115

Some Contributions of Integrated

Assessment Models of Global Climate
Change
John Weyant*
“. . .all models are wrong, but some are useful.”
—G. E. P. Box and N. R. Draper (1987)

Introduction
It is now clear that human-induced climate change is caused by (1) oil, gas, coal, and biofuel
combustion in utility and industrial boilers and land, sea, and air transportation systems that
produce emissions of carbon dioxide (CO2) and other radiatively active gases to the atmosphere;
and (2) land use and land use change activities that release CO2, methane, and/or nitrous oxide to
the atmosphere (IPCC 2013). Emissions of these substances lead to net increases in the accumu-
lations of these gases in the atmosphere (i.e., above those that occur naturally). Because these gases
allow more of the heat from the sun’s radiation through to the earth’s surface than from the earth’s
surface back out to deep space, they are generally referred to as greenhouse gases (GHGs).
Assessments of the effects of climate change on people and their property, wildlife, and ecosys-
tems indicate that these effects can be significant (IPCC 2014a). This has led to the consideration of
three main approaches for ameliorating the impacts of climate change: (1) mitigation of GHG
emissions, (2) adaptation to any climate changes that might occur, and (3) geoengineering to
influence the amount of solar energy reaching the earth’s surface and/or to influence the chemistry
of the oceans. Because the relationships within and between the various biogeochemical and
socioeconomic components of the earth system can be quite complex, a number of quantitative
models have been developed to study earth systemwide climate changes and the effect of various
types of public policies on projections of future climate change. These models have become known
as “integrated assessment of climate change” or simply integrated assessment models (IAMs).1

*Department of Management Science and Engineering, Room 260, Huang Engineering Center, Stanford
University, Stanford, CA 94305-4026, USA. Tel: +1 650 723 3506; Fax: +1 650 725 5362; e-mail:
[email protected].

The author thanks David Anthoff, Delavane Diaz, Karen Fisher-Vanden, Tom Hertel, Gib Metcalf, Chris Hope,
Suzy Leonard, Jia Li, William Nordhaus, Richard Richels, Steve Rose, Fran Sussman, James Sweeney, Richard
Tol, and three excellent anonymous reviewers for suggestions on this article.
1
Although the focus here is on the application of these modeling frameworks to climate change policy and
science program management, these same models have also been used to study related issues including air
Review of Environmental Economics and Policy, volume 11, issue 1, Winter 2017, pp. 115–137
doi:10.1093/reep/rew018

ß The Author 2017. Published by Oxford University Press on behalf of the Association of Environmental and Resource Economists.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License
(http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any
medium, provided the original work is properly cited. For commercial re-use, please contact [email protected]
116 J. Weyant

The objective of these models is to project alternative future climates with and without various
types of climate change policies in place in order to give policymakers at all levels of government
and industry an idea of the stakes involved in deciding whether or not to implement various
policies.
The literature on IAMs is now vast and continues to grow rapidly. However, this literature is
spread across many disciplines, with publications appearing in a wide range of journals, includ-
ing those that focus on earth sciences, biological sciences, environmental engineering, eco-
nomics, sociology, technological change, and other related fields. This has led to some
confusion about what IAMs are, what they have been developed to do, and what impacts
they have already had on decision making and people’s thinking about climate change problems
and solutions (cf., Pindyck 2013).
This article, which is part of a symposium on the use of IAMs for climate policy,2 examines the
use of IAMs of global climate change for policy analysis and research management. The objectives
are to (1) explain the models, (2) describe how they have been used, (3) assess the contributions
they have made, (4) identify the key challenges that remain in developing and using them, and
(5) suggest areas for improvement. Like all quantitative models used in policy analysis, these
models have strengths and weaknesses. Although the highly complex and very uncertain nature of
both the climate change problem and potential policy responses to it present model builders with
many formidable challenges, I believe these models have already proven quite useful in assessing
the magnitude of the climate change problem and the efficacy of potential solutions.
For the purposes of this article, I define an IAM of global climate change to be any model that
covers the whole world and, at a minimum, includes some key elements of the climate change
mitigation and climate impacts systems at some level of aggregation. Climate change studies
that focus on regional and national decision making (see, e.g., Knopf et al. 2013; Fawcett,
Clarke, Weyant 2014; Kraucunas et al. 2015) do not systematically cover the whole earth
system and thus are not included here. In addition, there is now an extensive and rapidly
growing literature on the use of benefit–cost analysis (BCA) to support climate change adap-
tation decision making on a regional scale (Li, Mullan, and Helgeson 2015), which is not
included here due to space limitations.
The article is organized as follows. In the next section I present an overview of the two major types
of global IAMs that have been used to study climate change. Then I describe applications of each
model type. This is followed by a discussion of the major research challenges concerning the use of
these models, highlighting areas where they can be improved. The final section presents conclusions
and recommendations for further development and use of IAMs to study climate change.

An Overview of IAMs
All IAMs include the economic and natural processes that produce GHG emissions. Those
emissions are used to drive a representation of the global carbon cycle and the chemical

quality, water scarcity, and food security. However, a discussion of direct applications in those areas is beyond
the scope of this article.
2
The other articles in the symposium are Metcalf and Stock (2017), which introduces the symposium and
examines the role of IAMs in estimating the social cost of carbon for use in U.S. regulatory analyses, and Pindyck
(2017), which examines the uncertainty that characterizes IAMs and proposes an alternative to an IAM-based
social cost of carbon.
Some Contributions of Integrated Assessment Models of Global Climate Change 117

composition of the atmosphere, which is then used to drive changes in climate and sea level.
The models then project how those changes impact natural systems on earth, some of which are
managed by and valuable to humans. However, IAMs differ tremendously in their level of detail
and the complexity and interconnections they consider. For example, some models represent
the whole earth system with a small number of fairly simple equations (e.g., Nordhaus 2014),
while others include thousands of equations drawn from physics, chemistry, biology, and
economics (e.g., Reilly 2012b). Approximately twenty global scale IAMs have been developed
(see the list of references for this article). However, there are two basic types: detailed process
(DP) IAMs and benefit–cost (BC) IAMs. Both types of IAMs have been applied to climate
change mitigation policy questions for several decades.3 Although both types of IAMs include
projections of GHG emissions and the costs of various ways to mitigate them (e.g., energy
conservation, changes in production processes, fuel switching), they handle climate change
impacts differently. DP IAMs are more disaggregated and seek to provide projections of climate
change impacts at detailed regional and sectoral levels, with some using economic valuation and
others using projections of physical impacts such as reductions in crop growth, land inundated
by sea level rise, and additional deaths from heat stress.
In contrast, BC IAMs provide a more aggregated representation of climate change mitigation
costs and aggregate impacts by sector and region into a single economic metric.4 The main
motivation for developing BC IAMs has been to use them to implement BCA to identify
“optimal” climate policies, but they have also been used to calculate the costs and benefits
associated with polices for which marginal costs and marginal benefits are not equal.
The DP IAMs provide more information than the BC IAMs on the physical impacts and
economic costs of climate change and the benefits of GHG emissions mitigation. This has two
important implications for the BC models. First, similar to the aggregation and calibration
required on the mitigation side of the BC IAMs, the more complex DP IAMs can also provide
projections of the economic costs net of endogenous adaptation to climate change for key
sectors (e.g., agriculture) by region. These estimates can then be aggregated for use in the BC
IAMs. Since most climate impacts take place at the watershed, agricultural growing region,
ecological zone, city, or similar level, this additional degree of geographical disaggregation is
essential for improving the validity of the damage functions used in the aggregate BC IAMs.
Second, the additional information on both the physical and economic impacts of climate
change may be vitally important to decision makers in some regions and sectors.5
Concepts and insights from the more complex DP IAMs can be used to calibrate the simple
aggregate BC IAMs and to identify where they might be improved. Concepts and results from
the DP IAMs can be important to the BC-oriented models because the cost of GHG emissions
mitigation functions in the aggregate BC IAMs are often calibrated to results from the more
disaggregated DP IAMs, which consider much more energy sector and land use detail
(including things like explicit agricultural and forestry activities, as well as changes in unman-
aged ecosystems). In fact, these more detailed models have frequently been used to examine the
“cost-effectiveness” of alternative policies for meeting various climate targets, such as limits on
3
For early examples of these models, see Nordhaus (1989, 1991, 1994b), Rotmans (1990), Manne and Richels
(1992); Edmonds et al. (1994); Edmonds, Wise, and MacCracken (1994), and Weyant et al. (1996).
4
These aggregate impacts functions are sometimes calibrated to detailed sectoral climate impact projections by
region.
5
This issue is discussed in more detail later.
118 J. Weyant

carbon concentrations or a maximum permissible global mean temperature increase at the

lowest possible societal cost.
In the next two sections I discuss applications of the more disaggregated DP IAMs and the
more aggregated BC IAMs, respectively.

Applications of Disaggregated DP IAMs

By about 1990, the early disaggregated DP IAMs were developed by adding a reduced form
carbon cycle, atmosphere, ocean, and other components to existing global energy, economic,
atmosphere, and ecosystems models (e.g., Edmonds et al. 1994; Edmonds, Wise, and
MacCracken 1994; Rotmans 1990). Over the years, research teams have been able to incorp-
orate sectors that are both impacted by and integral to the operation of the earth system as a
whole (e.g., land use and water), which have remained high priority objectives for most teams.
As described next, DP IAMs have been used in three main types of climate system and policy
analysis: mitigation analysis, climate impacts analysis, and integrated mitigation and impact
analysis.

