NGFS Climate Scenarios

Technical Documentation

V4.2

November 2023
Contents

Acknowledgements 3
Executive Summary 5

Introduction 6

Module 1: High-level overview 7

1. Introduction 7
2. NGFS scenario narratives 16
3. NGFS modelling approach 26
Box: MAGICC: A reduced complexity Earth system model 41
4. Main results of the NGFS scenarios 45
5. Phase III vs. Phase IV: what is new in the main results of NGFS scenarios? 57
Box: Towards a more disorderly climate transition 63

Module 2: IAM – REMIND-MAgPIE 66

1. Non-technical summary 66
2. Overview of model scope and methods 69
3. Key model inputs 81

Module 3: IAM – MESSAGE-GLOBIOM 85

1. Non-technical summary 85
2. Overview of model scope and methods 86
3. Key model inputs 91
4. Key model outputs 105
5. What is new in the NGFS Phase IV scenarios? 105

Module 4: IAM – GCAM 107

1. Non-technical summary 107
2. Overview of model scope and methods 108
3. Key model inputs 112
4. Key model outputs 120
5. What is new in GCAM modeling for NGFS Phase IV scenarios? 123

Module 5: Chronic physical risks 124

1. Non-technical summary 124
2. Macro-economic damage estimates 126
3. Chronic damages in post-processing and sources of uncertainty 128
4. Temperature downscaling 128
5. Scenarios with integrated transition and physical risks 129

Module 6: Acute physical risks 133

1. Non-technical summary 133
2. Differences relative to Phase III 133
3. Modelling of acute physical risk hazards 135
4. Modelling of acute risk macroeconomic effects (NiGEM) 152
5. Stochastic Implementation 155

Module 7: Country-level downscaling 160

1. Non-technical summary 160
2. Overview of method 161
3. Key inputs 163
4. Key outputs 170
5. What is new in the 2023 edition? 172

Module 8: NiGEM 174

1. Non-technical summary 174
2. Introduction to NiGEM 174
3. NiGEM integration into the NGFS scenarios 179
4. NiGEM NGFS output 183
5. NiGEM Technical references 185
6. IAM input variables 188

Appendix 192

Bibliography 231
Acknowledgements

The NGFS Scenarios were produced by NGFS Workstream on Scenario Design and Analysis in partnership with
an academic consortium from the Potsdam Institute for Climate Impact Research (PIK), International Institute
for Applied Systems Analysis (IIASA), University of Maryland (UMD), Climate Analytics (CA), Swiss Federal
Institute of Technology in Zurich (ETHZ), and National Institute of Economic and Social Research (NIESR). This
work was made possible by grants from Bloomberg Philanthropies and ClimateWorks Foundation.

The NGFS thanks the modellers of the academic consortium:

Oliver Richters1, Elmar Kriegler1, Alaa Al Khourdajie3, Jacob Anz2, Christoph Bertram4,1, David N. Bresch5,
Alessio Ciullo5, Ed Cornforth6, Ryna Cui4, Jae Edmonds4,7, Sophie Fuchs1, Philip Hackstock3, Ian Hurst6, Jarmo
Kikstra3, David Klein1, Kai Kornhuber2, Chahan M. Kropf5, Quentin Lejeune2, Jared Lewis3,8,9, Iana Liadze7,
Rahel Mandaroux1, Malte Meinshausen8,9, Jihoon Min3, Zebedee Nicholls3,8,9, Franziska Piontek1, Patricia
Sanchez Juanino6, Inga Sauer1, Carl-Friedrich Schleussner2, Jessie Ruth Schleypen2, Niklas Schwind2, Fabio
Sferra3, Miodrag Stevanović1, Bas van Ruijven3, Pascal Weigmann1, Michael Westphal4,7, Alicia Zhao4, Anne
Zimmer2, Matthew Zwerling4.

1. Potsdam Institute for Climate Impact Research (PIK), member of the Leibnitz Association, Potsdam,
Germany
2. Climate Analytics, Berlin, Germany
3. International Institute of Applied System Analysis (IIASA), Laxenburg, Austria
4. Center for Global Sustainability, School of Public Policy, University of Maryland, College Park, Maryland,
United States of America
5. Institute for Environmental Decisions, ETH Zurich, Zurich, Switzerland
6. National Institute for Economic and Social Research (NIESR), London, United Kingdom
7. Pacific Northwest National Laboratory (PNNL), Richland, State of Washington, United States of America
8. Climate Resource, Victoria, Australia
9. University of Melbourne, Victoria, Australia

3
This document was prepared under the auspice of the NGFS Workstream on Scenario Design and Analysis,
chaired by Livio Stracca (ECB).

Special thanks are given to the drafters and data reviewers:

Ruben Veiga (Banco de España), Rie Asakura (Financial Services Agency), Roman Marton (Oesterreichische
Nationalbank), Pietro Cova, Valentina Michelangeli, Maria Alessia Aiello (Banca d'Italia), Ed Byers (IIASA),
Ricardo Guerra Marques, Nuno Gonçalo Ribeiro (Banco de Portugal), Miroslav Petkov (Bank of International
Settlements), Hossein Jebeli (Bank of Canada), Mengmeng Miao (People’s Bank of China), Fernando Linardi
(Banco Central do Brasil), Robert Vermeulen (De Nederlandsche Bank), Gaya Aiche, Paul Champey, Valérie
Chouard, Li Savelin , Clément Payerols (Banque de France), Azusa Takeyama, Tomomi Naka (Bank of Japan),
Jihoon Kim (Bank of Korea) Laura Gutierrez (Banco de México), Michaela Dolk (World Bank), Daniel Ramos-
Garcia (European Investment Bank), George Krivorotov, Ding Du (Office of the Comptroller of the Currency),
Megha Arora (OSFI), Senne Aerts, Claudia Albers, Simone Boldrini, Paula Gonzalez Escribano, Yana Kostiuk,
Clemens Maria Lehofer, Mario Morelli, Laura Minuzi Nowzohour, Laura Parisi, Martina Spaggiari (European
Central Bank).

4
Executive Summary

Established in 2017, the Network for Greening the Financial System (NGFS) today represents a major hub for
the promotion of analytical work and best practices in the field of green finance. Currently (June 2023), the
NGFS consists of 127 central banks and supervisors (and 20 observers) from across five continents committed
to sharing best practices, contributing to the development of climate- and environment-related risk
management in the financial sector and mobilising mainstream finance to support the transition toward a
sustainable economy.1

One of the key initiatives of the NGFS is the development of climate-related scenarios that can be used by
financial institutions to assess and manage climate-related risks. These scenarios are intended to be forward-
looking and consider various climate-related factors, as well as policy and technology developments.
Hypothetical future pathways of climate change are used for analysing and assessing the potential impacts and
risks associated with different climate outcomes. The scenarios are not intended to predict the exact future
climate but rather provide a set of plausible pathways that can help policymakers, researchers, financial
institutions, and private sector businesses explore impacts and evaluate adaptation and mitigation strategies
in the face of climate change.

The NGFS climate scenarios have been developed in partnership with a consortium of academics from the
Potsdam Institute for Climate Impact Research (PIK), International Institute for Applied Systems Analysis
(IIASA), University of Maryland (UMD), Climate Analytics (CA) and the National Institute of Economic and
Social Research (NIESR). This work was made possible by grants from Bloomberg Philanthropies and
ClimateWorks Foundation.

1 See https://www.ngfs.net/en/about-us/membership

5
Introduction

This document provides technical information on the NGFS climate scenarios and the underlying
modelling infrastructure. It includes updated technical information from previously published material 2, and
expands the scope and the level of detail of the information provided to highlight key modelling assumptions
and comparisons between models, scenarios, and vintages for key variables. As accompanying material to the
NGFS scenarios, this document aims to answer conceptual and technical questions for a wide range of
stakeholders, from scenario users interested in performing analyses on the datasets themselves to interested
readers who would like to better understand the NGFS scenarios.

Following a layered structure, this document has been designed to target readers with different levels of
technical expertise. On the one hand, readers interested in gaining high-level information about the NGFS
scenarios will benefit from a comprehensive high-level overview, non-technical summaries prefacing each
section, as well as explainer boxes included throughout the document to provide relevant background
information. On the other hand, readers with advanced technical knowledge who are interested in detail will
find the extensive description of the NGFS modelling framework useful, with the specifics of each model being
presented in separate chapters and technical insights being highlighted in thematic boxes.

The remainder of this document is organised into eight modules. Module 1 provides a high-level overview
of the NGFS scenarios, their rationale, and their broader context, and presents the fourth vintage of NGFS
scenarios, including main characteristics and improvements. The following modules describe in detail the NGFS
modelling framework and methodology to generate the NGFS scenarios. Modules 2, 3 and 4 outline the three
Integrated Assessment Models (IAMs) used to generate the transition pathways for the NGFS scenarios:
REMIND-MAgPIE, MESSAGE-GLOBIOM, and the GCAM models respectively. Modules 5 and 6 describe the
modelling for physical risk and its relationship with transition policies, with focus on acute and chronic physical
risk respectively. Module 7 discusses the downscaling methodology applied to produce country-level results.
Module 8 covers the National Institute Global Econometric Model (NiGEM) and describes how this macro-
financial model has been specifically modified for the purpose of producing the NGFS scenarios to understand
the consequences of transition and physical risk on the key macro-financial fundamentals.

2 The previous version of the Technical Documentation for NGFS Phase III scenarios, published in September 2022, can
be found here.

6
Module 1: High-level overview

The NGFS scenarios have been developed to provide a common starting point for analysing climate
risks to the economy and financial system.

Key messages

 The NGFS scenarios have been created as a tool to shed light on potential future risks, and to
prepare the financial system for the shocks that may arise. Importantly, the NGFS scenarios are
not forecasts. Instead, they aim at exploring the bookends of plausible futures (neither the most
probable nor desirable) for financial risk assessment.

 The NGFS scenarios explore a range of plausible outcomes. They provide a common language for
how climate change (physical risk) and climate policy and technology trends (transition risk) could
evolve in different futures.

 The NGFS scenarios present unique features that make them particularly suitable for a wide
range of applications. They provide a common starting point for climate risk assessment, they
produce internally consistent results applicable at the global level that combine transition, physical
and macro-financial risks, and they are freely accessible through a public online platform.

1. Introduction
Since 2018, an increasing number of central banks and supervisors around the world have joined forces
in the Network for Greening the Financial System (NGFS) to help build a common understanding of
how climate change affects our economies and financial systems. While governments and legislators
are primarily responsible for the implementation of climate policies, central banks and supervisors can also
play an important role in addressing climate change within their mandates. In addition, in line with their
objectives and functions, central banks and supervisors need to be able to identify climate-related risks and
quantify their impact via rigorous analysis. The NGFS has thus developed, together with leading academic
climate institutions3, a common picture of what our economies might look like under different assumptions
in terms of transition policies and physical risks. These are called “climate scenarios”.

The NGFS climate scenarios4 have been created as a tool to shed light on potential future risks, and to
prepare the financial system for the shocks that may arise. They answer crucial questions like “what can
happen?” or “what should happen?” to enable a common understanding of how climate change and
climate mitigation can impact our economies in the long run (until 2100). Since its first vintage, published
in 2020, the NGFS scenarios have offered a useful guide to climate risks, as they combine the analysis of
transition, physical and macro-financial risks to reveal the long-term trade-offs between the costs of
climate mitigation and the consequences of unfettered climate change. The NGFS scenarios have three
essential features:

 Taking a long-term perspective, providing a common starting point for analysing climate-related
risks and their impact on the economy and financial system,

3 See modelling teams of the NGFS Academic Consortium in the Acknowledgements section.
4 Referred to as "NGFS scenarios" in the rest of the document.

7
  Covering the global economy, producing results that are internally consistent, applicable at the global
level and comparable across regions, and

 Representing a global public good as they are the product of an international collaboration among
leading academic institutions that (i) combine state-of-the-art climate models to capture the
interactions between transition, physical and macro-financial risk, and (ii) make the results available
as a set of climate pathways accessible to anyone, anywhere in the world on the NGFS Scenarios
Portal.

It is important to note that the NGFS scenarios are not forecasts. Instead, they aim at exploring plausible
futures (neither necessarily the most probable nor the most desirable) for financial risk assessment making
them particularly suitable for a wide range of applications. To reflect the uncertainty inherent to modelling
climate-related macroeconomic and financial risks, the NGFS scenarios use different models, and explore
a wide range of scenarios across regions and sectors.

The NGFS scenarios are regularly updated and enhanced in line with evolving expectations. The first
vintage of NGFS climate scenarios was released in 2020, and two more followed in fall 2021 and 2022. This
documentation has been published together with the fourth vintage of the scenarios. Over time, the NGFS
scenarios have become deeper, broader, and richer in terms of modelling tools, output results, risk
coverage and geographical scope. Continuous progress and refinements reflect the innovative nature of
climate scenario development, which lies at the frontier between climate science, macroeconomic
analysis, and policy assessment.

 In a world of uncertainty, a scenario is a hypothetical construct that describes a

path of development leading to a particular future outcome. Scenarios are not
forecasts or predictions, and do not provide a full description of the future, but rather
Explainer box 1
highlight central elements of a possible future.
What is a scenario and
scenario analysis, and  Scenario analysis is a tool to enhance critical strategic thinking. It is a process of
what are climate scenarios examining and evaluating possible future events and is used in a forward-looking
and climate modelling? assessment of risks and opportunities.

 Climate scenarios explore a different set of assumptions about how climate policy,
emissions, and temperatures evolve. They help to identify impacts from a changing
climate and the necessary policies for and opportunities from a green transition. They
can help our understanding of how climate-related risks could evolve and what the
implications might be for the economy and the financial system.

 Climate modelling refers to the use of quantitative methods to simulate and

analyse the interactions of climate variables, both to understand the dynamics of
the climate system and to project the future climate. Climate models may also be
qualitative to provide descriptive narratives of possible futures.

1.1 NGFS scenarios as a useful guide to climate risks

Climate change affects the way that our economy functions. In recent years, we have experienced a
multitude of climate disasters, from wildfires in North America, to floods in Brazil, to heatwaves with new
record temperatures in Europe. These are examples of more severe and more frequent extreme weather
events that are already visible today. However, in the transition to a less polluting economy, other events,
including less visible ones, could affect the profitability of businesses or the prosperity of households. Thus,

8
 climate change affects our economy and financial system through a range of different transmission
channels that can be classified into two types of risks:

 the physical risks of a changing climate, including more frequent or severe weather events like
floods, droughts, and storms, as well as other risks stemming from an increase in global
temperature, and

 the transition risks from moving towards a low-carbon economy, the timing and speed of which
will depend on policy and regulation, technology development and changes in consumer
preferences.

Policymakers and supervisors have identified climate change as a significant source of financial risk for
several years now5, but assessing its effects remains a daunting task for many, as they differ from the
traditional sources of financial distress. Capturing climate-related risks means considering their unique
and complex features, such as assessing an unprecedented combination of impacts spread over a long-
term horizon and bridging persisting climate data gaps6. While uncertainty is inherent in climate-related
risks, this is not reason enough to shy away from this fundamental challenge.

Scenario analysis is one approach to tackle this uncertainty. On the one hand, the NGFS scenarios
provide plausible future developments, because they are constructed with models designed to simulate
the complex and non-linear dynamics of the energy, economy, and climate systems. On the other hand,
they account for various possible policy and technology assumptions. Therefore, they allow a rich
exploration of various plausible future developments and an understanding of the trade-offs between
various policy and technology choices.

