages, the capital rental rate

or the allocation of capital across the regions since none of the profit-maximisation

conditions (or factor demands) is directly affected by the subsidy. Hence the extra subsidies

feed directly into profits which increase by much the same proportion in both regions. The

higher profits feed through to higher incomes, although the increase is substantially higher in

the coast (reflecting the greater importance of profits in income in the coast). In response to

the income changes, households in both regions change their consumption. Interior residents

increase their consumption of both goods while coastal residents increase their consumption

of their own good but reduce consumption of the interior good, reflecting the increase in the

14 Algebraically, we choose scI = scC so that the resulting value of e = -1 with gci endogenous in the central government

budget constraint (and gcI = gcC) and grhi endogenous to satisfy the regional government budget constraints.

21

relative price of interior goods. These changes affect utility which is also influenced by the

provision of government consumption goods, both central and provincial.

Given our assumption that the central government’s budget is balanced by changes in

GC, it is not surprising that this falls substantially in both regions. Regional government

consumption expenditure also falls in each region; this is surprising given the increase in tax

revenue generated by higher household incomes. The reason is that the considerable increase

in pollution abatement by firms also attracts subsidy payments from the regional governments

(even though the subsidy rate they pay has not increased) and this is financed by a fall in

regional government consumption expenditure, greater in the coast than in the interior

(coastal abatement subsidies per unit of abatement are about twice as high in the coast as they

are in the interior). The total provision of government consumption goods to households

therefore falls in both regions, although by considerably more in the coast than in the interior.

The considerable cuts in government expenditure offset any favourable private

consumption changes based on higher incomes so that welfare falls in both regions, although

by more in the coast than in the interior.

In the long run, labour moves from the coast to the interior in response to short-run

utility differences and this results in small corresponding output effects. Decreasing marginal

productivity results in a fall in wages in the interior and profits need to be shared with more

residents so profits per household also fall in the interior and rise in the coast (although

profits per firm move in the opposite direction). On the consumption side, coastal residents

improve their lot, increasing their consumption of both goods, while interior residents reduce

consumption of both goods. There are only small changes in government expenditure so that

the utility effects largely reflect changes in consumption – coastal residents benefit relative to

the short run and interior residents are worse-off relative to the short run but both are still

worse-off relative to the initial equilibrium.

22

We conclude this sub-section with a brief comparison of the tax and subsidy as a way

to reduce emissions. What really stands out in both the short and long runs is the difference

in the distribution of resources between the government and private sectors. This might

explain why governments prefer to tax pollution even though the subsidies are more popular

with industry! Secondly, welfare effects are quite different in the short run but quite similar

in the long run. In the short run the tax favours the interior since the proceeds are uniformly

distributed across regions while the subsidy has a greater effect in the coast since it receives

larger subsidies per unit of abatement and it has higher levels of abatement in the initial

equilibrium. But in the long run, when everything is allowed to adjust, there is not much

difference between the levels and regional distribution of utility changes. Although wages,

profits, returns to capital, incomes and private consumption are all much lower after a tax

change than a subsidy increase, the welfare effects of this seem to be largely offset by the

increase in government expenditure made possible by the tax. In summary, the effects of

taxes on real economic variables are much greater than the effects of a subsidy but this

largely disappears as far as welfare is concerned. In both cases there are long-run welfare

losses although this is not an argument against the policy of trying to reduce pollution since

we do not model the benefits in terms of direct welfare improvements from the pollution

reduction. There are short-run regional differences in welfare effects but these tend to wash

out in the long run.

4.3 An emissions tax offset by a fall in the VAT rate

It will be recalled that the VAT proceeds are shared with the regional governments so

that a change in the VAT rate will also affect the regional governments’ budget revenue. We

short-circuit this complication (which would make comparison to simulation 1 difficult) by

assuming the central government manipulates the share of the VAT revenue which it goes to

23

the regional governments so as to leave regional government VAT revenue unchanged. We

focus on the comparison to simulation 1 which will allow us to assess the importance of the

method of “financing” the emissions tax. Thus we focus on a comparison of the first and

third pairs of columns in Table 1.

Consider the short-run effects first. It is clear from a comparison of the two

simulations that the method of financing does not affect output, factor use or factor payments.

But the financing shift results in a move of resources from the government to the households.

Hence government provision of the consumption good is lower but household incomes are

higher and, so, consumption is proportionately higher under the VAT-financed case. Even

with central government consumption held constant under VAT-financing, government

consumption provision falls in both regions. This is due entirely to the fall in regional

government expenditure forced by the government budget constraint: the rise in abatement

activity by firms also draws additional subsidies from the regional governments which is

financed by a reduction in their consumption expenditure. The fall is larger for the interior

region since the government is faced with a loss of tax revenue as well as an increase in the

costs of subsidies while the coast receives an increase in revenue (since output increases) and

also faces higher costs. The final effect on utility is also influenced by relative price changes

(as in the GC-financed case) and the upshot is that the interior is better-off while the coast is

worse-off and, compared to the GC-financed case, the utility difference between the two

regions is wider. This reflects the fact that the coast benefits much more from the distribution

of the pollution tax revenue as government consumption than from its distribution as reduced

VAT.

In the long run, there is migration from the coast to the interior and, as in the case of

the expenditure-financed emissions tax increase, this serves to bring the two regional

24

outcomes closer. In welfare terms, the fall in utility in the coast is smaller and the short-run

increased in interior utility is reversed although the fall is not as large as that in the coast.

Thus, in all, the effects of the imposition of an emissions tax are not greatly

influenced by whether the tax revenue is used for government consumption or a reduction in

the VAT. There are no short-run effects on factor use, factor prices and output and the

principal effect is that VAT-financing rather than expenditure-financing shifts resources to

the households, increasing their income and consumption; but the effects on utility are offset

by a considerably lower government consumption. In the short run the shift to VAT-

financing makes the coast relatively worse-off because it benefitted most from the

distribution of pollution-tax revenue as government consumption. In the long run this

regional difference is ameliorated but not removed altogether.

We also ran a simulation for a subsidy financed by an increase in the VAT rate but, in

the interest of saving space (and the reader’s patience), we briefly summarise the results,

rather than provide a detailed discussion. The increase in the VAT reduces income and

consumption substantially compared to the GC-financed subsidy but leaves government

consumption expenditure (GH) relatively unchanged. In this case welfare is reduced in both

regions in the short run although the reduction is larger in the coast; this reflects the fact that

the coastal government pays greater subsidies to firms in its region so that when firms

respond to the central government subsidy increase by raising their abatement, the regional

governments also have to pay extra subsidies and this hits the coast harder than the interior.

Detailed results for this simulation are reported in Appendix 5.

We saw above that in terms of output, wages, profits and incomes, an increase in the

emissions tax has a greater adverse impact on the interior region than on the coast, even

though the utility effects are often the reverse. Add to this the fact that the interior region is

poorer to start with and it is not surprising that the central government has seriously

25

considered proposals to reduce pollution by more in the coast than in the interior. To assess

the possible effects of such a regionally-differentiated policy, we repeated the above

simulation under the constraint that the reduction in emissions in the coast is roughly twice

that in the interior; in particular, following a proposal in the Twelfth Five-Year Plan, we set

the policy shock so that in the short run eI = -0.67 and eC = -1.31 (which implied a value for e

of -1.0). The results were not greatly different. This is largely because the reduction in

pollution in the coast under a regionally-uniform policy is already considerably larger than in

the interior; in simulation 1, for example, the short-run value of eI is -0.7623 and the solution

for eC is -1.2225. We therefore do not discuss these simulations in detail but report the full

results in Appendix 5.

4.4 An increase in regional government subsidies financed by an increase in regional taxes

We now turn to our last set of simulations, of which we discuss two in some detail.

They all involve changes in subsidies by regional governments. In our model only the central

government can levy a pollution tax but both central and regional governments pay subsidies.

It is interesting to consider the effects of a regional government subsidy because the regional

government have instruments to balance their budget which have different effects to those

available to the central government. In particular, the central government has only

expenditure on the government consumption good and the VAT which can be adjusted to

balance its budget in response to an emissions tax or subsidy change while the regional

governments have not only consumption expenditure but also expenditure on infrastructure as

well as the output tax.

