Professional Documents
Culture Documents
of AlternativeClimate-Change Strategies
Although the issue of greenhouse warming checked, recent surveys indicate that over the
was first seriously studied a century ago, it next centurythe globally averagedsurfacetem-
has over the last decade emerged as the central peraturewill rise around3?C(degrees Celsius),
internationalenvironmentalquestion. Most na- which would produce climates that are unprec-
tions have adoptedthe FrameworkConvention edented duringthe entire span of human civili-
on Climate Change negotiated at the 1992 Rio zation. While wanning may seem benign, it has
Earth Summit. Under the Convention, nations major and unpredictableimpacts on weather
agreed to take steps to limit carbon dioxide patterns,ocean currents,sea-level rise, riverrun-
(CO2) and other greenhouse gas (GHG) emis- offs, storm and monsoonal tracks, desertifica-
sions before they reach "dangerous" levels. tion, and other geophysical phenomena.Many
Having increased its CO2 emissions at an av- scientistsand ecologists view these changes and
erage growth rate of almost 2 percent annually uncertaintieswith alarm.
for about a century, the United States has com- The other half of the calculus is the cost of
mitted itself to capping its emissions at 1990 slowing climate change. Even the most dra-
levels, and many other high-income countries conian policies (such as a virtual phaseout of
have made similar or even more ambitious fossil fuels) would only slow and not stop cli-
proposals (for a review of commitments, see mate change, and significant steps to slow the
Daniel M. Bodansky [1995] or International rate of increase of climate change would cost
Energy Agency (IEA) [1994]). hundreds of billions of dollars annually using
The climate-changeissue is so controversial today's energy technologies. Given the many
primarilybecause the stakes are so high. If un- economic issues facing humanity, it would re-
quire an unusually dire risk and uncommonly
*
statesmanlike behavior for nations to divert I
Nordhaus: Department of Economics, Yale Univer- or 2 percent of their national incomes today to
sity, 28 Hillhouse Ave., New Haven, CT 06511; Yang:
Center for Energy and Environmental Policy Research,
reduce conjecturalrisks that will not occur un-
MIT, Cambridge,MA 02137. This researchwas supported til well into the next millennium.
by the National Science Foundationand the U.S. Environ- In addition to the grave risks and huge costs,
mental Protection Agency. This research has benefited the issue of greenhouse warming is difficult
from discussions and comments of Richard Eckaus, because the problem is so complex. It involves
William Hogan, Alan Manne, Richard Richels, Herbert
Scarf, and two referees. All views and errors of omission a series of poorly understood systems, includ-
or commission are the sole responsibility of the authors. ing the carbon cycle, climate reactions, geo-
Correspondencecan be directedto W. D. Nordhaus. physical, ecological, and biological impacts of
74.
742 THE AMERICANECONOMICREVIEW SEPTEMBER1996
is equalized across regions. The preference * Market policies. The market approach is
function of each region is a utility function one in which there are no controls on the
which is the sum of discounted utilities of per emissions of greenhouse gases. This has
capita consumption times population, where been the approach followed by virtually all
the pure rate of social time preference (the dis- nations up to now.
count rate on utility) is 3 percent per year in * Cooperative policies. The second approach
each region. The utility function is logarithmic is the ideal one in which global environ-
in per capita consumption. mental concerns are treated cooperatively
The major contributionof the integratedap- through the efficient actions of all nations.
proaches like the RICE model is to integrate In this approach, nations agree to reduce
the climate-related sectors with the economic CO2 emissions in a globally efficient way.
model. This part of the model contains a num- This solution is efficient but requires an un-
ber of geophysical relationships that link to- realistically high degree of cooperation.
gether the different forces affecting climate * Noncooperative policies. In the third ap-
change, generate the greenhouse-gas emis- proach, individual nations undertake poli-
sions, and measure the impacts of climate cies that are in their national self-interests
change. RICE includes region-specific emis- and ignore the spillovers of their actions on
sions equations, a global concentrations equa- other nations. In the noncooperative ap-
tion, a global climate-change equation, and proach, to the extent that nations are small
regional climate-damage relationships. En- and the externality is truly global, efforts to
dogenous emissions are limited to CO2, while reduce CO2 emissions will be much smaller
other greenhouse gases are treated as exoge- than in the global cooperative solution. This
nous. Uncontrolled emissions are a slowly solution is inefficient but realistic.
declining fraction of gross output-a relation-
ship which is consistent with the observed B. Basic Structure
"decarbonization" in most countries over this
century that is also predicted by more detailed We outline here the major features and in-
energy models. CO2 emissions can be con- novations of the RICE model; the equations of
trolled by increasing the prices of factors or the model are contained in Appendix A.
outputs that are CO2 intensive, and we repre- The RICE model divides the global econ-
sent the C02-reduction cost schedule paramet- omy into 10 different regions. The first five are
rically by drawing upon a number of studies 1) the United States, 2) Japan, 3) China, 4)
of the cost of CO2reductions. Climate change the EuropeanUnion, 5) and the former Soviet
is representedby the realized global mean sur- Union (FSU). Each is treated as a single de-
face temperature,which uses relations based cision maker. The last five regions have dif-
on currentclimate models. The economic im- ferent numbers of countries, and each is
pacts of climate change are assumed to be in- treatedas multiple decision makers. These five
creasing along with the realized temperature regions are 6) India, 7) Brazil and Indonesia,
increase. The impacts of climate change are 8) 11 large countries, 9) 38 medium-sized
estimated from a number of different studies, countries, and 10) 137 small countries. (Basic
but it must be recognized that this is the most data on the major regions are contained in Ap-
uncertainpart of the model. pendix B.) To reduce the severe computational
The major economic choices faced by complexity of the solution, we sometimes ag-
nations (or the concert of nations) in this ap- gregate regions 6 through 10 into one region
proach are (a) to consume goods and services, as the "rest of the world" or "ROW."
