Z¥.

Energy Economics 23 2001 121
139

Subglobal climate-change actions and
carbon leakage: the implication of
international capital flows

Mustafa H. Babiker¥

MIT Joint Program on the Science and Policy of Global Change, Cambridge, MA 02139-4307,
USA


Abstract

The climate-change agreement negotiated in Kyoto obliges the OECD countries to
initiate the international effort of abating the anthropogenic greenhouse gas emissions. In
the event that such an initiative is taken, the associated competitive pressures may induce
significant offsetting increases in non-OECD emissions Z¥a process generally known as
leakage . .The current models used to study these competitive effects have adopted an
empirically inconsistent view of the world that international capital markets are perfectly
integrated. To the extent that restrictions on the international mobility of capital affect
regional investment decisions, it might be expected that the competitive impacts drawn from
these models could significantly be altered. This paper addresses this issue. The paper
suggests that these competitive effects are not really contingent on capital flows. With a
regionally disaggregated dynamic model of the world economy, we show a quite robust result
that carbon leakage is virtually unaffected by the presence of restrictions on the mobility of
international capital. ¥ 2001 Elsevier Science B.V. All rights reserved.

JEL classifications: F21; Q40

Keywords: Kyoto; Leakage; Capital flows; Competitive effects

1. Introduction
If OECD countries are to engage unilaterally in an effort to abate anthropogenicCO -emissions along the lines suggested by the Berlin Mandate Z¥1995, 1996 . and

E-mail address: babiker@mit.edu Z¥M.H. Babiker . .

0140-9883
01
$ -see front matter ¥ 2001 Elsevier Science B.V. All rights reserved.
PII: S0140-9883 Z¥ 00 . 00065-7


()

122 M.H. Babiker
Energy Economics 23 2001 121
139

Z¥.

the Kyoto Agreement 1997 , it is possible that significant offsetting increases in
global emissions may take place in non-OECD countries. The scope for such a
leakage effect as well as its manifestations on international trade flows, the
location of energy-intensive production, and the competitiveness of OECD
economies have been a subject of considerable interest among climate-change
policies analysts in the US and Europe. Nevertheless, almost all of the existingComputable General Equilibrium Z¥CGE. models used in studying these competitive
impacts have adopted an orthodox view of the world that international capital
markets are perfectly integrated.1 Unfortunately, the accumulated empirical
evidence over the last two decades seem to refute strongly such a view.

Z¥. Z¥

The findings in Feldstein and Horioka 1980 , and Gordon and Bovenberg 1996.
that the regional investment-savings patterns are largely independent of the rate of
return differentials are suggestive indicators that there have been some restrictions
on the international mobility of capital. The presence of such restrictions may be

Z¥.

due to risk aversion and capital market segmentation Feldstein, 1994 , to deliber


Z¥.

ate government policies Summers, 1988 , or due to imperfect information transfers.
To the extent that restrictions on the international mobility of capital affect
regional investment decisions, it might be expected that the competitive impacts
drawn from these models could significantly be altered. This paper addresses this
issue.

Specifically, we ask the question whether the impacts of a unilateral OECD
abatement action on the non-OECD emissions trajectories are affected by the
extent of capital flows. To put it more precisely: Do restrictions on international
capital mobility significantly affect the regional and the global leakage profiles?

We use a dynamic multi-regional disaggregated general equilibrium model of the
world economy to address the linkage between capital flows and carbon leakage.
Given the complexity of the interactions among regional economies, energy markets,
international trade, capital accumulations, and capital flows, such a detailed
computable general equilibrium framework is clearly warranted. Within this framework
we show an invariant result that carbon leakage is virtually unaffected by the
extent of capital flows.

For the current climate-change-policies regime, such a result has two crucial
implications. First, the policy recommendations to be drawn from the existing
Impact Assessment models are largely independent of the capital-flows assumption.
Second, in the event of a subglobal abatement action, policies that aim at
curtailing leakage through restricting the international mobility of capital are
virtually ineffective. Given the economic and political leverage of OECD it is
feasible that some forms of restrictions on capital flows, via the international
lending institutions, may be effected with the objective to curtail carbon leakage.
Our result suggests that exercising such a power does not yield any significant

1 Z¥. .

Examples of such impact assessment models are GREEN OECD , CRTM Z¥Rutherford , G-cubed
Z¥McKibbin and Wilcoxen , . IIAM Z¥Montgomery, Rutherford, and Bernstein , . and WorldScan Z¥CPB
Netherlands Bureau of Economic Policy Analysis . .For a description and comparisons among thedifferent Impact Assessment models see Gaskins and Weyant Z¥1993 . .