We deal very briefly, first, with the case where the regional governments balance their

budgets by changing their provision of the consumption good to households (GRH) since this,

not surprisingly, is very similar to the case where the central government pays a uniform

26

subsidy and balances its budget by expenditure on the government consumption good. In

order to make for easy comparison to the equivalent case of a central government subsidy, we

assume the changes in the regional government subsidies are such that they produce the same

reductions in emissions in each region in the short run. A comparison of the results reported

in the second pair of columns in Table 5.1 to those in the first pair of columns in Table 5.2 of

Appendix 5 shows that the effects are almost identical, with the difference driven by small

differences in the regional distribution of the government consumption good. We, therefore,

now proceed to the remaining two cases, starting with the case where the regional

governments increase their subsidies and finance this by an output tax. The results are

reported in the second pair of columns in Table 2. To facilitate comparison to the central

government policies reported in Table 1, we repeat the results of a central government

consumption financed pollution tax increase as the first two columns of numbers in Table 2.

[Table 2 about here]

The regional governments’ output tax affects firms’ decision-making since it changes

the “effective” factor price in the marginal-productivity conditions for profit maximisation.

We set the subsidies at the same level as in the GRH-financed subsidy above so that the

increases in abatement are the same in both cases but the changes in emissions are not since

the change in the output tax needed to balance the regional governments’ budgets also

influences output; in general, firms in both regions will reduce their factor demands. While

the increase in tax is greater for the interior region, the output tax is a smaller proportion of

total costs so that, on balance, they reduce their demands for factors by less. The

consequence is that capital moves from the coast to the interior. Labour is not mobile in the

short run so that the change in labour demand has only wage effects, with wages falling by

more in the coast than in the interior. Output is redistributed to the interior from the coast but

only by a small amount. The reduction in both wages and capital returns reduce incomes

27

while the subsidy increases profits and so incomes but not by enough to offset the reduced

factor income. Incomes and consumption therefore fall in both regions. Government

expenditure also falls; the higher abatement levels requires the central government to pay

higher subsidies the cost of which is offset by a reduction in the provision of the government

consumption goods. With both private and government consumption falling in both regions,

it is not surprising to find welfare lower as a result of the emissions-reduction policy, with the

coast being harder hit than the interior.

In the long run there is migration from the coast to the interior but the migration is

very small and narrows the gap in inter-regional welfare effects but only slightly.

Thus in comparison to the GC-financed subsidy, a subsidy financed by a regional

output tax affects factors demands and output in the short run, with a shift of factors and

output from the coast to the interior because the tax increase has a greater impact on the firms

in the coast. Nevertheless, on balance, the coast benefits from a shift from expenditure- to

tax-financing because it is more heavily reliant on central government expenditure. In the

long run there is a further redistribution of output from the coast to the interior but this

reduces interior income under the force of declining marginal products. Utility levels

therefore move closer to each other and also closer to those experienced in the long run for

the expenditure-financed subsidy.

4.5 An increase in regional government subsidies financed by a fall in infrastructure

expenditure

The final simulation is an increase in the subsidies paid by the regional governments

financed by a cut in infrastructure expenditure, GRF. As the final pair of columns in Table 2

show, there is a marked difference in the effects of this policy, essentially because now the

government good is a factor of production so that the decreases in its supply reduces output.

28

The subsidy increase is set at the same level as in the consumption-financed case so that the

abatement change is as in that case. The change in emissions however is very different since

now output also falls which results in greater emissions reduction which at the national level

is now -1.1283 rather than -1 as in previous cases. The reduction in the provision of

infrastructure also reduces the demand for labour and capital, resulting in a fall in the wage

and the capital rental rate in both regions. Both of these effects reduce household incomes.

The income reduction is offset, however, by the increase in profits as the subsidies flow into

profits,. Nevertheless, household income falls in both regions, as does consumption.

Regional government provision of the government consumption good is unchanged by

assumption but the central government reduces its provision of the consumption good since

its revenue falls and costs increase: revenue falls because incomes and so VAT proceed fall

and costs increase since it must pay subsidies on the increased abatement generated by the

regional government subsidy increases. Thus from a household point of view, private and

government consumption both fall and as a result utility of the representative household also

falls.

As in the other regional government shocks, the reduction in welfare of coastal

residents is harder-hit than interior residents and, in the long run, this generates migration

from the coast to the interior. As in previous cases, this serves to reduce but not remove the

inter-regional gap in the impacts of the policy.

A comparison of the tax- to the infrastructure-financed subsidy shows that the greatest

difference is in the output responses – with a reduction in the provision of infrastructure there

are significant falls in output in both regions rather than simply a small redistribution of

output. The overall fall in output means that aggregate emissions actually fall by more than

in the two previous cases. Despite this, the falls in income are similar; however, with larger

falls in central government provided consumption good because of the fall in VAT revenue,

29

utilities fall by more under infrastructure-financing than tax-financing. Compared to the tax-

financed case, the utility deterioration is greater for the interior. In the long run, internal

migration drives the two regions closer together with the utility losses being approximately

the same in each but larger than those experienced under the other two methods of financing

the regional government subsidy.

5. Conclusions

This paper has set out and derived a number of numerical solutions to a small two-

region model designed to have some features of the Chinese economy. The model has been

used to simulate the economic effects of a number of shocks specified to capture possible

ways in which policy might be implemented to reduce pollution. We have simulated both

pollution taxes as well as subsidies to abatement activities, we have varied the way in which

the government implementing the policy might finance the change and we have considered

policy both at the central and the regional levels.

The first substantive question we addressed was that of the effects of a pollution tax

v. an abatement subsidy as a method of reducing emissions. The most marked difference

between these two policies was found to be in the distribution of resources between the

government and private sectors. Wages, profits, returns to capital, incomes and private

consumption are all much lower under a pollution tax than an abatement subsidy. On the

other hand, government expenditure on the consumption good (which was assumed to vary to

balance the government’s budget) was much higher in the case in which a tax was used.

Welfare effects are quite different between the two policies in the short run but quite similar

in the long run. In the short run the tax favours the interior since the proceeds are uniformly

distributed across regions in the form of increased government consumption while the

subsidy benefits the coast more than the interior. But in the long run, when everything is

30

allowed to adjust, the utility changes are similar across financing methods and regions. In

summary, the effects of taxes on real economic variables are much greater than the effects of

a subsidy but this largely disappears as far as welfare is concerned.

Another question addressed was the effect of “financing” the imposition of a pollution

tax – in the first case the pollution-tax proceeds were assumed to be distributed as higher

government consumption and in the other as a reduction in the VAT rate. It was found that,

all in all, the effects of the imposition of an emissions tax are not greatly influenced by the

method of financing. There are no short-run effects on factor use, factor prices and output

whichever method is used; the principal effect is that VAT-financing rather than expenditure-

financing shifts resources to the households, increasing their income and consumption but the

effects on utility are offset by a considerably lower government consumption. In the short

run the shift to VAT-financing makes the coast relatively worse-off because it benefitted

most from the distribution of pollution-tax revenue as government consumption. In the long

run this regional difference is ameliorated but not removed altogether.

Finally we considered a number of shocks involving subsidies paid by regional

governments. These were of interest because of the different instruments which the regional

governments might use to finance the subsidy. We began by comparing the effects of a

subsidy financed by government consumption expenditure – by the central government on the

one hand and by the regional governments on the other. When the subsidy levels were set to

ensure equal short-run increases in abatement in both cases, it transpired, not surprisingly,

that there was little difference between the two policies.

We then moved to a regional government subsidy financed by an increase in output

taxes in each region and found that the tax-financed subsidy affects factors demands and

output in the short run, with a shift of factors and output from the coast to the interior because

the tax increase has a greater impact on the firms in the coast. Nevertheless, on balance, the

31

coast benefits from a shift from consumption-expenditure- to tax-financing because it is more

heavily reliant on government expenditure. In the long run there is a further redistribution of

output from the coast to the interior and utility levels move closer to each other and also

closer to those experienced in the long run for the expenditure-financed subsidy.