(b) to invest in productive capital, and (c) to The goal in creating the different regions is
slow climate change through reducing CO2 to structurethe problem so that the noncoop-
emissions. The new element introduced in the erative equilibriumis equivalent to the full but
RICE model and not present in other models enormous game with about 200 countries. This
of global warming is the possibility of differ- is done by allocating the smaller countries to
ent strategies undertakenby nations. We dis- groups so that within each group the national
tinguish three distinct approaches: benefits from slowing climate change are
744 THE AMERICANECONOMICREVIEW SEPTEMBER1996
roughly equal. We then mimic the free-riding TABLE 1-FUTURE LEVELS OF INCOMES,
temptationsof global public goods by dividing DIFFERENT REGIONS
emissions. Most studies are based on the The climate-change policies are character-
United States and Europe, and estimates for ized by "control rates" and "carbon taxes."
other regions have low levels of reliability. We Control rates are simply the percentage reduc-
have parametrizedthe cost function using the tions in CO2 emissions relative to a baseline
functional form from earlier studies but have or uncontrolled path. Carbon taxes represent
estimated the intercepts of the cost functions the marginal cost of reducing CO2 emissions.
on the basis of the international comparisons A carbon tax would equal the price of a
undertakenby the OECD and by Energy Mod- carbon-emissions permit if there were tradable
eling Forum 13.5 permits, and the prices of such permits in dif-
Estimates of the economic impacts or dam- ferent countries would obviously be equalized
ages from climate change are sparse at this (at market exchange rates) if permits were
stage. There are numerous studies of the esti- freely tradable.In the market solution, carbon
mated impact of climate change on the mar- taxes are zero. In the cooperative solution,
keted sectors for the United States, but few emissions are curtailed in a cost-effective
reliable studies for the nonmarket sectors or manner. The model does not deal explicitly
for developing countries. To estimate the im- with mechanisms by which winners might
pacts in different regions, we assume that the compensate losers, although we discuss some
damage function from climate change is iden- of the issues below.
tical for each industry across different regions,
and that the cost functions have the same pa- C. Algorithm to Calculate
rameters as those estimated for the United General Equilibrium
States. Impacts in different regions are cal-
culated by taking the estimated shares of dif- The RICE model presents a radically differ-
ferent sectors (agriculture, coastal activities, ent philosophy for estimating strategies to
and so on) in national output and then ag- cope with global warming from global-
gregating those up to obtain overall national optimization models used in many integrated
estimates. (This approach is described in assessments. The baseline calculation is cali-
Nordhaus [1994].) The results in the aggre- brated to a market equilibrium of the world
gate do not differ markedly from the other economy with all the differences in popula-
major estimates (see particularly Samuel tions, technologies, and incomes-the world
Fankhauser [ 1993] and the survey of experts is taken as it is for the purpose of the baseline
in Nordhaus [1994]), but it must be empha- calibration. We then calculate different strat-
sized that the distribution of climate impacts egies for global warming conditional on the
across countries is at this stage highly con- existing distributionof capital, labor, and tech-
jectural. Table 2 shows the major inputs as- nology. The strategies include doing nothing
sumptions for the different regions. (the market solution), finding an efficient
solution given the existing distribution of in-
come (the cooperative solution), and finding
the solution in which nations select policies to
' See Andrew Dean and Peter Hoeller (1992) and maximize national preferences alone (the non-
Darius W. Gaskins and John P. Weyant (1993). The func- cooperative or nationalistic equilibrium). This
tional form of the mitigation-cost function in the DICE public-choice approachis in sharp contrast to
model was estimated from studies of the cost of CO2 re- many of the debates on climate change today;
duction in nine families of models primarilybased on the in these, the distributionalissues of who shall
United States and takes the form Ci(t) = b, Mj(t)b2Yi(t),
where i is region i, Ci(t) = the cost of reducing CO2emis-
pay to slow climate change rise to the top of
sions, bj,j and b2 are parameters, pi(t) is the emission- the agenda.
control rate or fractional reduction in emissions from the We now describe the algorithm for finding
marketpath, Yi(t) is region i's gross regional product, and the cooperative solution in the RICE model.
t is the time period. The RICE model assumes that the The technique we employ originates with
exponents (b2)are the same across countries and calibrates
the intercepts (b,j) to estimates of the cost functions from
T. Negishi (1960), was discussed briefly in
the different countries or regional models mentioned Nordhaus (1990) in the context of global
above. warming, and has been used in similar models
746 THE AMERICANECONOMICREVIEW SEPTEMBER1996
by Manne and Thomas Rutherford (see, par- optimizing a global social welfare function of
ticularly, 1994). The theoretical basis for the the fonn:
algorithmis a theorem of Negishi which relies
N
on the second theorem of welfare economics.
Negishi suggested and proved that under cer- (1) W= E U'[c'(lI), cl(2),
i= I
tain conditions a competitive equilibrium can
be found by maximizing a social welfare func- c(t), ..,c'(T)]
tion of N agents in which the welfare weight
of each of the agents is adjusted to satisfy the where W is the value of the global social wel-
agent's budget constraint. We will call this fare function and 4. are the welfare or Negishi
equilibrium the Negishi solution. weights for country i, i = 1, ... , N. The U' are
What are the appropriatewelfare weights? the preference functions for the different coun-
In our calibration, we adopt the realistic ap- tries, and the c'(t) are the consumption bun-
proach by taking the welfare weights that re- dles of the countries.
flect the actual economic outcome across The relevant excess demand is found in the
regions. We do this not as a brief for the ex- intertemporalbudget constraintof each region.
isting international distribution of resources To find the competitive equilibrium, we add a
and income but because it is the startingpoint constraintto the problem that requireseach re-
for analyzing potential improvements in eco- gion to satisfy its intertemporal budget con-
nomic welfare that would arise from policies straint, which is represented by terminal net
that are imposed on the actual world economy. foreign assets, NFA'(T), T being the last
Hence, the weights are ones such that the ex- period:
cess demands in all markets are zero at the
given welfare weights and prices.6 More pre- (2) NFA'(T) = O, i = 1, ..., N.