Use of DP IAMs in Mitigation Analysis

There is a long history of using DP IAMs to study the energy–economy impacts of climate
change mitigation policies, including about a dozen intermodel comparison studies. The cur-
rent state of the art, including the results of the intermodel comparison studies, is captured
nicely in “Assessing Transformation Pathways” (Clarke et al. 2014), chapter 6 of the
Intergovernmental Panel on Climate Change (IPCC, 2014b) fifth assessment report, which
summarizes and contrasts some 1000 scenarios produced by these models.6 Based on the results
from these models, which include many climate policy and technology assumption sensitivity
analyses, the chapter demonstrates both the advantages of efficient policies and the availability
of a wide range of low GHG–emitting technologies to achieve stringent climate policies favored
by many nations, such as an increase in global mean temperature of no more than 2 C relative
to preindustrial times, a concentration of GHGs in the atmosphere of no more than 450 parts
per million (ppm) CO2 equivalent, or reducing emissions by 80% relative to 1990 levels by
2050. In addition to focusing on technology and more realistic policy architectures than were
available at the time of the fourth assessment report, this chapter reinforces several basic con-
clusions from previous assessments, which are summarized by Weyant (2008): (1) projections
of mitigation costs are very uncertain; (2) mitigation cost projections are very sensitive to the
specific assumptions made about how policies are formulated and implemented; (3) despite the
uncertainties, we can put rough bounds on the range of plausible mitigation cost estimates; and
(4) there are a number of general principles we can apply to the design of climate policies to help
make mitigation costs lower rather than higher. I discuss each of these conclusions in more
detail next.
6
This chapter focuses on studies produced since the fourth assessment report, including Bosetti and Frankel
(2012), Calvin et al. (2012), Krey et al. (2014), Kriegler et al. (2013a, 2013b, 2014a, 2014b), and Riahi et al.
(2015).
Some Contributions of Integrated Assessment Models of Global Climate Change 119

Uncertainty of mitigation cost projections

Given the huge range of mitigation cost projections in the literature (Weyant 1999, 2004; de la
Chesnaye and Weyant 2006; Edenhofer et al. 2006; Clarke et al. 2009; Bosetti and Frankel 2012;
Kriegler et al. 2013a, 2013b, 2014a, 2014b; Krey et al. 2014), it is clear that these projections are
inherently uncertain. Moreover, since most of these projections are model based, and the model
parameters are sometimes held within unrealistically narrow ranges, much of the literature consists
of mean projections from a number of different models with no, or at best only partial, uncertainty
analyses included. This level of uncertainty in mitigation cost projections is not very surprising. The
future of the world economy and the world energy system must be projected over many decades
using many assumptions about productivity growth, fuel prices, technology diffusion, the develop-
ment of new technologies, and the interest/discount rate used in making intertemporal investment
decisions. In addition to these projections about basic energy–economic conditions, projections
about future government energy–environment policies around the world must be made.

Sensitivity of projections to model assumptions about policy implementation

This leads to the second major conclusion regarding mitigation cost projections for achieving a
particular climate policy objective: that they are very sensitive to the assumptions made about the
specific policies used to achieve the emissions reductions and how those policies are implemented.
Clearly, assumptions about the possible range of policy instruments and how well they might be
implemented can lead to large differences (probably as large as an order of magnitude in either
direction) in mitigation cost projections. This policy dependence means that analysts cannot make
very precise mitigation cost projections without knowing the details of the policy regime that will
be used to achieve the mitigation target. Nevertheless, these cost projections can help policymakers
understand the trade-offs involved in selecting specific elements of any policy regime. For ex-
ample, projections can be used to examine the cost implications of different assumptions con-
cerning the availability of low GHG energy technologies (Kriegler et al. 2014b) and the
implications of different assumptions about the degree and timing of international cooperation
in achieving global climate policy objectives (e.g., Riahi et al. 2015). In addition, comparisons of
the costs of regulatory versus market approaches to GHG mitigation at the country and regional
level have emerged (see Knopf et al. 2013; Fawcett, Clarke, and Weyant 2014).

Ability to place bounds on estimated costs

Third, despite all the complexities and uncertainties involved, we can still place some rough
bounds on mitigation costs. More specifically, under wide ranges of assumptions about the
future of the world economy and possible policy architectures, we can conclude that stabilization
of GHG concentrations in the atmosphere during this century at 550 ppm will likely cost some-
where between 0.1 percent and 10 percent of gross world product (GWP) per year (e.g., Weyant
1999, 2004; de la Chesnaye and Weyant 2006; Edenhofer et al. 2006; Clarke et al. 2014). This range
of two orders of magnitude can probably be cut in half if the basic elements of the policy regime to
be used to stabilize GHG concentrations are specified.7 These bounds can help policymakers put
7
Although the scenario results of the last 5 years or so (Clarke et al. 2014) show a narrower range of cost
projections, in part because the cost of non-fossil-fuel-based alternative energy sources has come down, it is
critically important to recognize that these results indicate ranges of published results and the studies that
120 J. Weyant

climate change in perspective relative to other big societal issues such as national security, health,
education, and welfare. In addition, this range of cost estimates, and its dependence on the policies
assumed, provides a sense of how valuable economically efficient policies can be.

Policy design principles for lowering mitigation costs

Finally, and perhaps most importantly, several generations of mitigation cost studies have
allowed us to identify and analyze a number of general principles for improving the economic
efficiency of mitigation policies. The most general principle that has emerged is that broadly
targeted, flexible policies lead to lower costs than narrowly focused, inflexible ones (e.g., Weyant
2008). This suggests that allowing flexibility in what, where, when, and how GHG emissions
reductions are to be achieved results in lower mitigation costs across all economic assumptions
and all models. For example, flexibility in where emissions reductions take place can be facili-
tated by allowing trading of GHG emission rights across individuals, regions, sectors, and
countries. Flexibility in when emissions reductions are taken can be facilitated by allowing
borrowing and banking of GHG emission rights over time. Flexibility in how emissions reduc-
tions may be achieved can be promoted by including research and development (R&D) incen-
tives, information programs, and carefully chosen efficiency standards, along with traditional
market-based mechanisms like a carbon tax or a cap and trade program. One strong conclusion
from this recent work is that delays in implementing mitigation actions can cause increases in
the total discounted cost of meeting a particular global GHG concentration or temperature
target as emissions from new noncontrolled sources become harder and harder to offset, even if
future low GHG–emitting technologies become more efficient over time (Riahi et al. 2015).
Finally, allowing appropriately weighted emissions reductions in all GHGs and the use of
biologic sinks (like growing more trees or simply slowing deforestation) can reduce the overall
cost of achieving any climate outcome, such as a particular GHG concentration, radiative
forcing, or temperature target. Clarke et al. (2014) summarize the most recent work in this
area by considering the implications (across a number of recent model intercomparison pro-
jects) of key technology availabilities and alternative assumptions about when and who par-
ticipates in a global GHG mitigation regime.8

Use of DP IAMs for Climate Impact Analysis

Although the use of DP IAMs for both GHG mitigation and climate change impact analyses has
a long history (cf., Weyant et al. 1996), the use of DP IAMs for climate impacts analysis has thus
far been much less systematic, with research teams focusing on impacts in sectors they are most
interested in and/or able to study. Thus, for example, there have been DP IAM analyses of
climate change impacts on agriculture (Boehlert and Strzepek 2013; Tai, Val Martin, and Heald
2014), water (Blanc et al. 2014; Schlosser et al. 2014; Strzepek et al. 2010), biodiversity (Meller,
van Vuuren, and Cabeza 2013), coastal zone damages from sea level rise (Diaz 2016), and across
multiple impact sectors (Reilly 2012b; Stehfest et al. 2014). In order to enable IAMs to be used

include them were carried out for many different purposes and thus do not reflect the full range of uncertainties
over model inputs or parameters.
8
It is important to note that even in highly disaggregated DP IAM analyses of mitigation policies, the models
often do not consider options that might be important to decision makers, such as policies to stimulate energy
R&D or to compensate workers for dislocations resulting from the implementation of the policies.
Some Contributions of Integrated Assessment Models of Global Climate Change 121

for more systematic assessments and comparisons of impact projections from around the
world, a new effort is under way to create coordinated global scenarios for individual impact
assessments and mitigation assessments (Moss et al. 2010; O’Neill et al. 2014; Riahi et al. 2016).9

Use of DP IAMS for Integrated Mitigation and Impacts Analysis

The third and most advanced category of applications of disaggregated DP IAMs is integrated
mitigation and impact analysis, reflecting the fact that the way large-scale GHG mitigation or
adaptation is implemented can change the climate, and changes in climate can change the
efficacy of mitigation and adaptation strategies. Studies that consider these kinds of physical
and economic interactions between impact sectors and feedbacks from them to the climate
system are not possible with the more aggregated BC IAMs. However, a number of these inter-
actions and feedbacks have been addressed with DP IAMs, including the interaction between
climate and air pollution polices (Reilly et al. 2007; Chuwah et al. 2013; Nam et al. 2016), the
increased competition for water between agriculture and power plant cooling that would occur
in a hotter and dryer climate, and the impacts on water and land (and the resulting land emis-
sions) of global policies that rely on massive increases in biofuels (Reilly and Paltsev 2009; Reilly
et al. 2012a; Daioglu et al. 2014; Rose et al. 2014a). Additional examples of analyzing mitigation
and adaptation policies simultaneously include Hinkel et al. (2013); Calvin et al. (2014); Herrero,
Havlik, and Obersteiner (2013); Ermolieva et al. (2014); and Mosnier et al. (2014).