()

M.H. Babiker
Energy Economics 23 2001 121
139 123
reduction in leakage. The rest of the paper is organized as follows. Section 2
outlines the key features of our dynamic general equilibrium model. Section 3
presents the numerical results, and Section 4 concludes.

2. The model
The modeling framework adopted in this analysis is the conventional Ramsey
type growth model, in which along the baseline all real variables grow at a uniform
growth rate and the present values of all nominal variables decrease at a uniform
discount rate. The first year in the model is year 2000 and the horizon extendsthrough 2040 in time-intervals Z¥or periods .each of 5 years length. The model is
built on a comprehensive energy
economy dataset that accommodates a consistent
representation of energy markets in physical units as well as detailed accounts of
regional production and bilateral trade flows for 1992 Z¥for details on the dataset,
see Rutherford and Babiker, 1997 . .

There are seven regions in the model: OECD as defined in 1990 Z¥OEC ; oil .

Z¥. Z¥.Z¥.

exporting countries OEX ; former Soviet Union FSU ; China and India CHI ;

Z¥.

the dynamic Asian economies DAS ; the dynamic economies of South America
Z¥. ROW . . There are seven commodities in the

DSM ; and the rest of the world Z¥
model: five energy goods and two non-energy goods. The energy goods identified inZ¥COL , natural gas . .Z¥CRU , refined oil .

the model include coal Z¥GAS , crude oil
products Z¥OIL , and electricity .Z¥ELE . . This disaggregation is essential to distinguish
energy goods by carbon intensity and by the degree of substitutability. Thetwo non-energy goods are an energy intensive tradable good Z¥EIS , and a non-.

Z¥.

energy intensive good Y . The primary factors in the model include labor, physical
capital, and fossil-fuel resources. Fossil-fuel resources are sector-specific, but labor
and physical capital are treated as perfectly mobile across sectors. The key features
and equations in the model formulation are outlined in the following.

2.1. Consumer behaior
In each region, r, the intertemporal utility function of the infinitely lived
representative consumer is the discounted sum of the utility of consumption over
the horizon:

Ur Z¥.. ¥ Y¥
¥ z¥1 /tC1rt
¥
Z¥. 1

1 ¥ ¥ 1 ¥ 

t

In this equation, ¥ is the discount rate, ¥ is the coefficient-of-relative-risk-aversion
Z¥CRRA. that controls the intertemporal elasticity of substitution, and Crt is a
constant-elasticity-of-substitution Z¥CES.aggregate of energy and non-energy consumer
goods in the model:


124 ( )M.H. Babiker
Energy Economics 23 2001 121
139
C rt
E E¥ it it¥ C ¥ ¥ C¥
r irt r i rt
E1
t Z¥.2
iE iE

Where the elasticity of substitution between the energy and the non-energy
composites is given by Et ¥ 1
Z¥1 ¥ Et ., i indexes all goods, E indexes energy
goods, i s are the value shares, and where r and r are the requirement
coefficients in the CES function.2

In the standard manner, the representative consumer maximizes the utility

Z¥.

function in Eq. 1 subject to a life-time budget constraint that the present value of
consumption equals the present value of income:

max UCZ¥. s.t.

r rt
Crt


ck tR

Y¥Y¥PC ¥ PK 0 ¥ Y¥Pl Z¥1 ¥ G0.L0 ¥ Y¥Y¥ PR

irt irt r ,1 r rt r Frt Frt
it tFt


num g

Z¥.

Y¥Pt BOPDr t ¥ Y¥TRr t ¥ Y¥Y¥ PGirt 3

irt
t tit

In this expression Pc , Pk , Pl , PR , Pg, and P num are present-value price indices for
household consumption, capital, labor, rents on fossil-fuel resources, government
consumption, and a numeraire index, respectively. K 0 is the initial capital stock,
L0 is the initial labor supply, R is the fossil-fuel resource supply, TR is the total
revenue from taxes, BOPD are the baseline exogenous capital flows, G0 is the
exogenous growth rate in the labor-augmented technical change, and Grt is the
government consumption bundle and has the same form as Crt in Eq. Z¥.

2.
In this representation the household consumption is financed from labor income,
the value of the initial capital stock, the rents realized from the ownership of the
fossil-fuel resources, the tax revenues, and some exogenous capital flows. Taxes
apply to energy as well as non-energy demand, to production and incomes, and to
international trade. The exogenous capital flows represent the initial balance of
payment deficits in the dataset and we project them to grow along the baseline at
the GNP growth rate. The government consumption is financed through lump sum
levies and does not enter the representative consumer utility function, hence the
level of the government activity is fixed exogenously in our model.

