www.ijesci.

org InternationalJournalofEnergyScience(IJES)Volume3Issue3,June2013

MotivationofClimateChangeInvestment
andRiskExposure
NewPerspectivefromGameModel
CanWang*1,ZhugangJin2,WenjiaCai3
MinistryofEducationKeyLaboratoryforEarthSystemModeling,andCenterforEarthSystemScience,Tsinghua
University;StateKeyJointLaboratoryofEnvironmentSimulationandPollutionControl(SKLESPC),andSchoolof
Environment,TsinghuaUniversity
Room1004,SinoItalyEnvironmentandEnergyBuilding,TsinghuaUniversity,Beijing,China,100084
*1

canwang@tsinghua.edu.cn(CorrespondingAuthor);2xshs20071727@126.com;3wcai@mail.tsinghua.edu.cn

Abstract
The common and individual uncertainties are the main
factors influencing the motivation of climate change
investmentforeachcountry.Bydistinguishingandinvolving
the three effects of climate change investment on utility,
strategicgamemodelwithrationalityhypothesisisproposed
and simulated to evaluate the relationship between the
investment motivation and individual risk exposure due to
individualuncertainties.Thepositivecorrelationbetweenthe
risk exposure and investment motivations is confirmed. The
negative influences of risk exposure on utilities of game
equilibriumswhateverthedynamicstrategiesarealsotested.
Upgradeofthestructureofeconomyandenhancementofthe
antifluctuation capacity for the economy shock of global
warmingdeservetobethefocusandpriority.
Keywords
Investment;RiskExposure;ClimateChange;GameModel

Introduction
Over the past several decades, global warming and
climate negotiation have become more and more
severe in spite of the progress on the scientific and
sociological researches. Theres still a huge gap
between the attitudes towards the responsibility and
schedule of implementation. Apart from political and
economic concerns, the uncertainties of global
warming together with its impact are the major
obstacles on the way to reaching a widely acceptable
andbindingconsensusforallcountries.EUinsiststhat
actions should be taken immediately due to
irreversible risk of temperature increment on human
beings living environment, while the United States
believesthatitseconomicalandfeasibletomakeuseof
future technology progress to cope with the issue.
Different perspectives on the reaction to uncertainties
are the essential factors destabilizing the direction of
climatenegotiation.

156

Generally,theuncertaintiesasmentionedabovecanbe
classified into two categories: common uncertainties
and individual uncertainties. The global greenhouse
gas (GHG) concentration and the temperature
increment together with its trajectory under BAU
(businessasusual)scenariointhefuturebelongtothe
former type as its confronted in each country fairly.
Although the temperature increment due to the GHG
concentration is triggered by the combined effects of
individualmitigationfundamentally,theexternalityof
GHG results in the lack of motivation to conduct
carbonabatementforeachcountryalone.Chanderand
Tulkens (2006), Wood (2011) and other researchers
(Desombre, 2004; Hovi, Sprinz and Bang, 2012; Kroll
andShogren,2008)havealreadyrevealedthefailureof
cooperation from perspective of theoretical game
theory.
The other type of uncertainty is based on individual
propertiesincludinginitial GDP growth, the tolerance
oftemperatureincrementonindividualGDPetal.The
diversities of industry structures and choices of
competitive industries may affect the stabilities and
risk resistance capacities for each country under
different temperature increment scenarios. Compared
with developed countries, the developing countries,
especially the African and Southeast Asian nations
maysuffermoreseverlosswithhigheruncertaintyfor
their inadequate precaution and high dependence on
thetenderagricultureindustry(Hope,2006;Maddison,
2003; Nordhaus and Boyer, 2000; Tol, 2002). Such risk
exposures connected with individual properties may
attract more attention for their own interests and
trigger their subjective initiatives. Then the way how
the risk exposure will influence the negotiation
strategies is introduced. Such relationship is
investigated in this article with theoretical model and

InternationalJournalofEnergyScience(IJES)Volume3Issue3,June2013

simulations.Basedonrationalityhypothesis,thethree
effects of climate change investment on utility are
distinguished and involved in our dynamic game
model.Simulationtogetherwithcomparisoniscarried
out. More information about the model is introduced
in Section 2. Simulations together with evaluation are
presented in Section 3, followed by discussions and
conclusionsinSection4.
Theoretical Model

We posit a constant relative risk aversion (CRRA)

socialutilityfunctionforeachrationalcountry:

Utilityi e t U i ,t dt e t

i and

withthefollowingPDF(probabilitydensityfunction):
fT ( x; rT , T ,T )

T r

(rT )

( x T )rT 1 e T ( xT ) , x T .