Applications of Simple Aggregate BC IAMs

There have been three main applications of BC IAMs. First, BC IAMs have been used for several
decades to compute the optimal trajectory of global GHG emissions, and the corresponding
prices to charge for those emissions (cf., Weyant 2015). Such optimal policies equate the
marginal (discounted) benefits (in terms of climate damages avoided) with the marginal (dis-
counted) costs (in term of mitigation effort required) of the climate change policy. Second,
aggregate BC IAMs have been used to evaluate the costs (and benefits) of pursuing nonoptimal
climate policies (i.e., climate targets, such as emissions, concentration, or temperature targets)
rather than optimal policies. Because the aggregate mitigation components of these systems are
calibrated to the more disaggregated DP IAMs, many of the basic conclusions concerning DP
IAMs also hold for BC IAMs. However, in the case of BC IAMs, rough economic valuations of
changes in climate damages are included and can be traded off against the costs. Third, BC
IAMs have been used, especially in current U.S. climate policy deliberations, to compute the
social cost of carbon (SCC), which is defined as the incremental damage caused by one more
ton of carbon emissions. I discuss these three types of applications of BC IAMs in more detail
next.

9
Although the frameworks suggested in Moss et al. (2010), O’Neill et al. (2014), and Schellnhuber, Frieler, and
Kabat (2014) provide excellent tools for coordinating the efforts of the earth sciences community and for
benchmarking intermodel comparisons within and among them, they are not, in themselves, IAMs as defined
in this article.
122 J. Weyant

BC IAMs and “Optimal” Climate Polices

Optimal climate policies are those that balance the marginal cost of the last ton of emission
reductions against the marginal damages resulting from the last ton emitted — such policies
maximize welfare because they minimize the total cost of climate mitigation plus the remaining
damages that are incurred. One way to achieve this optimal policy outcome is to impose a
carbon tax that is equal to the optimal marginal mitigation cost/climate damages, which is
referred to as an optimal emissions tax. Recent projections of the optimal cost of carbon from
the three most widely used aggregate BC IAMs — Dynamic Integrated Climate-Economy
(DICE; Nordhaus 2014), Framework for Uncertainty, Negotiation, and Distribution (FUND;
Anthoff and Tol 2013), and Policy Analysis of the Greenhouse Effect (PAGE; Hope 2011) —
provide mean estimates of the optimal cost of carbon that range from $10 per ton for the FUND
model to $18 per ton for the DICE model to $71 per ton for the PAGE model. The differences in
estimates are due primarily to differences in the models’ climate damage functions, especially
their treatment of the potential for catastrophic outcomes like the melting of the Greenland ice
sheet or the release of trapped methane gas from thawing permafrost. The FUND estimate is
lower because, unlike DICE and FUND, it does not account for such possibilities. Moreover, the
potential for catastrophic outcomes (based on a survey in DICE) accounts for roughly 70
percent of global damages at 2.5  C (Wolverton et al. 2012). Another key driver of the lower
damage projections from FUND are its assumptions about the ability of those impacted by
climate change to adapt to those changes (e.g., farmers can change crops, adjust planting
schedules, incorporate irrigation, or apply fertilizer). PAGE also reports that the standard
deviation of the optimal carbon tax is $266 per ton if all 100,000 runs in their simulations
are included, but only $56 per ton if the highest 1% of the optimal results are eliminated—
which highlights the significance of the “thick tail” (i.e., a probability distribution of the
magnitude of climate damages that goes to zero very slowly at high damage levels) in their
probability distribution over damage outcomes. This means that the optimal carbon policies
can depend strongly on the modeling team’s world view concerning what impacts and what
range of climate outcomes to consider, and what types of adjustments those impacted are
assumed to make.
Thus model choice can lead to a large range in the projections of the optimal carbon tax, and
the plausible ranges of key input assumptions (many of which are discussed in more detail later)
can lead to a similarly large spread in the range of projections from a single model. In fact, the
combination of model choice (i.e., structure) and model input uncertainties can cause the
projected optimal carbon tax to vary by one to two orders of magnitude (Rose et al. 2014b).
Despite these uncertainties, however, the application of BC IAMs to climate change policy has
provided numerous insights. In particular, BC IAMs have helped identify the most important
drivers of the projected costs and benefits of climate policy, which has improved our ability to
quantify uncertainties about model outputs of interest. Perhaps the most important contribu-
tion of these models is their ability to highlight the critical issues (such as discounting, risk, and
damages) that arise in making projections of the costs and benefits of climate policies and to
incorporate new scientific findings into the projections of costs and benefits in a timely and
orderly fashion (e.g., Nordhaus 2013).
Some Contributions of Integrated Assessment Models of Global Climate Change 123

BC IAMs and Non-Optimal Climate Policies

The three main aggregate BC IAMs have also periodically been used to produce results for
nonoptimal climate policies. For example, Nordhaus (2007) considered a number of alternative
policy scenarios to an “optimal” scenario in which marginal costs equal marginal benefits, and
projected that for policies that are moderately more stringent than the optimal one (e.g., limits
of two times the GHG concentration or an increase of 2.5 C relative to preindustrial levels), the
abatement cost was between 0.1 and 0.25 percent of income (on a present value basis). These are
much lower costs than for some of the more ambitious emissions abatement programs that
have been proposed (e.g., a maximum temperature increase of 2 C). Under those programs,
abatement costs amount to around 1.5 percent of income.10 Thus the cost of these nonoptimal
policies exceeds the cost of the optimal policy by more than 1 percentage point of global income
while reducing aggregate climate damages by only about .3 percentage points of global income.

BC IAMs and the SCC

The third application of the BC IAMs has been to estimate the SCC. Estimates of the SCC by the
U.S. Interagency Working Group (IWG) on Social Cost of Carbon have been used for energy
efficiency and carbon emissions rulemaking by the U.S. government (see Greenstone, Kopits,
and Wolverton 2013). The IWG used the damage function modules of the three BC IAMs
mentioned in the last section to obtain a range of projections of the SCC.11 As discussed in Rose
et al. (2014b), the IWG SCC reports (IWG 2010, 2013) include a wide range of SCC estimates,
with the estimates again varying substantially across models and input assumptions. Rose et al.
(2014b) find a pattern of results that is similar to the optimal carbon tax projections discussed
in the last section (i.e., PAGE produces the highest average SCC estimates but considers more
uncertainty regarding climate damages than the other models, while FUND produces the lowest
average SCC estimates). For all the models used in the IWG SCC reports, SCC averages and
extremes decline with higher discount rates because of the greater discounting of the models’
estimates of positive annual damages over the next three centuries. The challenges discussed in
the next section provide additional sources of uncertainty to consider when interpreting and
assessing these SCC projections.

Major Challenges for Integrated Assessment Modeling

IAMs that have been used to study climate change policies and guide the setting of earth systems
research priorities have been constructively criticized (e.g., Ackerman and Stanton 2010, 2012;
Pindyck 2013, 2017; Stern 2013). The critiques have generally focused on model design choices
that are limited by data and/or understanding of key relationships in the physical and socioeco-
nomic world relevant to assessing the impacts of climate change and emissions mitigation
policies. Thus these critiques serve as both limitations on the current models and challenges
for improving them. In this section I summarize what the modelers, their critics, and experts in
relevant disciplines say are some of the biggest challenges for improving the design and use of
10
The Stern Review (Stern 2007) estimates the present value of abatement costs to be 1 percent of income.
11
See Johnson and Hope (2012), Tol (2008, 2009), and Nordhaus (2014) for more on these projections of the
SCC.
124 J. Weyant

IAMS. These challenges include (1) what to count and how to count it; (2) the inclusion of
extreme and discontinuous outcomes; (3) the treatment of regional, national, and international
equity; (4) the treatment of intertemporal discounting and intergenerational equity; (5) pro-
jections of baseline drivers; (6) capturing interactions between impact sectors and feedbacks to
the climate system; and (7) dealing with uncertainty and risk.

What to Count and How to Count It

The most fundamental challenge for the design of IAMs of global climate change is deciding
which policy-relevant emissions mitigation options and climate change impacts to include and
how to measure them. These considerations range from how many different impacts (aggregate
costs and benefits, poverty reduction, population at risk of starvation, etc.) to consider, to
whether to measure them in physical units or monetary equivalents, to whether to be more
concerned about mean or possible extreme outcomes. These choices pose challenges for both
DP IAMs and BC IAMs, but in somewhat different ways. For example, DP IAMs are able to put
economic values on some inputs and leave others in physical impact terms, while BC IAMs can
be so highly aggregated in terms of economic costs and benefits that it is not even clear which
impacts have been included and whether interactions between those impacts are being
considered.
The choices made about which climate change impacts to include and at what level of
aggregation clearly depend on what types of decisions (e.g., public policies versus private in-
vestments) and whose interests (e.g., power plant owners, farmers, automobile drivers, resi-
dents of coastal zones) are being considered. For example, a policymaker in a low income
country might be more interested in alleviating energy poverty by improving energy access and/
or reducing poor air quality than in reducing climate change impacts per se, while a decision
maker in a high income country might be more interested in trading off the reductions in
economic output that result from reducing emissions against the losses from increases in the
projected losses from extreme weather events in both poor and rich coastal regions. IAMs can
help quantify these trade-offs in both monetary and physical terms, but the production and
interpretation of the results necessarily requires a number of important value judgments con-
cerning whose preferences to count and how to weigh them against one another (cf., Sussman,
Weaver, and Grambsch 2015).