2.2. Production actiities
The model includes two types of production functions: those of fossil fuels
Z¥.and those of non-fossil fuels Z¥.

CRU, COL, OIL, and GAS ELE, EIS, and Y . An
index Yir t denotes the level of production of good i in region r in period t. Except
for crude oil, which is modeled as a perfectly homogenous good, good i is produced

2 To economize on notation, we shall use the symbols , ¥ and ¥ throughout to denote these
technology coefficients.


()

M.H. Babiker
Energy Economics 23 2001 121
139 125
as differentiated products for sale in the domestic and international markets. The
shares of sales at home and abroad are determined by relative prices. A constant-
elasticity-of-transformation Z¥CET.function characterizes the allocation of output
between domestic and export sales. Producers of the final good maximize profits
subject to the constraint:

¥ ¥ 1¥ ¥ Z¥.

Yir t ¥ ir Dirt ¥ ir X irt 4

In this equation the transformation elasticity is given by tr ¥ 1
Z¥1 ¥ ..

The production of the non-fossil fuel good, N, is associated with a nested CES
function based on non-energy intermediate inputs, Z, an energy component, E,
and a primary factor composite, V. Given the prices of these components, the
producer of good N operates to minimize the production cost for a given level of
output subject to the technology constraint:

E

1
t 5 Z¥.

tt

YNr t ¥ min 1/2min jNr ZjN r t 4,

¥ E¥ E ¥ ¥ V ¥ E

5

Nr Nrt Nr Nrt

j

In this function the non-energy intermediate inputs enter at the top nest in fixed
proportions among themselves as well as in relation to the energy-primary-factor
aggregate. In the second nest, we account for the substitution between energy andprimary factors through Et ¥ 1
Z¥1 ¥ Et .. In the third nest we characterize separately
the substitution possibilities among the components of the primary factor
composite and among the components of the energy composite. The primary factor
composite is represented by a Cobb
Douglas aggregate of labor and capital
services:

¥ L¥ Nr 1¥ Nr 6

VNrt Nrt KNrt Z¥.

In this equation labor is expressed in efficiency units and ¥ is the labor value
share.

The aggregate energy good, E, is produced by the linear technology:

Z¥.

ENr t ¥ EfN r t ¥ EbNr t 7

According to this equation there are two sources for the aggregate energy good
that are perfectly substitutable. There is a current low-cost fossil fuel source, f, and
there is a high-cost ÔbackstopÕ carbon free source that may be introduced in the
future.3 The fossil-fuel energy source is in turn associated with a nested CES

3Different from the conventional treatment, here we differentiate the backstop technology by sector.
This seems more appropriate than assuming a uniform technology, since it is more conceivable that theform of backstop for producing electricity Z¥e.g. from biomass .may be quite different from the form
suitable for producing chemical products. In our model, these technologies are calibrated according to
cost, market share, and the date of entry.


RR FrFr 4Y R min Z ,LZ¥.F r t F r F rt F r jF rt F r t
()

126 M.H. Babiker
Energy Economics 23 2001 121
139

function based on refined oil, gas, coal, and fossil-fuel based electricity:

o o

f

Nr
o 1Nr Z¥.

¥ ELEN r t 5 8
E Nrt ¥ 1/2

Nr O ¥ GN r t ¥ NrCOLNr t

In this expression electricity enters in a Cobb
Douglas form with oil, gas, and coalat the top nest, with a value share defined by Z¥1 ¥ .. At the second level the

oo

oil
gas composite substitutes with coal according ¥ ¥ 1
Z¥1 ¥ ¥ .. The oil
gas
composite, on the other hand, is assumed to have a simple Cobb
Douglas representation.
Hence, in our model, carbon abatement may be achieved in three ways:
changing the fossil fuels mix, reducing the amount of energy per unit of output,
and by investing in the carbon free source.

In contrast, the production of fossil fuel F is associated with a nested CES
function based on a fuel-specific resource, labor, and intermediate inputs. Given
the prices of these inputs, mine managers minimize production costs subject to the
technology constraint:

R

1
Fr

Z¥.