cisely, the algorithmicprocedureis the follow-
ing. We first solve the RICE model by Next, define qi'(T) as the dual variable of
NFA'(T), which in economic terms is the
marginal utility of consumption or income in
the last period. Given condition (2), ,r'(T) is
6 A brief but illuminating discussion of the Negishi ap- a function of the welfare weights and we can
proach is in contained in Andreu Mas-Colell et al. (1995 write these functions as i'(T) = G'(4.', 42,
pp. 630-31). ...,ON), i = 1, ... , N. Without condition (2),
VOL 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGESTRATEGIES 747
an arbitrary set of social welfare weights virtually no difference for the results of the
would generate a set of nonzero NFA (T), analysis below.
which implies that at least one region does not Given these constraints on international
live within its budget. However, when the dual capital flows, our algorithm will not produce
variables are equalized across all countries, the the necessary complete price equalization for
welfare-weighted marginal utilities of income carbon-tradingpermits, which are assumed to
are equal and the intertemporal budget con- be fully tradable and reach price. To ensure
straints is therefore satisfied. price equalization for carbon-emission rights,
Hence, the algorithmworks by searchingfor we adjust the Negishi weights across regions
the welfare weights, as a function of the dual for every period. We call this new algorithm
variables i"(T), so that the marginal utilities the time-dependentNegishi solution. It differs
of consumption are equalized: from the pure Negishi solution because it in-
corporates the constraints on capital flows so
(3) qi(T) = G(O', (p2,)",2 ')N) = # that the regional budget constraintsare binding
for every period. As a result, carbon-emissions
for all i = 1,..., N. permits have equal prices in all regions in each
time period (at marketexchange rates). Under
Combining ( 1), (2), and (3), we know that this revised algorithm, we seek the time-
each of the country budget constraints is sat- dependent Negishi weights, 4i(t). To find
isfied and that no region can gain from a these, we first solve the model with an arbi-
change in the resulting allocation. Hence, by trary set of welfare weights while continuing
the Negishi theorem, we know that this op- to impose (2). Following the Negishi theorem,
timized outcome using the welfare weights we then reset the welfare weights for all coun-
generated in (3) represents a competitive tries and time periods according to the follow-
equi-librium consistent with the initial en- ing formula:
dowments, technology, and preferences. The
equilibrium thus found is the "pure Negishi 1
solution."
Unsatisfactory aspects of the solution led (4) n1(t)=
to the following refinements of the pure
Negishi solution. The major problem with the
pure Negishi solution was that it generatedex-
tremely large capital flows among regions This equation sets the welfare weights equal
(this is a common feature in intertemporally to the inverse of the marginal utilities of con-
optimized models).' Because these are un- sumption. The search algorithm based on (4)
realistic, we took one further step which was very quickly converges to a solution that sat-
to impose certain flow and stock constraints isfies (2) and (3). We have conducted a
on debt and currentaccounts to ensure that net number of experiments and have found no in-
foreign investment does not exceed certain dication of multiple equilibria.
limits. These limitations limited the export- What is the underlying economic rationale
GDP ratio to 1, limited the ratio of net foreign for this algorithm? The solution represents a
assets to output to 0.1, and limited the current competitive equilibrium under the assumption
account deficit to GDP ratio to 0.1 (see Ap- that the preferences or technological con-
pendix A for details). These constraints were straintslimit the internationalflows of capital.
based on observed limitations, but they made For example, there may be strong home-
country preferences in portfolios because of
limitations of the marketabilityof human cap-
ital. The limitation of this approach is that
to the extent that the constraints on capital
7 The difficulty raised by unrealistically high capital
flows is not related to the use of the Negishi technique;
flows have nonmarket-clearingelements due
the same issue would occur if fixed-point methods were to rationing, the excess demands will not be
employed. zero and we may depart from the market
748 THE AMERICANECONOMICREVIEW SEPTEMBER1996
equilibrium (which would in any case be dif- regions by optimizing for each region holding
ficult to compute). the control rates and resulting emissions, con-
Once we have obtained a competitive equi- centrations, and impacts in other regions from
librium, we then perturb various elements, the previous iteration fixed. We continue to
such as the climate parameters or the cost cycle through this sequence until the set of
functions, and resolve the maximization in control rates are unchanged given the set of
( 1). We do this holding the welfare weights noncooperative strategies of other countries,
constant across runs. This resolves the index which is then the Nash equilibrium. The out-
number problem of changing prices by calcu- come matches well the theoretical predictions
lating the welfare changes at the market wel- and is in our simulations invariant to initial
fare weights.8 conditions, which suggests that the Nash equi-
librium is unique.
D. Finding the Noncooperative Equilibrium How reasonable is this solution concept?
While the pure Nash equilibrium is a sensi-
The algorithm just described provides the ble assumption for small countries like Chad,
solutions for both the market and the cooper- whose global warming policies will hardly
ative equilibrium.A different approachis nec- make the front pages, it may lack realism for
essary to find the noncooperative equilibrium. large or influential countries. Large coun-
The noncooperative or nationalistic equilib- tries like China and influential countries like
rium exists as the equilibrium of the strategies the United States would probably want to
of the different countries. We hence need an take into account the effect of their policies
assumption about strategies and a method of on other countries' policies. The ambivalent
finding the equilibrium. policy on global warming by the United
As for strategies, we assume that each na- States over the last decade has undoubtedly
tion determines its policies by maximizing its strengthened the hand of those in other coun-
domestic intertemporalutility function assum- tries who want to do little. An alternative
ing that other nations' strategies are unaffected approach would be for countries to posit
by its policies. The noncooperative strategies conjectural variations or reactions of other
are hence dynamic, full-information, Nash countries to their policies. For example, the
strategies, and we are seeking the Nash equi- United States might assume that Japan or Eu-
librium. Technically, our solution is a Nash rope would be a follower in terms of carbon-
equilibrium in a finite game with perfect in- tax policies or tradable emissions policies.
formation, and it is therefore time consistent. Another possibility would be to model coa-
Such games have pure strategy Nash equilibria litions of different countries. We have not
which can be calculated throughbackwardin- explored these alternative solution strategies
duction, which is essentially what our algo- in the present paper. Once we admit nonzero
rithm does (for a discussion, see Mas-Colell conjectural variations, we are in a deep
et al. [1995 Chapter9]). thicket and the possibilities become unlim-
More precisely, we assume that each nation ited. Future research will examine the pos-
sets its own control rate over time [u' = sibility of coalitions of countries.