The NGFS scenarios provide a common starting point for understanding how climate change (physical
risk) and climate policy and technology trends (transition risk) could evolve in different futures. In the
newly released Phase IV, the NGFS scenario framework explores a set of seven scenarios characterised by
different levels of physical and transition risk, primarily driven by the level of policy ambition, policy timing,
coordination, and technology levels. The main technical features include:

 different climate pathways that depict potential future trajectories of greenhouse gas emissions and
global temperature increases,

 macro-economic variables that are influenced by climate change, such as GDP growth, inflation,
interest rates and employment,

 sectoral breakdown, including energy, transportation, and agriculture,

 geographical coverage, accounting for regional and country-level variations in climate risk,

 time horizon spanning multiple decades and long-term perspectives to capture the gradual nature of
climate change impacts, and

5 Network for Greening the Financial System (2018), “First progress report”, Banque de France, October.
Carney, M., Villeroy de Galhau, F. and Elderson, F. (2019), “Open letter on climate-related financial risks”, Bank of
England, April.
6 Baranović, Ivana, Busies, Iulia, Coussens, Wouter, Grill, Michael and Hempell, Hannah S., (2021), “The challenge of
capturing climate risks in the banking regulatory framework: is there a need for a macroprudential response?”, ECB
Macroprudential Bulletin, 15, issue, number 1.

9
  policy assumptions, such as the implementation of carbon pricing mechanisms, renewable energy
targets and other mitigation and adaptation measures.

The NGFS scenarios combine the analysis of transition, physical and macro-financial risks. To make this
possible, the NGFS scenarios bring together a global, harmonised set of transition pathways, physical climate
change impacts and economic indicators. A combination of models is used to capture separately but
consistently climate, macroeconomic, and financial contingencies. This methodology will later be referred to
as the suite-of-model approach. As shown in Figure 1, the models used to derive the NGFS scenarios can be
classified into three broad categories: physical risk models, transition risk models and a macro-financial
model7.

Figure 1. NGFS suite-of-models approach

 Physical risk models include all models that are participating in the Inter-Sectoral Impact Model
Intercomparison Project (ISIMIP)8 and CLIMADA9 and provide climate and economic indicators
accounting for changes in climate.

 Transition risk models include three Integrated Assessment Models (IAMs), specifically REMIND-
MAgPIE, GCAM and MESSAGEix-GLOBIOM, that derive the impacts of different policy ambitions on
the energy sector, emissions, and land use.

 The macro-financial model consists of the NiGEM model (a version specifically modified for the
purpose of producing the NGFS scenarios), to understand the consequences of transition and physical
risk on the key macro-financial fundamentals.

The NGFS suite of models produces a range of internally consistent data on transition risks, physical risks,
and economic impacts. The NGFS scenarios consist of a set of climate-related and macro-financial variables
available for each model, scenario, and geography (Figure 2). The data can be accessed freely online:

7 More details on the NGFS modelling approach are provided in NGFS modelling approach of this module.
8 More information about ISIMIP can be found here.
9 CLIMADA stands for climate adaptation and is a probabilistic natural catastrophe damage model that also calculates
averted damage (benefit) thanks to adaptation measures of any kind (from grey to green infrastructure, behavioural,
etc.). More information here.

10
 Figure 2. Overview of the range of data provided by NGFS scenarios.
Note: this visual does not contain the full list of variables and is for illustrative purposes only. The names of the variables do
not necessarily correspond to the ones used in the databases. The number of countries/regions available varies significantly
depending on the variable. Downscaled climate-related and macro-financial variables are available for 180+ and 50+
countries, respectively.

 Physical risk variables can be explored through the NGFS Climate Impact Explorer hosted by Climate
Analytics. More granular data are available via the ISIMIP project. Physical risk analysis was supported
by Climate Analytics, ETH Zurich and PIK.

 Transition pathway and macro-financial impact variables are made available in the NGFS Scenarios
Database hosted by IIASA. The transition pathways were produced by three IAM teams: PIK
(REMIND-MAgPIE model), IIASA (MESSAGEix-GLOBIOM model) and UMD (GCAM model). Economic
variables were produced by the National Institute for Economic and Social Research (NIESR) (NiGEM
model).

 Key data and resources can be explored interactively on the NGFS Scenarios Portal.

1.2 Comparison with other existing climate scenarios

The NGFS scenarios share some commonalities with other existing climate scenarios, such as the ones
developed by the Intergovernmental Panel on Climate Change (IPCC) and the International Energy Agency
(IEA). For instance, all three sets of climate scenarios rely on Integrated Assessment Models (IAMs) to provide
transition pathways for various narratives (with different but consistent results). Moreover, another shared
feature of the IPCC, the IEA and the NGFS scenarios is that they are all neither predictions nor forecasts but
instead explore a range of future climate pathways and/or green transition policies to estimate their future
economic implications.

Despite having similar objectives, the NGFS scenarios have some unique features that make them
particularly suitable for a wide range of applications. The main differences to other existing climate scenarios
can be summarized in three categories:

 Scope. The NGFS scenarios assess the consequences of both transition and physical risks globally,
while the IEA scenarios 10 focus on transition risk only and IPCC focus on the possible evolution of
greenhouse gas emissions11. The NGFS Scenarios also include more macroeconomic details.

10 Based on the Global Energy and Climate Model

11 IPCC (2000), Special Report on Emissions Scenarios, Working Group III of the Intergovernmental Panel on Climate
Change, see here.

11
  Time horizon. The NGFS scenarios look at the trade-offs between a green transition and a no-
transition scenario until the end of the century (2100), while the IEA scenarios focus on the implications
of a green transition until 2050.

 Applications. For the reasons above, the NGFS scenarios are mostly used by central banks,
supervisory authorities, and financial institutions to assess the costs and benefits of a green transition
for the financial sector, while the IEA scenarios are mostly used to better understand the implications
of different green policies in the short run.

The combination of transition, physical and macroeconomic models has been confirmed by scenario users
as one of the key strengths of the NGFS scenarios. The results of the first public NGFS survey on climate
scenarios underline that scenario users rate the NGFS scenarios framework positively compared to other
existing climate scenarios and highlight the number and relevance of output variables as an additional unique
selling point of the NGFS scenarios13.

More technically, the NGFS, IEA and IPCC scenarios also differ in terms of modelling
and narrative:
Explainer box 2
 The NGFS produces a wider set of scenarios. The NGFS scenarios explore seven
Modeling structure
possible future pathways, looking at both transition and physical risk, while the IEA
example: how do
NGFS scenarios scenarios explore only three scenarios, abstracting from the implications of physical
compare with IEA12 risk.
and IPCC scenarios?
 The NGFS offers sets of scenarios that have been created using the REMIND-
MAgPIE model that integrates the macro-economic climate damages into the
optimization procedure.

 The NGFS scenarios combine three model categories in a consistent manner to

assess the costs and benefits of a green transition: the Integrated Assessment
Models (IAMs) to assess the economic implications of a green transition; climate
damage models to understand the consequences of physical risk; a macro-economic
model (NiGEM) to assess the macroeconomic implications of climate policies and
unfettered climate change. The IEA scenarios, instead, rely on a single model, i.e.,
the World Energy model. The NGFS suite-of-model approach allows for exploring the
uncertainty related to model structures and techno-economic and potentially other
assumptions.

 Moreover, the types of models and variables (e.g., endogenous vs exogenous)

are different from IEA scenarios, which lead to different results.

12 There are three main IEA scenarios in the World Energy Outlook 2023: Stated Policies scenario (STEPS) (“the trajectory
implied by today’s policy settings”), Announced Pledges Scenario (APS) (“all aspirational targets announced by
governments are met on time and in full“), and Net Zero Emissions by 2050 (NZE) Scenario (“a way to achieve a 1.5 °C
stabilization in the rise in global average temperatures, alongside universal access to modern energy by 2030”).

13 NGFS (2023), “NGFS Survey on Climate Scenarios: key findings”, June

12
 1.3 Scenario applications

The NGFS scenarios have become a key ingredient for exploratory stress test and scenario analysis
exercises worldwide. Originally designed as a tool to advise policymakers on potential future risks, their user
community continues to grow substantially beyond central banks and supervisors. Since their first vintage in
2020, the NGFS scenarios have been repeatedly refined with the release of three improved vintages and made
available as a public good. While evolving to cater for new needs, the NGFS scenarios’ unique features as a
financial risk assessment tool have made them particularly well suited for an increasing range of applications.
In other words, not only the number of users continues to grow, but also the variety of their applications:

 Risk assessment, scenario analysis and stress testing. Central banks, supervisors and financial
institutions can use the scenarios to assess the resilience of portfolios, individual institutions or the
entire financial system under different climate scenarios. This helps to identify potential vulnerabilities
and allows for the appropriate risk management strategies, as well as assessing the trade-offs of
different options.

Since the NGFS scenarios were developed for risk assessment purposes, they do not
always have equivalents in the IEA or IPCC models, as the latter focus on exploring
Explainer box 3
transition pathways. To illustrate this, let us look at the example of carbon pricing.
Carbon pricing
example: how do
 Carbon prices are structurally different in the NGFS and IEA scenarios. In the case
NGFS scenarios of the NGFS scenarios, the carbon price is calculated endogenously within each IAM,
compare with IEA whereas in the case of the IEA, the carbon price is set exogenously depending on
and IPCC scenarios? national carbon pricing policy and commitments and the degree of emission
reductions in each scenario.

 In other words, in the NGFS scenarios, carbon prices are shadow prices that
reflect the policy ambition specified by the scenario (e.g., Net Zero by 2050) and
serve as a measure of overall policy intensity. They are sensitive to factors such as
the level of ambition to mitigate climate change, the timing of policy
implementation, the distribution of policy measures across sectors and regions, and
assumptions regarding technology (e.g., the availability and feasibility of carbon
dioxide removal).

 In addition to (actual) carbon pricing, the scenarios developed by the IEA

separately consider a wide range of other policy measures that can contribute to
emission reductions, and the carbon price is not a marginal abatement cost that
is derived through an optimization calculation. Carbon prices that are linked to
emission reductions through formulation under IAMs and carbon prices that are set in
a situation where policy measures other than carbon pricing are in place are different
in nature. In the presence of other policy measures, the carbon prices implicit in the
IAMs tend to be higher (CRIEPI, 2022).

 Climate disclosures. Granular data on transition pathways, climate impacts and macro-financial
indicators can enhance strategic thinking and form a key part of climate-related financial disclosures.
Climate scenarios support harmonisation efforts in this field.

13
  Strategy and policy alignment. While many actors in the private and public sectors are revising their
strategies and policies to align with particular goals, the NGFS scenarios highlight some key themes
that can be used to help guide decision-making and set more granular targets. For example, NGFS
scenarios can help financial institutions develop their net zero transition plan and manage associated
risks, as well as support the alignment of climate targets.

 Investor engagement: Investors can use the scenarios as a basis for dialogue with companies and
assess the long-term sustainability of their investment portfolios.

 Further academic research: The NGFS scenarios can be used as a starting point for researchers and
technical specialists who wish to extend them to include higher granularity and other channels and/or
feedback effects.

The NGFS scenarios are helping a wide range of public and private sector players to identify climate risks
globally. The results of a stocktaking exercise14 on climate scenarios, models, data, and metrics used by
members of the Financial Stability Board (FSB) and the NGFS show that the vast majority of the 53 members
that have completed, are conducting, or plan to conduct a climate scenario analysis exercise rely on the NGFS
scenarios. The report argues that the NGFS scenarios are at the core of these exercises, with most of the
sampled institutions worldwide making use of them, either with or without adjustments in some of their
components or outcomes. (See FSB-NGFS, 2022).15

Furthermore, survey results16 confirm that the NGFS scenarios have become an essential tool among both
private and public sector actors in understanding the financial risks stemming from a changing climate, as well
as the opportunities of climate mitigation action. The NGFS survey finds that over 70% of 213 respondents from
57 countries use them, mostly to better understand the impacts of climate risks and to build internal capacity.
In addition, the richness and granularity of the scenarios make them useful also for a wide range of audiences
including consultancies, academics, international organisations, or civil society organisations, among others.
(See NGFS,2023).17

The NGFS scenarios can also help policymakers and central banks understand the impacts of transition
policies on the macroeconomic outlook, which in turn can feed into relevant policy decisions. Alongside the
impact of climate change and climate policies on the key macroeconomic indicators (such as GDP, commodity
prices, inflation, and interest rates), the NGFS scenarios also give insights that can inform policymakers. For
example, they provide estimates of the investments needed across energy sectors to reach the climate targets.
The scenarios show how much all these variables differ across regions and can thus support the calibration of
country-specific climate policies and risk assessment exercises. The NGFS scenarios are also an important tool
to assess the forward-looking impact of climate change on macroeconomic fundamentals, which can support
central banks’ policy decisions.

14 Information on a total of 66 climate scenario analysis exercises was obtained.

15 Climate Scenario Analysis by Jurisdictions: initial findings and lessons, November 2022

16Most of the respondents come from financial institutions, central banks, and consulting firms; and use NGFS scenarios
to better understand the impact of climate risks on the respective organisation, individual financial institutions, or
financial stability. In addition, 95% of the respondents that have already concluded exercises based on NGFS scenarios
are (at least partially) satisfied with the outcome.

17 NGFS Survey on Climate Scenarios: key findings, June 2023

14
The NGFS scenarios have evolved from a policy tool for selected users to a common language for climate
risks for all. Although they were originally developed by central banks and supervisors to help inform and guide
policy across the globe, they are used, by now, for an increasing variety of applications by financial institutions,
policymakers, and other key stakeholders to assess the financial risks associated with climate change and
support the transition to a more sustainable and resilient economy.

15
2. NGFS scenario narratives

The NGFS scenarios explore the impacts of climate change and climate policy with the
aim of providing a common reference framework.
Key messages
 The NGFS scenarios explore a set of seven climate scenarios that can be grouped
into four categories (quadrants): orderly transition, disorderly transition, hot house
world, and too little, too late. Each scenario is characterized by its overall level of physical
and transition risk, which are driven by the level of policy ambition, policy timing,
coordination, and technology levers.
 In this fourth vintage, the NGFS scenarios have been enriched and updated, including
both narrative-related and technical improvements, as well as increased usability.
o The NGFS scenarios narratives have been updated and expanded to reflect the
most recent developments, including geopolitical (e.g. the Russian invasion of
Ukraine) and technological ones (e.g. the availability of CDR), as well as delays in
policy implementation.
o The technical refinement of the modelling framework has improved the design of the
NGFS scenarios, especially in the area of physical risk.
o Improvements in the usability of the accompanying materials have increased the
transparency of the NGFS scenarios and their underlying models. The
documentation has been adapted to different user needs, making it more accessible
to both expert and non-expert users, based on user feedback collected.

The NGFS scenarios framework explores a set of possible transition pathways, depending on different
levels of ambition and coordination in terms of climate policies. As shown in the NGFS scenarios framework
(Figure 3), the NGFS scenarios can be grouped into four quadrants: orderly transition, disorderly transition, hot
house world, and too little, too late scenarios.

Figure 3. Overview of the NGFS scenario framework

16
  Orderly scenarios assume that ambitious climate policies are introduced early and become gradually
more stringent. Both physical and transition risks are relatively subdued.

 Disorderly scenarios assume that climate policies are delayed or divergent across countries and
sectors. These scenarios are associated with subdued physical but high transition risks, as, for instance,
carbon prices might need to rise sharply and abruptly.

 Hot house world scenarios assume that global warming cannot be limited due to insufficient global
efforts. As a result, critical temperature thresholds are exceeded, leading to severe physical risks and
irreversible impacts like sea-level rise.

 Too little, too late scenarios assume that a late and uncoordinated transition fails to limit physical
risks.