Finally, we moved to a consideration of a regional government subsidy financed by a

reduction in infrastructure expenditure. The comparison of tax- and infrastructure-financed

subsidies shows that the greatest difference is in the output responses – with a reduction in

the provision of infrastructure, there are significant falls in output in both regions rather than

simply a small redistribution of output. Despite this, the falls in income in the two

simulations are similar. However, utilities fall by more under infrastructure-financing than

tax-financing since there are larger falls in central government provided consumption good

because of the fall in VAT revenue. Consequently the utility deterioration is greater for the

interior. In the long run, internal migration drives the two regions closer together with the

utility losses being approximately the same in each but larger than those experienced under

the other two methods of financing the regional government subsidy.

In summary, tax or subsidy and the method of financing either matter more for

economic variables such as wages, output, income and consumption than they do for welfare.

The similarity in welfare effects is stronger in the long run than in the short run. Hence in

choosing policy governments should not focus only on standard economic effects but also

consider the more balanced effects on welfare and recognise the differences in the effects of

different policies and between different regions although they are likely to dissipate over time.

32

References

Aunan, K., Berntsen, T., O’Connor, D., Persson, T.H., Vennemo, H. and Zhai, F.(2007),

“Benefits and Costs to China of a Climate Policy”, Environment and Development

Economics,12, 471–497.

BP(2012), The BP Statistical Review of World Energy 2012, at:

http://www.bp.com/statisticalreview

Beladi, H. and Frasca, R. (1996), “Regional Pollution and Multinational Firms”, Ecological

Economics¸ 17, 117-125.

Beladi, H. and Rapp, J. (1993), “Urban Unemployment and the Backward Incidence of

Pollution Control”, Annals of Regional Science, 27, 153-163.

Bohringer C. and Rutherford, T. F. (1997), “Carbon Taxes with Exemptions in an Open

Economy: A General Equilibrium Analysis of the German Tax Initiative”, Journal of

Environmental Economics and Management, 32, 189–203.

Boucekkine, R. and Germain, M. (2009), “The Burden Sharing of Pollution Abatement Costs

in Multi-Regional Open Economies”, The B. E. Journal of Macroeconomics, 9, 1-32.

Bye B. and Jacobsen K. (2011), “Restricted Carbon Emissions and Directed R&D Support:

An Applied General Equilibrium Analysis”, Energy Economics, 31, 543-555.

Callan T., Lyons, S., Scott, S., Tol, R. S. J. and Verde, S. (2009), “The Distributional

Implications of a Carbon Tax in Ireland”, Energy Policy, 37, 407–412.

Chen A.( 2010), “Reducing Regional Disparities in China: Is There a Growth Cost?”, China

Economic Review, 21, 2-13.

Chen A. and Groenewold, N. (2012), “Does Investment Allocation Affect the Inter-Regional

Output Gap in China? A Time-Series Investigation”, China Economic Review, at

http://www.sciencedirect.com/science/article/pii/S1043951X1200017X

Chen A. and Groenewold, N. (2013), “The Regional Economic Effects of a Reduction in

Carbon Emission and an Evaluation of Offsetting Policies in China”, Papers in

Regional Science, forthcoming.

Devarajan S, Go, D., Robinson, S. and Thierfelder, K. (2011), “Tax Policy to Reduce Carbon

Emissions in a Distorted Economy: illustrations from a South Africa CGE Model”,

The B.E. Journal of Economic Analysis and Policy, 11, 1-22.

Fischer C. (2008), “Emissions pricing, spillovers, and public investment in environmentally

friendly technologies”, Energy Economics, 30, 487–502.

Fisher-Vanden, Karen and Ho, Mun S. (2007), “How Do Market Reforms Affect China’s

Responsiveness to Environmental Policy?”, Journal of Development Economics, 82,

200– 233.

Fleisher B., Li H. and Zhao M. (2010), “Human Capital, Economic Growth, and Regional

Inequality in China”, Journal of Development Economics, 92, 215-231.

Gilbert E. and Metcalf, G. E. (2009), “Designing a Carbon Tax to Reduce U.S. Greenhouse

Gas Emissions”, Review of Environmental Economics and Policy, 3, 63-83.

Garbaccio R. F., Ho, M. S. and Jorgenson, D. W. (1999), “Controlling Carbon Emissions in

China”, Environment and Development Economics, 4, 493-518.

Galinato G. I. and Yoder, J. K. (2010), “An integrated tax-subsidy policy for carbon emission

reduction”, Resource and Energy Economics, 32, 310–326.

Groenewold, N. and Hagger, A. J. (2007), “The Effects of Fiscal Equalisation in a Model

with Endogenous Regional Governments: An Analysis in a Two-Region Numerical

Model”, Annals of Regional Science, 41, 353-374.

Hadjiyiannis, C., Hatzipanayotou P and Michael., M. S. (2009), “Public Pollution Abatement,

Regional Capital Mobility, and Tax Competition”, Southern Economic Journal, 75,

703-719.

33

He, C., Wei, Y. D. and Xie, X. (2008), ‘Globalization, Institutional Change, and Industrial

Location: Economic Transition and Industrial Concentration in China’, Regional

Studies, 42, 923-945.

Hoel, M.(2006), “Should a carbon tax be differentiated across sectors?”, Journal of Public

Economics, 59, 17-32.

Hosoe, M. and Naito, T. (2006), “Trans-Boundary Pollution Transmission and Regional

Agglomeration Effects”, Papers in Regional Science, 83, 99-119

Krugman, P. R. (1991), “Increasing Returns and Economic Geography”, Journal of Political

Economy, 99, 483-99.

Li, S. and He, J. (2010), “Regional disparity and mitigation cost for carbon policy in China:

Assessment based on multi-regional CGE model”, 13th Annual Conference on Global

Economic Analysis.

Liang Q. M., Fan, Y. and Wei, Y. M. (2007), “Carbon Taxation Policy in China: How to

Protect Energy- and Trade-Intensive Sectors?”, Journal of Policy Modeling, 29, 311–

333.

Liang Q. M. and Wei, Y. M. (2012), "Distributional Impacts of Taxing Carbon in China:

Results from the CEEPA Model”, Applied Energy, 92, 545–551

Lin, P., Lin, C. and Ho, I. (2013), “Regional Convergence or Divergence in China? Evidence

from Unit Root Tests with Breaks”, Annals of Regional Science, 50, 223-243.

Liu, Z. (2005), “Institution and Inequality: The Hukou System in China”, Journal of

Comparative Economics, 33, 133-157.

Lu C., Tong, Q. and Liu, X. (2010), “The Impacts of Carbon Tax and Complementary

Policies on Chinese Economy”, Energy Policy, 38, 7278-7285.

Mansoorian, A. and Myers, G.M.(1993), “Attachment to Home and Efficient Purchases of

Population in a Fiscal Externality Economy”, Journal of Public Economics, 52, 117–

132.

Meng S., Siriwardana, M. and McNeill, J. (2013), “The Environmental and Economic Impact

of the Carbon Tax in Australia”, Environmental and Resource Economics, 54, 313–

332

National Development and Reform Commission (2010), Contract Management and

Financial Subsidy on Energy Consumption, at:

http://www.sdpc.gov.cn/zcfb/zcfbqt/2010qt/t20100609_353606.htm

Pu Z.and Hayashiyama, Y. (2012), “Effects of Carbon Dioxide Control Policy in China by

Multi-Regional CGE Model”, Frontier of Economics in China, 7(4), 580-603.

Rosendahl, K. E. (2008), “Incentives and Prices in an Emissions Trading Scheme with

Updating”, Journal of Environmental Economics and Management, 56, 69-82.

State Statistical Bureau (2005), China Civil Affairs Statistical Yearbook 2005, Statistical

Publishing House of China, Beijing.

State Statistical Bureau (various issues), Statistical Yearbook of China, Statistical Publishing

House of China, Beijing.

State Statistical Bureau (2010), Comprehensive Statistical Data and Materials on 60 Years of

New China, Statistical Publishing House of China, Beijing.

State Statistical Bureau (various issues), China Energy Statistical Yearbook, Statistical

Publishing House of China, Beijing.