9

In this equation production is characterized by the presence of a resource in
fixed-supply that substitutes with the rest of inputs at the top level nest accordingto the elasticity R ¥ 1
Z¥1 ¥ R .. This substitution elasticity is controlled by the

Fr Fr

supply price elasticity Z¥.for the particular fuel according to the formula
Fr

1 ¥ Fr R


¥ Fr , with Fr being the resource value share. At the second nest, the

Fr

rest of the inputs enter in fixed proportions. On the other hand, since the refinery
activity does not require a sector-specific resource, the production technology for
refined oil collapses to the fixed proportion part of Eq. Z¥.

9.
2.3. Supplies of final goods and foreign trade
Except for crude oil, intermediate and final consumption goods are differentiated
following the standard Armington convention. Accordingly, for each type, the
total supply of the good is a CES composite of a domestic variety and an imported
one. Given the domestic and the import prices, firms in the distribution sector
maximize profits subject to the constraints:

Zir t ¥

ir ZD¥
ir t
D ¥ ir ZM¥
ir t
D ,

1
¥ D

,

Cir t ¥

ir CD¥
ir t
D ¥ irCM¥
ir t
D

and

1
¥ D . Z¥.

10

Gir t ¥

ir GD¥
ir t
D ¥ irGM¥
ir t
D


() 127

M.H. Babiker
Energy Economics 23 2001 121
139
In these expressions the Armington elasticity between the domestic and the
imported varieties is controlled by D .

All goods are traded in world markets subject to export taxes, tariffs, and
transport costs. Crude oil is exported and imported as a perfectly homogenous
good, whereas all other goods are characterized by product differentiation with
explicit representation of bilateral trade flows. Given the regional export prices,
tariffs, and transport costs, firms operating in the import sector minimize costs by
allocating their import orders across the different trading partners subject to the
constraint:

1
¥ M Z¥.

Y¥¥ X¥ M

11

Mir t ¥

isr isrt

s

In this equation Xisr t is the amount imported by region r from region s, and M
controls the extent of the product differentiability among the trading partners.

2.4. Capital accumulation
The region aggregate investment is the sum of the sectoral investments in the
region. Part of the investment in period t is assumed to mature in the same period
and the rest in period t ¥ 1. The capital stock evolves according to the standard
rule:

Kr ,t1 ¥ Z¥1 ¥ .Krt ¥ Z¥1 ¥
.Irt ¥ ¥ Ir ,t1,

Z¥.

Kr ,1 ¥ K 0 12

r

In this expression, ¥ is the capital depreciation rate, ¥ is the investment own-period
maturation rate, and K 0 is the initial capital stock.

2.5. Market clearance conditions
Output for the domestic market in period t is either consumed or invested:
Dir t ¥ ZDi rt ¥ CDi r t ¥ GDirt ¥ Investirt Z¥.
13
The output for the export market in period t has to meet the regional demands:

¥ 14

Xir t Y¥Xirst Z¥.
s

The import supply in period t has to satisfy the domestic demands for the imported
good:

Z¥.

Mir t ¥ ZMir t ¥ CMir t ¥ GMir t 15


()

128 M.H. Babiker
Energy Economics 23 2001 121
139

Finally, international markets have to clear for each good and in each period:

Z¥.

Y¥X ¥ Y¥M 16

irt irt
rr

2.6. Emissions and carbon leakage
CO2 emissions are generated in fixed proportions via the consumption of fossil
fuels by the industry and the final demand sectors. Accordingly, the carbon
emissions in region r in period t are given by:

Z¥. Z¥.

Emissions ¥ Y¥CO Coeff Z ¥ C ¥ G 17

rt 2 F Frt Frt Frt
F

In which, CO Coeff is the carbon content expressed in the heat units Z¥b ton


2

exajoule . .Under a subglobal abatement action, the carbon leakage rates in the
non-abating regions are defined as the deviations in their emissions from their
baseline trajectories divided by the corresponding amount of abatement in the
colluding regions. The global leakage rate is then simply the sum of the regional
leakage rates.

2.7. Balance of payments and capital flows
The net balance of payment deficit in region r in period t is given by the
expression:

¥ X PXX ¥ Z¥Oil m .NBOBD Y¥Y¥PM ¥ Y¥Y¥ ¥ Oil xPcrude

rt i srt i srt ir st i r st rt rt t
s i s i
num ¥ P BOPDt r t Z¥.18

In this expression PM is the cif present-value price of imports, PX is the fob
present-value price of exports, Oil m is crude oil imports, Oil x is crude oil exports,
and Pcrude is the international present-value price of crude oil. As before, Pnum is
a present-value numeraire price and BOPD is the baseline balance of payment
deficit. Along the baseline the last term in Eq. Z¥.represents the net capital

18
inflows. It is critical, however, which price to use as a numeraire for denominating
the baseline balance of payment deficits. In principle the price of any homogenous
commodity traded in the international market can serve the role of a numeraire.
Unfortunately, the only homogenous good in our model Z¥crude oil .has its price
directly affected by the carbon abatement action. Hence we canÕt use crude oil as a
numeraire and instead we decide to use the OECD labor price index.