Inclusion of Extreme and Discontinuous Outcomes

One of the most troubling concerns about climate change is the potential for abrupt, irrevers-
ible, or catastrophic climate changes or climate change impacts (see National Research Council
2002; Lenton et al. 2008; Ackerman and Stanton (2010); IPCC 2013, 2014a). Projections of the
economic costs of such extreme events are directly included in the PAGE model and captured
conceptually in the damage estimates in the DICE-2013R model. However, these two models
do not explicitly address tipping point concerns, primarily because tipping points have not been
reliably identified. It is important to emphasize that other than expert opinion, there is not yet
much basis for determining the size, timing, or probability of such tipping points or the
economic damages that would result from them. Although this lack of empirical evidence is
Some Contributions of Integrated Assessment Models of Global Climate Change 125

inescapable, these discontinuities may result in some of the most serious climate change
impacts.
To date, the use of expert elicitations to characterize the functional form and uncertainty of
these tipping points has been rare. Early expert elicitations on the probability of extreme climate
outcomes (Morgan and Keith 1995) and impacts (Nordhaus 1994a) have been extremely
valuable. However, these assessments have not addressed the issue of the “independence” of
experts. It can easily be the case that experts are not independent if many of them have received
identical training, use the same approaches to modeling and analysis, and maybe even have the
same mentors. The issue of independence can be addressed by “combining expert opinions”
(Morris 1977, 1983; Clemen and Winkler 1986), which, although difficult to accomplish, is
almost certainly necessary.
A further challenge for the use of IAMs that has not yet been fully resolved is the potential for
highly skewed distributions (i.e., fat tails) of uncertain variables, an issue raised by Weitzman
(2009) and discussed extensively in the literature (cf., Weitzman 2013). A simplified way to state
Weitzman’s argument is that the combination of fat tails and strong risk aversion may lead to
large losses in expected welfare. If this is the case, then the true BC IAM damage projections
could become unbounded or extremely large.
Another important limitation of the simple BC IAMs, and to a lesser extent the DP IAMs, is
that the whole climate system is represented with a very small number of equations drawn from
highly reduced-form climate models (e.g., the Model for the Assessment of Greenhouse Gas
Induced Climate Change [MAGICC] in Meinshausen, Raper, and Wigley 2011). These models
generally include only energy balance equations, thus excluding the conservation of momen-
tum and mass balance equations included in a full-scale climate or earth system model
(Washington and Parkinson 2005). In actual tests of temperature projections, these models
have been found to match the behavior of full-scale climate models for scenarios that move
gradually from one steady state to another, but not in cases where more rapid changes are
considered (van Vuuren et al. 2011; Schaeffer et al. 2013; Zickfeld et al. 2013; Rose et al. 2014b).

Treatment of Regional, National, and International Equity

The application of IAMs to climate policy has generally been efficiency focused; equity con-
siderations have rarely been addressed directly, except for occasional ex post reporting of
economic impacts across various socioeconomic strata within and between nations. This is
unfortunate because equity and fairness issues often dominate the political debate concerning
what to do about climate change. Moreover, this shortcoming is not easily fixed after a model
with an efficiency-based optimizing architecture has been run, because of the many equity
trade-offs and approximations that are typically made when such models are constructed in the
first place. In addition, the global or regional representative consumer formulations, in which
the choices of—and impacts on—a single “average” consumer are used to represent all con-
sumers in an economy, that are included in some BC IAMs can make it hard to assess impacts
across income classes at the national or international level. Consider the plight of the approxi-
mately two billion people on earth who live in areas without functioning goods and financial
markets. Since they have no markets, these people have no (measured) income and make no
(measured) expenditures. Thus, if the climate changes, they absorb no (measured) impacts and
would not benefit from any reductions in those impacts that result from any policy designed to
126 J. Weyant

reduce carbon emissions. This means that if an optimal carbon tax is computed for a market or
country based on global costs and benefits, the impacts on the world’s poorest people will not be
included.12
A related challenge for applying IAMs to climate policy concerns aggregation across coun-
tries: whose damages should be considered and how should they be weighted in computing an
aggregate impact metric? This raises some daunting analytic as well as deep philosophical issues.
For example, should market exchange rates or purchasing power parity weights be used to value
economic damages in developing countries relative to those in developed countries? This by
itself can change the valuation of Chinese impacts vis-à-vis those in the United States or
European Union by a factor of two or more. But the difficulty does not stop with dollar
equivalent issues: weighting losses in wealthy versus poor countries is not straightforward
(Adler 2012), and it is not lost on poor countries that putting less weight on their impacts
because they produce less marketed economic output can lead to climate policy recommen-
dations that are unfair to them.
Another related challenge for DP IAMs is how to include nonclimate benefits or costs, which
can be a major issue in policy debates, especially in developing countries, where air pollution
and energy access can often be higher priorities than climate change per se. Recent research with
two major DP IAMs highlights the significance of considering this broader set of objectives
(Pachauri, Van Ruijven, and Nakicenovic 2013), which are just starting to be incorporated into
DP IAMs but do not yet (and maybe cannot) figure explicitly in BC IAMs.
A final practical challenge for global BC IAMs is that even if all the site-specific physical
climate change impacts can be projected and valued and aggregated with impacts in other
sectors and other regions at a specific point in time, all the aggregation methods would need to
be revised over time. Although this could arguably be done well for small changes, it could be
problematic for the types of large systematic variations that are the biggest concern as the
climate changes. This is also a challenge for DP IAMs, but it is easier to include nonlinearities
in these models.

Treatment of Inter-temporal Discounting and Intergenerational Equity

In addition to the issue of treating countries consistently and fairly, it is necessary to weigh costs
and benefits over time, which ultimately means considering the preferences of as-yet unborn
generations. Clearly, this cannot be done directly. Thus people making decisions today on
behalf of those not yet alive need to make collective ethical choices about what kind of
opportunities (usually characterized as a particular state of the climate system measured by
global mean temperature, GHG concentration, or maximum climate damages allowable by
some future date) they want to leave future inhabitants of planet Earth (Portney and Weyant
1999; Arrow et al. 2013). BC IAMs use the discount rate to trade off the benefits and costs of
climate polices, while the DP IAMs typically use the discount rate to pick cost-effective miti-
gation technologies and to make ex post comparisons of any valued climate damages with those
mitigation costs.

12
See Adler (2012) for more on this type of critique of BCA, as well as a well-developed argument for using much
more nuanced social welfare functions (SWFs) instead.
Some Contributions of Integrated Assessment Models of Global Climate Change 127

There are two conflicting views in the economics literature concerning how to think about
discounting (Nordhaus 2013). The first is the prescriptive view taken by Cline (1992) and Stern
(2007), which leads to the conclusion that any positive pure rate of time preference is unethical.
The second is the descriptive view, advocated by Lind (1995) and Nordhaus (1994a), which
assumes that investments to slow climate change must compete with investments in other areas.
To summarize, intertemporal equity is extremely important in determining the appropriate
rate of implementation of policies designed to reduce carbon emissions—explicitly in the BC
IAMs and implicitly in the DP IAMs. Low discount rates generally make rapid implementation
of such policies much more urgent than high discount rates because damages are projected to
grow steadily over time at a much more rapid rate than mitigation costs.

Projections of Baseline Drivers

As mentioned earlier, the optimal levels of GHG emissions in BC IAMs depend on the GHG
emissions baseline used. Historically the main drivers of baseline GHG emissions scenarios have
been projections of future population by region, economic growth and its composition by
sector and region, and future fuel prices, as well as technology costs and performance. Now that
many countries have adopted climate policies in one form or another, an additional challenge is
projecting the effect of these policies and measures on future emissions, which would serve as a
baseline from which to examine the impacts of any new targets or new policies and measures we
might want to evaluate. A further major challenge for both the mitigation and impacts/adap-
tation analyses is projecting the impacts of technological change on costs and benefits over 50 to
100 years or more.13
This same baseline dependence also holds for the mitigation and climate change impacts
calculated by the DP IAMs, thus making it desirable to consider a range of baselines when
making assessments. Climate change impact assessments can be especially sensitive to the
baseline assumed, especially in cases where “tipping points” may be reached; the closer the
tipping point is to the baseline the more likely it is to be triggered.