2.1 Description of narratives

In this fourth vintage, the NGFS scenarios explore a set of seven possible transition pathways, depending
on different levels of ambition and coordination in terms of climate policies. The scenarios are mapped in
the NGFS scenario framework in Figure 4 and can be summarised as follows.

Orderly - Low Demand explores the global efforts needed to be able to limit global warming to below
1.5°C by 2050 in an orderly fashion, aligned with the Paris Agreement, driven by lower energy
demands. Given the policy delays, this orderly scenario shows that achieving these targets
will require even greater ambition in future compared with the previously published ‘orderly
transition’ scenarios.

- Net Zero 2050 limits global warming to 1.5°C through stringent climate policies and
innovation, reaching global net zero CO2 emissions around 2050. Some jurisdictions such as
the US, EU, UK, Canada, Australia, and Japan reach net zero for all GHGs.

- Below 2°C Below 2°C gradually increases the stringency of climate policies, giving a 67%
chance of limiting global warming to below 2°C. Additionally, countries with net zero targets
reach them partially (80% of the target).

Disorderly - Delayed Transition assumes annual emissions do not decrease until 2030. Strong policies are
needed to limit warming to below 2°C. Negative emissions are limited.

Hot house - Nationally Determined Contributions (NDCs) includes all pledged targets even if not yet
world backed up by implemented effective policies.

- Current Policies assumes that only currently implemented policies are preserved, leading to
high physical risks.

Too little, too - Fragmented World assumes a delayed and divergent climate policy response among
late countries globally, leading to high physical and transition risks. Countries without zero targets
follow current policies, while other countries achieve them only partially (80% of the target).

17
 Figure 4. The NGFS scenario framework in Phase IV

The transition pathways for the NGFS scenarios show a range of higher and lower risk outcomes that
explore a different set of assumptions about the evolution of climate policy, emissions, and temperatures.
Each scenario is based on several key design choices relating to policy ambition (captured by specific end-of-
century temperature targets or policy packages), short-term policy, overall policy coordination and technology
availability. Table 1 highlights the various assumptions underlying these design options, which are explained in
more detail below. Table 3 compares the key assumptions across scenarios.

The main driving forces of the scenarios are the evolution of carbon prices and the evolution of CO2
emissions, which are strictly related.

18
Table 1. Overview of NGFS scenario narratives

Quadrant NGFS scenario Narrative explained

Orderly Low Demand This scenario includes significant behavioural changes in our energy
(1.5°C) generation and consumption activities to ensure an orderly, Paris-
aligned transition18.

 Global CO2 emissions reach or approach net zero in 2050.

Countries with a political commitment to a net zero target defined
before February 2023 reach net zero at their target year or earlier.

 Some jurisdictions such as the US, EU, UK, Canada, Australia, and
Japan reach net zero for all GHGs.

 Additional levers in end-use sectors (e.g., behavioural changes,

reducing energy demand, inducing faster electrification, and
substitution through renewables) mitigate the pressure on carbon
taxes to induce the transition and are the distinguishing feature of
this scenario compared to the Net Zero scenario by 2050.

Net Zero 2050 Global warming is limited to 1.5°C (with a 50% chance) through
(1.5°C) stringent climate policies and innovation, reaching global net zero CO 2
emissions around 2050.

 Global CO2 emissions reach or approach zero in 2050. Countries

with a political commitment to a net zero target defined before
end of March 2023 reach net zero at their target year or earlier.

 Some jurisdictions such as the US, EU, UK, Canada, Australia, and
Japan reach net zero for all GHGs.

Below 2°C The stringency of climate policies is gradually increased, giving a 67%
(2°C)
chance of limiting global warming to below 2°C by the end of the
century.

 Global CO2 emissions evolve such that the end-of-century temperature

goal of 2°C warming is reached (with a 67% chance).

 Countries who have net zero targets follow through on 80% of them,
others follow less ambitious trajectories.

18 “Paris-aligned” refers to achieving the 1.5-degree target of the Paris Climate Agreement by reducing greenhouse gas
emissions: “limiting the global temperature increase to no more than 1.5 degrees Celsius by the year 2100 compared
to pre-industrial times.”

19
 Disorderly Delayed Annual emissions do not decrease until 2030. Strong policies are needed
Transition
to limit warming to below 2°C.

 Countries stick to current policies until 2030 and experience a “fossil

recovery”, after which they transition such that the end-of-century
temperature goal of 2°C warming is reached. This change of regime in
2030 is unanticipated and therefore disruptive. Countries with net-zero
policy target commitments are assumed to follow-through on 80% of
them. Negative emissions are limited.

Hot house Nationally All pledged targets are assumed to be implemented, even if they are not
world Determined
yet backed up by effective policies.
Contributions
(NDCs)  Countries implement pledged policies in addition to current policies
and keep their level of ambition beyond the NDC horizon. The cut-off
date for targets being considered here is those published by the
UNFCCC until end of March 202319.

Current Policies Only currently implemented policies are preserved, leading to high
physical risks.

 Existing climate policies remain in place but there is no strengthening

of ambition level of these policies 20.

Too little, Fragmented A delayed and divergent climate policy response among countries
too late World globally leads to high physical and transition risks.

 Only currently implemented policies are maintained until 2030

(delayed transition); thereafter, countries that have set themselves a
net zero target only reach an 80% reduction by 2050, while others
continue with current policies (divergent transition).

19 See https://unfccc.int/NDCREG

20 The detail of policy representation differs across models and across different sectors. Policy implementation has been
included in as much detail as possible, but due to limited granularity of sector representation, all models also
represent some policies as proxies, for example via aggregate final energy reductions instead of explicit
implementation of efficiency standards, or a carbon price.

20
Table 2. Overview of NGFS scenarios by key assumptions. The table maps out key features of the scenario narrative and their
macro-financial risk implications stemming from transition or physical risk. Green means “low risk”, yellow means “medium
risk”, red means “high risk”.

Category Scenario End of Policy Technology Carbon Regional

century reaction change dioxide policy
(peak) removal - variation +
warming –
model
average

Orderly Low Demand 1.4°C Immediate Fast change Medium Medium

(NEW) (1.6°C) and smooth use Variation

Net Zero 2050 1.4°C Immediate Fast change Medium- Medium

(1.6°C) and smooth high use Variation

Below 2°C 1.7°C Immediate Moderate Medium Low

(1.8°C) and smooth change use variation

Disorderly Delayed 1.7°C Delayed Slow/ Fast Low- High

Transition (1.8°C) change medium variation
use

Hot house Nationally 2.4°C NDCs Slow change Low- Medium

world Determined (2.4°C) medium variation
Contributions use
(NDCs)

Current 2.9°C None - Slow change Low use Low

Policies (2.9°C) current variation
policies

Too-little- Fragmented 2.3°C Delayed and Slow/ Low- High

too-late World (NEW) (2.3°C) Fragmented Fragmented medium variation
change use

21
2.2 What is new in the NGFS scenario framework?

In this fourth vintage, all NGFS scenarios have been substantially enriched in their key features to better
reflect a more pronounced disorderly future considering recent developments. Delayed implementation of
climate policies, persistently high emissions, and the consequences of the war in Ukraine on energy system
trajectories are contributing to an overall increase in disorderliness of the NGFS scenarios. Figure 5 displays the
movements in scenario mapping in the NGFS scenario framework. The main changes are described as follows:

Figure 5. NGFS scenario framework: from Phase III to Phase IV. Movements in the scenario mapping are represented by
arrows, new scenarios introduced in Phase IV are indicated with a plus (+) symbol, and the phased out scenario is marked with
a cross (x).

 Net Zero 2050 (1.5°C) has shifted upwards and right in the framework given the higher baseline emissions
(2021-2025), leading to increased disorderliness with higher physical and transition risks. This scenario
assumes higher peak warming temperatures, and steeper increases in carbon tax programmes to reach
global net zero CO2 around 2050. CDR availability is kept “medium” as in Phase III, but with lower Bioenergy
with Carbon Capture and Storage (BECCS) use.

 Below 2°C has shifted upwards in the framework, showing an increased transition risk and slightly lower
physical risk. It assumes that countries limit global warming to +2°C in 2100 (with 67% probability).
Compared to Phase III, country net zero targets have been added, assuming an emissions reduction of at
least 80% compared to 2020 would be reached by countries with such targets. This inclusion slightly lowers
peak warming and increases disorderliness, resulting in less physical risk but more transition risk.

 Delayed transition has been updated without a change in its narrative. It assumes annual emissions do not
decrease until 2030. Strong policies are needed to limit warming to below 2°C. Negative emissions are
limited.

22
 Divergent Net Zero (1.5C) scenario, previously included in Phase III, has been phased out in this new fourth
vintage given the reduced likelihood of a successful uncoordinated transition. This streamlining decision
also contributes to the ongoing efforts to improve the usability of NGFS scenarios for its broad user
community, striking the right balance between usability and complexity of its framework.

 Nationally Determined Contributions (NDCs) foresees that currently pledged conditional NDCs are
implemented fully, and respective targets on energy and emissions in 2025 and 2030 are reached in all
countries, even if not yet backed up by implemented effective policies. The cut-off date for targets being
considered here is until end of March 2023, as published by the UNFCCC. This results in a slight decrease in
long-term physical risk compared to Phase III due to newly announced commitments. Nevertheless,
physical risk remains high.

 Current Policies existing climate policies remain in place, but there is no strengthening of the ambition
level of these policies. The scenario has been updated to take new policies into account (EU Fit-for-55, US
IRA, etc.). This results in a slight decrease in long-term physical risk compared to Phase III due to newly
implemented policies. Nevertheless, physical risk remains very high.

Second, the NGFS scenarios framework has been further expanded in the set of scenarios to capture more
and less adverse disruptions. To account for “too-little-too-late” climate policies, as well as the possibility of
still meeting the Paris Agreement, two new scenarios have been designed and introduced.

 First, there is a risk of further delays and fragmentation in the international climate policy landscape, made
more severe by the energy crisis following the war in Ukraine. A new too-little-too-late scenario
Fragmented World has been added to explore such adverse developments. It entails delayed policy action
until 2030 and divergence in policy ambition thereafter, with countries having net zero targets supposedly
following these, and countries without net zero targets supposedly following the current policies.

Given its design features, the FW scenario can be particularly useful for regulators, as well as central banks
and supervisors. It emphasizes the critical role played by international policy coordination (or the lack
thereof), and it explores more adverse impacts if we fail to implement climate mitigation policies in a timely
and globally coordinated manner. Therefore, it can be used as a baseline for climate stress tests.

 Second, the +1.5°C end-of-century warming limit, which the world has committed to in the Paris
agreement, becomes ever more challenging to achieve due to persistently high and rising global emissions.
This is reflected in the increasing disorderliness of the Net Zero 2050 scenario. Therefore, a new Paris-
aligned orderly transition scenario Low Demand has been added.

This scenario includes lower energy demand and stronger behavioural changes, mapping out the
challenging path to still reach the Paris goals in an orderly way.

Low Demand shows that greater energy sobriety, deep electrification, and improved energy efficiency will
be needed to reach the 1.5°C target – all of which will, however, be less costly than inaction or disorderly
transition in the long term. This scenario can be particularly informative to policymakers and regulators, as
it illustrates the challenges as well as the feasibility and benefits of carrying on with the net zero transition.

Third, as in each iteration, the NGFS scenarios have been brought up to date with new economic and
climate data, policy commitments and model versions. All NGFS scenarios have been updated to account for
changes in the broad geopolitical and climate policy situation, including the war in Ukraine, delays in
government climate action and lock-in of fossil fuel technologies in many jurisdictions. Updates have also been
made to the use of Carbon Dioxide Removal (CDR) technologies, which has been limited compared to the Phase

23
III scenarios. These updates have been applied consistently across scenarios and models to improve
comparability of results. Table 3 summarises the main changes in Phase IV.

Table 3. Overview of main changes in Phase IV compared to Phase III

NGFS
Quadrant Main changes in Phase IV (compared to Phase III)
scenario

All scenarios are more disorderly in Phase IV reflecting delays in policy action
and the current geopolitical situation/energy crisis

 Updates reflecting energy market disruption due to the Russian war in

Ukraine. All scenarios and their underlying models have been updated,
e.g., including recent policy actions, and including energy sector
ramifications of the Russian war in Ukraine.
All NGFS scenarios
 Limits to the use of carbon dioxide removal (CDR) technologies. All
scenarios assume limited availability of BECCS and dropping the
availability of direct air carbon capture and storage (DACCS). This makes
scenarios more adverse in the sense that reductions of fossil fuel use need
to be more severe. The carbon tax implementing these demand
reductions would be higher.

Orderly Low Demand NEW scenario

(1.5°C)

Net Zero New net zero targets considered, less CDR

2050
(1.5°C) Update of net zero targets (see table 38)

Below 2°C Countries with net zero targets follow through on 80% of pledged reductions,
(2°C) less CDR

NDCs partial achievement: net zero targets are assumed to be 80% fulfilled, to
better reflect the actual policy ambition globally

Disorderly Delayed Net zero targets updated

Transition
(2°C)  Update of net zero targets (see table 38)

Hot house Nationally New NDCs considered

world Determined
Contributions  Updated NDCs, including IRA and Fit for 55.
(NDCs)

Current Inclusion of additional policies (IRA, Fit for 55), updated assumptions regarding
Policies technical change and GDP.

24
 Too little, Fragmented NEW scenario
too late World

2.3 Shared model input assumptions

All scenarios share the same underlying assumption on key socio-economic drivers, such as harmonised
population and economic developments, which are taken from the Shared Socioeconomic Pathway SSP2
(Dellink et al., 2017; Fricko et al., 2017; KC & Lutz, 2017; O’Neill et al., 2017; Riahi, van Vuuren, et al., 2017). Thus,
all NGFS scenarios are to a great extent aligned with the Middle of the Road Shared Socioeconomic Pathway,
which is neither optimistic, nor very pessimistic.

Further drivers such as food and energy demand are also harmonised, though not at a precise level but in terms
of general patterns.

To account for the COVID-19 pandemic and its impact on economic systems including the near-term recovery
until 2025, the GDP and final energy demand trajectories have been adjusted based on projections from the
IMF (IMF 2021, Koch & Leimbach 2023).

The transition pathways do not incorporate the anticipation of potential future economic damages from
physical risks (except for REMIND-MAgPIE scenarios with integrated damages). In other words, damages to
infrastructure systems and the economy in the future, caused by emissions today, have no feedback
mechanism that affects current choices. We provide acute and chronic physical damages for each scenario, but
they are not incorporated in the transition models.

25
3. NGFS modelling approach
This section explains the modelling choices made by the NGFS, and describes the
models used to generate the NGFS Scenarios.
Key messages
 To reflect the uncertainty inherent to modelling climate-related macroeconomic and financial
risks, the NGFS scenarios use different models, and explore a wide range of scenarios
across regions and sectors. This is called the NGFS suite-of-model approach.
 The NGFS suite-of-models is internally aligned in a coherent way and produces a range of
data on transition risks, physical risks, and economic impacts.

The NGFS scenarios bring together a global, harmonised set of transition pathways, physical climate
change impacts and economic indicators. They combine the analysis of transition, physical and macro-
financial risks to shed light on the long-term trade-offs between the costs of climate mitigation and the
consequences of unfettered climate change. They take a long-term perspective, which is necessary to assess
the benefits of reduced physical risk over the next decades driven by effective climate policies: a scenario
focussed on the next few years would, in fact, only capture the costs of climate action, while omitting the future
but long-lasting benefits of meeting the Paris temperature targets – which is the very reason why climate action
is needed.