Numerically, the model is formulated and solved as Mixed Complementarity
Problem Z¥MCP.using GAMS
MPSGE system described in Rutherford Z¥1995,
.

1997 . In terms of size and complexity, the model proved to be quite challenging
since it requires the solution of a highly non-linear system of approximately 10 000
equations. The full code of the model is available from the author upon request.


() 129

M.H. Babiker
Energy Economics 23 2001 121
139
3. Results
3.1. Parameters and policy scenarios
The essential parameters and elasticities used in the model are summarized in
Table 1. The carbon emissions limit in OECD is set according to the Kyoto
Agreement, and is assumed to take effect starting from year 2005 on.4 To assess
the impact of capital controls on carbon leakage in the presence of an OECD
abatement policy, we implement a scenario in which all regions, except OECD, are
constrained in the international capital market. According to this scenario each
non-OECD region is subject to global capital restrictions that its BOP deficit must
not exceed its baseline trajectory throughout the model horizon. The baseline Z¥or
the Business as Usual, BaU.is in turn defined as the situation where there is no
carbon abatement anywhere.

A natural way of modeling capital controls and financial repression is by treating
them as implicit taxes Z¥Giovannini and de Melo, 1993; Alm and Buckley, 1997;
Manne and Stephan, 1997 . .Provided the absence of financial instruments and
government deficits in our model, BOP controls are equivalent to current accountcontrols. Therefore, the capital controls are modeled as endogenous taxes Z¥or
equivalently, foreign exchange premiums.on international merchandise trade
subject to the constraint that the net current account deficits should not exceed
their target levels in the scenario.

3.2. Numerical results ¥ central case
In the central case the Armington elasticities Z¥D and M .are, respectively, 8
Z¥.is 5%, and the crude oil supply elasticity Z¥. is

and 16, the utility discount rate ¥

1.
Fig. 1 shows the implications of carbon Policies on regional BOP deficits with
and without capital flow constraints. Fig. 1a displays the magnitudes of the BAU
projected exogenous capital flows in present-value dollars. It is clear that the oilexporting countries Z¥OEX.are the major suppliers of international capital and thatOECD and the dynamic Asian economies Z¥DAS.are the major demanders. It isalso evident that all other regions, except the rest of the world Z¥ROW.and the

Z¥.

dynamic South American economies DSM , are net borrowers in the international
capital market. In disposable income terms, these flows amount to 0.7% in OECD,
1.3% in FSU, 1.9% in CHN, 15% in OEX, 0.5% in DSM, 5.5% in DAS, and 1% in
ROW for the year 2000. Fig. 1b shows how the regional BOP deficits might evolve
over time in the absence of capital controls if OECD were to adopt the carbon
abatement policy. Fig. 1b suggests significant capital outflows from OECD to the
other regions during the first 15 years of the horizon. In particular OECD appears
to move from a major importer of international capital to a major exporter,

4 We estimate the carbon limit implied by Kyoto to 92.75% of the 1992 OECD emissions level.


130 ()

M.H. Babiker
Energy Economics 23 2001 121
139
Table 1

A summary of the modelÕs parameters

Parameter Description Value
Low Medium High
Comments
¥ tr The transformation elasticity 1 2 ¥ 1 for fossil fuels
2 for non-fossil fuels
¥ E
t Substitution between energy and
non-energy in intermediate and
final demand sectors
0.25 0.35 0.5 Increases gradually
from the short run to
the long run
O Substitution between the oil
gas
composite and coal in production
¥ 0.5 ¥ Non-fossil fuel
production only
¥ D Armington substitution between
domestic and imports
¥
4
2
8
¥
16
Energy goods
Non-energy goods
¥ M Armington substitution across
imports
¥
8
4
16
¥
32
Energy goods
Non-energy goods
Fr Fossil fuels supply elasticities 0
¥
¥
1
0.5
1
5
¥
¥
Crude oil
Coal
Natural gas
G0 Labor supply growth rate ¥
¥
2.5%
2.5%
¥
3.5%
Uniform growth case
Different growth case:
2.5 for OECD and
3.5 for non-OECD
¥ The depreciation rate of capital
stock
¥ 7% ¥ Uniform across regions
¥ Investment own-period
maturation rate
¥ 0.4 ¥ Uniform across regions
¥ The utility discount rate 0.03 0.05 0.07 Uniform across
regions
¥ The coefficient of relative risk
Z¥aversion CRRA.
¥ 2 ¥ Uniform across
regions