Capturing Interactions Between Impact Sectors and Feedbacks to the Climate

System
A formidable challenge for all IAMs is capturing important interactions between impacted
sectors and regions, and feedbacks that can occur between the impacted sectors and atmos-
pheric concentrations of GHGs, temperatures, and precipitation. Other than the reduction in
global economic output resulting from climate change damages and climate mitigation, the BC
IAMs are too aggregated to study these interactions and feedbacks. The DP IAMs have made
slow, but steady, progress on these effects over the last 20 years, with great emphasis increasingly
being placed on interactions among the global energy, water, land, and food systems. According
to the IPCC (2014a, chap. 19), many of the most severe impacts of climate changes in specific
sectors in particular regions will result from interactions between these particular regions and
sectors and other regions or impact sectors. These interactions can occur at the physical level;
13
See Clarke et al. (2014) for a review of the sensitivity of mitigation cost projections to technology assumptions.
Neumann and Strzepek (2015) and Li, Mullan, and Helgeson (2014) discuss impacts and adaptation technology
dependencies.
128 J. Weyant

for example, climate change can cause a particular region to become hotter and dryer, thus
increasing competition for limited water between agricultural, power plant cooling, and house-
hold uses (Taheripour, Hertel, and Liu 2013). These shortages can result in abrupt economic
adjustments, with both food and energy becoming more expensive in the region, which could
also affect other economic sectors in the region (Baldos and Hertel 2014). Trade and transfers of
water, food, and energy from other regions could ameliorate impacts in regions that are
experiencing shortages, but would also lead to price increases in the regions where the trade
and transfers originate. A particularly pernicious result—that is already occurring in a number
of places—is increases in temperature and decreases in water in key growing regions. This has
led to more irrigation being fed largely by depleting groundwater resources (Grogan et al. 2015).
If the groundwater aquifers in these regions approach depletion, the cost of water will increase,
which will put incredible pressure on farms in regions that are already very vulnerable and
poverty stricken (Zaveri et al. 2016). Although capturing these types of interactions is not yet
standard in most DP IAMs (Reilly et al., 2012b), there has been a lot of research showing how
these interactions would likely occur in particular regions and sectors. For example,
Diffenbaugh et al. (2012) examine the effects drought and biofuel mandates can have on
both grain and energy price volatility.

Dealing with Uncertainty and Risk

The complexities of both human-induced climate change and the policies designed to address it
mean that there are vast uncertainties regarding key model inputs and parameters (e.g., baseline
rates of economic growth and technological change, the lag in the rate of heat transfer from the
atmosphere to the deep ocean) and important model outcomes (e.g., changes in projected
temperatures and precipitation amounts).14 There are a number of ways to address all of this
uncertainty, starting with sensitivity analysis on key parameters and model inputs of the type
discussed earlier.15 In addition, Monte Carlo simulations (in which input and parameter values
are selected from probability distributions) can be run through the models in order to produce
probability distributions over important model outputs—such as changes in temperatures and
sea levels—as well as total climate change damage and mitigation costs.
Although sensitivity analyses and Monte Carlo simulations are a good place to start, two
other crucial dimensions of the climate change problem should be included in any compre-
hensive attempt to inform climate policy decisions. First, decisions made today can be revisited
and modified at any point in the future as new information on climate change damages and
mitigation costs becomes available (see, e.g., Manne and Richels 1992; Ha Duong, Grubb, and
Hourcade 1997; Rozenberg et al. 2010). Thus decision making about climate change is one of
sequential decision making under uncertainty. This means that society can hedge against bad
climate outcomes by adopting mitigation policies that are more stringent than would be sug-
gested by expected damages and mitigation costs without going all the way to the mitigation
levels suggested by the worse possible damage and mitigation cost outcomes. The second crucial
dimension of climate change uncertainty that has yet to be systematically addressed by
14
See InterAcademy Council (2010) for a particularly noteworthy critique of the IPCC and, by extension, the
entire global climate change research community on this issue.
15
Clarke et al. (2014) summarize the results of numerous intermodel comparison studies, which include sen-
sitivity analyses of alternative model structures as well as selected model input assumptions.
Some Contributions of Integrated Assessment Models of Global Climate Change 129

researchers are assumptions about the decision makers’ attitudes toward risk. Much of the
application of modern decision theory targets decisions where the stakes are important, but not
sufficiently important to have a major influence on the individual’s overall level of wealth. Thus,
since at least Raiffa (1968) and Howard (1984), researchers have used exponential utility
functions with constant relative risk aversion to represent the preferences of individual decision
makers because this approach tends to provide good approximations when the outcomes of
decisions are not expected to have a major influence on the individual’s overall level of wealth.16
However, in the case of climate change, where the stakes can be extremely high and highly
correlated with other big societal decisions, and good experts may be hard to find, special care
must be exercised in interpreting model results. Current IAMs generally make one fairly homo-
geneous set of simplifying assumptions about risk attitudes and the implications of alternative
assumptions are generally not explored in any depth,17 which may lead to results that under-
state society’s degree of risk aversion towards bad outcomes and lead to less emissions mitiga-
tion than would be desirable.

Conclusions and Recommendations for Further

Development and Use of IAMs
This article has summarized some of the contributions of IAMs to the development of climate
policy and the setting of research priorities on global climate change. It has also identified some
of the key challenges that remain in developing and using these models. Given the challenges
involved in producing IAM results and the many uncertainties underlying those results, I
believe it is best to view these models as providing a good place to start in terms of basic
principles and rough numbers to use in developing short-term (say through the next 5 to 10
years) policies and research priorities, but a poor place to finish in the design of specific longer-
term global policies. Nevertheless, as these models become more refined and spatially explicit,
their results will provide more relevant information and more reliable guidance to policy
advisers and science program managers to help them make better decisions over the long
time period (decades to centuries) required to effectively address the climate change problem.
To some economists working on climate change policy, however, the challenges I have
discussed, as well as the complexities and uncertainties inherent in the climate problem and
its possible solutions, make the IAMs described here worse than useless. In particular, Pindyck
(2013, p. 870) concludes that IAMs “are of little or no value for evaluating alternative climate
change policies and estimating the SCC.” Although Nordhaus (2014, p. 301) agrees with many
of Pindyck’s individual criticisms, he reaches a different conclusion regarding the usefulness of
the current generation of climate policy BC IAMs:
Pindyck’s criticism of IAMs touches both empirical and conceptual issues.
Beginning with the empirical questions, he highlights (1) the social preference
function, particularly the discount rate, (2) the damage function, (3) the potential
16
The assumption of constant relative risk aversion implies that a decision on climate change can safely be
analyzed without looking at the correlations between that decision and all the other decisions a decision maker
has made for which the payoffs are not yet known with certainty.
17
Lempert (2015) and Toman (2015) provide guidance on what to do when these simplifying assumptions do
not hold.
130 J. Weyant

for catastrophic changes, and (4) the temperature sensitivity to greenhouse gas
increases. While Pindyck’s observations about the empirical weaknesses of IAMs
or calculations of the SCC are worthy of careful study, the conclusion that IAMs are
therefore useless fundamentally misconceives the enterprise.
Following Nordhaus (2014), I would argue that the main contribution of IAMs thus far,
including aggregate BC IAMs, has been to provide conceptual frameworks for developing
insights about highly complex, nonlinear, dynamic, and uncertain systems, even though
these models have not yet been able to provide high precision forecasts. Even at their current
stage of development, IAMs allow us to conduct “If. . ., then. . .” analyses of the impacts of
different factors that are internal or external to their operation. More importantly, these models
have provided important insights into many aspects of climate change policy. I review and
summarize some of these insights and contributions here and then present some recommen-
dations for further development and use of IAMs.
First, aggregate BC IAMs have improved our understanding of the importance of cost-
effectiveness in designing climate policies; the value of using market-based policy instruments
(like emissions taxes) versus command-and-control regulations; the value of information about
new technologies and improvements in the science of climate change, GHG mitigation, and
climate impacts; the importance of broad participation in mitigating carbon emissions; the
potential volatility in carbon prices that can result from systems that cap emissions; and the
costs of alternative approaches to reducing emissions. Perhaps the most valuable contribution
of systematic models such as aggregate BC IAMs is in highlighting the importance of critical
issues and uncertainties (such as discounting, risk assessment, and damages) in evaluating
alternative policies and regulations, and the ability they provide to incorporate new scientific
findings into the assessments in a timely and orderly fashion.
Second, projections from DP IAMs provide much more detail than BC IAMs by identifying
key energy technologies and impact sectors/regions, including energy, water, land, agriculture,
forestry, and ecosystem impacts in exceptionally hot and cold or wet and dry regions. In
addition, DP IAMs are increasingly being used to study interactions between mitigation and
impacts/adaptation by accounting for competition for water and land that could have large
joint impacts on the agricultural, forestry, energy, and ecosystem sectors as well as on the global
carbon cycle and the amount of solar energy that gets reflected back into deep space. Results at
this level can be extremely useful to international negotiators as well as national and subnational
decision makers because they represent impacts that can be directly measured, many of which
have already been observed (IPCC 2014a).
With these contributions and challenges of IAMs in mind, my first recommendation con-
cerning the further development and use of IAMs in the climate policy arena is to consider an
even more comprehensive set of sensitivity analyses than has typically been included thus far—
one that includes alternative treatments of concepts such as equity, attitudes toward risk, and
the amount of technological optimism. In this way, numbers and policies that are robust (see
Lempert 2015 for a working definition) across all these dimensions can be identified and used to
find solutions that are more stable and are characterized by less uncertainty.
Second, given all the uncertainties concerning the inputs to, structures of, and parameter
values included in IAMs, it is important to continue to develop decision support tools based
on—and supplementary to—conventional IAMs. For example, tools could include robust
Some Contributions of Integrated Assessment Models of Global Climate Change 131