Su J. and Jefferson, Gary H.(2012), “Differences in returns to FDI between China's coast and

interior: One country, Two economies?”, Journal of Asian Economics,23, 259-269.

Vennemo H., Aunan, K., He, J., Hu, T. and Li, S. (2009), “Benefits and Costs to China of

Three Different Climate Treaties”, Resource and Energy Economics, 31, 139–160.

Whalley, J. and Zhang, S. (2007), “A Numerical Simulation Analysis of (Hukou) Labour

Mobility Restrictions in China”, Journal of Development Economics, 83, 392-410.

34

Woodland, A.D. and Yoshida, C.(2006),“Risk Preference, Immigration Policy and Illegal

Immigration”, Journal of Development Economics, 81, 500– 513.

Zhang D., Rausch S., Karpulus V. and Zhang, X. (2012), “Quantifying Regional Economic

Impacts of CO2 Intensity Targets in China”, MIT Joint Program on the Science and

Policy of Global Change, Report No. 230.

35

Table 1: Simulation Results for Central Government Policy Shocks

Variable

Emissions Tax

(gc endogenous)

Subsidy

(gc endogenous)

Emissions Tax

(tv endogenous)

Cut in Both Regions SR LR SR LR SR LR

vI 0.1119 -0.0361 -0.0138 -0.0608 0.1879 -0.0043

vC -0.3154 -0.1648 -0.1494 -0.1016 -0.3685 -0.1725

cII -0.4661 -0.5455 0.0925 0.0673 -0.0544 -0.1508

cCI 0.1356 -0.1488 0.1682 0.0780 0.5474 0.1829

cIC -1.2320 -0.9218 -0.0038 0.0946 -0.8202 -0.4072

cCC -0.6302 -0.5251 0.0719 0.1052 -0.2185 -0.0735

ghI 0.9481 0.9707 -0.5440 -0.5368 -0.1675 -0.1581

ghC 1.6020 1.6264 -0.8085 -0.8007 -0.0556 -0.0537

jI -0.9019 -0.8329 0.0377 0.0596 -0.4902 -0.3924

jC -0.3239 -0.3231 0.1104 0.1106 0.0879 0.0964

yI -0.0389 -0.0014 0.0000 0.0119 -0.0389 0.0101

yC 0.0233 -0.0135 0.0000 -0.0117 0.0233 -0.0248

lI 0.0000 0.0560 0.0000 0.0178 0.0000 0.0732

lC 0.0000 -0.0727 0.0000 -0.0231 0.0000 -0.0951

kI -0.1753 -0.1295 0.0000 0.0145 -0.1753 -0.1154

kC 0.1087 0.0803 0.0000 -0.0090 0.1087 0.0716

eI -0.7623 -0.7205 -0.7235 -0.7102 -0.7623 -0.7076

eC -1.2225 -1.2673 -1.2588 -1.2730 -1.2225 -1.2811

e -1.0000 -1.0030 -1.0000 -1.0009 -1.0000 -1.0039

wI -0.7776 -0.7961 0.0000 -0.0059 -0.7776 -0.8018

wC -0.4936 -0.4577 0.0000 0.0114 -0.4936 -0.4466

p 1.3676 0.9017 0.1720 0.0242 1.3676 0.7583

grhI -0.4333 -0.4241 -0.0408 -0.0379 -0.2680 -0.2530

grfI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

grhC -0.3400 -0.3398 -0.1321 -0.1320 -0.1212 -0.1169

grfC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

gcI 3.2517 3.2966 -1.3830 -1.3688 0.0000 0.0000

gcC 3.2517 3.2966 -1.3830 -1.3688 0.0000 0.0000

bI 6.2510 6.2510 6.2906 6.2906 6.2510 6.2510

bC 5.6410 5.6410 5.6767 5.6767 5.6410 5.6410

scI 0.0000 0.0000 16.2393 16.2393 0.0000 0.0000

scC 0.0000 0.0000 16.2393 16.2393 0.0000 0.0000

srI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

srC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

sI 0.0000 0.0000 12.6131 12.6131 0.0000 0.0000

sC 0.0000 0.0000 10.3655 10.3655 0.0000 0.0000

Notes: The symbols in the first column and at the head of the columns of results are the proportional changes of

their upper-case counterparts; thus, for example, vI is the proportional change in VI. SR and LR are abbreviations

of “short run” and “long run”.

36

Table 2 Simulation Results for Regional Government Policy Shocks

Variable

Central Government Subsidy

(gc endogenous)

Regional Subsidy

(tj endogenous)

Regional Subsidy

(grf endogenous)

Cut in Both Regions SR LR SR LR SR LR

vI -0.0138 -0.0608 -0.0651 -0.0786 -0.0860 -0.0949

vC -0.1494 -0.1016 -0.1039 -0.0902 -0.1115 -0.1024

cII 0.0925 0.0673 -0.0150 -0.0221 -0.0258 -0.0305

cCI 0.1682 0.0780 -0.0035 -0.0293 -0.0306 -0.0476

cIC -0.0038 0.0946 -0.0296 -0.0013 -0.0198 -0.0012

cCC 0.0719 0.1052 -0.0181 -0.0085 -0.0245 -0.0183

ghI -0.5440 -0.5368 -0.2663 -0.2648 -0.2899 -0.2888

ghC -0.8085 -0.8007 -0.3841 -0.3819 -0.4181 -0.4166

jI 0.0377 0.0596 -0.0233 -0.0168 -0.0224 -0.0181

jC 0.1104 0.1106 -0.0123 -0.0122 -0.0270 -0.0270

yI 0.0000 0.0119 0.0049 0.0082 -0.0857 -0.0834

yC 0.0000 -0.0117 -0.0029 -0.0062 -0.1300 -0.1322

lI 0.0000 0.0178 0.0000 0.0050 0.0000 0.0033

lC 0.0000 -0.0231 0.0000 -0.0065 0.0000 -0.0043

kI 0.0000 0.0145 0.0219 0.0261 0.0273 0.0301

kC 0.0000 -0.0090 -0.0136 -0.0162 -0.0170 -0.0187

eI -0.7235 -0.7102 -0.7181 -0.7143 -0.8191 -0.8165

eC -1.2588 -1.2730 -1.2623 -1.2664 -1.4177 -1.4203

e -1.0000 -1.0009 -0.9992 -0.9995 -1.1283 -1.1284

wI 0.0000 -0.0059 -0.0815 -0.0830 -0.0857 -0.0867

wC 0.0000 0.0114 -0.1169 -0.1137 -0.1300 -0.1279

p 0.1720 0.0242 0.0261 -0.0165 -0.0108 -0.0388

grhI -0.0408 -0.0379 0.0000 0.0000 0.0000 0.0000

grfI 0.0000 0.0000 0.0000 0.0000 -0.6410 -0.6403

grhC -0.1321 -0.1320 0.0000 0.0000 0.0000 0.0000

grfC 0.0000 0.0000 0.0000 0.0000 -1.3926 -1.3926

gcI -1.3830 -1.3688 -0.7104 -0.7063 -0.7732 -0.7705

gcC -1.3830 -1.3688 -0.7104 -0.7063 -0.7732 -0.7705

bI 6.2906 6.2906 6.2906 6.2906 6.2906 6.2906

bC 5.6767 5.6767 5.6767 5.6767 5.6767 5.6767

scI 16.2393 16.2393 0.0000 0.0000 0.0000 0.0000

scC 16.2393 16.2393 0.0000 0.0000 0.0000 0.0000

srI 0.0000 0.0000 56.4846 56.4846 56.4846 56.4846

srC 0.0000 0.0000 28.6577 28.6577 28.6577 28.6577

sI 12.6131 12.6131 12.6131 12.6131 12.6131 12.6131

sC 10.3655 10.3655 10.3656 10.3656 10.3656 10.3656

Notes: The symbols in the first column and at the head of the columns of results are the proportional changes of

their upper-case counterparts; thus, for example, vI is the proportional change in VI. SR and LR are abbreviations

of “short run” and “long run”.

37

Appendix 1 Variable definitions

Vi = utility of the representative household, region i

CIi= real private consumption of interior output per household, region i

CCi= real private consumption of coastal output per household, region i

GHi = real government-provided consumption per household, region i.