whereas both DSM and ROW are now net importers of capital. In contrast, Fig. 1c
shows how the regional BOP deficits might evolve in the presence of global
restrictions on capital mobility. It is clear from Fig. 1c that international capital
flights are virtually sterilized in the first two decades during which the BOP
constraints are binding.

Next we investigate the patterns of global and regional leakage rates under the
regime of capital flows vs. the regime of capital controls.


() 131

M.H. Babiker
Energy Economics 23 2001 121
139
Z¥. Z¥ .

Fig. 1. Carbon tax, BOP deficits and capital flows: a baseline BOP deficits BAU ¥ no carbon tax ;

Z¥. Z¥.

b BOP deficits: capital flows; and c BOP deficits: capital controls.


132 ()

M.H. Babiker
Energy Economics 23 2001 121
139
Fig. 2. Capital flows and global carbon leakage.

Fig. 2 depicts the global carbon leakage profiles under these two regimes. In
both regimes the leakage rate is high at the beginning of the horizon, but declines
gradually to reach a level of 10% by the end of the horizon. Such a time profile of
leakage may be explained by the fact that the unilateral emissions cut in OECD
puts a downward pressure on the international crude-oil price in the early periods
and thereby induces non-OECD regions to become more energy intensive. However,
the strength of this pressure declines as the backstop technologies start to
kick in gradually after year 2020. Compared to other studies, our average leakage
rate of 14% for the capital flows case is 2% higher than that reported in Babikerand Rutherford Z¥1998 for the static version of the model, and 3% higher than that.
reported in OECD Z¥1992. for a similar parameterization of an EU unilateral
abatement initiative. With the capital controls in effect, in spite of the tightness ofthe restrictions, Fig. 2 indicates that Z¥in absolute terms . the global leakage rate is
at most only 1% lower than that in the case of capital flows. Furthermore, the
figure shows a virtually complete convergence to the no-controls leakage profile by
year 2025. Hence, the upshot of Fig. 2 is the result that the global leakage profile is
largely unaffected by the restrictions on capital flows.

Then, are the magnitudes and time profiles of the regional leakage rates affected
by the presence of capital controls? The answer is provided in Fig. 3, where the
time profiles of regional leakage rates under the two regimes are depicted in Fig.
3a,b, respectively. Overall, the plots reveal only small downward shifts in leakage


() 133

M.H. Babiker
Energy Economics 23 2001 121
139
Z¥. Z¥.

Fig. 3. Capital flows and regional leakage rates: a regional leakage rates: capital flows; and b
regional leakage rates: capital controls.


134 ()

M.H. Babiker
Energy Economics 23 2001 121
139
profiles from their unrestricted levels during the first two decades. Indeed, without
a careful look, it may be even hard to discern the differences between the two
plots. Therefore, not only the global leakage profile but also the regional leakage
profiles are largely unaffected by the presence of capital controls.

Next we ask whether our result of the ineffectiveness of capital controls on
carbon leakage is in fact dictated by the parameter values used in the central case.
The following subsection addresses this question. We shall see that this result holds
regardless of the particular assumptions on the trade structure, the oil supply
elasticity, the utility discount rate, and the pattern of regional growth rates.