Climate risks and their transmission channels

Climate risks could affect the economy and financial system through a range of different transmission
channels. To understand the potential macroeconomic impacts of climate risks, we can distinguish between
risks related to the transition to a lower-carbon economy (transition risks) and risks related to the physical
impacts of climate change (physical risks), as shown in Figure 6.

Figure 6. Transmission channels: climate risks to financial risks. Source: NGFS (2022)

26
  Transition risks affect the profitability of businesses and wealth of households, creating financial risks
for lenders and investors. They also affect the broader economy through investment, productivity, and
relative price channels, particularly if the transition leads to stranded assets.

 Physical risks affect the economy in two ways: Acute impacts from extreme weather events can lead
to business disruption and damages to property. There is some evidence that with increased warming
they could also lead to persistent longer-term impacts on the economy. These events can increase
underwriting risks for insurers, possibly leading to lower insurance coverage in some regions, and
impair asset values. Chronic impacts, particularly from increased temperatures, , may affect labour,
capital, land, and natural capital in specific areas. By affecting individual businesses, households and
the broader macroeconomy, climate risks could translate into financial risks and affect the financial
system.

Seven different academic institutions or initiatives joined forces under the aegis of the NGFS to ensure the
overall consistency of the scenario framework while still relying on state-of-the-art and peer-reviewed
academic literature. The NGFS therefore uses existing models, each of them being specialised and advanced
in capturing one single part of the framework. This allows deep diving in the reactions of economic sectors to
climate change and/or climate policies with a higher level of granularity, coverage, and precision than otherwise
possible. The models chosen for the NGFS scenarios also inform the IPCC reports, thus ensuring a high level of
consistency between the NGFS and the IPCC frameworks. Furthermore, this collaboration has facilitated
dialogues across specialised institutions, that has allowed a cross-fertilisation of ideas and skills to improve
existing methodologies and advance our understanding of climate scenarios further.

NGFS suite-of-model approach

The NGFS scenarios are based on a modular, suite-of-model approach to capture separately but
consistently climate, macroeconomic, and financial contingencies. The models used to derive the NGFS
scenarios can be classified into three broad categories:

 Physical risk models include all physical risk models that are participating in ISIMIP and CLIMADA and
provide climate and economic indicators because of changes in climate.

 Transition risk models include three Integrated Assessment Models (IAMs), specifically REMIND-
MAgPIE, GCAM and MESSAGEix-GLOBIOM, that derive the impacts of different policy ambitions on
the energy sector, emissions, and land use.

 The macroeconomic model consists of the NiGEM model (a version specifically modified for the
purpose of producing the NGFS scenarios), to understand the consequences of transition and physical
risk on key macro-financial fundamentals.

The transition pathways for the NGFS scenarios have been generated with these three well-established
IAMs and linked to a macroeconomic model (NiGEM) to extend the macro-economic information. The IAMs
have been used in a vast number of peer-reviewed scientific studies on climate change mitigation and their
results feature in several assessment reports (Clarke et al., 2014; Forster et al., 2018; Jia et al., 2019; Rogelj,
Shindell, et al., 2018; UNEP, 2018, IPCC 2022a). They allow the estimation of global and regional mitigation
costs (Kriegler et al., 2013, 2014, 2015; Luderer et al., 2013; Riahi et al., 2015; Tavoni et al., 2013), the analysis of
emissions pathways (Riahi, van Vuuren, et al., 2017; Rogelj, Popp, et al., 2018, Riahi et al., 2021), associated
land use (Popp et al., 2017) and energy system transition characteristics (Bauer et al., 2017; GEA, 2012; Kriegler
et al., 2014; McJeon et al., 2014), the quantification of investments required to transform the energy system
(GEA, 2012; McCollum et al., 2018; Bertram et al., 2021) and the identification of synergies and trade-offs of
sustainable development pathways (Bertram et al., 2018; TWI2050, 2018). In short, they optimize energy

27
systems and land/water use in the face of long-run population and production trends. To shed light on the
potential macroeconomic impacts, the IAMs have been linked to the macro-economic model NiGEM. Put
simply, this mid-term global econometric model takes in energy-related and carbon tax inputs from the IAMs
and generates macro-financial series such as inflation, unemployment, and house price index (HPI) that would
be more typically used in financial modelling. The NGFS suite of models is aligned in a coherent way to produce
results that are internally consistent.

Figure 7. Interactions between the three model categories in the NGFS framework

Figure 7 illustrates the NGFS suite of models and how models interact with each other. Transition and physical
risk models have been aligned in terms of temperature pathways, to ensure that both their impacts on the
economy are consistent and comparable. The way transition and physical models are combined with the
macro-financial model NiGEM is instead sequential. Energy- and emission-related variables produced by the
three IAMs are used as input by NiGEM. Similarly, NiGEM obtains input variables from the physical risk models.

28
 The NGFS suite of models contrasts with the single model approach in that it
builds on the strengths of each type of model, but at the same time there are
Explainer box 4 challenges to be considered.
What are benefits and
challenges of the NGFS Main benefits
suite of models?
 The strength of the NGFS suite of models is in their global coverage and
integrated assessment of risks. Where possible, multiple models have been
used for each scenario and warming level to represent uncertainty.

 Comparable outputs. The three NGFS IAMs produce comparable outputs for
energy and land-use series, although sometimes with varying methodologies,
and can be equally interchanged when used downstream in NiGEM.

 Flexibility. Users can choose from comparable IAM outputs in macroeconomic

models.

 Openness. Users can examine the sensitivity of using different methodological

assumptions at the energy and land/water use system level on lower-level
economic drivers, increasing transparency and robustness.

Main challenges

 This modular approach makes sacrifices in terms of unity (although there is a

reconciliation process between common endogenous outputs).
 While significant research advances have been made recently, care should be
taken in using the results, particularly at the most granular levels.

Modelling the macro-financial effects of climate scenarios is an exercise that sits in the nexus of two distinct
fields 1) climate, energy, and land use, and 2) the macroeconomy. The first is usually modelled using an
engineering approach since energy systems (and emissions) dynamics are determined by long-run investment
in different vintages of technologies that convert resources into energy, such as coal-fired power plants or
windmills, and are constrained by physical resource endowments inherent in each region due to factors such as
geology or wind patterns. On the other hand, higher-frequency macroeconomic dynamics are typically
modelled by understanding the relationship of purely economic series in history and are agnostic on energy mix
or land/water use. Producing climate scenarios most useful for financial analysis would entail integrating these
two contrasting frameworks' data and key methodological elements. Modelers can address this challenge by
either designing a holistic model that can represent both long-term energy use and mid-term macroeconomic
outcomes or utilize a modular approach by linking two separate frameworks. Both methods have their upsides
and downsides to consider. While a modular approach gains in specificity, openness, and flexibility, it also
sacrifices in consistency. A single unified model on the other hand may need to make sacrifices in terms of detail
on certain, potentially key, subcomponents, but would find it easier to maintain conceptual soundness and
overall consistency.

29
The NGFS scenarios take the modular approach as a starting point, following the trend in the energy systems
field to tackle complexity by leveraging the individual strengths of a variety of models. 21 Indeed, the IAMs also
follow this strategy, as they were developed to model phenomena that were studied in different disciplines. For
example, within the IAMs, separate macroeconomic modules exist mainly to generate long-run aggregate
energy and resource demand, with energy and land-use modules calculating the optimal structure of these
systems to fulfil this demand and climate modules calculating the subsequent effects on temperature. Key
inter-module interactions, such as price feedback, are modelled with hard links.

Utilizing the IAMs (GCAM, MESSAGEix-GLOBIOM, and REMIND-MAgPIE) for key energy, land/water use, and
carbon tax series ensures these components are modelled most accurately, providing energy transition and
other dynamics that have been used in hundreds of peer-reviewed scientific studies on climate change
mitigation. On the other side, the modelling of the impact of these series on the macroeconomy with NiGEM,
the leading global macroeconomic model, leverages its considerable strengths in this area as well. Both
policymakers and private sector organizations across the globe rely on these models for economic forecasting,
scenario building, and stress testing.

The modular approach also carries the benefit of flexibility, additionally leading to further transparency and
robustness. If intermediate series are consistent in interpretation, models that interact with them can be
switched in or out. A concrete example is in the presence of 3 options of IAMs to generate energy and land-use
series. They all produce comparable outputs, although sometimes with varying methodologies, and can be
equally interchanged when used downstream in NiGEM. This enables users to examine the sensitivity of using
different methodological assumptions at the energy and land/water use system level on lower-level economic
drivers, increasing transparency and robustness. In addition, users can directly use IAM outputs in
macroeconomic models of their choice. This contrasts with a single model approach which would be more
difficult to disentangle modularly.

While the outputs of the two modelling frameworks mostly do not intersect, with energy and land/water use
on the IAM side and macro-financial variables from NiGEM, both sides of the framework do generate
endogenous or semi-endogenous GDP estimates. IAMs’ baseline GDP is exogenous and based on SSP2
variables. GDP then changes semi-endogenously due to transition costs in each scenario. To ensure
consistency, NiGEM sets demand and supply-side shocks such that its baseline GDP matches the GDP target
from the IAMs (GDP then deviates from baseline in each scenario). In addition to the above, each scenario also
includes chronic physical risks shocks via demand and supply-side shocks (see Figure 8). Based on scenario
temperature outcomes, GDP shocks for chronic physical risks are computed using a damage function, and
subsequently fed as input to NiGEM.

21 See https://papers.ssrn.com/sol3/papers.cfm?abstract_id=318680.

30
 Integrated assessment models (IAMs) are simplified representations of complex
physical and social systems, focusing on the interaction between economy, society,
Explainer box 5
and the environment.
What are Integrated
Assessment Models?  IAMs represent the coupled energy-economy-land-climate system to varying
degrees. In some ways, IAMs can differ from each other: there can be significant
variation in geographical, sectoral, spatial and time resolution; they rely on different
technological representation; they can use partial or general equilibrium
assumptions; and they can assume perfect foresight or recursive-dynamic
methodology. The difficulty in fully representing the extent of climate damages in
monetary terms may be the most important and challenging limitation of IAMs.

 IAMs integrate knowledge from two or more domains into a single framework.
They are one of the main tools for undertaking integrated assessments. The IAMs
used by the NGFS include representations of multiple sectors of the economy, such
as energy, land use and land-use change; interactions between sectors; the economy
as a whole; associated GHG emissions and sinks; and reduced representations of the
climate system. This class of model is used to assess linkages between economic,
social, and technological development and the evolution of the climate system.
Other types of IAMs additionally include representations of the costs associated with
climate change impacts. These can be used to assess impacts and mitigation in a
cost-benefit framework and have been used to estimate the social cost of carbon.
The NGFS does not use such models and relies on other methods to compute the
cost of transition and physical risks.

3.1 The NGFS IAMs – process-based modelling of the energy transition

Even though the NGFS IAM models were developed by different research groups, with each model having
its own unique features, strengths, and limitations, they all have in common a similar modular structure
and set of shared assumptions, with scenarios based on the widely used Shared Socio-economic Pathways
(SSPs), and harmonized population and economic development trajectories. They combine macro-economic,
agriculture and land-use, energy, water, and climate systems into common numerical frameworks that enable
the analysis of the complex and non-linear dynamics in and between these components.

In contrast to simpler cost-benefit IAMs like Dynamic Integrated Climate-Economy models (DICE) and Regional
Integrated Climate-Economy models (RICE), the IAMs used by the NGFS cover more systems with a finer
granularity and process detail, which leads them to be referred to as process-based IAMs. For instance, they
offer highly granular representations of the energy system, taking an engineering-style approach to modelling
the conversion of raw natural resources into energy using various competing technologies. An overview of their
high-level structure can be pictured in Figure 8.

31
 Figure 8. Simplified diagram of the elements that integrate the IAM models

As one of the key outputs from the IAMs is the energy transition path induced by a policy change, such as a
carbon tax, we will outline the key modules used in all the IAMs to produce this output and highlight where
differences occur. Since all the models produce a cost-minimizing or approximately cost-minimizing energy
mix given demand derived in the macroeconomy, we trace through the system from the top down.

 Macroeconomy: This module provides assumptions or modelling of high-level GDP or population

trends at the regional level. IAMs’ baseline GDP is exogenous and based on SSP2 variables. GDP then
changes semi-endogenously due to transition costs in each scenario. MESSAGEix and REMIND utilize
Ramsey-type growth models where capital investment and energy are chosen to maximize
intertemporal welfare given energy costs (model parameters are calibrated to match exogenous GDP).
In contrast, GCAM GDP is set to an exogenous baseline and adjusts according to the labour force,
population, and energy price changes.

o Sectoral energy demand is determined by the level of economic activity in the

macroeconomic modules which in turn, is determined within the macroeconomic
optimization problem given energy costs (in REMIND), calibrated to exogenous baseline
projections with endogenous deviations due to changes in energy costs (in MESSAGE and
GCAM), or set according to exogenous projections (in GCAM).22 When demand is
endogenous, assumptions are made regarding substitutability of sectors in aggregate
production using constant elasticity of substitution (CES) production functions. Sectors differ
between models but include, at the very least, Buildings, Transport, Industry, and
Agriculture/Land-use.

22 MESSAGE can also determine energy demand fully endogenously using price feedback between the MACRO
andMESSAGE modules (see MESSAGEix-GLOBIOM documentation, page 49). However, in Phase IV scenarios, energy
demand is generated from exogenous material demand projections. Keeping energy demand endogenous allows for
substitution between inputs.

32
  Energy Systems. In all 3 IAMs, these are represented in a separate module. Sectoral final energy
demand is derived in the previous step and used here as an input. This module then calculates the
lowest-cost method of supplying this energy. REMIND-MAgPIE features a feedback mechanism where
this derived cost is fed back into the macro model to generate a new sectoral energy demand, iterating
until convergence.

o Energy conversion technologies. This is the process of converting raw resources (also known
as primary energy) into secondary energy, such as electricity. Secondary energy is then
subsequently also converted into final energy. All 3 IAMs model technological investment in
vintages. Each vintage is associated with fixed costs associated with the initial investment,
variable costs associated with the running of the investment, costs from early retirement, and
depreciation from age. This feature introduces frictions that prevent instantaneous switching
between technologies and represents capital stock inertia. MESSAGEix and REMIND
explicitly model cost minimization between these technologies either using constant
elasticity of substitution (CES) production functions in REMIND or linear substitution in
MESSAGE. In contrast, GCAM models the discrete choice using logit functions. This means
that even if the cost is unambiguously lower in one technology vs. the other, a certain share
will mechanically remain with the less efficient technology depending on logit exponent.

o Raw energy supply curves. This determines the cost of extracting raw resources both in the
case of renewables as well as fossil fuel and other non-renewable resources. In each model,
higher grades of resources reflect those that are easiest to exploit, such as land with high solar
irradiance or easily accessible fossil fuel reserves. In the case of the models that linearly
optimize energy cost at some level, REMIND and MESSAGEix, this introduces convexities
that further prevent corner solutions in cost minimization, i.e., 100% adoption of the lowest
cost technology.

Carbon prices are the primary channel for which transition scenarios are implemented in the IAMs, where,
roughly speaking, the price is set such that the emissions constraints applied to the scenario are satisfied. This
occurs through substitution effects, where the emissions price raises the cost of operating the given emitting
technology. In REMIND-MAgPIE, MESSAGEix-GLOBIOM and GCAM this triggers an endogenous shift in
investment in energy conversion technologies, with increased capacity for cheaper technologies and reduced
usage and early retirement for more expensive technologies.