(4)

For the individual uncertainty, i.e , the marginal

effectoftemperatureincrementonGDPgrowthcanbe
calibrated based on the individual exponential loss
function information obtained from IAMs researches
as there is a linear relationship between and

theoretically1 . Given the trajectory of temperature

UtilityandUncertainties

where Utilityi, Ui,t,

www.ijesci.org

Ci1,t i
1 i

dt.

incrementTtproposedbyWeitzman(2009)andapplied
by Pindyck(2012), Figure 1 shows the change of loss
function value due to the temperature increment at
horizonyear,parameter andtimet.

Loss(T ) e (T ) . (5)
2

(1)

refer to the net present

utility, utility at time t, index of risk aversion and

discounting rate for country i respectively. Ci,t means
the disposable consumption at time t which can be
connectedwithGDPgrowthrategt:

1
Tt 2TH [1 ( )t / H ].
2
(6)

1
0.95

Ct exp ( g s ds ). (2)

0.9

Loss

0.85

In the absence of global warming, real GDP and

consumption can keep a constant growth g0, but the
temperatureincrementwouldhaveasimplifiedlinear
effect on real GDP growth as Dell, Jones and
Olken(2009)estimated:

0.75
0.7

temperature increment and the marginal effect

parameter belong to the two categories of
uncertainties respectively. Although lots of researches
have calibrated these parameters, the results are not
alwaysconsistentorpreciseduetovariousrestrictions
ofeachmodel.Oneofthewidelyacceptedsolutionsis
to involve probability theory and to estimate the
expectedutility.
For the common uncertainty, i.e. Tt, the temperature
increment at horizon year (H=100) has been fitted by
Pindyck(2012), Newbold and Daigneault(2009) with
assumptions that the mean, 5% and 1% points
temperature
increments
are
3,
7and
10respectively. Threeparameter (i.e. rT , T , T )
Gamma function fits the uncertainty distribution well

Beta=0.0169

0.65
0

100

gt g 0 Tt . (3)
Themarginaleffectoftemperatureincrementattimet
(Tt) on GDP growth is . As just discussed, the

Beta=0.0082
0.8

TH

50

4
6

FIG.1THERELATIONSHIPSOFLOSS(TT),T,THAND

EffectsofClimateChangeInvestmentand
Motivations
The three effects of climate change investment on
utility are distinguished: the positive, negative and
spillover effects. If invi represents the percentage of
climate change investment to GDP for country i, then
theamountofdisposableconsumptionwillbereduced
whichisdefinedasthenegativeeffectofinvionutility.
On the other hand, climate change investment will
boost the original GDP growth with marginal
parameter as its implemented as extra investment.

WithEq.(1)to(6),thetheoreticalrelationbetween

and canbesolvedout:
TH

1
]
H [2
ln(0.5)

157

www.ijesci.org InternationalJournalofEnergyScience(IJES)Volume3Issue3,June2013

Therefore the modified GDP growth and disposable

consumption due to the positive and negative effects
areadjustedas:

g gi ,0 (1 i invi ) iTt .
*
i ,t

(7)
t

Ci*,t Ci ,t (1 invi ) (1 invi ) exp( gi*, s ds).

0

(8)

Ontheotherhand,theexternalityofglobalwarming
determines the spillover effect of climate change
investment. Supposing that the relationship between
theupperboundaryoftemperatureincrementTUpand
globalclimatechangeinvestmentcanbequantifiedas
function , then each countrys investment decision
inviwillpartiallyinfluencetheTUpwithitsGDPrelated
weighting factor w indirectly. Compared to the
temperature increment uncertainty under no
investment scenario, TUp should promote each
countryspresentnetutilityastheoriginaldistribution
ofcommonuncertaintywillbeupdatedwithBayesian
theorem. In other words, the original PDF for the
temperature increment in the future fT ( x; r , , ) is

fT ( x; r , , ) asEq.(10)shows:

Bayesianupdatedto

TUp = (invworld )= ( wi invi ). (9)

i

f ( x; r , , )
fT ( x; r , , ) TUp T

, x TUP . (10)

fT ( x; r , , )dx

Utilityi = i (invi , f i ; inv j )=

0 i T

(Ci*,t )1i
1 i

fT f i dTd dt.