3.3. Sensitiity analysis
In assessing the sensitivity of the relationship between leakage and capital flows
to the alternative model parameters, we limit our focus to the global leakage rate.
To ensure comparability across the different cases and the different parameter
values, we turn off the backstop technologies in all the sensitivity simulations. In
relation to carbon leakage, the three crucial parameters in the model are the utility
discount rate, the two Armington elasticities, and the crude-oil supply elasticity.
The utility discount rate affects the consumption profiles and the rates of return to
capital accumulation. By moving resources from consumption to investment, a low
discount rate is likely to reduce the global leakage rate whereas a high discount
rate is likely to increase it. The Armington elasticities determine the scope for
trade-induced carbon leakage: the higher the elasticities the more homogenous are
the traded goods and the greater the scope for leakage. The crude-oil supply
elasticity determines the extent to which the international crude oil price falls in
response to the OECD carbon abatement policy. The lower the supply elasticity
the greater the downward pressure on the international crude oil price and the
greater the scope for carbon leakage. Three experiments, one on each of these
parameters, are included in the sensitivity exercise. In each experiment the global
leakage rate is computed for a low value and a high value of the particular
parameter in both capital-flows regimes. For the Armington elasticities the low
values are 4 and 8 and the high values are 16 and 32, for the utility discount rate
the low value is 3% and the high value is 7%, and for the crude-oil supply elasticity
the low value is 0 and the high value is 5. In addition a fourth experiment is
included to test the sensitivity of the result on leakage to the differences in
regional growth rates. In this experiment, we relax our baseline assumption that
both OECD and non-OECD regions grow at the same rate. Since differences in
growth rates may affect leakage rates through altering the regional fossil fuel
consumption profiles, it is possible that our result on the effect of capital restrictions
on leakage may no longer hold. To test this conjecture, we calibrate the
model on a new baseline with 2.5% growth rate in OECD and 3.5% growth rate in
all other regions.

Fig. 4 reports the sensitivity results of the four experiments. Fig. 4a shows the
sensitivity of global leakage to the value of the utility discount rate in the case ofcapital flows Z¥CF. vs. the case of capital controls Z¥CC . We see that, in absolute .


() 135

M.H. Babiker
Energy Economics 23 2001 121
139
terms, the maximum difference between the CF and the CC leakage profiles is only
approximately 0.8% for either value of the discount rate. Hence, our basic result is
invariant to the value of the utility discount rate.

Fig. 4b depicts the CF and the CC global leakage profiles for the low and the
high oil-supply elasticities. As is apparent from the graphs, the maximum discrepancy
between the FC and the CC leakage trajectories is approximately 1.5% for the
low elasticity and approximately 1% for the high elasticity value. In addition, for
both elasticity values, the two leakage profiles appear to converge to the same
trajectory by year 2030. This suggests that the crude-oil supply elasticity has no
leverage on the relationship between carbon leakage and capital flows and therefore
does not alter our basic result.

Fig. 4c reports the sensitivity story for the Armington elasticities. The maximum
discrepancy between the CF and the CC leakage profiles is approximately 1
percentage point for the high elasticity values, and approximately 0.5 percentage
points for the low elasticity ones. Furthermore, for both elasticity values the global
leakage rates converge to the same trajectory by year 2035. Hence, we may
conclude that the relationship between leakage and capital flows is largely invariant
to the degree of homogeneity among the traded goods.

Finally, Fig. 4d shows, for the CF and the CC regimes, the global leakage profiles
under the baseline assumption that non-OECD regions grow at a higher rate than
OECD. Not surprisingly, Fig. 4d indicates that the magnitude of the maximum
divergence between the CF and the CC leakage profiles is only approximately 1%.

Hence the result that carbon leakage is largely unaffected by the presence of
restrictions on capital flows seems to be quite robust.

3.4. Some intuition
At first this result may appear to be counter intuitive. The general intuition
would have it that carbon restrictions increase the costs of production and lower
the rate of return to capital accumulation in OECD. This would cause capital to
outflow into non-OECD regions where investment rates would boom and as a
result fossil fuel consumption profiles and hence leakage rates would also boom.
This intuition would mean that constraints in the international capital markets
would stop these investment flows and cause leakage rates to fall drastically from
their otherwise unconstrained levels.

This reasoning misses at least three important points. First, the source of carbon
leakage is not capital flows but mainly the movement of regional terms of trade
that follows the OECD abatement action. The presence of restrictions on capital
flows does not reduce this initial effect and therefore does not affect the competitive
incentives that lead to the carbon leakage in the first place. Second, in a world
with forward-looking optimizing agents, domestic regional savings will adjust and
undo any capital flow constraints without much effect on consumption profiles. An
increase in investment rates funded by capital inflows merely imply a present
liability of high saving rates to maintain these high investment rates, as well as a
future liability of repaying the debt plus the accrued interest. Therefore, these


136 ()

M.H. Babiker
Energy Economics 23 2001 121
139
Z¥.

Fig. 4. Global leakage and capital flows: sensitivity to model parameters: a sensitivity to the discount

Z¥.Z¥. Z¥.

rate no backstop ; and b sensitivity to oil supply elasticity no backstop .


() 137

M.H. Babiker
Energy Economics 23 2001 121
139
Fig. 4.