IA } as given. Beginning with an initial set of Using the algorithmsjust described, we will
control rates, we iterate through the different analyze the three different strategies as de-
scribed in Table 3: market, cooperative, and
noncooperative. In addition, for reference we
sometimes compare the results of the RICE
8The RICE model runs on the GAMS software (see
Anthony Brooke et al., 1988). The full model including model to those of its parent, the DICE model,
searching for welfare weights takes approximately6 hours which is essentially a one-region efficient or
on a 486-66 processor. cooperative solution.
VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGESTRATEGIES 749
TABLE 3-ALTERNATIVE SOLUTION CONCEPTS FOR THE regional outputs are shown in Figure 1; these
RICEMODEL indicate that the projected relative sizes of
the Chinese and ROW economies grow
1. Market RICE: This strategy assumes that there sharply over the next century. The output
is no correction for the climate-change
externality and that there is therefore no growth in the RICE model is significantly
abatementof CO2 emissions. larger than that in many projections prepared
2. Cooperative RICE: In this strategy, countries by international study groups, most of which
undertakepolicies that reduce greenhouse-gas envision a stability of current relative in-
emissions efficiently. The reduction of CO2 come differentials rather than the projected
emissions is efficient across countries and partial convergence in the RICE model. Note
across time. as well that we use market exchange rates
3. Noncooperative RICE: This strategic concept because we will want to find the equilibrium
assumes that each country sets its CO2 in which the prices of internationally-traded
emissions controls to maximize its own carbon-emissions permits are equalized.
economic welfare assuming that other
countries' control strategies are invariantto a Emissions are also considerably higher in
country's policies. RICE than in the many other projections. For
example, CO2 emissions in the RICE model
reach 38 billion tons of carbon by the year
2100 in the market or uncontrolled run. This
II. Results compares with an estimated 21 billion tons in
the DICE model and a range of 5 to 35 billion
We now reportthe results of the policies and tons in the IPCC projections (see T. M. L.
strategies described above. As in all modeling Wigley [1994] for a description). CO2 emis-
efforts of this kind, they should be interpreted sions grow substantially faster in the RICE
with caution as this study is the first empirical model partially because of the projected rapid
application of noncooperative game theory to growth in output and partially because of the
global environmental policy. On the other rising output share of regions with high
hand, the major results concerning the level of emission-output ratios.
stringency of climate-change policies have Figure 2 shows the resulting CO2emissions
been relatively stable over a wide variety of under the different solution concepts and also
models and alternative specifications of the compares estimates from this study with the
RICE model, so we have considerable con- earlier DICE model. Model estimates (not
fidence in these estimates (conditional, of shown) indicate that the share of CO2 emis-
course, on the assumptions underlyingthe ma- sions will rise sharply in China, region S1
jor components, such as those concerning the (India), region S3 (middle-sized developing
long-run growth projections, the costs and countrieslike Thailand), and region S4 (smalier
damages, and the discount rate). developing countries). These four regions ac-
counted for about one third of CO2 emissions
A. Output.Emissions, and Climate Change in 1990 but are projected in the marketruns of
the RICE model to comprise three quartersof
The projections for the major economic emissions by 2100.
and environmental variables are shown as CO2 concentrationsare shown in Figure 3.
Figures 1 through 4. One important outcome Given the higher emissions rates in the RICE
of this study is that the RICE model has sub- model, its concentrationsrise more rapidlythan
stantially higher projected world output and in the DICE model. It is useful to examine the
emissions by the end of the next century than date of doubling of CO2concentrationsrelative
do many other integrated assessments, such
as the earlier DICE model.9 Projections for
emmental Panel on Climate Change (IPCC) (1990), the re-
sults in Nordhaus(1994), and preliminaryresultsof the sur-
9This statementis based on a comparisonof the results vey of models by the Energy Modeling Forum 14 directed
in the RICE model with the projections of the Intergov- by John Weyant,EnergyModeling Forum(EMF) (1995).
750 THE AMERICANECONOMICREVIEW SEPTEMBER1996
100.0 -
Rest of World
P.-O ~~~~~~~~~~~~~Europe
0
O 10.0 - X - ------------
1.0
? 1 990 2010 2030 2050 20&0 2090
FIGURE 1. REGIONAL OUTPUTS: COOPERATIVE SCENARIO
40
Market
solution
35 --
Noncooperative
equilibrium
d 25 -
-- - - - - - - - - - - - - - - - - - - - - - - - --- -- -- -- -- --
0
40 20 -- - - - ---------- - - ---------------- ---- - - - - - -
1800 -
Market
solution
1600 ---- --------------- --------
Noncooperative
104 0 - equilibrium
0 200 --------
'1 -
these relativecontrolrateswould be roughly with the averageof 9.7 percentin the cooper-
proportional to thoseshownhereif the overall ativecase.Thereasonforthelowercontrolrate
level of controlswereraisedor lowered.) is completelyintuitive:it resultsfromthe free-
One immediateconclusionthatcomesfrom ridingwhereineachnationignoresthe impacts
this resultis thatcurrentapproachesto com- of its CO2emissionson the welfareof other
batingglobalwarmingmakeno sensefromthe nations(as well, of course,as assumingthat
pointof view of pureeconomicefficiency.The othernations'effortsareunaffectedby its own
currentFramework Conventioncalls for major behavior).The size of the free-
self-interested
emissionsreductionsin the OECDregionwith ridingeffectis the majornew resulthere.
no immediatereductionsin the developing The second interestingconclusion in the
countries-this being exactly the oppositeof noncooperative approachis the distributionof
theefficientsolution.Theonlypotentialration- controlrates.Thismodelpredictsthatthelarg-
ale fortheFramework Conventionis thatit puts est (albeit small) effortswill be takenby the
a veryhighweighton equity(by relievingpoor largest regions-particularly by the United
countriesof obligationsto reduceemissions) Statesand the EuropeanUnion. This predic-
and rulesout the possibilityof side payments tion seems quiteon the mark.It also correctly
(say throughallocationof emissionspermits). suggests that developing countries,particu-
The controlratesin the noncooperativesolu- larly small and poor countriessuch as Benin
tion are markedlylower (not shownbut avail- andKyrgyzstan,will not be in the forefrontof
able from the authors).Thereare two major global-warming politics.