planning methods that use extensive sensitivity analyses to identify decisions that lead to good
outcomes over a wide range of input assumptions or decision theory–based approaches that
focus on likely dependencies among model inputs and the sequential characteristics of climate
policy decisions where a decision can be taken today and revised periodically based on new
information.
Finally, although most of the major uncertainties confronting the models will not be resolved
for decades, society cannot afford to wait to make climate policy decisions until these uncer-
tainties are resolved and the BC or DP IAM calculations are further refined. Nevertheless, as
suggested in several assessments (e.g., Lempert 2015; Toman 2015), the analysis of these deci-
sions can be placed in a broader context that recognizes both that the risks associated with these
uncertainties can be quantified and managed and that the world includes many different re-
gional and sectoral decision makers with different perceptions of the severity of the climate
change problem, different stakes in climate outcomes, different preferences, different attitudes
toward equity, and different attitudes toward risk. This challenge is made ever more difficult by
the interactions between climate change and other major global issues such as air and water
quality, poverty, food security, national security, and human health. Welfare measured across
climate mitigation and impacts/adaptation around the globe is a large challenge, but even this
conception of welfare needs to be placed in a broader context. IAMs can provide very useful
information, but this information needs to be carefully interpreted and integrated with other
quantitative and qualitative inputs in the decision-making process. As astutely observed by one
reviewer of this article: “the models can be very useful, but not usually on their own.”

References
Ackerman, F., and E. A. Stanton. 2010. The Social productivity and climate change. Australian Journal
Cost of Carbon: A Report for the Economics for of Agricultural and Resource Economics 58:1–18.
Equity and the Environment Network. Cambridge, Bosetti, V., and J. Frankel. 2012. Politically feasible
MA: Stockholm Environment Institute. emissions targets to attain 460 ppm CO2 concentra-
———. 2012. Climate risks and carbon prices: tions. Review of Environmental Economics and
Revising the social cost of carbon. Economics: The Policy 6:86–109.
Open-Access, Open-Assessment E-Journal 6(2012- Box, G. E. P., and N. R. Draper. 1987. Empirical
10):1–25. http://dx.doi.org/10.5018/economics- Model Building and Response Surfaces. New York:
ejournal.ja.2012-10 John Wiley & Sons.
Adler, M. 2012. Well-Being and Fair Distribution: Blanc, É., K. Strzepek, and J. Reilly. (2014).
Beyond Cost-Benefit Analysis. New York: Oxford Modeling U.S. water resources under climate
University Press. change. Earth’s Future 2(4): 197–224.
Anthoff, D., and R. S. J. Tol. 2013. The uncertainty Boehlert, B., and K. Strzepek. 2013. Commentary
about the social cost of carbon: A decomposition XV: Competition for water for agriculture through
analysis using FUND. Climatic Change 2050. In UNCTAD Trade and Environment Review.
117(3):515–30. Geneva: United Nations, 82–85.
Arrow, K., M. Cropper, C. Gollier, B. Groom, G. Calvin, K., L. Clarke, V. Krey, G. J. Blanford, K.
Heal, R. Newell, W. Nordhaus, R. Pindyck, W. Jiang, M. Kainuma, E. Kriegler, G. Luderer, and P.
Pizer, P. Portney, T. Sterner, R. S. J. Tol, and M. R. Shukla. 2012. The role of Asia in mitigating cli-
Weitzman. 2013. Determining benefits and costs mate change: Results from the Asia modeling exer-
for future generations. Science 341:349–50. cise. Energy Economics 34(Suppl 3):S251–S260.
Baldos, U. L. C., and T. W. Hertel. 2014. Global Calvin, K., M. Wise, P. Kyle, P. Patel, L. Clarke,
food security in 2050: the role of agricultural and J. Edmonds. 2014. Trade-offs of different land
132 J. Weyant

and bioenergy policies on the path to achieving cli- Edenhofer, O., C. Carraro, J. Kohler, and M.
mate targets. Climatic Change 123(3–4):691–704. Grubb. 2006. Endogenous technological change
Chuwah, C., T. van Noije, D. P. van Vuuren, W. and the economics of atmospheric stabilisation. In
Hazeleger, A. Strunk, S. Deetman, A. M. Beltran, Special Issue 1 of The Energy Journal. Cleveland,
and J. van Vliet. 2013. Implications of alternative OH: International Association for Energy
assumptions regarding future air pollution control Economics.
in scenarios similar to the Representative Edmonds, J., H. Pitcher, N. Rosenberg, and T.
Concentration Pathways. Atmospheric Environment Wigley. 1994. Design for the Global Change
79:787–801. Assessment Model. Presented at the International
Clarke, L., J. Edmonds, V. Krey, R. Richels, S. Rose, Workshop on Integrative Assessment of
and M. Tavoni. 2009. International climate policy Mitigation, Impacts, and Adaptation to Climate
architectures: Overview of the EMF 22 Change, International Institute for Applied Systems
International Scenarios. Energy Economics Analysis, Laxenburg, Austria, October 13–15.
31:S64–S81. Edmonds, J., M. Wise, and C. MacCracken. 1994.
Clarke, L., K. Jiang, K. Akimoto, M. Babiker, G. Advanced energy technologies and climate change:
Blanford, K. Fisher-Vanden, J.-C. Hourcade, V. An analysis using the Global Change Assessment
Krey, E. Kriegler, A. Löschel, D. McCollum, S. Model (GCAM). Presented at the Air and Waste
Paltsev, S. Rose, P.R. Shukla, M. Tavoni, B.C.C. Management Meeting, April 6, Tempe, AZ.
van der Zwaan, and D.P. van Vuuren. 2014. Ermolieva, T., Y. Ermoliev, P. Havlik, A. Mosnier,
Assessing transformation pathways. In: Climate S. Fuss, D. Leclere, N. Khabarov, and M.
Change 2014: Mitigation of Climate Change. Obersteiner. 2014. Systems analysis of strategic de-
Contribution of Working Group III to the Fifth cisions for planning secure food, energy, water pro-
Assessment Report of the Intergovernmental Panel on vision based on stochastic GLOBIOM. Journal of
Climate Change. Cambridge: Cambridge University Cybernetics and Systems Analysis 1:125–33.
Press.
Fawcett, A. A., L. E. Clarke, and J. P. Weyant.
Clemen, R., and R. Winkler. 1986. Combining eco- 2014. The EMF 24 study on U.S. technology and
nomic forecasts. Journal of Business & Economic climate policy strategies. Special Issue 1 of the
Statistics 4(1):39–46. Energy Journal. Cleveland, OH: International
Cline, W. 1992. The Economics of Global Warming. Association for Energy Economics.
Washington, DC: Institute of International Greenstone, M., E. Kopits, and A. Wolverton.
Economics. 2013. Developing a social cost of carbon for US
Daioglou, V., B. Wicke, A. P. C. Faaij, and D. P. regulatory analysis: A methodology and interpret-
van Vuuren. 2014. Competing uses of biomass for ation. Review of Environmental Economics and
energy and chemicals: implications for long-term Policy 7(1):23–46.
global CO2 mitigation potential. GCB Bioenergy
Grogan, D. S., F. Zhang, A. Prusevich, R. B.
7(6):1321–34.
Lammers, D. Wisser, S. Glidden, C. Li, and S.
de la Chesnaye, F., and J. P. Weyant (eds.). 2006. Frolking. 2015. Quantifying the link between crop
Multi-greenhouse gas mitigation and climate production and mined groundwater irrigation in
policy. Special Issue 3 of The Energy Journal. China. Science of the Total Environment
Cleveland, OH: International Association for 511:161–75.
Energy Economics.
Ha-Duong, M., M. J. Grubb, and J.-C. Hourcade.
Diaz, D. B. 2016. Estimating global damages from 1997. Influence of socioeconomic inertia and un-
sea level rise with the Coastal Impact and certainty on optimal CO2-emission abatement.
Adaptation Model (CIAM). Climatic Change Nature 390:270–73.
137(1):143–56.
Herrero M., P. Havlik, and M. Obersteiner. 2013.
Diffenbaugh, N. S., T. W. Hertel, M. Scherer, and Biomass use, production, feed efficiencies, and
M. Verma. 2012. Response of corn markets to cli-
greenhouse gas emissions from global livestock sys-
mate volatility under alternative energy futures.
tems. Proceedings of the National Academy of
Nature Climate Change 2:514–18.
Some Contributions of Integrated Assessment Models of Global Climate Change 133