138 ()

M.H. Babiker
Energy Economics 23 2001 121
139
capital inflows do not represent any direct real increase in national purchasing
power and accordingly the regional consumption profiles and hence the regional
emissions profiles should not be much affected by their presence. Third, to the
extent that the non-OECD consumption profiles were actually constrained by the
presence of restrictions on international capital mobility, it is the non-fossil rather
than the fossil fuel profiles that would mostly likely to bear the effect. This is
simply because the presence of carbon constraint in OECD lowers the relative
prices of fossil fuels in non-OECD. In effect, these arguments suggest that leakage
is a self-financed process and does not really require any capital flows.

The irrelevance of capital flows to the issue of climate-change policies was alsofound by Manne and Stephan Z¥1997. in a somewhat different context. They have
looked at the determination of the optimal greenhouse gas abatement in a
North
South Integrated Assessment model. With a ÔdescriptiveÕ view of the world,
they have shown an invariant result that the Pareto-optimal amount of greenhouse
abatement is virtually independent of the restrictions on capital flows. Though
their result mainly relates to the optimal spilt of investment between physical and
environmental capital rather than to the consumption-saving decision, it has the
same implication to the current climate-change integrated-assessment models as
ours to the current climate-change impact-assessment models. Both results imply
that the presence of restrictions on capital flows does not significantly alter the
results obtained from these models.

4. Conclusion
The current climate-change-policies regime obliges the OECD countries to
initiate the international effort of abating the anthropogenic greenhouse gas
emissions. If such an effort were to be taken, many observers have argued that the
competitive effects could vitiate any reduction of emissions to be achieved through
the carbon leakage. Nevertheless, almost all of the models that are used to study
these competitive impacts have adopted an empirically inconsistent view of the
world that international capital markets are perfectly functioning. To the extent
that the presence of restrictions on international capital mobility affects the
regional investment decisions, it might be expected that the results from these
models would be significantly altered. This paper has suggested that the competitive
effects of such an abatement action are not really contingent on the international
mobility of capital. With a regionally disaggregated dynamic model of the world
economy, we have shown a robust result that carbon leakage is largely unaffected
by the extent of capital flows.

Acknowledgements

I thank Thomas Rutherford for comments and helpful discussions.


() 139

M.H. Babiker
Energy Economics 23 2001 121
139
References

Alm, J., Buckley, R., 1997. Are government revenues from financial repression worth the costs? PublicFinance Rev. Z¥forthcoming . .

Babiker, M., Rutherford, T., 1998. A unilateral OECD climate-change-policy: carbon leakage, competitiveness,
and the optimal border adjustments. Working Paper, University of Colorado, Boulder.

Feldstein, M., 1994. Tax Policy and International Capital Flows. Wetwirtschaftliches Arch. 130 Z¥.4,
675
697.

Feldstein, M., Horioka, C., 1980. Domestic saving and international capital flows. Econ. J. 90, 314
329.

Gaskins, D., Weyant, J., 1993. Tentative conclusions from energy modeling forum study number 12 oncontrolling greenhouse gas emissions. In: Kaya, Y., Nakicenovic, N., Nordhaus, W., Toth, F. Z¥Eds. , .
Costs, Impacts, and Benefits of CO 2 Mitigation. IIASA CP-93-2, Austria, pp. 235
246.

Giovannini, A., de Melo, M., 1993. Government revenue from financial repression. Am. Econ. Rev. 83

Z¥.

4 , 953
963.
Gordon, R., Bovenberg, A.L., 1996. Why capital is so immobile internationally? Possible explanations
and implications for capital income taxation. Am. Econ. Rev. 86 Z¥.5 , 1057
1075.
Manne, A.S., Stephan, G., 1997. Climate-change policies and international rate-of-return differentials.
Working Paper, Stanford University.
OECD, 1992. The costs of reducing CO 2 emissions: evidence from GREEN. Economics Department
Working Paper, No. 115, Paris.
Rutherford, T., 1995. Extensions of GAMS for Complementarity Problems Arising in Applied
Economics. J. Econ. Dynamics Control 19 Z¥.8 , 1299
1324.
Rutherford, T., 1997. Applied general equilibrium modeling with MPSGE as a GAMS subsystem: anoverview of modeling framework and syntax. Comput. Econ. Z¥forthcoming . .

Rutherford, T., Babiker, M., 1997. Input
output and general equilibrium estimates of embodied carbon:
a dataset and static framework for assessment. Working Paper 97-2, University of Colorado,
Boulder.

Z¥.

Summers, L., 1988. Tax policy and international competitiveness. In: Frenkel, J. Ed. , International
Aspects of Fiscal Policies. Chicago, pp. 1162
1170.