findingshere.First,the aggregateglobalemis- Figures6 through7 show the resultsfor es-
sions controlratefor the noncooperative equi- timatedcarbontaxes. Lookingfirst at Figure
libriumis in 2000 only2.3 percentas compared 6, we can comparethe aggregatecarbontaxes
VOL 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGESTRATEGIES 753
3.5 -
Market
solution
3 .0 ------------------------------------------- <-------------equilibrium
V0.5 -------------------------------------------------------------------
0.0 I I I
190 2 010 2 030 2 050 2 070 209 0
FIGURE 4. TEMPERATURE CHANGE
under different strategies. Note that the coop- to have significantly more (but not very) strin-
erative RICE model looks quite similar to gent controls as compared to small countries.
the older DICE model (which also found The noncooperative carbon taxes are highest
the global optimum). The carbon tax starts in the EuropeanUnion ($0.86 per ton in 2000)
slightly higher and grows more rapidly be- and the United States ($0.65 per ton in 2000).
cause of the steeper trajectory for emissions. The difference reflects the slightly larger out-
The first-periodcarbon tax in the cooperative put in the European Union. For smaller coun-
case is $6.19 per ton carbon in 2000 versus tries, the tax rates are much smaller: 10 cents
$5.94 in the DICE model. (Here and through- per ton in India, and only 1 cent per ton in the
out, all dollar figures refer to prices in 1990 S4 group of countries.
U.S. dollars at 1990 market exchange rates.) It seems appropriateto conclude that outside
The cooperative tax rates are significantly the United States, Europe, and Japan, the ra-
higher than the noncooperative or nationalistic tional noncooperative strategy would be sim-
policies for all regions and periods. The ply to ignore global warming at the present
weighted average carbon tax for the nonco- time. Even by the end of the 21st century, no
operative policy is 24 cents per ton carbon for country acting in a noncooperative framework
the noncooperation equilibrium in 2000. The would have carbon taxes above $2 per ton C.
distribution of carbon taxes for the noncoop- If we define the "cooperation ratio" as the
erative policy is shown in Figure 7. For the ratio of the noncooperative carbon tax to the
noncooperative strategies, large countries tend cooperative carbon tax, we can calculate that
754 THE AMERICANECONOMICREVIEW SEPTEMBER1996
0.25 -
China
0.20 - -----
FSU
0.1
'-4 ~~~ROW
p O 5~~~~Jpa
' ~ 0.10 -- - - -X - - - - -- - - - - - - - - - - - - -
USA
Europe
0
0.00
1990 2d01 2030 2d50 2070 209o
FIGURE5. CO2 CONTROLRATES:
COOPERATIVE SCENARIO
30
Cooperative
equilibrium
DICE model
_1 0 --- --- --- --- --- ------_-__-----_------------------------- 1 - - -- - -
= _/ ~~~~~~~~~~~~~Noncooperative
0
5n equilibrium
-~~~~~~~~~solutionMlarket ------
o
0 w;'$ t t
1990 2010 2030 2050 2070 2090
FIGURE 6. AVERAGE CARBON TAXES
smaller developing countries) dominate CO2 is assigned its optimal policy without any
emissions by the middle of the next century side paymentsfrom other countries.This is
and will have to behave cooperatively if the equivalentto each countryreceivingin the co-
gains from cooperation are to be realized. operativeequilibriuma quotaof tradableemis-
sions permitsequalto the quantityof its own
C. Welfare Effects by Region emissions.
The resultingimpactsupon economicwel-
What are the overall economic effects by fare are shown in Table4. Note firstthatthe
region? The gain to cooperation is calcu- overallresultsfromthe cooperativeRICEso-
lated as the present value of the change in lution are quite close to those of the original
consumption valued using the region-specific DICEmodel.The formeris aboutone quarter
discount rates on consumption (not to be con- higherbecauseof the highergrowthrates in
fused with the pure rates of social time pref- the RICEmodel.By contrast,the noncooper-
erence, or discount rates on utility, which are ative,six-regionRICEmodelshowsextremely
equal across regions). The discount rates slim net benefits-only $43 billion in dis-
in this calculation are region and time spe- countedbenefitsas opposedto $344 billionfor
cific, and they average about 41/2 percent per the cooperativeRICEor $271 for the coop-
year (in real terms) oveF the next century. In erativeDICEmodel.
these runs, there are no internationaltransfers, Figure8 showsthegainsto differentregions
which essentially means that each country for the cooperativeandnoncooperativecases.
756 THE AMERICANECONOMICREVIEW SEPTEMBER1996
1.5
Europe
USA
Japan
FSU
\rO-
05 X . 0
0.0
1990 2010 2030 2050
FIGURE 7. NONCOOPERATIVE CARBON TAXES
Table 4 and Figure 8 present a number of sur- benefits in the cooperative strategy because it
prises in the regional results. The noncooper- is required to undertake significant mitigation
ative solution produces positive net benefits efforts and has few benefits because of its
relative to the market solution for all regions. northerly location. By contrast, the ROW re-
This result is expected because the noncoop- gion reaps major net benefits from the coop-
erative policies improve welfare while the ex- erative solution because the mitigation efforts
ternal interactions among countries are ones are undertaken primarily in the high-income
that are beneficial relative to the market case. countries early in time while the majorbenefits
The net benefits in the noncooperativecase are in terms of damages avoided accrue to the de-
relatively uniform across the different regions, veloping countries in several decades.
with most of the positive effects coming from These results indicate that the cooperative
the reductions in damage from climate change. solution-one in which nations are allocated
The major surprisein these results is the lop- emissions equals to their efficient emissions-
sided benefits from the cooperative strategy. might well not emerge as the outcome of a
The United States actually loses in the coop- bargaining process in which nations will only
erative solution relative to the noncooperative sign on to an agreement that improves their
equilibrium. The reason is that, with its rela- economic welfare. Of course, the patternof net
tively large emissions, the United States would gains can in principle be altered through dif-
be slated to incur major costs today, while its ferent schemes for allocating emissions rights
benefits would be relatively small given its de- to countries (that is, by adding side payments
clining share of the world economy. Similarly, to the program analyzed here); the gains and
the former Soviet Union has quite modest net losses could be made much more equal over
VOL. 86 NO. 4 NORDHAUS AND YANG:CLIMATE-CHANGESTRATEGIES 757
TABLE 4-NET BENEFITS OF DIFFERENT STRATEGIES BY REGION RELATIVE TO THE MARKET EQUILIBRIUM
(BILLIONS OF 1990 U.S. DOLLARS, DISCOUNTED TO 1990)
Note: Each entry indicates the net benefits for a region relative to the marketor uncontrolled strategy. NA is not available.
a From the aggregate DICE model in Nordhaus (1994).