Water and land-use systems such as GLOBIOM in MESSAGEix, MAgPIE in REMIND, and integrated modules in
GCAM operate parallel to the core energy and economic modules, optimizing agriculture, forestry, and other
land use (AFOLU) in concert with policy changes. For example, greater production and population demands
would negatively impact forest cover, reducing carbon uptake in forests and changing the supply of biomass
for bioenergy and lumber. This, in turn, affects emissions and estimates of the carbon price for any given
scenario.

33
 US$2010/t World, Price|Carbon, Net Zero 2050 scenario
2500

2000

1500

1000

500

0
2020 2030 2040 2050 2060 2070 2080 2090 2100

GCAM 6.0 NGFS MESSAGEix-GLOBIOM 1.1-M-R12 REMIND-MAgPIE 3.2-4.6

Figure 9. Carbon price IAM comparison in Net Zero Scenario.

A visualization of the derived carbon tax trajectories for the Net Zero 2050 scenario is shown in Figure 9, with
REMIND and GCAM broadly consistent in initial years and MESSAGEix diverging on the entire path. The
MESSAGE price path here is strongly determined by the peak-warming/net-zero definition of the scenarios.
The prices increase strongly until net-zero CO2 emissions are reached, after which the emissions remain stable
at net-zero, which happens at lower costs. The following section identifies the most critical shared drivers in
determining how each IAM would model this transition.

3.2 The NGFS IAMs – key dynamics driving an energy transition

Each model has different effective substitutability of energy sources, which is crucial in determining the speed
of energy transition and carbon prices for each scenario. This is both due to explicit assumptions and differing
structure.

For instance, MESSAGEix and REMIND determine energy technology through optimization using constant
elasticity of substitution (CES) production functions. The CES parameters of this function, as well as the
nesting structure of the CES goods, determine the substitutability of the energy sources. For example, REMIND
assumes perfect substitution at the lowest level of energy good, such as power, following the intuition that
electrons from fossil fuel or renewables are indistinguishable in their ability to fulfil electricity demand and
similar for liquids, gases, and others. However, this energy demand is still segmented by sector. So, power to
the transport sector cannot be substituted for power to stationary energy-use sectors, as seen in Figure 10.
Conversely, imperfect substitution is only explicitly modelled at the sectoral energy demand level in MESSAGE.
Instead, more detailed process-based modelling of frictions at the lower level drives imperfect substitution of
energy sources.

34
 Figure 10: REMIND production structure. σ refers to the CES substitution parameter.

In contrast, GCAM uses a vintage capital model of capital investment and utilization. Existing vintages are
assumed to operate throughout their useful life as long as the vintage is able to cover its operating costs. Costs
include energy and other operating costs such as labor, water and new investments compete based on expected
levelized cost of production. The distribution of investments into new capital is modelled using a discrete
choice logit model that determines the share of new investment based on expected cost. Technologies with
the lowest expected cost of production receive the largest share of the portfolio, with more expensive
technology options garnering a smaller share of the portfolio, with the size of this portion depending on an
exogenously set exponent parameter. In this framework, this parameter explicitly controls substitution
between technologies, although other factors would also modify this implicitly. The scale of investment in the
new vintage is set by the expected need for new capacity, which in turn is determined by difference between
expected demand for the sector’s output and capacity available from existing vintages of capital. For example,
see Figure 11 for an illustration of hydrogen transmission, distribution, and end-use.

35
 Figure 11: GCAM hydrogen technology nesting structure, including approximate energy requirements at each stage.

Even with full substitutability assumed for an energy input, several frictions at the lower level in the IAMs would
continue to impede instantaneous switching to the lower-cost option. The first is associated with the vintage-
style modelling of energy conversion technologies. This feature represents the inertia in the energy system
due to its long-lived capital stock. It includes both fixed costs in installing new capacity and hard constraints on
the early retirement of old capacity. The second is the presence of convex resource supply curves, representing
the increasing marginal costs of extracting resources as quantity increases. For example, using larger quantities
of specific resources, such as coal, would entail mining in more difficult-to-access regions, represented as
grades. These energy resource endowment curves are derived from the bottom up, using sources such as the
US Geological Survey and various energy institutes and agencies. An example of oil in MESSAGE is given in
Figure 12.

Figure 12: Oil resource endowment curve, MESSAGE.

A final component essential for the determination of endogenous energy mix is technological change and
diffusion. In REMIND, this is partly endogenous through a model of “learning-by-doing,” with global learning
curves and internalized spillovers for wind and solar power, and advanced vehicle and energy storage
technology. This causes capital costs to decrease while cumulated capacity increases. In MESSAGE, there is a

36
modelling of technology diffusion that produces a similar effect, with dynamic constraints that relate the
construction of a technology added in period t to the construction or activity in period t-1. In all models,
exogenous paths for cost and efficiency parameters in each scenario follow assumptions implied by the SSPs.

EJ/yr World, Secondary Energy|Electricity|Gas Net zero 2050 Scenario

30

25

20

15

10

0
2020 2030 2040 2050 2060 2070 2080 2090 2100

GCAM 6.0 NGFS MESSAGEix-GLOBIOM 1.1-M-R12 REMIND-MAgPIE 3.2-4.6

Figure 13: Gas-powered electricity production, net zero scenario.

Other mechanisms that would affect transition dynamics between the IAMs include negative emissions
technologies like carbon capture and storage (CCS), which is incorporated in differing sectors depending on the
IAM. As expected, the CCS option allows for a certain amount of fossil fuels to be burned even in a net zero
scenario. For example, the prominence of CCS in MESSAGEix explains the high level of gas-powered electricity
production in the net zero scenario shown in Figure 13 in combination with high energy demand and the role
of gas power plants to balance the load of variable renewables). Generally, though, it is typically represented
as a costly option. Other negative emissions technologies typically go through the land-use modules, including
soil carbon management, direct ocean capture, and forest restoration. However, IAMs vary depending on which
technologies they include.

In addition, there are various other ways through which energy technologies materially affect transition speed.
For example, two IAMs model the intermittency and lack of flexibility inherent in some energy sources,
particularly wind and solar, and require additional investment in energy storage capacity to bridge the gap. This
is parametrized as a reliability factor in MESSAGE, while GCAM models the costs of intermittent and non-
intermittent conversion technologies separately.

Emissions are calculated in separate modules and are typically a function of sectoral energy usage, technology
choice and emissions mitigation, interacting with the land/water-use modules. This includes emissions from
power generation, limestone production in cement, steel production, and other industrial CO2 production. In
all IAMs, these are then fed into separate climate modules that represent the global carbon cycle and
atmospheric chemistry and produce estimates of atmospheric composition, radiative forcing, and mean global
surface temperature. MAGICC performs the endogenous climate modelling in MESSAGEix-GLOBIOM and
REMIND. HECTOR produces the endogenous estimates in GCAM. The IAMs also model other pollutants to
some extent, such as sulphur dioxide (SO2), nitrogen oxides (NOx), and ammonia (NH3). The NGFS scenario

37
 database includes harmonized climate modelling results for all three models from the latest version of MAGICC
(see Box: MAGICC: A reduced complexity Earth system model)

Table 4. Key model characteristics.

MESSAGEix-GLOBIOM
REMIND-MAgPIE 3.2-4.6 GCAM 6.0
1.1-M-R12

Hosting Institution PIK IIASA PNNL

Economic Equilibrium General equilibrium (GE) GE Partial equilibrium (PE)

Agriculture Sector PE with recursive dynamic GE with intertemporal PE with recursive dynamic
Modelling optimization

Solution method Inter-temporal Inter-temporal Recursive dynamic

optimization optimization

Energy technology Modelling of supply Modelling of supply Modelling considering choice

diffusion energy based on cost energy based on cost function (cost plus penalty cost
optimization optimization due to inconvenience)

Cross sectors Primary energy supply, transformation, manufacturing, and end uses, including residential,
commercial, transport, construction, agriculture, and forestry

Model specific (singular)

sectors

Transport Road, rail, air, and sea; Aggregated Road, rail, air, and sea;
breakdown by freight and breakdown by freight and
passenger and type of passenger and type of vehicle
vehicle

Buildings Residential and Aggregated Residential floor space and

commercial floor space commercial floor space
determine scale of demands for
building services (e.g. heating,
cooling, cooking)

Industry Cement, chemicals and Cement, chemicals (high Cement, chemicals (ammonia)
steel value chemicals) non- non-ferrous metal and steel;
ferrous metal and steel breakdown by technology
investment and type of fuel

Regions 12 12 32

Technological change Endogenous Exogenous Exogenous

Behavioural change Only in Low Demand Yes No

38
Number of policy 9 7 14
instruments

Demand side mitigation 15 16 14

options

Supply side mitigation 17 20 18

options

AFOLU options 7 8 8

Freight electrification No No Yes

Technology substitution high substitutability with High substitutability with Mixed high and low
increment cost increment cost substitutability with growth
constraints

Technology lifetime Fixed lifetime and early Fixed lifetime and early Fixed lifetime and early
retirement retirement retirement

CCS Technologies Included in electricity Included in electricity Included in electricity, refining,

technologies, bioenergy technologies, bioenergy hydrogen production, and
with CCS, afforestation, with CCS, reforestation manufacturing technologies,
direct air capture and and afforestation bioenergy with CCS,
enhanced weathering reforestation and afforestation

One notable difference in approaches is in the level of foresight. As mentioned previously, GCAM assumes that
investors have myopic foresight and assume that current prices, including for example, the price of carbon, will
persist throughout the future life of the investment. Hence, costs for capacity installation are assumed to be
relative to current prices and not expected future prices. In contrast, REMIND and MESSAGEix assume perfect
foresight in that the full path of the carbon price is known to agents while investing. Ceteris paribus, one could
expect a sharper reaction to a net zero announcement in the perfect foresight case. However, this is not
immediately clear when looking at a between-IAM comparison due to other model differences.

In addition, the solution method differs between the IAMs, with MESSAGEix and REMIND being general
equilibrium models. One of the consequences is that carbon revenues can be recycled back into the economy
for MESSAGEix and REMIND, increasing production, while this is not feasible in GCAM. On the other hand,
GCAM clears all energy, agriculture, land, and water markets, but does not model other markets explicitly.
However, GCAM is able to model over twice the number of regions and several more key sectors.

For a summary of high-level differences between the IAMs, see Table 4.

3.3 The NGFS IAMs – implications of differences and similarities.

While the previous section illustrates some of the key structural differences between IAMs, the conceptual
underpinnings and goals of the models are broadly aligned. Indeed, most variables have an estimate from
each IAM, with differences confined to small variations in sectoral and regional granularity. This means that
variability observed between models can safely be interpreted as a measure of confidence. This is especially
useful as the models otherwise are deterministic and do not have a concept of uncertainty. In addition, as time
horizons are in the distant future, there is no possibility of direct validation. Hence, sensitivity and robustness
analyses are even more central to ensuring model trust.

39
Suppose a model like MESSAGE, which relies strongly on low-level process-based modelling, agrees on a
variable with a model like REMIND, which relies more on a nested CES structure. The convergence of these two
methodologies would provide additional assurance on robustness for downstream users. On the other hand, if
they disagree, one can pinpoint the methodological divergence that causes the difference, providing insights
into potential areas of uncertainty.23 A straightforward example is related to the CCS cost differences between
models, which causes significant differences in fossil fuel electricity production in the net zero scenario.
However, on many of the most critical metrics, the IAMs are closely aligned – Figure 14 shows an overview for
many of the main energy series.

Figure 14 Energy series dynamics comparison between the IAMs. Metric shown is the absolute difference between the net zero
and current scenario for the year 2050.

A final benefit to the suite-of-model method lies in the differing granularity of the estimates. By necessity,
models with more complicated estimation routines must make trade-offs in terms of granularity. On one end,
GCAM estimates significantly more regions and sectors while maintaining broadly consistent dynamics with
the other IAMs. Conversely, MESSAGEix-GLOBIOM has a less granular representation of sectors and regions
but is grounded in sophisticated process-based modelling at the low-level and conceptually sound
macroeconomic modelling at the high level. REMIND-MAgPIE provides a middle ground in sectoral
representation, highly reliant on a CES nesting structure, with a geographic representation like MESSAGE, but
leveraging a partial equilibrium approach to interact with the agricultural module. This wide range of
approaches leads to different coverage and provides a flexible menu of options for various potential use cases.
However, this flexibility increases the importance that users deep-dive into the methodology of each model to
understand which provides the best fitness for purpose for their specific area of analysis.

23 Multi model analysis for climate and integrated assessment models is a common approach in the literature. See
https://iopscience.iop.org/article/10.1088/1748-9326/aaf8f9/meta. This is also the approach taken for the climate
models prepared for the IPCC, which resulted in the Coupled Model Intercomparison Project, currently in its sixth
phase (CMIP6).

40
 Box: MAGICC: A reduced complexity Earth system model

What is the MAGICC model?

MAGICC is a reduced complexity Earth system model that has been widely used in climate science for over
three decades, most notably in multiple IPCC reports. It is most often used in a probabilistic setup, providing
information not only about our best estimate of future climate change but also the uncertainty that arises from
interactions between the Earth system’s many components.

 MAGICC is used for evaluating the impacts of greenhouse gas emissions on global climate change. It
combines climate science, atmospheric chemistry, and radiative forcing calculations to estimate the
relationship between greenhouse gas emissions and changes in temperature, as well as other related
climate variables under different emissions scenarios.

 MAGICC is also used as the climate component in multiple integrated assessment models (IAMs), including
REMIND and MESSAGE24. The strength of MAGICC is that it is sufficiently flexible to closely emulate the
large and complex climate models and sufficiently physically based to allow credible interpolations and
indicative extrapolation near the calibration range.

 The key outputs of the climate model include climate feedbacks which are processes that amplify or
dampen the initial response to greenhouse gas emissions, radiative forcing caused by greenhouse gases,
aerosols, and other factors, and climate sensitivity, a crucial parameter in estimating the temperature
response to greenhouse gas emissions.

 For the NGFS scenarios25, MAGICC 7.5.3 is applied in a probabilistic setup as used in the Sixth Assessment
Report of the Intergovernmental Panel on Climate Change (IPCC) (IPCC WG3 2022; Kikstra et al. 2022).
This ensures comparability of the NGFS scenario climate outcomes with the latest IPCC report and
assessment.

What are the key model inputs?

MAGICC’s key input are anthropogenic emissions that impact the climate system, primarily greenhouse gases
but also aerosol precursors and emissions that influence other gas cycles such as carbon monoxide. Based on
these inputs, MAGICC provides projections of a number of key quantities, including atmospheric greenhouse
gas concentrations, effective radiative forcing for different species, temperature change, Earth system heat
uptake and sea-level rise. Global-mean quantities are the key output, but outputs at the hemispheric level can
also be used in more specialized applications.

24 See post-processing sections under the IAM chapters in Module 5: Chronic physical risks.

25 The “Temperature|Global Mean” variable that existed in Phase 3 data for REMIND and GCAM and was not harmonized
across models was superseded by “AR6 climate diagnostics|Surface Temperature (GSAT)|MAGICCv7.5.3|50.0th
Percentile”.

41
 Figure 15. Schematic overview of MAGICC calculations showing the key steps from emissions to
global and hemispheric climate responses.

MAGICC represents the Earth with four boxes: one for land and one for the ocean, in each hemisphere. The
ocean component is an upwelling-diffusion model with multiple layers. The atmospheric component is based
on the energy balance equation, modified to support MAGICC’s four box structure. In addition to its core energy
balance components, MAGICC includes models of the carbon cycle, methane cycle, impact of anthropogenic
aerosol emissions and sea-level rise.