(11)

If the three effects of climate change investment are

combined, every rational negotiator has incentives to
choose the optimal investment to maximize its own
utility. Apart from common uncertainties, individual
uncertainties will also affect their motivations. By
means of integration of the Eq. (1) to (10), the present
net utility of country i can be further expressed
as i (invi , f ; inv j ) . Given the other countries
i

investment
uncertainty

information

Furthermore, the noncooperative game equilibriums

can also be worked out by involving the dynamic
decision steps for each country under complete
information hypothesis. The influence of individual
uncertainty

f i on the game equilibriums can be

examined as well. With the backward induction

method (Aumann, 1995), the theoretical game
equilibriumsofdynamicgamemodelfortwocountries
involve three scenarios as table 1 shows.
First, second and simul mean the three strategies,
and OptInvi

first / second / simul

, Utilityi

first / second / simul

represent

climate change investment and utility of equilibriums

respectively.
Simulation and Comparisons
For the individual uncertainty, three groups with
differenthypothesesaresimulatedandcompared.The
probability density function

f i is fitted by taking the

mean loss for TH=2.5 to be 3% GDP, and the 17%

point and 83% point to be 1% and 4.8% of GDP
respectively for Group 1. The risk exposures are
adjusted by means of increasing and decreasing the
variance with the same expectation for Group 2 and
Group3respectively.2 Figure2showsthedistribution
ofparameter .Generally,theriskexposuremeetsthe
order:Group2>Group1>Group3.

TUp

the relationship of climate change investment

motivationandriskexposurethencanbeevaluated.

invj

and

individual

f i , the optimal amount of climate change

In comparison with the optimal reactions of three

groupsasFigure3shows3,theoptimalreactiontohigh
riskexposurecountrytendstobemorethanthatwith
lowriskexposure.Wheninvj=3%,theoptimalreaction
of Group 1 is 1.248% GDP, higher than 1.246% of
Group3andlowerthan1.251%ofGroup2respectively.
The simulation results are rational and in accordance
with the reality because the optimal reaction refers to
anoptimalbalanceofthethreeeffectsofclimate

ForthreeparameterGammadistribution,themean,
varianceandskewness
are E ( x ) r / , Var ( x ) r / 2 and

Skew( x) 2r / 3 respectively.

investment for county i should meet the target of

maximum i .Hence, OptInvi = i (f ; inv j ) isemployed

TheothercommonparametersincludeinitialGDP
growthg0=4.5%,riskaversion 2 ,marginalpromoting

to represent the optimal reaction which is determined

effectofclimatechangeinvestmentonGDPgrowth 1 ,
timepreference 0 andinitialtemperatureuncertainty

by invj and individual uncertainty

f i . By comparing

the gap of OptInvi under different assumption of

158

f i ,

(rT , T ,T ) 3.8, 0.92, 1.13 whichisinaccordancewith

Pindyck(2012).

InternationalJournalofEnergyScience(IJES)Volume3Issue3,June2013

www.ijesci.org

TABLE1THEORETICALEQUILIBRIUMSFORDYNAMICGAMEMODELWITHTWOCOUNTRIES

Scenarios

CountryImovesfirstly

Countryjmovesfirstly

Simultaneously

Countryi

OptInvi first ,Utilityi first

OptInvisecond ,Utilityisecond

OptInvisimul ,Utilityisimul

Countryj

OptInv second
,Utility second
j
j

OptInv first ,Utility first

OptInv simul
,Utility simul
j
j

j
j

Notes:Thistableonlysimulatestwocountryscenariowhichcanbeadjustedtothemulticountrysituation.

TABLE2HYPOTHESESANDCALIBRATIONRESULTSFORINDIVIDUALUNCERTAINTY

Hypothesesof(1Loss(2.5))

Mean

Group1
Group2
Group3
Notes:theparametervaluesfor

Threeparametersfor

17%Point

83%Point

3%
1%
4.8%
3%
0.5%
5.3%
3%
1.5%
4.3%
areinaccordancewiththehypothesesrespectively.

11.141
6.935
20.72

22329
13815
41624

2.85E04
2.87E04
2.85E04

TABLE3NONCOOPERATIVEDYNAMICGAMEMODELEQUILIBRIUMSOFTHREESTRATEGYSCENARIOS

imovesfirstly

OptInvi
OptInvj
Utilityi
Utilityj

Group1
1.342%
0.455%
21.7745
66.4760

jmovesfirstly
Group2
1.343%
0.455%
21.7756
66.4760

G3
1.341%
0.455%
21.7738
66.4760

G1
1.337%
0.560%
21.7745
66.4760

Group 1
Group 2
Group 3

3000

PDF

2500

1500
1000
500

-2

2
4
Parameter r

10
-4

x 10

FIG.2PROBABILITYDENSITYDISTRIBUTIONOFPARAMETER
FORGROUP1TO3

0.0127

Group 1
Group 2
Group 3

0.0126

G1
1.341%
0.455%
21.7745
66.4760

G2
1.339%
0.454%
21.7756
66.4760

G3
1.337%
0.454%
21.7738
66.4760

Inaddition,dynamicgameequilibriumsinTable3also
show the differences due to the risk exposures 4 .
Although the equilibriums of investment are not
consistentineachdynamicstrategyscenario,theutility
differences are significant. Whatever the dynamic
strategy scenario is, the utility of country i meets the
order constantly: Group 3>Group 1> Group2. The
simulation results are also inspiring. It indicates that
the risk exposure itself has much more determining
powerthanthegamestrategiesundernoncooperative
scenarios in influencing the utilities of each country,
which means that we should give the priority to the
enhancement of the antifluctuation capacity for the
economyshockofglobalwarming.