P = price of interior output in terms of coastal output

Ji= real household income (net of VAT), region i

Wj= real wage income, industry j

ΠHi = real profit distribution per household, region i

RKj= capital rental rate, industry j

Kj= capital stock, industry j

K = national capital stock

TEj= emission tax, industry j

Ej= emission permit, industry j

E = national emission permit

Bj= abatement, industry j

Sj= subsidy per unit of abatement, industry j

SCj= subsidy per unit of abatement by the central government, industry j

SRj= subsidy per unit of abatement by the regional government, industry j

Dj= productivity parameter, industry j

Yj = real output, industry j

Lj= employment, industry j

L= national population

ΠFj = firm profit, industry j

Tv = value added tax rate

Tj = output tax rate, industry j

TRi = lump-sum transfer from the central government to regional government i

GRHi= real regional government-provided consumption good per household, region i

GRFj= real regional government-provided public good, industry j

GCi= real central government-provided consumption good per household in region i

θ= share of valued tax to the central government

μ = hukou parameter

38

Appendix 2: Data base

Variables

CC

(100

million

yuan)

CI

(100

million

yuan)

W*L

(100

million

yuan)

L

(10,000)

GRH

(100

million

yuan)

GRF

(100

million

yuan)

GC

(100

million

yuan)

Coastal 34079.60 22719.73 32491.98 29737.89 7558.78 6998.46 8897.81

Interior 22719.73 17206.99 22487.58 38628.87 7032.19 6599.20 4217.09

Variables

K*RK

(100

million

yuan)

E*TE

(100

million

yuan)

TR

(100

million

yuan)

TE

(yuan

per ton)

SC

(yuan

per

ton)

SR

(yuan

per

ton)

B

(100

million

ton)

Coastal 16489.06 2615.63 3741.39 83.00 64.00 36.00 6.99

Interior 10226.74 2447.82 8066.29 83.00 64.00 18.00 3.39 Sources: Comprehensive Statistical Data and Materials on 60 Years of New China (SSB, 2010), China

Energy Statistical Year Book (SSB, various issues), China Statistics Year Book (SSB, various issues),

Contract Management and Financial Subsidy on Energy Consumption (NDRC, 2010), and State and

Trends of the Carbon Market (World Bank, various issues).

39

Appendix 3 Linearised model

The model is linearised in terms of proportional differences by taking logarithms and

differentials of each equation. The linearised form of equations (1) to (19) (excluding

equations (14) which are redundant) of the model are as follows, with the linearised form

having the same number as the original equation but being distinguished by a prime.

The linearised utility function is:

(1’) i cIiv Ii cCiv Ci ghiv iv c c gh , i=I, C

where lower-case letters represent the proportional changes (log differential) of their upper-

case counterparts and

Ii IicIiv

Ii Ii Ci Ci i i

C

C C GH

,

Ci CicCiv

Ii Ii Ci Ci i i

C

C C GH

,

i ighiv

Ii Ii Ci Ci i i

GH

C C GH

.

The linearised consumption demand functions are:

(2a’) II I cII elasc j p p p ,

where 1

1 1

1

1 ( )

cII

CI

II

P

, 1

1elas

, and

(2b’) IC C cIC elasc j p p ,

where 1

1 1

1

1 ( )

cIC

CC

IC

P

(2c’) CI I cIIc j p p

(2d’) CC C cICc j p

The linearised definitions of real household income are:

(3a’) 1 1 ( )tv v I j h h j hwI I j hkI KI It j h w r k l

( )jrkC KC Cp r k l

where 1

vtv

v

T

T

, 1

1

1(1 )j h h

V

H

T J

,

1

,(1 )

Ij hwI

V

W

T J

1

/,

(1 )

KI Ij hkI

V

R K L

T J

1

1

/

(1 )

KC CjrkC

V

P R K L

T J

(3b’) ( )tv v C j h hC C j hwC C j hkC KC Ct j h w r k l

[ ]jrkI KI Ip r k l

where, (1 )

Cj h hC

V C

H

T J

, ,

(1 )

Cj hwC

V C

W

T J

/,

(1 )

KC Cj hkC

V C

R K L

T J

/

(1 )

KI IjrkI

V C

PR K L

T J

40

The linearised migration equilibrium condition corresponding to equation (4) is:

(4’) /

* log( ) ( )/

C CC I C I

I I

L Av v l l

L A

where μ* = dμ/μ and we have made the obvious assumption that area is constant.

The linearised production functions are:

(5’) j j Lj j Kj j Gj jy d l k grf , j=I, C.

The emission equation

(6’) j EYj j EBj je y b

where j j

EYj

j j j

Y

Y B

and

j

EBj

j j j

B

Y B

The linearised profit definitions are given by:

(7’) ( )j y fj j tj y fj j w fj j jf y t w l

2( ) ( ) ( ) ( )k fj j kj ry fj Ej j rb fj Ej j b fj j sb fj j jk r t y t b b s b

where (1 )

,j j

y fj

j

T Y

F

1

j

tj

j

T

T

,

j j

w f j

j

W L

F

, ,

Kj j

k f j

j

R K

F

j Ej j

ry f j

j

T Y

F

,

Ej j

rb f j

j

T B

F

,

2

2

2 j j

b f j

j

B

F

,

j j

sb f j

j

S B

F

The manufacturing industry’s profit-maximisation condition in linear form is:

(8a’) j ttj j trj Ej j jy t t w l , j=I, C

(8b’) j ttj j trj Ej Kj jy t t r k , j=I, C

(8c’) j bbrj Ej bbsj jb t s , j=I, C

where 1

j

ttj

j Ej j

T

T T

,

1

Ej j

trj

j Ej j

T

T T

,

Ej

bbrj

Ej j

T

T S

,

j

bbsj

Ej j

S

T S

The capital allocation equilibrium condition is:

(9’) KI KCr r .

The central government’s budget constraint is linearised as:

(10’) ( ) ( ) ( )gcIgc I I gcCgc C C gctrI I gctrC C gcreI EI Il gc l gc tr tr t e

csc csc( ) ( ) ( )gcreC EC C g bI I I g bC C Ct e sc b sc b * ( ) ( )v jIj I I jCj C Ct l j l j

where I IgcIgc

I I C C I C EI I EC C I I C C

L GC

L GC L GC TR TR T E T E SC B SC B

,

C CgcCgc

I I C C I C EI I EC C I I C C

L GC

L GC L GC TR TR T E T E SC B SC B

IgctrI

I I C C I C EI I EC C I I C C

TR

L GC L GC TR TR T E T E SC B SC B

CgctrC

I I C C I C EI I EC C I I C C

TR

L GC L GC TR TR T E T E SC B SC B

EI IgcreI

I I C C I C EI I EC C I I C C

T E

L GC L GC TR TR T E T E SC B SC B

41

EC CgcreC

I I C C I C EI I EC C I I C C

T E

L GC L GC TR TR T E T E SC B SC B

cscI I

g bI

I I C C I C EI I EC C I I C C

SC B

L GC L GC TR TR T E T E SC B SC B

cscC C

g bC

I I C C I C EI I EC C I I C C

SC B

L GC L GC TR TR T E T E SC B SC B

I IjIj

I I C C

L J

L J L J

, C C

jCj

C I C C

L J

L J L J

, θ* = dθ/θ,

The regional governments’ budget constraints are linearised as:

(11a’) ( ) ( )grhIgr I I grfIgr I grtrI I grsrI I Il grh grf tr sr b

( ) ( * )tIgr I I tvIgr V I It y t l j

where I IgrhIgr

I I I I I I

L GRH

L GRH GRF TR SR B

, I

grfIgr

I I I I I I

GRF

L GRH GRF TR SR B

,

IgrtrI

I I I I I I

TR

L GRH GRF TR SR B

, ,I I

grsrI

I I I I I I

SR B

L GRH GRF TR SR B

1

,

(1 )

I ItIgr

I I V I I

T Y

T Y T L J

,

(1 )

(1 )

V I ItvIgr

I I V I I

T N J

T Y T L J

, and

(11b’) ( ) ( )grhCgr C C grfCgr C grtrC C grsrC C Cl grh grf tr sr b

( ) ( * )tCgr C C tvCgr V C Ct y t l j

where C CgrhCgr

C C C C C C

L GRH

L GRH GRF TR SR B

,

CgrfCgr

C C C C C C

GRF

L GRH GRF TR SR B

, C

grtrC

C C C C C C

TR

L GRH GRF TR SR B

C CgrsrC

C C C C C C

SR B

L GRH GRF TR SR B

(1 )

C CtCgr

C C V C C

T Y

T Y T L J

,

(1 )

(1 )

V C CtvCgr

C C V C C

T L J

T Y T L J

.