Sciences of the United States of America Knopf, B., E. De Cian, I. Keppo, D. P. van Vuuren,
110(52):20888–93. and J. P. Weyant. 2013. The EMF 28 study on
Hinkel, J., D. van Vuuren, R. J. Nicholls, and R. J. scenarios for transforming the European energy
T. Klein. 2013. The effects of adaptation and miti- system. Climate Change Economics 4(1):.
gation on coastal flood impacts during the 21st Kraucunas, I. P., L. E. Clarke, J. Dirks, J.
century. An application of the DIVA and IMAGE Hathaway, M. Hejazi, K. Hibbard, M. Huang, C.
models. Climatic Change 117(4):783–94. Jin, M. Kintner-Meyer, K. Kleese van Dam, R.
Hope, C. 2011. The social cost of CO2 from the Leung, H.-Y. Li, R. Moss, M. Peterson, J. Rice, M.
PAGE09 model. Working paper, Judge Business Scott, A. Thomson, N. Voisin, and T. O. West.
School, Cambridge University. 2015. Investigating the nexus of climate, energy,
Howard, R. A. 1984. Risk preference (paper 34). water, and land at decision-relevant scales: the plat-
In: The Principles and Applications of Decision form for regional integrated modeling and analysis
Analysis, vol. II, ed. R. A. Howard and J. E. (PRIMA). Climatic Change 129(3–4):573–88.
Matheson. Menlo Park, CA: Strategic Decisions Krey, V., G. Luderer, L. Clarke, and E. Kriegler.
Group. 2014. Getting from here to there—energy technol-
InterAcademy Council. 2010. Climate Change ogy transformation pathways in the EMF27 scen-
Assessments. Review of the Processes and arios. Climatic Change 123:369–82.
Procedures of the Intergovernmental Panel on Kriegler, E., O. Edenhofer, L. Reuster, G. Luderer,
Climate Change. Committee to Review and D. Klein. 2013a. Is atmospheric carbon dioxide
Intergovernmental Panel on Climate Change. removal a game changer for climate change mitiga-
Amsterdam: InterAcademy Council. tion? Climatic Change 118:45–57
IPCC. 2013. Climate Change 2013: The Physical Kriegler E., N. Petermann, V. Krey, V. J. Schwanitz,
Science Basis. Contribution of Working Group I to G. Luderer, S. Ashina, V. Bosetti, J. Eom, A.
the Intergovernmental Panel on Climate Change Kitous, A. Méjean, L. Paroussos, F. Sano, H.
Fifth Assessment Report. Cambridge: Cambridge Turton, C. Wilson, and D. P. van Vuuren. 2014a.
University Press. Diagnostic indicators for integrated assessment
———. 2014a. Climate Change 2014: Impacts, models of climate policies. Technological Forecasting
Adaptation, and Vulnerability. Contribution of and Social Change 90:45–61.
Working Group II to the Fifth Assessment Report of Kriegler, E., M. Tavoni, T. Aboumahboub, G.
the Intergovernmental Panel on Climate Change. Luderer, K. Calvin, G. Demaere, V. Krey, K. Riahi,
Cambridge: Cambridge University Press. H. Rösler, M. Schaeffer, and D. P. Van Vuuren.
———. 2014b. Climate Change 2014: Mitigation of 2013b. What does the 2  C target imply for a global
Climate Change. Contribution of Working Group III climate agreement in 2020? The LIMITS study on
to the Fifth Assessment Report of the Durban Platform scenarios. Climate Change
Intergovernmental Panel on Climate Change. Economics 4(4):1–30.
Cambridge: Cambridge University Press. Kriegler E., J. Weyant, G. J. Blanford, V. Krey, L.
IWG. 2010. Technical support document: Social cost Clarke, J. Edmonds, A. Fawcett, G. Luderer, K.
of carbon for regulatory impact analysis under Riahi, R. Richels, S. K. Rose, M. Tavoni, and D. P.
Executive Order 12866. Washington, DC: US van Vuuren. 2014b. The role of technology for
government. achieving climate policy objectives: Overview of the
———. 2013. Technical support document: EMF 27 study on global technology and climate
Technical update of the social cost of carbon for regu- policy strategies. Climatic Change 123:353–67.
latory impact analysis under Executive Order 12866. Lempert, R. J. 2015. Embedding (some) benefit-
Washington, DC: US government. cost concepts into decision support processes with
Johnson, L., and C. Hope. 2012. The social cost of deep uncertainty. Journal of Benefit–Cost Analysis
carbon in U.S. regulatory impact analyses: An 5(3):487–514.
introduction and critique. Journal of Environmental Lenton, T., H. Held, E. Kriegler, J. W. Hall, W.
Studies and Sciences 2(3):205–21. Lucht, S. Rahmstorf, and J. Schellenhuber. 2008.
134 J. Weyant

Tipping elements in the Earth’s climate system. pollution and carbon emissions control:
Nature 105(6):1786–93. comparing China and the U.S. Energy Economics
Li, J., M. Mullan, and J. Helgeson. 2015. Improving 56:422–31.
the practice of economic analysis of climate change National Research Council, Committee on
adaptation. Journal of Benefit–Cost Analysis Abrupt Climate Change. 2002. Abrupt Climate
5(3):445–67. Change: Inevitable Surprises. Washington, DC:
Lind, R. 1995. Intergenerational equity, discount- National Academies Press.
ing, and the role of cost benefit analysis in evaluat- Neumann, J. E., and K. Strzepek. 2015. State of the
ing global climate policy. Energy Policy 23:379–89. literature on the economic impacts of climate
Manne, A. S., and R. G. Richels. 1992. Buying change in the United States. Journal of Benefit–Cost
Greenhouse Insurance. Cambridge, MA: MIT Press. Analysis 5(3):411–43.
Meinshausen, M., S. Raper, and T. Wigley. 2011. Nordhaus, W. D. 1994a. Expert opinion on cli-
Emulating coupled atmosphere-ocean and carbon matic change. American Scientist 82:45–51.
cycle models with a simpler model, MAGICC6 – ———. 2007. A review of the Stern Review on the
Part 1: Model description and calibration. economics of climate change. Journal of Economic
Atmospheric Chemistry and Physics 11(4):1417–56. Literature 45(3):686–702.
Meller, L., D. van Vuuren, and M. Cabeza. 2013. ———. 2013. Integrated economic and climate
Quantifying biodiversity impacts of climate change modeling. In Handbook of Computable General
and bioenergy: the role of integrated global scen- Equilibrium Modeling, ed. Peter B. Dixon and
arios. Regional Environmental Change Dale W. Jorgenson. Oxford: North-Holland.
15(6):961–71. ———. 2014. Estimates of the social cost of
Metcalf, G. E., and J. Stock. 2017. Integrated carbon: concepts and results from the DICE-2013R
assessment models and the social cost of carbon: a model and alternative approaches. Journal of the
review and assessment of U.S. experience. Review of Association of Environmental and Resource
Environmental Economics and Policy 11(1):80–99. Economists 1(1/2):273–312.
Morgan, G., and D. Keith. 1995. Subjective judg- Nordhaus, William D. 1989. The economics of the
ments by climate experts. Environmental Science & greenhouse effect. Paper presented to the June
Technology 29:468A–76A. 1989 meeting of the International Energy
Morris, P. A. 1977. Combining expert judgments: Workshop, Laxenburg, Austria.
A Bayesian approach. Management Science ———. 1991. To slow or not to slow: The eco-
23(7):679–93. nomics of the greenhouse effect. Economic Journal
———. 1983. An axiomatic approach to expert 101:920–37.
resolution. Management Science 29(1):24–32. ———. 1994b. Managing the Global Commons:
Mosnier A, M. Obersteiner, P. Havlik , E. Schmid, The Economics of Climate Change. Cambridge, MA:
N. Khabarov, M. Westphal, H. Valin, and S. Frank. MIT Press.
2014. Global food markets, trade and the cost of O’Neill, Brian C., Elmar Kriegler, Keywan Riahi,
climate change adaptation. Food Security Kristie L. Ebi, Stephane Hallegatte, Timothy R.
6(1):29–44. Carter, Ritu Mathur, and Detlef P. van Vuuren.
Moss, R. H., J. A. Edmonds, K. A. Hibbard, M. R. 2014. A new scenario framework for climate
Manning, S. K. Rose, D. P. van Vuuren, T. R. change research: the concept of shared
Carter, S. Emori, M. Kainuma, T. Kram, G. A. socioeconomic pathways. Climatic Change
Meehl, J. F. B. Mitchell, N. Nakicenovic, K Riahi, S 122:387–400.
J. Smith, R. J. Stouffer, A. M. Thomson, J. P. Pindyck, R. 2013. Climate change policy: What do
Weyant, and T. W. Wilbanks. 2010. The next gen- the models tell us? Journal of Economic Literature
eration of scenarios for climate change research 51(3):860–72
and assessment. Nature 463:747–56. Pindyck, R. S. 2017. The use and misuse of models
Nam, K.-M., C. J. Waugh, S. Paltsev, J. M. Reilly, for climate policy. Review of Environmental
and V. J. Karplus. 2016. Synergy between Economics and Policy 11(1):100–114.
Some Contributions of Integrated Assessment Models of Global Climate Change 135