350
u, 250 - ---
0
150 - -
0
0 50 L
gains and losses, with the majorlong-run gains uncertainties are often critical to determining
coming to developing countries while the net policies, formal techniques for determining
benefits to the United States and the former the uncertaintyof future trajectoriesor of im-
Soviet Union are minimal, is a most surprising pacts have been rarely applied to major policy
and troubling finding. issues.14
A full-scale analysis of the uncertaintiesas-
D. Sensitivity Analysis sociated with the RICE model-including un-
certainty about model structure as well as
To understand the full range of outcomes about individual parameters-is beyond the
and policy responses to the threat of global scope of the current article. Many of the cen-
warming, we must assess the fact that many of tral uncertainties have been examined in the
the underlying processes are imperfectly un- context of the DICE model (see Nordhaus,
derstood. Social scientists have developed a
variety of tools to incorporateuncertaintyinto
quantitative modeling, and these can help put
" One notable and controversial example of the sys-
boundson potentialfutureoutcomes.'3Although
tematic application of statistical techniques is the Ras-
mussen report (Nuclear Regulatory Commission, 1975),
which estimated the risk of accidents of different levels of
13 See M. Granger Morgan and Max Henrion (1990) severity in commercial nuclear power plants. An exem-
for a recent survey of tools for the analysis of uncertainty plary study used probabilistic assessments for ozone de-
in quantitativerisk and policy analysis. pletion (National Academy of Sciences, 1979).
VOL 86 NO. 4 NORDHAUS AND YANG:CLIMATE-CHANGESTRATEGIES 759
10 -
China Japan
XOI Europe
.tg 0 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -- - ----
Rest of World
- 9 30 2010
I5 2630 2650
FIrGURE9. CUMULATIVEDISCOUNTEDCONSUMPTION:
VERSUS MARKETrSTRATEGY(TOyrAL
COOPERATIVE GAINS FORCONSlJMPTlON
THROUGH
THE GIVENDATE, DISCOUNTEDBACKTO 1990 )
1994 Chapters6-8), and those resultsapply in Nordhaus( 1994 Table6.1), andthe reader
equally well to the RICE model. To under- is referredto thatreferencefor a fullldiscussion.
standthe extentof sensitivityof the modelwe Figure10 showstheresultsof thesensitivity
presentherea limitedsensitivityanalysiswith analysis.Thatfigureshows the sensitivityof
respect to the importantparametersof the three importantvariablesin the cooperative
model. For each of the importantparameters equilibrium:the carbontax in 2000, the effi-
of the model(see the descriptionin Appendix cientreductionof CO2emissionsin 2000, and
A), we havevariedtheparameter by changing the change in global mean temperaturein
it from the subjective50th percentileto the
subjective90th percentile.'5The exact deri-
vationof the uncertaintyrangewas developed
the sensitivityanalyses,we estimatethe (subjective)90th
percentileof the distribution,17'. Figure 10 shows the
ratioof differentoutcomesfor the 90th percentileof a
5 Symbolically, we can representthe RICE model as a variablesto the 50th percentileof that variable;that is,
mapping, Y, = F(X, - ; 1), where Y, is the vector of en- Ai = F(X,_7; rV0)/F(X,7;rU), whereA, is the ratioof
dogenous and policy variables, X, iS a vector of current
T outcomesfor variablesof interestwhen varyingthe ith
and lagged exogenous variables, and r is the set of un- parameter, rJ is the vectorof r with all variablesset at
certain parameters.The base run estimates outcomes for their50thpercentilewhileU?ois thevectorof parameters
the "best-guess" parameters (IF, which represents the withall variablesbuttheithset atthe50thpercentilewhile
50th percentile of the distribution of the parameters). In the ith parameteris set at its 90thpercentile.
760 THE AMERICANECONOMICREVIEW SEPTEMBER1996
4.1
2.0
o9
0 1.
10
05
00 I -_
p (I A TI (AA/A)O rT2 b2 AZ. M To (AP/P)oTo* M0 K b1 (J0
Parametervaried
x Temperature
(2100) * Carbon
tax(2000) * Controlrate(2000)
FIGURE10. SENSITIVITY
TESTSFORPARAMETERS
Note: The variables on the horizontal axis are parametersof the RICE model as defined in Appendix A. The markers
indicate the ratio of the outcome variable in the sensitivity case to the outcome for the base case. The sensitivity cases
set the values of the variables at the subjective 90th percentile. The outcome variables are the optimal cooperative carbon
tax in 2000, the optimal cooperative reduction rate for CO2 emissions in 2000, and global mean temperaturein 2100.
2100. For each of the three variables, we have variables are relatively unimportant for the
displayed in Figure 10 the ratio of the value of results. 1
the variable in the sensitivity run to the value In analyzing model sensitivity, it is easy to
of the variable in the base case. become lost in the details. For policy purposes,
Figure 10 indicates that the results are ex- however, the single most critical question is
tremely sensitive to the pure rate of social time how an uncertainty affects current policy,
preference. The low rate of time preference which is best seen in the effect on the carbon
(equal to 1 ratherthan 3 percent per year) in- tax. By this standard,the two crucial parame-
creases the carbon tax by a factor of 4 and the te-rs are the- diiscounntrnte {whic-h indicatesc the-
control rateby a factor of almost 2. In addition,
the damage intercept (which is the fraction of
output lost from a doubling of atmospheric 6 These results parallel closely the findings of other
C02) leads to a marked increase in both the studies on the sensitivity of policy to uncertainties about
carbon tax and the control rate. The other major variables.