The definition of GHi is linearised as:

(12’) i grhigh i gcigh igh grh gc , i=I, C

where igrhigh

i

GRH

GH , i

gcigh

i

GC

GH .

The definition of subsidy

(13’) sscj j ssrj j jsc sr s , j=I, C

where / , /sscj j j ssrj j jSC S SR S

Equations (14), the goods markets clearing conditions, are dropped from the model due to the

redundancy result explained in section 2.

The profit distribution conditions can be linearised to give:

(15a’) I I If l h ,

(15b’) C C Cf l h .

The balance of trade condition in linear form is:

42

(16’) C IC I CIl p c l c .

The national employment constraint results in the following linearised condition:

(17’) lI I lC Cl l l

where / , /lI I lC CL L L L .

The national capital constraint results in the following linearised condition:

(18’) kI I kC Ck k k

where / , /kI I kC CK K K K .

The national emission permits constraint results in the following linearised condition:

(19’) eI I eC Ce e e

where / , /eI I eC CE E E E .

43

Appendix 4 Calibration

The linearised model contains a number of parameters which have to be evaluated before the

model can be put to work to simulate the effects of various shocks. These parameters fall into

two groups. The first are parameters which appear in model relationships; γji, δi and ρ appear

in the utility function (1) and αGj,αKj, and αLj appear in the production function (5). The

remainder, on the other hand, are linearisation parameters which are all shares of some sort.

The model parameters were evaluated as follows. For the parameters of the utility function

we broadly followed the method set out in Mansur and Whalley (1984) in which the

substitution elasticity σ = 1/(1+ρ) is derived from the equation:

1

i i

i

where i is the (uncompensated) own-price elasticity, values for which were derived as

averages from Table 4 in Mansur and Whalley, and i

can be derived from ratios of

consumption expenditure and our assumption that Ii + Ci + i = 1.

The production parameters, αGj, αKj and αLj. were calibrated as follows. Using the

firm’s first-order condition for profit-maximisation, equation (8a)-(8c), and the assumption

that the firm can choose the government expenditure to maximise profit, we can write:

(1 )

j j

Lj

j j Ej j

W L

Y T T

,

(1 )

Kj j

Kj

j j Ej j

R K

Y T T

, and

(1 )

j

Gj

j j Ej j

GRF

Y T T

and use data for the wage bill, capital rental income, government infrastructure expenditure

and output net of tax to compute the parameters.

The linearisation parameters can be evaluated directly from their definitions, given values

forCji,P, θ, μ, η, IIHi, RKj, Kj, TEj, Ej, Wj, Tv, Tj, Yj, ΠFj, Lj,GCi, Ji,GRHi, GRFi, GHi, Bj, SCi,

SRi, Si and TRi. We normalise P and η at unity and also set the immigration parameter, μ, at

unity; θ is set at 0.75 to reflect the current division of VAT revenue between the central and

regional governments. We then use these assumed values and the data for Cji, GRHi, GRFj,

RKjKj, TEjEj, GCi, LjWj, Bj, SCi, SRi, TRi together with the model definitions to calculate the

value of all other variables. The use of the model definitions ensures that the parameter

values used in the simulations are consistent with the model constraints.

We therefore need data for two regions, the interior and the coast, for the variables Cji, GRHi,

GRFj, RKjKj, TEjEj, GCi, LjWj, Bj, SCi, SRi, TRi. The data we use are based on those for the

Chinese provinces which we have allocated to the two regions as follows. The coastal region

consists of Beijing, Tianjin, Hebei, Guangdong, Hainan, Shandong, Fujian, Zhejiang, Jiangsu,

Shanghai, Liaoning and Guangxi with the remaining provinces being allocated to the interior

region. The interior therefore consist of: Shanxi, Inner Mongolia, Jilin, Heilongjiang, Anhui,

Jiangxi, Henan, Hubei, Hunan, Sichuan, Chongqing, Guizhou, Yunnan, Shaanxi, Gansu,

Qinghai, Ningxia, Tibet, Xinjiang. For each region we use data averaged over the 11-year

period 2000-2010 to avoid cyclical influences on the share parameters. The data for

emissions were generated as follows. We first computed the CO2 emission in each province

using the energy consumption data for coal, gas and oil and their emission factor index and

44

then use the world market CO2 trading price as a measure of carbon tax. The energy

consumption data come from Comprehensive Statistical Data and Materials on 60 Years of

New China (SSB, 2010) and China Energy Statistical Year Book (SSB, various issues), the

emission factor indexes for coal, gas and oil come from IPCC (2006), and the CO2 trading

price come from State and Trends of the Carbon Market (World Bank, various issues). We

computed the abatement data for each year as Bt=GDPt*(E2000/GDP2000)-Et. We let the model

compute the εi parameter with εi=(Ej+Bj)/Yj, and computed the γi parameters for the abatement

cost function from the first-order condition for B. For the subsidy data, we collected from

government’s documents of Contract Management and Financial Subsidy on Energy

Consumption (NDRC, 2010). All the other data come from China Statistics Year Book (SSB,

various issues) except for data on area used to compute population density for the migration

equilibrium condition, equation (4’), which come from China Civil Affairs Statistical

Yearbook 2005 (SSB, 2005).

45

Appendix 5: Table 5.1, Central Government Policy

Variable

Carbon Tax

(gc endogenous)

Subsidy

(gc endogenous)

Carbon Tax

(tV endogenous)

Cut in Both Regions

Subsidy

(tV endogenous)