MAGICC’s development is led by Prof. Malte Meinshausen at the University of Melbourne. A full description of
MAGICC can be found in Meinshausen et al. (2011), with updates as described in Meinshausen et al. (2020) and
Nicholls et al. (2021).

42
3.4 Physical risk modelling approach

Physical risk modelling is split between chronic and acute risk. Chronic risk macro-economic impacts are
calculated with the damage function developed by Kalkuhl and Wenz (2020), used to quantify the effect of a
change in climate-related variables (temperature) on economic output. This approach has several advantages,
including its easy implementation for a large range of countries, but it is still an area of research and most likely
does not capture the full extent of climate change yet. The modelling approach has not changed compared to
the previous phase; hence additional impacts derive from updated temperature paths.

Figure 16. Gross Domestic Product deviation due to chronic physical risk between Phase 3 and 4. Nigem with REMIND inputs

Acute risk is modelled instead focusing on specific climate risks, or hazards, and their potential increase for a
given temperature pathway. This approach captures better countries idiosyncrasies and allows to better
identify the channel through which climate risk might affect the economy. Probabilities of damages are
estimated on the basis of various data sources (mentioned in the dedicated section) and macroeconomic
impacts estimated via stochastic simulations (to also control for the uncertainty surrounding the estimates) in
NiGEM on the basis of the relevant transmission channels. However, challenges are posed by the lack of reliable
and complete datasets and improvements in modelling should be matched by an increasing effort in collecting
data and observation.

43
Figure 17. implementation of climate shock in the NiGEM macreconomic model

44
4. Main results of the NGFS scenarios26
Key messages
 Reaching global net zero CO2 emissions by 2050 will require ambitious transition efforts
across all sectors of the economy. The NGFS scenarios, however, show that immediate
coordinated transition will be less costly than inaction or disorderly transition in the long run.
 More precisely, physical risks in hot house world scenarios (Current Policies or Nationally
Determined Contributions scenarios) will lead to the strongest negative impacts on GDP
with economic cost diverging significantly after 2040.
 Reducing greenhouse gas emissions continues to affect all sectors of the economy and
gives rise to transition risks for the economy and financial system
 Policy intensity increases as timelines for net zero 2050 scenarios shorten. Precisely, the
shadow emissions prices continue to rise drastically in respective scenarios.

 Energy is a key sector. Extensive energy investments pathways continue to rise. Until 2030
all scenarios predict a rapid scale-up of spending on overall energy supply.
 Global mean temperatures strongly depend on the policy assumptions of the respective
scenarios. In general, the modelled pathways reach from 1.5 to above 3 degrees.

This section presents key results from the seven NGFS scenarios described above and corresponding to the four
quadrants of the NGFS scenarios framework as follows:

 Orderly scenarios: Low Demand, Below 2ᵒC and Net Zero 2050

 Disorderly scenario: Delayed Transition

 Hot house world scenarios: Current policies and NDCs

 Too-little-too-late scenario: Fragmented World

The results can differ based on the model and the type of climate risk considered (transition and/or physical
risk). The macro-financial impacts of transition and physical risks are expressed as deviations from a baseline
scenario, where there is no climate-related risk. The NiGEM model is able to determine the contribution of each
type of risk (transition or physical) in deviation from this hypothetical scenario.

The time horizon covers a period from 2020 until 2050 or 2100, depending on the variable. The geographic
sample includes more than 180 countries and more than 30 regions. The following sub-sections describe the
key output variables available in the NGFS scenarios distinguishing between three categories: macro-financial
impacts, transition risk and physical risk.

26 The plots of this section have been generated with this EnTry script.

45
4.1 Key macro-financial impacts

Gross domestic product

NGFS scenarios differ markedly in their economic impact, with some difference across models and
significant variation across regions. Impacts on GDP are specified relative to a forecast representing prior
trends but also incorporating most recent impacts (e.g., the consequences of the Russian war in Ukraine).
Transition risks have a moderately negative impact on world GDP in Net Zero 2050 as negative impacts on
demand from higher carbon prices and energy costs are partially offset by the recycling of carbon revenues into
government investment and lower employment taxes. GDP impacts from transition risks are more markedly
negative in the Disorderly scenarios as the speed of the transition combined with investment uncertainty
affects consumption and investment. GDP losses from physical risks vary in line with different temperatures
projected for each scenario. Chronic physical risk becomes gradually more important over time – both in
absolute and relative terms – but acute physical risk remains the main source of risk until 2050. In all scenarios,
the impact of physical risk rapidly outweighs the impact of transition efforts. Indeed, the GDP deviation from
baseline due to the combined impact of transition, chronic and acute risks ranges from 5 to 7 pp over the next
30 years in the Net Zero 2050 scenario, while it reaches almost 10 pp in 2040 and 14 pp in 2050 in the Current
Policies scenario. Stringent mitigation initiatives that are in line with the Net Zero 2050 scenario will already be
beneficial by 2040 and strongly reduce risks towards the end of the century. This also underlines the need to
add investments on adaptation. Finally, impacts diverge even more thereafter. By 2100 impacts are highest in
the Current Policies scenario (up to 20 % of GDP relative to prior trend) as temperature targets and the
corresponding decarbonization efforts are missed.

46
 Figure 18. GDP deviation due to transition, chronic and acute risks - REMIND model

The NiGEM model provides economic impacts per country and region, giving estimates of country’s
exposure to transition and physical risks. In the NGFS scenarios, both transition and physical risk impacts vary
across countries based on several factors. Transition risk depends, among others, on the structure of the
economy, the reliance on fossil fuels and the trade composition. Physical risk depends on the exposure and
vulnerability to temperature increase and extreme weather events, with tropical and subtropical regions facing
larger risk increases. NiGEM provides country and regional pathways for GDP. Impacts are higher for countries
and regions that face higher emissions reduction, higher carbon prices, lower fossil fuel exports or higher
physical risk damages. Impacts also vary across models, depending on model structure and assumptions. To
estimate GDP impacts, the NiGEM model is calibrated based on inputs from the three IAMs (REMIND,
MESSAGE, GCAM). Results from MESSAGE are more adverse because of lower CDR use, more ambitious
temperature outcomes and stronger decarbonization strategies needed due to structurally higher energy
demand in MESSAGE, therefore inducing higher carbon prices.

47
 Figure 19. GDP deviation relative to Baseline across countries and models in the Net Zero 2050 scenario.

Inflation and unemployment

The scenarios include a wide range of macroeconomic variables, capturing structural relationships
between key aggregates such as unemployment and inflation. In many countries, the implementation of
carbon prices in the transition scenarios tends to raise energy costs in the short-term, initially weighing down
on prices (as lower demand and financial market losses hit outputs). Rising carbon prices subsequently induce
modest increases in inflation and unemployment before returning to prior trends. In some countries and
periods, the offsetting growth effects stemming from carbon revenue recycling leads to a reduction in
unemployment. In some scenarios this leads to a potential monetary policy trade-off. The NGFS modelling

48
framework assumes a ‘two-pillar’ strategy, targeting a combination of inflation and nominal GDP as a default.
This can be adjusted in the NiGEM model alongside fiscal policy assumptions. The negligible impacts in the
Current Policies scenario reflect not only limited transition risk, but also the fact that only one potential chronic
physical risk transmission channel (productivity) has been modelled. More research is needed on the potential
for climate impacts to raise inflation (e.g., through supply-side shortages) and/or unemployment (e.g., due to
displacement). The impact is much more sizable in the Too-little-to-late scenario, given the higher transition
risk.

Figure 20. Inflation rate and unemployment rate: Europe vs. China

Financial Markets

Climate change and transition policies create significant financial fluctuations. The macro-financial results
reflect both risks and opportunities. Long-term interest rates tend to increase in the transition scenarios,
reflecting the inflationary pressure created by carbon prices, as well as the increased investment demand that
the transition spurs. A disorderly transition can affect real financial asset valuations significantly, with
considerable differences across regions. Although the NiGEM model’s results cannot be disaggregated into
individual sectors, it is likely that sectors that can decarbonize to a lesser extent will be affected more than other
sectors. The NGFS is working to further develop sectoral impacts. In the disorderly transition scenarios, it is
assumed that policy uncertainty leads to a higher investment premium for two years, with the premium
gradually returning to the baseline thereafter. This occurs in a period ranging between 2030 and 2031 in both
the Delayed Transition and the Fragmented World scenarios.

49
 Figure 21. Long-term interest rate in Europe vs. Long-Term interest rate in ChinaTransition risk

Understanding transition risk

Cutting greenhouse gas emissions affects all sectors of the economy and gives rise to transition risks for
the economy and financial system. Transitioning away from fossil fuels and carbon intensive production and
consumption requires a significant shift towards emission-neutral alternatives in all sectors. Policy makers can
induce this transition by increasing the implicit cost of emissions. As it takes time to develop and deploy
alternative technologies, climate policies may lead to higher costs in the interim. The transition pathways have
been modelled using three detailed Integrated Assessment Models (IAMs) 27. They can be used to assess the
changes in energy, land-use and policy needed to meet a particular temperature outcome or carbon budget.
The shadow carbon price underpinning those changes has been derived for each model. This price is distinct of,
and may differ from, the social cost of carbon, which depends on an assessment of avoided damages and
valuing impacts on present versus future generations. The increased cost of emissions, combined with a
consistent (re-)allocation of investments and the employment of Carbon Dioxide Removal technologies, results
in a reduction in GHG emission in all sectors under the Net Zero scenario. The energy supply sector is projected
to experience the largest drop in those emissions, followed by industry.

27 These models have been used extensively to inform policy and decision makers and feature in several climate
assessment reports, c.f. IPCC, 2018, IPCC, 2022, UNEP, 2018.

50
 Figure 22. Sectoral GHG (Kyoto Gases) emissions – Net Zero 2050 scenario, REMIND

Emissions prices

A key indicator of the level of transition risk is the shadow emissions price, a proxy for government policy
intensity and changes in technology and consumer preferences. In the IAMs used to produce the NGFS
scenarios, a higher emissions price28 implies more stringent policy. Models suggest that a carbon price of
around $(2010)200/ton would be needed and assumed to be applied by all the countries in the next decade to
encourage a transition towards net zero by 2050. The largest increase in carbon price is projected under the Net
Zero 2050 scenario, reaching over 600$(2010)/ton in 2050. The different IAMs’ 2050 projections of the carbon
prices under the Net Zero scenario vary within a 1.5°C range of temperature and $(2010) 1150 /ton range of carbon
price.

This shadow price is a measure of overall policy intensity. Governments are pursuing a range of fiscal and
regulatory policies, which have varying costs and benefits. Shadow emission prices are sensitive to:

 The level of ambition to mitigate climate change. Higher ambition translates into higher emissions
prices.

 The timing of policy implementation. Higher emissions prices are needed in the medium to long-
term if action is delayed.

 The distribution of policy measures across sectors and regions, which are assumed to be
differentiated in the Fragmented World scenario.

 Technology assumptions such as the availability and viability of carbon dioxide removal

28 Emissions prices are defined as the marginal abatement cost of an incremental tonne of greenhouse gas emissions.
Prices are influenced by the stringency of policy as well as how technology costs will evolve. Prices tend to be lower in
emerging economies as there tends to be a greater number of low-cost abatement options still available.

51
 Figure 23. Carbon price development

Figure 24. Carbon price across models – Net Zero 2050 scenario

Carbon dioxide removal

Large-scale Carbon dioxide removal (CDR), also known as negative emissions, or carbon drawdown, aims
at addressing the primary human source of climate change by removing carbon dioxide (CO2) permanently
from the atmosphere. CDR encompasses a w.ide array of approaches, including removing carbon from the
atmosphere through increasing forest cover and soil sequestration (land use) or growing crops for bioenergy
(BioEnergy with Carbon Capture and Storage, BECCS).

CDR assumptions play an important role in IAMs. If deployed effectively, lower warming outcomes could be
achieved, or targets could be reached sooner given the practical difficulty of eliminating all (gross) emissions in
the near term. For example, under the Net Zero 2050 scenario, approximately 5 GtCO2 per year should have
been removed via CDR by 2050. However, these strategies currently take place on a limited scale only and face
several challenges.

The NGFS scenarios assume limited availability of these technologies. The patterns vary strongly across
models depending on cost assumptions. They also vary substantially across countries depending on the costs
and availability of CDR options. […]

52
 Figure 25. CO2 emissions in Net Zero 2050

Figure 26. CO2 removals across models

Energy Investment

Significant investment flows would need to be directed towards green energy in the coming decades to
achieve net zero. Transitioning to a net zero economy would require investment flows to be allocated towards
mass deployment of green electricity and electricity storage. There is some legacy capital investment in fossil
fuel extraction, which is a measure of all investments in mining, shipping and ports for fossil fuels, transmission,
and distribution for gas as well as the transport and refining of oil to maintain the infrastructure while
decreasing the overall capacity. Average annual investment by 2050 in renewable electricity and storage
amounts to about 1.8 trillion US$ under the Net Zero scenario, about 0.5 trillion more than under the current
policies. Given its high CO2 emissions relative to other fossil fuel alternatives, the share of coal is rapidly
dwindling in the energy mix from 28% in 2020 to 7% in 2030 and close to 0% in 2050 in the Net Zero 2050
scenario. By 2050, renewables and biomass will deliver roughly 75% of global primary energy needs in the same
scenario. This is in marked contrast to the Current Policies scenario where fossil fuels continue to be the
dominant source of primary energy, even after accounting for current technology trends.

53
 Figure 27. Current and expected annual energy investments until 2050

Figure 28. Energy mix per scenario

4.2 Physical risk

Temperature increases

Mean temperatures rise in all scenarios, exceeding 3°C in Current Policies. Changing climate conditions
affect physical labour productivity and lead to severe impacts. Projected annual average temperatures are
estimated to increase across scenarios, however with different magnitudes, exceeding 3°C in later decades of

54
the 21st century in the current policies scenario. At present, global mean temperatures have increased by
around 1.1°C from pre-industrial levels and have accelerated since 1975 at a rate of approximately 0.15 to 0.20
°C per decade (according to NASA’s Goddard Institute for Space Studies (GISS)).

Deep reductions in emissions are needed to limit the rise in global mean temperatures to below 1.5°C or
2°C by the end of the century. This does not occur in the Current Policies scenario, leading to a temperature
rise exceeding 3°C with severe impacts. Temperatures are increasing unevenly across the world with dry land
warming faster than oceans and regions located at high latitudes experiencing greater warming. Temperature
changes lead to chronic changes in living conditions affecting health, labour productivity, agriculture,
ecosystems, and sea-level rise. It is also changing the frequency and severity of acute weather events such as
heatwaves, droughts, wildfires, hurricanes, tropical cyclones, and flooding. Figure 29 shows estimated global
temperature dynamics across various scenarios.