2000

0
-4

G3
1.334%
0.560%
21.7738
66.4760

more attractive and significant, which will trigger the

higher investment of climate change. In other words,
high risk exposure will promote the motivation of
climatechangeinvestment.

4000
3500

Simultaneously
G2
1.336%
0.560%
21.7756
66.4760

0.0126

Discussions and Conclusion

OptInvi

0.0125

The effects of risk exposures on optimal reactions

together with game equilibriums are evaluated. By
distinguishing the three effects of climate change
investmentonutility,themotivationofclimatechange
investmentareinvestigatedwithrationalityhypothesis.
Simulationsresultsrevealthatthehigherriskexposure

0.0125

0.0124

0.0124

0.0123
0.026

FIG.3

0.027

0.028

0.029

0.03
invj

0.031

0.032

0.033

OptInvi =i (f i ; inv j ) FORGROUP1TO3

change investment. With higher risk exposure, the

room of improvement in spillover effect would be

Theothercountry/sectorjmeetthefollowinginitial
hypotheseswhichisbasedonreality:
g0,j=1.5%, 1 , 0 and

( r , , ) 103.8, 411600, 1.745e 4 .

159

www.ijesci.org InternationalJournalofEnergyScience(IJES)Volume3Issue3,June2013

is,themoreinvestmentonenationisincentivetomake,
whichmeansthatthefragilityofeconomymayhavea
significanteffectonthemotivations.Forcountrieswith
highdependence on agriculture and other industries
susceptible to global warming, its rational and
motivated to invest more. Therefore, its advisable for
developingcountries,especiallyAfricanandSoutheast
Asiannationstoupgradetheindustrystructureandto
takeprecautionstocopewiththefluctuationofclimate
change, which will not only be conductive to the
stability of future utility, but also enhance their
bargainingpowerinclimatenegotiation.Moreover,the
results of game equilibriums under different dynamic
strategies also suggest the benefits of reduction of the
risk exposure. Countries with higher risk exposure
mayhavetofacemoreseveredecreaseofutility.Such
decrement is significant whatever the adopted
dynamic strategies are, leading to the cognition that
more attention should be paid to the improvement of
economy structure and antifluctuation capacity
comparedtothenegotiationstrategiesthemselves.

Hope, C.The marginal impact of CO2 from PAGE2002: An

integrated assessment model incorporating the IPCCs
five reasons for concern.Integrated assessment (2006):
6(1).
Hovi,J.Sprinz,D.F.Bang,G.WhytheUnitedStatesdidnot
become a party to the Kyoto Protocol: German,
Norwegian, and US perspectives.European Journal of
InternationalRelations(2012):129150.
Kroll, Stephan Shogren, Jason F.Domestic politics and
climate change: international public goods in twolevel
games.Cambridge Review of International Affairs
(2008):563583.
Maddison, D. The amenity value of the climate: the
household production function approach.Resource and
EnergyEconomics(2003):155175.
Newbold, Stephen C. Daigneault, Adam. Climate response
uncertaintyandthebenefitsofgreenhousegasemissions
reductions. Environmental and Resource Economics

ACKNOWLEDGMENT

(2009):351377.

This research was funded by the National Natural

ScienceFoundationofChina(No.71273153).

Nordhaus, W. D. Boyer, J. G. Warming the world: the

economics of the greenhouse effect. Cambridge, MA:
MITPress,2000.

REFERENCES

Pindyck, R. S. Uncertain outcomes and climate change

Aumann,R.J.Backwardinductionandcommonknowledge
of rationality. Games and Economic Behavior (1995):
619.
Chander, P. Tulkens, H. Cooperation, Stability and
SelfEnforcement

in

International

Environmental

Agreements:AConceptualDiscussion.CORE,2006.
Dell, MelissaJones, Benjamin F. Olken, Benjamin A.
Temperature

and

income:

reconciling

new

policy. Journal of Environmental Economics and

Management(2012):289303.
Tol,R.Estimatesofthedamagecostsofclimatechange.Part
1: Benchmark estimates. Environmental and Resource
Economics(2002):4773.
Weitzman, M. L. Additive damages, fattailed climate
dynamics, and uncertain discounting. Economics: The

crosssectional and panel estimates. National Bureau of

OpenAccess,

EconomicResearch.2009.

200939.

Desombre, E. R.Global warming: more common than

tragic.EthicsandInternationalAffairs(2004):4146.

160

OpenAssessment

EJournal

(2009):

Wood,P.J.Climatechangeandgametheory.Annalsofthe
NewYorkAcademyofSciences(2011):153170.