SR LR SR LR SR LR SR LR

vI 0.1119 -0.0361 -0.0138 -0.0608 0.1879 -0.0043 -0.0461 -0.0740

vC -0.3154 -0.1648 -0.1494 -0.1016 -0.3685 -0.1725 -0.1268 -0.0984

cII -0.4661 -0.5455 0.0925 0.0673 -0.0544 -0.1508 -0.0826 -0.0966

cCI 0.1356 -0.1488 0.1682 0.0780 0.5474 0.1829 -0.0070 -0.0598

cIC -1.2320 -0.9218 -0.0038 0.0946 -0.8202 -0.4072 -0.1790 -0.1191

cCC -0.6302 -0.5251 0.0719 0.1052 -0.2185 -0.0735 -0.1033 -0.0823

ghI 0.9481 0.9707 -0.5440 -0.5368 -0.1675 -0.1581 -0.0695 -0.0681

ghC 1.6020 1.6264 -0.8085 -0.8007 -0.0556 -0.0537 -0.1034 -0.1031

jI -0.9019 -0.8329 0.0377 0.0596 -0.4902 -0.3924 -0.1375 -0.1233

jC -0.3239 -0.3231 0.1104 0.1106 0.0879 0.0964 -0.0648 -0.0635

πhI

-0.5043 -0.5235 0.2590 0.2529 -0.5043 -0.5294 0.2590 0.2554

πhC

-0.2311 -0.1943 0.2531 0.2648 -0.2311 -0.1830 0.2531 0.2601

yI -0.0389 -0.0014 0.0000 0.0119 -0.0389 0.0101 0.0000 0.0071

yC 0.0233 -0.0135 0.0000 -0.0117 0.0233 -0.0248 0.0000 -0.0070

lI 0.0000 0.0560 0.0000 0.0178 0.0000 0.0732 0.0000 0.0106

lC 0.0000 -0.0727 0.0000 -0.0231 0.0000 -0.0951 0.0000 -0.0138

rKI -0.6024 -0.6107 0.0000 -0.0026 -0.6024 -0.6132 0.0000 -0.0016

rKC -0.6024 -0.6107 0.0000 -0.0026 -0.6024 -0.6132 0.0000 -0.0016

kI -0.1753 -0.1295 0.0000 0.0145 -0.1753 -0.1154 0.0000 0.0087

kC 0.1087 0.0803 0.0000 -0.0090 0.1087 0.0716 0.0000 -0.0054

tEI 12.4704 12.4704 0.0000 0.0000 12.4704 12.4704 0.0000 0.0000

tEC 12.4704 12.4704 0.0000 0.0000 12.4704 12.4704 0.0000 0.0000

eI -0.7623 -0.7205 -0.7235 -0.7102 -0.7623 -0.7076 -0.7235 -0.7156

eC -1.2225 -1.2673 -1.2588 -1.2730 -1.2225 -1.2811 -1.2588 -1.2673

e -1.0000 -1.0030 -1.0000 -1.0009 -1.0000 -1.0039 -1.0000 -1.0006

πfI

-0.5043 -0.4675 0.2590 0.2707 -0.5043 -0.4562 0.2590 0.2660

πfC

-0.2311 -0.2670 0.2531 0.2417 -0.2311 -0.2781 0.2531 0.2463

wI -0.7776 -0.7961 0.0000 -0.0059 -0.7776 -0.8018 0.0000 -0.0035

wC -0.4936 -0.4577 0.0000 0.0114 -0.4936 -0.4466 0.0000 0.0068

p 1.3676 0.9017 0.1720 0.0242 1.3676 0.7583 0.1720 0.0837

grhI -0.4333 -0.4241 -0.0408 -0.0379 -0.2680 -0.2530 -0.1112 -0.1090

grfI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

grhC -0.3400 -0.3398 -0.1321 -0.1320 -0.1212 -0.1169 -0.2252 -0.2245

grfC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

gcI 3.2517 3.2966 -1.3830 -1.3688 0.0000 0.0000 0.0000 0.0000

gcC 3.2517 3.2966 -1.3830 -1.3688 0.0000 0.0000 0.0000 0.0000

tI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

tC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

q 0.0000 0.0000 0.0000 0.0000 -0.6224 -0.6336 0.2647 0.2631

tV 0.0000 0.0000 0.0000 0.0000 -1.8672 -1.9009 0.7941 0.7893

bI 6.2510 6.2510 6.2906 6.2906 6.2510 6.2510 6.2906 6.2906

bC 5.6410 5.6410 5.6767 5.6767 5.6410 5.6410 5.6767 5.6767

scI 0.0000 0.0000 16.2393 16.2393 0.0000 0.0000 16.2393 16.2393

scC 0.0000 0.0000 16.2393 16.2393 0.0000 0.0000 16.2393 16.2393

srI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

srC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

sI 0.0000 0.0000 12.6131 12.6131 0.0000 0.0000 12.6131 12.6131

sC 0.0000 0.0000 10.3655 10.3655 0.0000 0.0000 10.3655 10.3655

46

Appendix5: Table 5.2 Regional Government Policy

Variable

Subsidy

(grh endogenous)

Regional Subsidy

(tj endogenous)

Subsidy

(grf endogenous)

Cut in Both Regions SR LR SR LR SR LR

vI -0.0149 -0.0612 -0.0651 -0.0786 -0.0860 -0.0949

vC -0.1486 -0.1015 -0.1039 -0.0902 -0.1115 -0.1024

cII 0.0925 0.0677 -0.0150 -0.0221 -0.0258 -0.0305

cCI 0.1682 0.0792 -0.0035 -0.0293 -0.0306 -0.0476

cIC -0.0038 0.0932 -0.0296 -0.0013 -0.0198 -0.0012

cCC 0.0719 0.1048 -0.0181 -0.0085 -0.0245 -0.0183

ghI -0.5490 -0.5419 -0.2663 -0.2648 -0.2899 -0.2888

ghC -0.8050 -0.7974 -0.3841 -0.3819 -0.4181 -0.4166

jI 0.0377 0.0593 -0.0233 -0.0168 -0.0224 -0.0181

jC 0.1104 0.1106 -0.0123 -0.0122 -0.0270 -0.0270

πhI

0.2590 0.2530 0.1792 0.1776 0.1750 0.1740

πhC

0.2531 0.2646 0.1388 0.1421 0.1260 0.1281

yI 0.0000 0.0117 0.0049 0.0082 -0.0857 -0.0834

yC 0.0000 -0.0115 -0.0029 -0.0062 -0.1300 -0.1322

lI 0.0000 0.0175 0.0000 0.0050 0.0000 0.0033

lC 0.0000 -0.0228 0.0000 -0.0065 0.0000 -0.0043

rKI 0.0000 -0.0026 -0.1034 -0.1041 -0.1131 -0.1135

rKC 0.0000 -0.0026 -0.1034 -0.1041 -0.1131 -0.1135

kI 0.0000 0.0143 0.0219 0.0261 0.0273 0.0301

kC 0.0000 -0.0089 -0.0136 -0.0162 -0.0170 -0.0187

tEI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

tEC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

eI -0.7235 -0.7104 -0.7181 -0.7143 -0.8191 -0.8165

eC -1.2588 -1.2728 -1.2623 -1.2664 -1.4177 -1.4203

e -1.0000 -1.0009 -0.9992 -0.9995 -1.1283 -1.1284

πfI

0.2590 0.2705 0.1792 0.1826 0.1750 0.1773

πfC

0.2531 0.2419 0.1388 0.1356 0.1260 0.1239

wI 0.0000 -0.0058 -0.0815 -0.0830 -0.0857 -0.0867

wC 0.0000 0.0113 -0.1169 -0.1137 -0.1300 -0.1279

p 0.1720 0.0262 0.0261 -0.0165 -0.0108 -0.0388

grhI -0.5432 -0.5403 0.0000 0.0000 0.0000 0.0000

grfI 0.0000 0.0000 0.0000 0.0000 -0.6410 -0.6403

grhC -1.0950 -1.0950 0.0000 0.0000 0.0000 0.0000

grfC 0.0000 0.0000 0.0000 0.0000 -1.3926 -1.3926

gcI -0.5587 -0.5446 -0.7104 -0.7063 -0.7732 -0.7705

gcC -0.5587 -0.5446 -0.7104 -0.7063 -0.7732 -0.7705

tI 0.0000 0.0000 1.4191 1.4167 0.0000 0.0000

tC 0.0000 0.0000 1.2464 1.2465 0.0000 0.0000

q 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

tV 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

bI 6.2906 6.2906 6.2906 6.2906 6.2906 6.2906

bC 5.6767 5.6767 5.6767 5.6767 5.6767 5.6767

scI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

scC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

srI 56.4846 56.4846 56.4846 56.4846 56.4846 56.4846

srC 28.6577 28.6577 28.6577 28.6577 28.6577 28.6577

sI 12.6131 12.6131 12.6131 12.6131 12.6131 12.6131

sC 10.3656 10.3656 10.3656 10.3656 10.3656 10.3656

47

Appendix 6: Table 6-1, Central Government Policy (Regionally-Differentiated Policy)

Variable

Carbon Tax

(gc endogenous)

Subsidy

(gc endogenous)

Carbon Tax

(tV endogenous)

Cut in Both Regions

Subsidy

(tV endogenous)