Portney, P., and J. Weyant (eds.). 1999. Discounting Bosetti, J. Eom, D. Gernaat, T. Masui, J. Rogelj, J.
and Intergenerational Equity. Washington, DC: Strefler, L. Drouet, V. Krey, G. Luderer, M.
Resources for the Future Press. Harmsen, K. Takahashi., L. Baumstark, J.
Pachauri, S., B. J. Van Ruijven, and N. Doelman, M. Kainuma, Z. Klimont, G. Marangoni,
Nakicenovic. 2013. Pathways to achieve universal H. Lotze-Campen, M. Obersteiner, A. Tabeau, M.
household access to modern energy by 2030. Tavoni. The Shared Socioeconomic Pathways and
Environmental Research Letters 8(2):024015. their energy, land use, and greenhouse gas emis-
Raiffa, H. 1968. Decision Analysis: Introductory sions implications: An overview (2016) Global
Lectures on Choices under Uncertainty. New York: Environmental Change, Article in Press. DOI:
Addison-Wesley. 10.1016/j.gloenvcha.2016.05.009
Reilly, J., S. Paltsev, B. Felzer, X. Wang, D. Rose, S., E. Kriegler, R. Bibas, K. Calvin, A. Popp,
Kicklighter, J. Melillo, R. Prinn, M. Sarofim, and C. D. P. van Vuuren, and J. Weyant. 2014a. Bioenergy
Wang. 2007. Global economic effects of changes in in energy transformation and climate management.
crops, pasture, and forests due to changing climate, Climatic Change 123(3–4):477–93.
carbon dioxide, and ozone. Energy Policy Rose, S., D. Turner, G. Blanford, J. Bistline, F. de la
35(11):5370–83. Chesnaye, and T. Wilson. 2014b. Understanding
Reilly, J., and S. Paltsev. 2009. Biomass energy and the Social Cost of Carbon: A Technical Assessment.
competition for land. In Economic Analysis of Report 3002004657. Palo Alto, CA: Electric Power
Land Use in Global Climate Change Policy, ed. T. Research Institute.
W. Hertel , S. K. Rose, and R. S. J. Tol, 184–207. Rotmans, Jan. 1990. Image: An Integrated Model to
New York: Routledge. Assess the Greenhouse Effect. Dordrecht: Kluwer
Reilly, J., J. Melillo, Y. Cai, D. Kicklighter, A. Academic.
Gurgel, S. Paltsev, T. Cronin, A. Sokolov, and A. Rozenberg, Julie, Stephane Hallegatte, Adrien
Schlosser. 2012a. Using land to mitigate climate Vogt-Schilb, Olivier Sassi, Celine Guivarch, Henri
change: Hitting the target, recognizing the trade- Waisman, and Jean-Charles Hourcade. 2010.
offs. Environmental Science and Technology Climate policies as a hedge against the uncertainty
46(11):5672–79. on future oil supply. Climatic Change
Reilly, J., S. Paltsev, K. Strzepek, N. Selin, Y. Cai, 101(3–4):663–68.
H.-M. Nam, E. Monier, S. Dutkiewitz, J. Scott, M. Schaeffer, M., L. Gohar, E. Kriegler, J. Lowe, K.
Webster, and S. Sokolov. 2012b. Valuing climate Riahi, and D. van Vuuren. 2013. Mid- and long-
impacts in integrated assessment models: the MIT term climate projections for fragmented and
IGSM. Climatic Change 117:561–73. delayed-action scenarios. Technological Forecasting
Riahi, K., E. Kriegler, N. Johnson, C. Bertram, M. and Social Change 90(Pt A):257–68.
den Elzen, J. Eom, M. Schaeffer, J. Edmonds, M. Schellnhuber, H. J., K. Frieler, and P. Kabat. 2014.
Isaac, V. Krey, T. Longden, G. Luderer, A. Méjean, Global climate impacts: A cross-sector, multi-
D. L. McCollum, S. Mima, H. Turton, D. P. van model assessment special feature. Proceedings of the
Vuuren, K. Wada, V. Bosetti, and P. Capros. 2015. National Academy of Sciences of the United States of
Locked into Copenhagen pledges—Implications of America 111(9):3225–97.
short-term emission targets for the cost and feasi- Schlosser, C. A., K. Strzepek, X. Gao, C. Fant, É.
bility of long-term climate goals. Technological Blanc, S. Paltsev, H. Jacoby, J. Reilly, and A.
Forecasting and Social Change 90(Pt A):8–23. Gueneau. 2014. The future of global water stress:
Riahi, K., D. van Vuuren, E. Kriegler, J. Edmonds, An integrated assessment. Earth’s Future
B. O’Neill, S. Fujimori, N. Bauer, K. Calvin, R. 2(8):341–61.
Dellink, O. Fricko, W. Lutz, A. Popp, J. Cuaresma, Stehfest, E., D. van Vuuren, T. Kram, and L.
S, KC, M. Leimbach, L. Jiang, T. Kram, S. Rao, J. Bouwman. 2014. Integrated Assessment of Global
Emmerling, K. Ebi, T. Hasegawa, P. Havlik, F. Environmental Change with IMAGE 3.0; Model
Humpenöder, L. Da Silva, S. Smith, E. Stehfest, V. Description and Policy Applications. Bilthoven, The
136 J. Weyant

Netherlands: PBL Netherlands Environmental Weitzman, M. 2009. On modeling and interpreting

Agency. the economics of catastrophic climate change.
Stern, N. 2007. The Economics of Climate Change: Review of Economics and Statistics 91(1):1–19.
The Stern Review. Cambridge: Cambridge ———. 2013. Tail-hedge discounting and the
University Press. social cost of carbon. Journal of Economic Literature
———. 2013. The structure of economic modeling 51(3):873–82.
of the potential impacts of climate change: Grafting Weyant, J. P., O. Davidson, H. Dowlatabadi, J.
gross underestimation of risk onto already narrow Edmonds, M. Grubb, E. A. Parson, R. Richels, J.
science models. Journal of Economic Literature Rotmans, P.R. Shukla, R. S. J. Tol, W. R. Cline,
51(3):838–59. and S. Fankhauser. 1996. Integrated assessment of
Strzepek, K., G. Yohe, J. Neumann, and B. climate change: An overview and comparison of
Boehlert. 2010. Characterizing changes in approaches and results. In Climate Change 1995:
drought risk for the United States from climate Economic and Social Dimensions – Contribution of
change. Environmental Research Letters Working Group III to the Second Assessment Report
5(4):044012. of the Intergovernmental Panel on Climate Change,
chap. 10. Cambridge: Cambridge University Press.
Sussman, Fran, Christopher P. Weaver, and Anne
Grambsch. 2015. Challenges in applying the para- Weyant, J. P. (ed.). 1999. The costs of the Kyoto
digm of welfare economics to climate change. protocol: a multi-model evaluation. Energy Journal
Journal of Benefit–Cost Analysis 5(3):347–76. (Special Issue).

Taheripour, F., T. W. Hertel, and J. Liu. 2013. The Weyant, J. P. (ed.). 2004. EMF 19: alternative tech-
role of irrigation in determining the global land use nology strategies for climate change policy. Energy
Economics 26(4):501–755.
impacts of biofuels. Energy, Sustainability and
Society 3(4):1–18. Weyant, J. P. 2008. A critique of the Stern Review’s
mitigation cost analyses and integrated assessment.
Tai, A. P. K., M. Val Martin, and C. L. Heald.
Review of Environmental Economics and Policy
2014. Threat to future global food security from
2:77–93.
climate change and ozone air pollution. Nature
Climate Change 4:817–21. Weyant, J. P. 2015. Integrated assessments of cli-
mate change: State of the literature. Journal of
Tol, R. 2008. The social cost of carbon: Trends,
Benefit–Cost Analysis 5(3):377–409.
outliers and catastrophes, economics. Economics:
The Open-Access, Open-Assessment E-Journal Wolverton, A., E. Kopits, C. Moore, A. Marten, S.
2(25):1–22. Newbold, and C. Griffiths. 2012. The social cost of
carbon: Valuing carbon reductions in policy ana-
———. 2009. The economic impact of climate
lysis. In Fiscal Policy to Mitigate Climate Change:
change. Journal of Economic Perspectives
A Guide for Policymakers, ed. I. Parry , M. Keen,
23(2):29–51.
and R. de Mooij. Washington, DC: International
Toman, M. 2015. The need for multiple types of Monetary Fund.
information to inform climate change assessment.
Zaveri, E., D. S. Grogan, K. Fisher-Vanden, S.
Journal of Benefit–Cost Analysis 5(3):469–85.
Frolking, R. B. Lammers, D. H. Wrenn, A.
van Vuuren, D., J. Lowe, E. Stehfest, L. Gohar, A. Prusevich, and R. E. Nicholas. 2016. Invisible
Hof, C. Hope, R. Warren, M. Meinshausen, and G. water, visible impact: Groundwater use and Indian
K. Plattner. 2011. How well do integrated assess- agriculture under climate change. Environmental
ment models simulate climate change? Climatic Research Letters 11:084005.
Change 104:255–85
Zickfeld, K., M. Eby, A. J. Weaver, K. Alexander, E.
Washington, W., and C. Parkinson. 2005. An Crespin, N. R. Edwards, A. V. Eliseev, G. Feulner,
Introduction to Three-Dimensional Climate T. Fichefet, C. E. Forest, P. Friedlingstein, H.
Modeling, 2nd ed. Sausalito, CA: University Goosse, P. B. Holden, F. Joos, M. Kawamiya, D.
Scientific Books. Kicklighter, H. Kienert, K. Matsumoto, I. I.
Some Contributions of Integrated Assessment Models of Global Climate Change 137

Mokhov, E. Monier, S. M. Olsen, J. O. P. Pedersen, K. S. Tokos, M. Yoshimori, N. Zheng, and Z.

M. Perrette, G. Philippon-Berthier, A. Ridgwell, A. Zhao. 2013. Long-term climate change commit-
Schlosser, T. Schneider Von Deimling, G. Shaffer, ment and reversibility: An EMIC intercomparison.
A. Sokolov, R. Spahni, M. Steinacher, K. Tachiiri, Journal of Climate 26(16):5782–809.