Figure 29. Global temperature dynamics

GDP loss estimates stemming from acute physical risks

Observed temperature increase of 1.1°C has already more than doubled both the global land area and the global
population annually exposed to riverine flooding, crop failures, hurricanes, tropical cyclones, wildfire, droughts,
and heatwaves (Lange et al., 2020). Global temperature increases of 2°C relative to preindustrial conditions are
projected to lead to a fivefold increase in exposure to all types of natural hazards worldwide. The most
pronounced increases are expected to stem from droughts and heatwaves. Changes in exposure are unevenly
distributed across the globe, with tropical and subtropical regions facing larger increases than regions situated
at higher latitudes. The NGFS scenarios now include estimates of global GDP impacts emanating from acute
physical risks29 for three NGFS scenarios. Information from the international disaster database Emergency
Events Database (EM-DAT) is used to approximate historic damages from weather-related extreme events to
derive stochastic shocks as inputs to the NiGEM model. Projections for selected Climate Impact Explorer (CIE)
indicators30 are used to derive changes to projected damages for the three NGFS scenarios in the CIE. GDP is
projected to fall compared to the baseline scenario more rapidly under Current Policies. While the drop in GDP
is projected also under the Net Zero and Delayed Transition scenarios, it is overall milder. Figure 30 shows GDP
loss estimates across scenarios:

29 The impact of acute physical risk on macro-financial variables other than GDP is not available in this iteration of the
NGFS Scenarios
30 Acute risks for these projected damages include cyclones and riverine flooding based on the CLIMADA model. See
https://climate-impact-explorer.climateanalytics.org/methodology/

55
 Figure 30 . GDP loss estimates from acute risk across scenarios (NiGEM model with REMIND input).

GDP loss estimates stemming from chronic physical risks

Estimates of GDP losses from chronic risks vary considerably depending on assumptions about climate
sensitivity and the method used to estimate the damages. Estimates suggest a global GDP impact of up to 18%
relative to a prior trends baseline in the current policies scenario; GDP decrease would be more contained, but
still sizeable, in the Net Zero scenario. Losses are much higher in tropical regions under any scenario. The NGFS
estimates have been updated to account for model uncertainty and now include higher damages. GDP losses
were calculated based on the methodology set out in Kalkuhl and Wenz (2020) at the country level for the
change in average temperature in each scenario compared to the previous year. The methodology does not
include impacts related to extreme weather, sea-level rise or wider societal impacts from migration or conflict.
For given countries these would likely strongly increase the physical risk. These estimates also do not fully
capture adaptation, which would reduce impacts but require significant investment.

Figure 31. GDP loss estimates by country – Current Policies scenario

56
5. Phase III vs. Phase IV: what is new in the main results of NGFS
scenarios?

Key messages
 The NGFS scenarios have been brought up to date with new economic and climate data,
policy commitments and model versions, as usual. Phase IV accounts for the post-covid
recovery and the Russian war in Ukraine.

 The possibility of using Carbon Dioxide Removal (CDR) technologies has been limited in
this vintage by switching off Direct Air Carbon Capture and Storage (DACCS) technologies
and limiting Bioenergy Carbon Capture and Storage (BECCS) capacities.

 The NGFS scenarios have been upgraded with improved and more granular estimates of
the macroeconomic impact of acute physical risks, based on new models and now
accounting for four hazards, therefore expanding the set of risk drivers considered.
 These changes, combined with the shortening of the time available to reach the temperature
targets, result in more disorderly scenarios and more adverse GDP impacts.

The NGFS scenarios have been brought up to date with new economic and climate data, policy
commitments and model versions. The NGFS Phase IV scenarios reflect the most recent economic events
that have occurred after the implementation of Phase III and the end of 2022, notably the resumption of global
economic growth after the Covid-19 crisis and Russia’s war on Ukraine. The scenarios also include data to reflect
the latest trends in renewable energy technologies (e.g., solar and wind), and key mitigation technologies. As
in previous phases, data for short-term GDP and final energy demand trajectories have been updated using
the latest snapshot from the IMF World Economic Outlook 2023. These trajectories are used as exogenous
variables in the IAMs. Similarly, Phase IV reflects the new country-level commitments made until March 2023.
It also considers new climate policy announcements, for example as part of European Union's Fit for 55
package or the Inflation Reduction Act in the USA.

The mapping of the NGFS scenarios has been updated accordingly within the NGFS scenario framework.
As described in NGFS scenario narratives, the new scenario positioning reflects an increased overall
disorderliness to account for the latest macroeconomic and geopolitical developments, as well as the
shortening of the time available to reach the temperature targets that are characterising and driving these
scenarios.

5.1 Accounting for the post-covid recovery and the Russian war in Ukraine

Since the implementation of Phase III, new shocks have modified the short-term macroeconomic outlook,
notably the post-covid economic recovery and the Russian war in Ukraine. As a result, the calibration of the
models has been modified for Phase IV to consider these shocks and their consequences on inflation and energy
markets.

The main consequence of the post-covid economic recovery is a rebound in Fossil Fuel and Industrial (FFI)
emissions. Indeed, FFI CO2 emissions in 2022 returned to 2019 levels of ~37 Gt CO 2 after the Covid shock with
its largest dip in 2020 (to ~35 Gt CO2) (GCP, 2022). As a result, this rebound generated higher FFI emissions than

57
expected by Phase III, and thus weakened the likelihood of achieving the Phase III emissions reductions at equal
cost. More concretely, this translates into higher shadow carbon price through 2025, relative to Phase III. 31

Emissions

In the Net Zero 2050 scenario, Phase IV assumes higher emissions in the short/medium term compared to Phase
III, in line with the latest trends, and slightly lower emissions in the long term for all IAMs. The latter are
necessary to achieve the (unchanged) temperature targets despite differences in the short-term development
and the shortening of the time available. In the Current Policies scenario, emissions in the second half of the
century significantly lower in REMIND and MESSAGE models, reflecting the implementation of new policies
and the translation of previous commitments into action (Figure 32).

Figure 32. Phase III vs. Phase IV: Global CO2 emissions in Net Zero 2050 and Current Policies scenarios

Carbon prices

As described in the previous sections, NGFS (shadow) carbon prices are strongly linked to the emission
pathways. In the Net Zero 2050 scenario, carbon prices have much steeper trends compared to Phase III, mainly
due to adverse changes in the starting points coupled with the need to still reach global net zero CO2 around
2050. As for Current Policies scenarios, the new GCAM 6.0 model version uses shadow carbon constraints to
match the current policy trajectory with estimates from Climate Action Tracker, but does not account for these
in the carbon prices, as the majority of existing policies globally are not using carbon pricing. (Figure 33).

31 The use of
the updated GCAM 6.0 version resulted in slightly different model near-term dynamics, which had a bigger
impact than the updated historic record and near-term expectations.

58
 Figure 33. Phase III vs. Phase IV: Carbon Price in Net Zero 2050 and Current Policies scenarios

Energy-related variables

The Russian war in Ukraine has strongly disrupted the energy markets by rationing the supply of Russian gas to
Europe. The modelling of Phase IV has therefore reinforced the constraints on European gas supply in the long
term, assuming limits to European gas and fossil consumption to capture the energy-market disruptions.
Overall, global fossil energy consumption in Phase IV is lower in all models and scenarios, exception made for
MESSAGE in Net Zero (due to limitations in emission reduction in the year 2025) 2050 and GCAM in Current
Policies, where the impact is more than offset by changes in the modelling assumptions and dynamics. (Figure
34).

Figure 34. Phase III vs Phase IV: Fossil fuels consumption in Net Zero 2050 and Current Policies scenarios

59
At the same time, the consumption of renewables under both Net Zero 2050 and Current Policies scenarios is
higher in Phase IV (Figure 35).

Figure 35. Phase III vs Phase IV: Consumption of renewables in Net Zero 2050 and Current Policies scenarios.

When it comes to secondary energy, the major update is related to the electricity consumption. While showing
some heterogeneity in the Net Zero 2050 scenario, the models agree on a stronger electrification by the end of
the century with respect to Phase III in the Current Policies scenario (Figure 36).

60
 Figure 36. Phase III vs Phase IV: Electricity consumption in Net Zero 2050 and Current Policies scenarios

Phase IV features updated final energy prices. While the average electricity price for the industry sector differs
in GCAM and REMIND, the models agree on its increase with respect to Phase III. This applies almost to the
entire horizon, with differences in the shortterm being more marked, especially in the Current Policies scenarios
(Figure 37).

Figure 37. Phase III vs Phase IV: Industry-related energy prices

61
5.2 Updates on Carbon Dioxide Removal availability

Phase IV also includes limits to the availability of Carbon Direct Removal Technologies based on new
projections, resulting in a lower overall availability of these technologies compared to Phase III. This is
modelled via explicit constraints on the process level (time-dependent maximum area available for
afforestation, maximum yearly injection rate for geological sequestration, maximum yearly bioenergy
potentials). More concretely, Direct Air Carbon Capture and Storage (DACCS) technologies were switched off
in all scenarios and Bioenergy Carbon Capture and Storage (BECCS) capacities have been limited. Figure 38
shows the reduced BECCS capacity for REMIND and MESSAGE. The same changes in GCAM are offset by the
need to compensate for higher fossil fuel emissions compared to the other models.

Figure 38. BECCS availability in Orderly scenarios

5.3 Updates in the data model

An effort has been made to keep the NGFS Phase 4 data model as much in line with Phase 3 as possible.
Nonetheless, there are a few changes:

 The MESSAGE model (IAM and Downscaling) does not report GDP for the Low demand scenario as
they are consistent with the 1.5c scenario.
 The Divergent net Zero scenario has been terminated while a new one, Fragmented World has been
created.
 From the MAGICC climate model, we now report more percentiles ranging from 5th to 95th of
previously only the 50th percentile.
 In addition we also report atmospheric concentrations from MAGICC for CO2, CH4, and N2O.
 The damages post processing was moved under the downscaling model as it reports country level
data. Previously it was filed under the native IAM models.
 Due to limitation in Excel data size the Downscaling data have now been split into three files, one for
each IAM.

62
 Box: Towards a more disorderly climate transition
Since their inception, every update of the NGFS scenarios has shown more adverse macro-economic
impacts. There are two main reasons for this.

 The NGFS has been modelling more and more climate impact channels in the NGFS Scenarios, leading to
increasing estimates of physical risk impacts.

 The delay in political action makes it more and more difficult to reach climate targets in an orderly way,
leading to higher transition risks in each iteration.

More climate impact channels

Over time, the NGFS scenarios are becoming more and more detailed, but also increasingly adverse, with
the inclusion of additional transmission channels.

 In 2021, the NGFS scenarios included for the first time an assessment of chronic physical risks.

 In 2022, the modelling of chronic physical risks was improved to account for model uncertainty, leading to
higher estimates in the Hot-house world scenarios. For the first time, they additionally included a basic
assessment of acute physical risks, adding some adversity to all scenarios.

 In this 2023 edition, the modelling of acute physical risks was improved to include more hazards and
provide a country-level breakdown The inclusion of new hazards directly translates into an increased
adversity of the NGFS scenarios, while the country-level breakdown shows that the results are very
heterogeneous, with tropical countries being most exposed to both chronic and acute physical risks.

However, it is certain that the NGFS scenarios still underestimate the impact of physical risks on the
macro-economy, as many hazards and transmission channels are not modelled yet, and our understanding
of the links between climate change and the economy is still partial. The NGFS scenarios will continue to
evolve as common knowledge is being built in the academic and central banking communities.

Trend toward more adversity in a context of inaction

The increasing adversity of the orderly transition scenarios stems from persistently elevated levels of global
emissions against the backdrop of a limited carbon budget associated with reaching any particular climate goal
(e.g. net zero by 2050). In other words, as ambitious transition efforts at the global level are delayed and levels
of emissions remain elevated, carbon budget runs out, and more rapid and stringent action is needed to achieve
the same emission reductions over a shorter period. In addition, the IAMs calculate cost-efficient future
pathways, which usually represent smooth adjustments. Thus, not only are current emissions higher than what
would have been projected as an optimal pathway for reaching net zero by 2050 in Phase III, but also near-
future emissions will be higher until 2035, likely driven by the high costs that would be necessary for stronger
changes in the energy sector at an even shorter time scale.

Figure 39 shows the net zero emission pathways in the three most recent NGFS scenario vintages. The
pathways of this fourth iteration assume higher emissions in the mid-term future, and intensified emission
reduction in the long term compared to previous phases. The changes are quantitatively significant: the
remaining ‘carbon budget’ for the period 2025-2050 in the Net Zero 2050 scenario decreased from 275 Gt C02
in Phase III to 231 Gt C02 in Phase IV, corresponding to a 16% drop. In line with these changes to the emissions
pathways, the global carbon price in the NGFS Net Zero by 2050 scenario is higher at each point in time in the

63
future in Phase IV as compared to Phase III, indicating more pressure on high-emitting firms to mitigate their
emissions (

Figure 40).

Figure 39. Current emissions vs Phase I or II emission pathways in the Net Zero 2050 scenario.

Figure 40. Shadow carbon price, Phase III vs Phase IV, Net Zero 2050 scenario.

The energy sector is central for the low-carbon transition and likely to shoulder the strongest adjustments and
associated costs by a less orderly transition in Phase IV as compared to Phase III. Specifically, while overall
energy demand is projected to be higher in Phase IV as compared to Phase III, reflecting smooth trajectories
starting at higher base levels in 2025 as mentioned above, the aggregate development masks strong changes
in the energy mix over time. Specifically, fossil fuel demand reductions are much more intense in Phase IV as
compared to Phase III starting in 2040. The increase in energy demand is thus mostly driven by an increase in
primary energy based on Biomass or Non-Biomass Renewables. The fossil fuel demand reductions would be
felt most directly by companies in sectors highly reliant on fossil fuels, such as mining, manufacturing, and
utilities, who now have less time in switching their business processes to renewables-based activities.

64
 Figure 41. Energy investments by source, Phase III vs Phase IV, Net Zero 2050 scenario.

Facilitating the previously described changes in energy demand will require a faster and much more extensive
investment push in the energy sector in Phase IV as compared to Phase III (Figure 41). While overall energy
investments in Phase IV are projected to lag what was projected as needed in 2025 under Phase III, they are
much higher starting in 2030, reflecting the need for a strong investment push. The strongest increase in
investments across phases occurs between 2035 and 2055 and is focused on renewables-based electricity
generation, electricity storage, electricity transmission and distribution and hydrogen. The overall cumulative
increase in investments is quantitatively significant, in that, it requires an additional 20.2tn USD (578bn
USD/year) by 2050 and 34.4tn USD (530bn USD/year) by 2100 compared to phase III.

As a result, scenarios that were previously labelled as orderly shift towards the disorderly category. Net Zero
2050, which is the most ambitious scenario in terms of temperature target (<1.5°C), can hardly be called an
orderly scenario anymore, leaving room for a new Low Demand orderly scenario. This scenario uses new
hypotheses (some aligned with SSP1) and describes a world with rapid and important changes in consumer
behaviours (e.g., reduced plane transportation, meat consumption, etc.), leading to a reduced demand in
energy and CO2 emissions. The decrease in energy demand is achieved via improvement in energy efficiency
and a redirection of consumption towards less carbon-intensive goods and services. This consumer-led
decrease in energy demand (as opposed to a carbon price-led decrease) makes the transition to the 1.5°C target
more orderly.

65
Module 2: IAM – REMIND-MAgPIE

1. Non-technical summary

What is the REMIND-MAgPIE model?

REMIND and MAgPIE are two models developed at the Potsdam Institute for Climate Impact Research (PIK)
that were created over a decade ago (Leimbach et al., 2010a; Lotze-Campen et al., 2008) and are continually
being improved to provide up-to-date scientific evidence.
REMIND (REgional Model of INvestment and Development) is a numerical model that generates
projections for the future evolution of the world economies with a special focus on the development of the
energy sector and the implications for our world climate. The goal of REMIND is to find the optimal mix of
investments in the economy and the energy sectors of each of the 12 model regions given a set of population,
technology, policy, and climate constraints. It also accounts for regional trade characteristics on goods, energy
fuels, and emissions allowances. The most relevant greenhouse gas emissions due to human activities are
represented in the model.