SR LR SR LR SR LR SR LR

vI 0.0222 -0.0626 -0.0039 -0.0579 0.0972 -0.0314 -0.0363 -0.0711

vC -0.2227 -0.1364 -0.1598 -0.1049 -0.2752 -0.1440 -0.1372 -0.1016

cII -0.4718 -0.5173 0.0942 0.0652 -0.0654 -0.1300 -0.0815 -0.0990

cCI -0.0759 -0.2389 0.1906 0.0868 0.3305 0.0866 0.0149 -0.0512

cIC -0.9758 -0.7980 -0.0286 0.0847 -0.5694 -0.2930 -0.2043 -0.1294

cCC -0.5798 -0.5195 0.0679 0.1063 -0.1734 -0.0764 -0.1078 -0.0816

ghI 0.9759 0.9888 -0.5466 -0.5384 -0.1251 -0.1189 -0.0706 -0.0689

ghC 1.5563 1.5703 -0.8126 -0.8037 -0.0796 -0.0783 -0.1052 -0.1049

jI -0.7586 -0.7190 0.0243 0.0495 -0.3522 -0.2868 -0.1514 -0.1337

jC -0.3782 -0.3778 0.1170 0.1173 0.0282 0.0338 -0.0587 -0.0572

πhI

-0.4422 -0.4532 0.2411 0.2341 -0.4422 -0.4590 0.2411 0.2366

πhC

-0.2556 -0.2345 0.2625 0.2760 -0.2556 -0.2234 0.2625 0.2712

yI -0.0213 0.0001 0.0000 0.0137 -0.0213 0.0115 0.0000 0.0089

yC 0.0128 -0.0083 0.0000 -0.0134 0.0128 -0.0194 0.0000 -0.0087

lI 0.0000 0.0321 0.0000 0.0204 0.0000 0.0490 0.0000 0.0133

lC 0.0000 -0.0417 0.0000 -0.0265 0.0000 -0.0636 0.0000 -0.0172

rKI -0.5928 -0.5976 0.0000 -0.0030 -0.5928 -0.6001 0.0000 -0.0020

rKC -0.5928 -0.5976 0.0000 -0.0030 -0.5928 -0.6001 0.0000 -0.0020

kI -0.0961 -0.0699 0.0000 0.0167 -0.0961 -0.0561 0.0000 0.0109

kC 0.0596 0.0434 0.0000 -0.0104 0.0596 0.0348 0.0000 -0.0067

tEI 11.2699 11.2699 0.0000 0.0000 11.2699 11.2699 0.0000 0.0000

tEC 13.1709 13.1709 0.0000 0.0000 13.1709 13.1709 0.0000 0.0000

eI -0.6735 -0.6495 -0.6735 -0.6583 -0.6735 -0.6369 -0.6735 -0.6636

eC -1.3056 -1.3313 -1.3056 -1.3219 -1.3056 -1.3448 -1.3056 -1.3162

e -1.0000 -1.0017 -1.0000 -1.0011 -1.0000 -1.0026 -1.0000 -1.0007

πfI

-0.4422 -0.4211 0.2411 0.2545 -0.4422 -0.4100 0.2411 0.2498

πfC

-0.2556 -0.2762 0.2625 0.2494 -0.2556 -0.2871 0.2625 0.2540

wI -0.6890 -0.6995 0.0000 -0.0068 -0.6890 -0.7051 0.0000 -0.0044

wC -0.5332 -0.5126 0.0000 0.0131 -0.5332 -0.5017 0.0000 0.0085

p 0.8999 0.6328 0.2192 0.0491 0.8999 0.4922 0.2192 0.1088

grhI -0.3634 -0.3581 -0.0423 -0.0390 -0.2002 -0.1902 -0.1129 -0.1102

grfI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

grhC -0.3893 -0.3892 -0.1357 -0.1356 -0.1733 -0.1705 -0.2291 -0.2283

grfC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

gcI 3.2092 3.2349 -1.3876 -1.3712 0.0000 0.0000 0.0000 0.0000

gcC 3.2092 3.2349 -1.3876 -1.3712 0.0000 0.0000 0.0000 0.0000

tI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

tC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

q 0.0000 0.0000 0.0000 0.0000 -0.6143 -0.6218 0.2656 0.2636

tV 0.0000 0.0000 0.0000 0.0000 -1.8427 -1.8653 0.7968 0.7907

bI 5.6492 5.6492 5.8561 5.8561 5.6492 5.6492 5.8561 5.8561

bC 5.9578 5.9578 5.8876 5.8876 5.9578 5.9578 5.8876 5.8876

scI 0.0000 0.0000 15.1177 15.1177 0.0000 0.0000 15.1177 15.1177

scC 0.0000 0.0000 16.8426 16.8426 0.0000 0.0000 16.8426 16.8426

srI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

srC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

sI 0.0000 0.0000 11.7419 11.7419 0.0000 0.0000 11.7419 11.7419

sC 0.0000 0.0000 10.7506 10.7506 0.0000 0.0000 10.7506 10.7506

48

Appendix 6: Table 6-2, Regional Government Policy (Regionally-Differentiated Policy)

Variable

Subsidy

(grh endogenous)

Subsidy

(tj endogenous)

Subsidy

(grf endogenous)

Cut in Both Regions SR LR SR LR SR LR

vI 0.0000 -0.0563 -0.0671 -0.0794 -0.0793 -0.0853

vC -0.1625 -0.1052 -0.1025 -0.0899 -0.0964 -0.0903

cII 0.0942 0.0640 -0.0154 -0.0219 -0.0236 -0.0268

cCI 0.1906 0.0825 -0.0074 -0.0310 -0.0335 -0.0449

cIC -0.0286 0.0894 -0.0256 0.0002 -0.0111 0.0014

cCC 0.0679 0.1079 -0.0176 -0.0089 -0.0209 -0.0167

ghI -0.5292 -0.5206 -0.2665 -0.2651 -0.2569 -0.2562

ghC -0.8245 -0.8152 -0.3844 -0.3823 -0.3705 -0.3695

jI 0.0243 0.0506 -0.0212 -0.0153 -0.0165 -0.0136

jC 0.1170 0.1173 -0.0135 -0.0135 -0.0260 -0.0260

πhI

0.2411 0.2338 0.1706 0.1692 0.1469 0.1462

πhC

0.2625 0.2765 0.1425 0.1455 0.1148 0.1163

yI 0.0000 0.0143 0.0065 0.0096 -0.0684 -0.0668

yC 0.0000 -0.0140 -0.0039 -0.0069 -0.1208 -0.1222

lI 0.0000 0.0213 0.0000 0.0046 0.0000 0.0022

lC 0.0000 -0.0276 0.0000 -0.0060 0.0000 -0.0029

rKI 0.0000 -0.0032 -0.1038 -0.1044 -0.1007 -0.1010

rKC 0.0000 -0.0032 -0.1038 -0.1044 -0.1007 -0.1010

kI 0.0000 0.0174 0.0291 0.0330 0.0323 0.0342

kC 0.0000 -0.0108 -0.0181 -0.0205 -0.0201 -0.0212

tEI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

tEC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

eI -0.6735 -0.6576 -0.6735 -0.6701 -0.6735 -0.6718

eC -1.3056 -1.3226 -1.3056 -1.3092 -1.3056 -1.3073

e -1.0000 -1.0011 -1.0000 -1.0002 -1.0000 -1.0001

πfI

0.2411 0.2551 0.1706 0.1738 0.1469 0.1484

πfC

0.2625 0.2489 0.1425 0.1395 0.1148 0.1134

wI 0.0000 -0.0070 -0.0746 -0.0760 -0.0684 -0.0690

wC 0.0000 0.0137 -0.1218 -0.1189 -0.1208 -0.1193

p 0.2192 0.0420 0.0182 -0.0207 -0.0224 -0.0412

grhI -0.5100 -0.5065 0.0000 0.0000 0.0000 0.0000

grfI 0.0000 0.0000 0.0000 0.0000 -0.5273 -0.5268

grhC -1.1344 -1.1343 0.0000 0.0000 0.0000 0.0000

grfC 0.0000 0.0000 0.0000 0.0000 -1.2829 -1.2829

gcI -0.5612 -0.5442 -0.7109 -0.7071 -0.6852 -0.6834

gcC -0.5612 -0.5442 -0.7109 -0.7071 -0.6852 -0.6834

tI 0.0000 0.0000 1.3326 1.3305 0.0000 0.0000

tC 0.0000 0.0000 1.2894 1.2894 0.0000 0.0000

q 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

tV 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

bI 5.8561 5.8561 5.9188 5.9188 5.1935 5.1935

bC 5.8876 5.8876 5.8663 5.8663 5.2223 5.2223

scI 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

scC 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

srI 52.5832 52.5832 53.1460 53.1460 46.6334 46.6334

srC 29.7223 29.7223 29.6148 29.6148 26.3638 26.3638

sI 11.7419 11.7419 11.8676 11.8676 10.4133 10.4133

sC 10.7506 10.7506 10.7117 10.7117 9.5358 9.5358