VOL. 86 NO. 4 NORDHAUS AND YANG:CLIMATE-CHANGESTRATEGIES 761
relative importance of the future compared to sions, and CO2 concentrations as compared
the present) and the damages from climate with the earlier DICE model is largely offset
change (which measure the willingness to by revisions in estimated effects of other
pay to prevent or slow climate change). It is greenhouse gases. As a result the estimated
interesting to note that both major uncer- extent of global warming in the market case
tainties involve human preferences rather by the year 2100-approximately 3?C-
than pure questions of 'fact" about the nat- differs little between the RICE model and
ural sciences. other estimates.
Second, the efficient or cooperative policies
III. Conclusions in the regional model confirm estimates made
in globally aggregated models, such as the
To summarize, this paper has presented the DICE model. The best summary variable for
RICE model, which is a new dynamic, multi- efficient controls is the carbon tax, which is
region, general-equilibrium model of climate calculated to be about $6 per ton carbon in
and the economy. It differs from earlier work, 2000, a number that is virtually identical to
which focussed on a globally aggregated ap- estimates for the efficient policy in the DICE
proach, by introducing production, consump- model.18The estimated degree of control in the
tion, emissions, and damages for different RICE model is, however, estimated to grow
regions. This approachcompares three differ- somewhat more rapidly than in the DICE and
ent strategies for the control of global warm- other models, with estimated efficient carbon
ing: a market approach in which no climate taxes at the end of the next century near $27
change policies are taken, a global coopera- per ton carbon.
tive approach in which all countries choose Third, the RICE model provides estimates
climate-change policies to maximize global in- of the efficient control rates in different
comes, and a noncooperative or nationalistic regions as well. In the efficient solution, car-
approachin which each country takes policies bon taxes are identical in all regions. The con-
to maximize its own national income. These trol rates will differ, however, because of
results are tentative and subject to revision. different costs of reducing CO2emissions. The
Furtherwork will be necessary to test their ro- estimates presented here indicate that the ef-
bustness against alternative assumptions, to ficient emissions control rates will be highest
appraisethe results for differentcoalitions, and in China and the former Soviet Union and low-
to compare the results against other models. est in Japan and Europe, with the differences
Subject to these reservations,the following are being at least a factor of two. These results
the major conclusions. indicate that there will be substantial ineffi-
First, the model produces results for the ciencies in any policy (such as that currently
baseline (market or uncontrolled) which dif- in force under the Framework Convention)
fer significantly from other projections.17 that equalizes emissions control rates across
Output and emissions in the RICE model are countries or does not allow trading of emis-
estimated to grow much more rapidly than sions permits.
in the DICE model or than in many inter- Fourth, a major contribution of this study
national projections (such as that of the In- is to estimate the difference between the ef-
tergovernmental Panel on Climate Change). ficient policy and the noncooperative policy.
The more rapid growth comes largely from The noncooperative or nationalistic policy
a view of the growth process in which there is one in which countries maximize their
is considerable but incomplete convergence economic welfare taking policies of other
of per capita incomes of countries. The countries as given. This implies that small
higher projected growth of output, emis- countries, whose climate-change policies
have little effect on their own economic
welfare, will have little incentive to reduce Japan show reductions in cumulative dis-
emissions while the largest countries will counted consumptionuntil after the middle
have greatlyattenuatedincentivesto engage of the next century.
in costly reductionsin CO2emissions. The Seventh,the resultsindicatethat there are
calculationshereindicatethatthe controlsin majorgains to takingan efficientcooperative
the noncooperativecase (as measuredby the approachto coping with global warmingas
averagerateof carbontax) will be only '/25 of opposedto the noncooperativeapproach.We
thelevel of thecooperativecase.Thatis, while estimatethat the net economicgain from an
the averagecarbontax in 2000 is estimated efficientpolicyhas a discountedvalueof $344
to be about $6 per ton carbonin the coop- billion relativeto the marketscenario,while
erativecase, it is calculatedto be about$0.24 the noncooperativepolicy has a gain of only
per ton in the noncooperativecase. More- $43 billion.Hence,therearecleargainsto at-
over, the divergence between the coopera- taining a cooperativepolicy (assuming, of
tive and the noncooperative policies is course,thatthe policy is itself efficient). The
calculated to increase over time as the in- gains from cooperationwould be even larger
equality of county sizes decreases, and this if climatechangeprovedto have catastrophic
divergence would increase furtherif large consequences thatareveryunevenlyfelt across
countrieslike China, India, Russia, Canada nations.
or the United States splinter into smaller In sum, the results of this new integrated
countriesor decision-makingunits. model of climate and the economy empha-
Fifth,theseresultsindicatethatthe stakesin sizes the implicationsof the fact that while
controllingglobal warmingare modestin the climate change is a global externality, the
contextof overalleconomicactivityover the decision makersare national and relatively
next century.If our estimates are accurate, small. These inherentdifficultiesinvolved in
theyindicatethatthelosses fromglobalwarm- planningover a horizonof a centuryor more
ing will be in the range of 1 to 2 percentof about so uncertainand complex a phenom-
globalincome over the next century.The net enon are compoundedby the dispersedna-
costs (that is, climate-changedamages less tureof the decisions andthe strongtendency
mitigationcosts) can be reducedby perhaps for free-riding by nonparticipantsin any
'A,percentof income by a judicious choice global agreement.Countriesmay therefore
of climate-changepolicies-although, to be be triply persuadednot to undertakecostly
sure, the impactis much greateron our de- efforts today-first because the benefitsare
scendantsthan on ourselves. According to so conjectural,secondly because they occur
RICE, successful cooperationwould lead to so far in the future,and thirdbecause no in-
net gains, but the failureto cooperateis un- dividual countrycan have a significantim-
likely to lead to economic disasterover the pact upon the pace of global warming.The
next century. presentstudyindicatesthatthe thirdof these,
Sixth, the pattern of gains and losses the dispersednatureof the decision making
from differentstrategiesis quite surprising. andthe consequentdilutedincentivesto act,
All countriesgain from the noncooperative is a powerful hindranceto setting efficient
approach,although the amount of gain is climate-changepolicies.
relatively small. The net gains from coop-
APPENDIX A: EQUATIONS OF THE RICE MODEL
eration without internationaltransfers are
quite unevenly distributed,with the major This appendix gives the details of the RICE model. We
gains accruingto developing countrieswith first list and define the variables and then provide the com-
low and rapidly growing emissions. High- plete equation listing.
income countrieshave but modest gains to
cooperation,but the United States actually 1. Variables
loses from cooperatingrelative to a nonco-
The variables are as follows. In the listing, t always
operativestrategy.In addition,the time path refers to time (t = 1990, 2000, ...) while i refers to the
of gains and losses indicates that even in region (i = 1, ... n = USA, Japan, Europe, ...). The
the cooperativescenario, all regions except regional definition is given in Appendix B.
VOL. 86 NO. 4 NORDHAUS AND YANG:CLIMATE-CHANGESTRATEGIES 763
Endogenous Variables.
(A8) M(t) = /3 z Ei(t) + (1 - 6M)M(t - 1)
Ci(t) = total consumption i=l
ci (t) = per capita consumption
CAi (t) = currentaccount balance (A9) T(t) = T(t- 1)
Di (t) = damage from greenhouse warming
Ei (t) = C02 emissions - 1)]-'2[T(t -1) -T*(t -1)]
+ r,[F(t) -AT(t
EX,j(t) = exports from region i to region j r3
F(t) = radiative forcing from all greenhouse gas
concentrations 1)
(AIO) T*(t) = T*(t -1) + T(t - 1)- T*(t -
Qi (t) = output scaling factor due to emissions controls 74
and to damages from climate change
K1(t) = capital stock
IMi i(t) = imports from region i to region]
(All) F(t)= 4.1 log[M(t)IM(O)I + O(t)
log(2)
M(t) = increase in mass of CO2 in atmospherefrom pre-
b2
industriallevel - bii (t)
(A12) ?i (t) = 2, =1 , n.
NFAi (t) = net foreign assets of county i
= welfare weight on country i
n
Qi (t) = gross domestic or regional product
R(t) = net rate of returnon capital
I yQ1(t)
T(t) = atmospheric temperaturerelative to preindustrial (A13) R(t) = =n' - -Q
level I Ki(t)
i=l
T*(t) = deep ocean temperaturerelative to preindustrial
level (A14) NFAi (t) = NFAi (t - 1) + CAi (t - 1)
ui (t) = ui[c i(t)] = utility of per capita consumption
W = social welfare function determined by country con- (A15) CAi (t) = R(t)NFAi (t)
sumption levels n n
Yi(t) = gross national or regional product (net of climate + I IMi,Jt) - Y EXiJt)
damage and mitigation costs) j*i j*i
Gaskins,DariusW. and Weyant,John P. "EMF-12: Negishi,T. "Welfare Economics and the Ex-
Modeling Comparisonsof the Costs of Re- istence of an Equilibriumfor a Competitive
ducingCO2Emissions."AmericanEconomic Economy." Metroeconomica, 1960,12, pp.
Review, May 1993, 83(2), pp. 318-23. 92-97.
Intergovernmental Panel on Climate Change Nordhaus, William D. "An Intertemporal
(IPCC). Climate change: The IPCC scien- General-Equilibrium Model of Economic
tific assessment, J. T. Houghton, G. J. Growth and Climate Change," in Wood,
Jenkins, and J. J. Ephraums,eds. New York: David, ed., Proceedings of a Conference at
Cambridge University Press, 1990. MIT, October 1990.
International Energy Agency (IEA). Climate .__ "To Slow or Not to Slow: The Eco-
change policy initiatives: 1994 update, Vol. nomics of the Greenhouse Effect." Eco-
1. Paris: Organization of Economic Coop- nomic Journal, Juty 1991, 101(407), pp.
eration and Development, 1994. 920-37.
Manne, Alan S. and Richels, Richard. Buying "
"An Optimal Transition Path for
greenhouse insurance: The economic costs Slowing Climate Change." Science, No-
of CO2 emission limits. Cambridge, MA: vember 20, 1992, pp. 1315-19.
MIT Press, 1992. . Managing the global commons: The
Manne, Alan S. and Rutherford, Thomas F. "In- economics of the greenhouse effect. Cam-
ternational Trade, Capital Flows and Sec- bridge, MA: MIT Press, 1994.
toral Analysis: Formulation and Solution NuclearRegulatoryCommission.Reactor safety
of IntertemporalEquilibrium Models," in study: An assessment of accident risks in
W. W. Cooper and A. B. Whinston, eds., U.S. Commercial nuclear power plants,
New directions in computational econom- NUREG-75/014 WASH-1400. Washing-
ics. Dordrecht, Netherlands: Kluwer, 1994, ton, DC: 1975.
pp. 191-205. Peck, Stephen C. and Teisberg, Thomas J.
Mas-Colell, Andreu; Whinston, Michael D. and "CETA: A Model for Carbon Emissions
Green,JerryR. Microeconomic theory. New Trajectory Assessment." Energy Journal,
York: Oxford University Press, 1995. 1992, 13(1), pp. 55-77.
Morgan, M. Granger and Henrion, Max. Uncer- Wigley,T. M. L. "The Contributionfrom Emis-
tainty: A guide to dealing with uncertainty sions of Different Gases to the Enhanced
in quantitative risk and policy analysis. Greenhouse Effect," in Ted Hanisch, ed.,
New York: Cambridge University Press, Climate change and the agenda for re-
1990. search. Boulder, CO: Westview Press,
National Academy of Sciences. Stratospheric 1994, pp. 193-222.
ozone depletion by halocarbons: Chemistry Yang, Zili. "Essays on the International As-
and transport. Report of a committee of the pects of Resource and Environmental
National Research Council, National Re- Economics." Ph.D. dissertation, Yale Uni-
search Council, Washington, DC, 1979. versity, 1993.