MAgPIE (Model of Agricultural Production and its Impacts on the Environment) is a global land use
allocation model, which is in turn connected to the grid-based dynamic vegetation model LPJmL (Lund-
Potsdam-Jena managed Land) to simulate the interactions between the land surface and the atmosphere
as well as the impact of human activities on the environment. As a partial equilibrium model, the objective
function of MAgPIE is the fulfilment of agricultural demand for each region at minimum global costs under
consideration of biophysical and socioeconomic constraints. The MAgPIE results are consolidated to the 12
REMIND regions through a process called spatial aggregation or regional harmonization. This process involves
grouping or merging the individual regions into larger and more manageable units for analysis and modelling
purposes. The specific method of consolidation can vary depending on the specific requirements of the
modelling framework and the research objectives. Common approaches include geographical proximity,
economic similarities, administrative boundaries, and model requirements.

REMIND-MAgPIE aims to help policy and other decision makers to plan ahead by understanding the roles,
synergies and trade-offs between various factors, including population, resources, technologies, policies
and the environment. Using REMIND-MAgPIE, research and policy-relevant questions related to sustainability
can be explored: Which technologies should we use in the future? What is the impact of policy proposals that
are meant to prevent (mitigate) climate change? What are the consequences on economic development, air
pollution, and land use? For some questions, REMIND is used in connection with other models to allow the
analysis of other environmental impacts such as water demand, air pollution, health effects and climate
impacts. (see four main components of the REMIND-MAgPIE framework in Figure 42). One such model is
MAGICC (Model for the Assessment of Greenhouse Gas Induced Climate Change). This is a climate model,
which accounts for changes in climate-related variables like global surface mean temperature. The linkage to
MAGICC analyses the complex interactions between agriculture, land-use, greenhouse gas emissions, and
climate change. More details on MAGICC are provided in a dedicated box (Module 1).

66
 Figure 42. Overview of REMIND-MAgPIE framework

REMIND-MAgPIE is well equipped to capture the interactions between the energy transformation in
response to climate policies and economic development. Full macroeconomic integration is particularly
valuable for the assessment of effects of climate policies on the scarcity of energy carriers, demand response,
structural changes, investments, macroeconomic costs, and their regional distribution. Changing crucial
parameters in REMIND (such as the climate target or the availability of technologies or resources) can have
significant impact on GHG prices and bioenergy demand. Thus, REMIND and MAgPIE can be run in an iterative
soft-coupled mode, where REMIND updates MAgPIE's assumptions regarding bioenergy demand and GHG
prices, and MAgPIE, in turn, updates REMIND's assumptions regarding bioenergy prices and land-use emissions
and agricultural production costs. The iteration is continued until changes between iterations become
negligible. The resulting scenarios are consistent regarding the price and quantity of bioenergy and GHG
emissions.

The central strength of REMIND with its perfect foresight is its ability to calculate first-best mitigation
strategies that provide benchmark development scenarios with detailed representation of the key dynamics
related to the scale up of novel technologies and integration constraints in the power sector. These benchmark
scenarios allow for comparison with mitigation scenarios under second-best policy settings (regional or sectoral
fragmentation) or technology constraints. Within some numerical restrictions, the flexible spatial resolution of
REMIND enables the exploration of transformation pathways of the energy-economic system for specific
countries or global regions.

Figure 43. Regions in REMIND

67
What are the key model inputs?
Key model inputs are the available historical data, for example population, GDP, fossil resources, energy use,
emissions, land-use and vegetation patterns, capital stocks, and investments. They come from both
international organisations (World Bank, OECD, UN, IEA, …) and the academic literature. Additionally,
projections for future development are used, such as UN population projections, short-term IMF or World Bank
GDP projections, long-run projections on technological parameters and prices, again drawing on academic
literature. These datasets are used to calibrate the model to determine key model parameters.

What are the key model outputs?

The output of these models (with a given set of population, technology, policy and climate constraints) can help
policymakers and other relevant stakeholders evaluate the effectiveness and efficiency of different policy
interventions and identify optimal pathways for achieving sustainable development goals. It also accounts for
regional trade characteristics on goods, energy fuels, and emissions allowances and all greenhouse gas
emissions due to human activities.

What is new in the 2023 edition?

The REMIND-MAgPIE version 3.2-4.6 contains new datasets from UNFCCC, IEA WEO, UBA, IRENA, EDGAR7
and EEA to improve the quality of the calibration. The policy database for NDC and net Zero targets was
updated. Compared to Phase 3, the flexibility of the model to adapt its 2025 value was reduced. To reflect the
energy crisis, import restrictions on fossil gas in Europe were implemented.

68
2. Overview of model scope and methods

REMIND32 is a modular multiregional general equilibrium model linking a macro-economic growth model
with a bottom-up engineering-based energy system model. It uses non-linear optimization to derive welfare-
optimal regional transformation pathways of the energy-economic system subject to climate and sustainability
constraints for the time horizon from 2005 to 2100. REMIND operates at a time resolution of 5 years until 2060,
and 10 years thereafter to derive long-term projections. Using different scenario analysis techniques which
cover a wide array of factors, simulations can explore a range of possible futures. The resulting solution
corresponds to the decentralized market outcome under the assumptions of perfect foresight of economic
agents. In the integrated damage runs, external effects from climate damages are internalized into the
optimization function.

Figure 44: Structure of REMIND model

A Ramsey-type33 growth model with perfect foresight serves as a macro-economic core projecting growth,
savings and investments, factor incomes, energy, and material demand. The macroeconomic production
factors are capital, labour, and final energy. A nested production function with constant elasticity of
substitution (CES) determines the final energy demand. REMIND uses economic output for investments in the
macro-economic capital stock as well as for consumption, trade, and energy system expenditures.

32 Phase IV NGFS scenarios have used version 3.2 of REMIND. The last release of the model is available at
https://github.com/remindmodel/remind and https://doi.org/10.5281/zenodo.7852740. The REMIND Documentation is
available at: https://rse.pik-potsdam.de/doc/remind/3.2.0/.

33 In the Ramsey growth model, the investment share of economic output is determined endogenously to maximize inter-
temporal welfare. See Barro and Sala-i-Martin (2004) for an overview of these models.

69
The energy system representation differentiates between a variety of fossil, biogenic, nuclear, and renewable
energy resources. More than 50 technologies are available for the conversion of primary energy into secondary
energy carriers as well as for the distribution of secondary energy carriers into final energy.

The macroeconomic core and the energy system part are hard linked via final or useful energy demand (input
to the economy) and the costs incurred by the energy system (output of the economic part). Economic activity
results in demand for energy in different sectors (transport, industry, and building sectors) and of different
types (electric and non-electric). See Figure 44.

The model accounts for crucial drivers of energy system inertia and path dependencies by representing full
capacity vintage structure, technological learning of emergent new technologies, as well as adjustment costs
for rapidly expanding technologies.

Tax revenues are redistributed as a lump sum; thus, net taxes converge to zero in the optimal solution. REMIND
considers the trade of coal, gas, oil, biomass, uranium, the composite good (aggregated output of the
macroeconomic system) It assumes that renewable energy sources (other than biomass) and secondary energy
carriers are non-tradable across regions.

The emissions of greenhouse gases (GHGs) and air pollutants are largely represented by source and linked to
activities in the energy-economic system. Several energy sector policies are represented explicitly, including
energy-sector fuel taxes and consumer subsidies. The model also represents trade in energy resources.

2.1 Macro-Economic Module

The macroeconomic core of REMIND (Leimbach et al., 2010b, a; Bauer et al., 2012b; Luderer et al., 2012)
features a multiregional general equilibrium representation of the Ramsey growth model (i.e., the investment
share of economic output is determined endogenously to maximize intertemporal welfare). This approach is
well suited for describing patterns of long-term economic growth (Barro and Sala-i-Martin, 2004), which are
key drivers of energy demand and, thus, emissions.

Physical capital is a major driver of economic growth, and related investments are endogenous in such models.
In each period, the representative agent, endowed with perfect foresight, has to make the choice of using
output for consumption or for investment, which can be used to produce consumption goods tomorrow.
Perfect foresight is a standard assumption in economic models and widely used in IAMs (e.g., DICE/RICE,
Nordhaus and Yang, 1996; MERGE, Manne et al., 1995; MESSAGE, Fricko et al., 2017; WITCH, Bosetti et al.,
2007). While in the real-world agents rarely have perfect foresight, using this concept is a useful approximation
in a context of models with long planning horizons. When using the perfect foresight assumption to formulate
an intertemporal optimization problem, the model is completed by components (technically: side constraints)
that help to reproduce real-world dynamics caused by imperfectly foresighted decision-making (e.g.,
adjustment costs for the increase of the macroeconomic capital stock). In REMIND, each region maximizes its
welfare subject to a budget constraint.

The model explicitly represents trade in final composite good, primary energy carriers, and, if certain climate
policies are enabled, emissions allowances. Thus, equilibrium refers to the balance in goods markets and
international trade, such as the global oil market. It is a valid assumption for the decadal timescales considered
in scenarios and, thus, does not compromise the validity of the model dynamics.

70
2.2 Production

The macroeconomic production factors are capital, labour, and final or useful energy. A nested production
function with constant elasticity of substitution (CES) determines a region’s gross domestic product (GDP) and
its energy demand.

Generated economic output (GDP) is used for consumption, investments in the macroeconomic capital stock,
and energy system expenditures, as well as trade, non-energy-related greenhouse gas abatement costs, and
agricultural costs delivered by the land-use model MAgPIE (see Figure 45).

Figure 45: Production function of REMIND model

Inputs at the upper layer of the production function include labour, capital, and energy services. Labour is
represented by the population at working age (exogenous). Investments increase capital stocks which
depreciate according to the depreciation rate and energy is produced at a cost. Energy services at the upper
level are the output from a CES tree combining sectoral energy inputs from transportation, the building sector,
and industry. In turn, the demand for specific energy carriers at the sectoral level is also depicted through
individual CES nests. Each production factor in the various macroeconomic CES functions has an efficiency
parameter. These three sectors present slightly different structures.

Transport demand composition is calculated for light-duty vehicles (LDVs), electric trains, and heavy-duty
vehicles (HDVs), an aggregate category including passenger non-LDVs and freight modes (Pietzcker et al.,
2014a). The three corresponding nodes in the CES transport branch represent aggregated transportation
demands in terms of useful, i.e., motive, energy. The LDV node in the CES tree is supplied by either electricity,
hydrogen, or liquid fuels with different conversion efficiencies, accounting for vehicles with internal combustion
engines, fuel cell cars, or battery electric vehicles.

The final energy demand is determined for the aggregated industry sector and subdivided into four industry
subsectors: cement production, chemicals production, iron, and steel production, as well as all remaining
industry energy demand (denoted “other Industry”) using region-specific shares that are kept constant at 2005
levels. Fuel switching (e.g., electrification) is enabled based on final energy prices and elasticities of substitution
of the final energy carriers in the CES function.

The energy demand from industry is modelled explicitly for the four subsectors (cement, chemicals, and iron
and steel, and all remaining industry energy demand (denoted “other industry”) in the nested CES production
function. The iron and steel sector is subdivided into primary steel (from iron ore) and secondary steel (from

71
scrap). The production of cement and steel as well as the value added from chemicals are derived via
econometric regressions models based on per capita GDP at the country level. Steel demand is projected
following the approach of Pauliuk et al. (2013). In all realizations of the industry module, three marginal
abatement cost (MAC) curves have been derived from the literature for CCS in the cement, chemicals, and iron
and steel sectors (Kuramochi et al., 2012).

The heterogeneity of the building demand is rendered through a nested CES function with a high degree of
substitutability among non-electric fuels (e.g., heating oil and natural gas) and a low degree of substitutability
between non-electric fuels and electric demand. The distinction between the non-electric and electric energy
carriers is motivated by the different uses that can be made of these energy sources. While non-electric fuels
are mostly used for heating purposes (e.g., space, water, and cooking), electricity consumption covers a wider
range of purposes (e.g., lighting, appliances, and cooling).

2.3 Trade

REMIND considers the trade of coal, gas, oil, biomass, uranium, and the composite good (aggregated output
of the macroeconomic system). It assumes that renewable energy sources (other than biomass) and secondary
energy carriers are non-tradable across regions.

REMIND models regional trade via a common pool. While each region is an open system – meaning that it can
import more than it exports – the global system is closed. The combination of regional budget constraints and
balanced international flows ensures that the sum of regional consumption, investments, and energy-system
expenditures cannot be greater than the global total output in each period. In line with the classical Heckscher–
Ohlin and Ricardian models (Heckscher et al., 1991), trade between regions is induced by differences in factor
endowments and technologies. REMIND also represents the additional possibility of intertemporal trade. This
can be interpreted as capital trade or borrowing and lending. Capital trade is linked to the export and import of
goods and energy and is accounted for in the inter-temporal trade balance. By directing the goods trade, the
capital market implementation affects the consumption.

To reconcile modelled capital flows and currently observed patterns (Lucas paradox 34; Lucas, 1990), REMIND
represents capital market imperfections. The default setting includes limitations on the growth of debts and
surpluses that each region can accumulate within a 5-year period. Alternative realizations with capital market
imperfections are available (Leimbach and Bauer, 2021)

2.4 Taxes

REMIND includes different types of taxes (see Table 5), representing existing energy taxes, emulating climate
policies via carbon prices or additional externalities for some technologies and processes. The overall tax
revenue is the sum of various components.

34 i.e. the fact that capital do not always flows from developed countries to developing countries despite the fact latter
have lower levels of capital per worker.

72
Table 5: Types of taxes within REMIND, the reason for their inclusion, and the approach to their implementation.

Tax type Rationale Implementation

Bioenergy tax Represents negative externalities Determined as the tax rate

of bioenergy plantation on land (as multiple of bioenergy
price) times primary energy
use of purpose-grown
biomass

Greenhouse gas tax Main policy instrument for Calculated as the tax rate
achieving mitigation targets times GHG emissions

CCS (carbon capture Represents performance Calculated as the tax rate

and storage) tax difference of carbon stored in fuel (defined as fraction of
vs. carbon in the form of CO2 in operation and maintenance,
geological storage O&M, costs) times the
amount of CO2
sequestration

Net-negative Represents marginal damages of Calculated as the tax rate

emissions tax overshoot in emissions budget (defined as fraction of
carbon price) times net
negative emissions

Final energy taxes in Status quo of fuel taxation, with Calculated as the effective
Transports different assumptions regarding tax rate (tax minus subsidy)
convergence times final energy (FE) use in
transport

Final energy taxes in Status quo of fuel taxation, with Calculated as the effective
Buildings/Industry different assumptions regarding tax rate (tax minus subsidy)
convergence times FE use in the sector

Final energy taxes in Status quo of fuel taxation, with Calculated as the effective
the sectors with different assumptions regarding tax rate (tax minus subsidy)
energy service convergence times FE use in the sector
representation

Resource extraction Status quo of extraction subsidies Calculated as the subsidy

subsidies rate times fuel extraction

Primary to secondary Non-explicitly represented Calculated as the effective

energy technology externalities of different tax rate (tax minus subsidy)
technologies (water use, emissions

73
 taxes, specified by of substances beyond SO2 and times the SE output of
technology CO2) technology

Export taxes Represent export barriers Calculated as the tax rate

times the export volume

SO2 tax Represents air pollution externality Calculated as the tax rate
times emissions

High implicit discount Mirror the overvaluation of initial Calculated as the additional
rates in energy investments vs. runtime costs by discount rate times the input
efficiency capital customers of capital at different levels

Regional subsidy on Internalizes the positive externality Sum over the regional
learning technologies of the learning spillovers to other capitalized benefits of
regions so as to arrive at a globally learning which corresponds
optimal solution to the shadow price of the
equation that describes the
capacity build-up of this
technology.

2.5 Energy Module

The energy system module includes a detailed representation of energy supply and demand sectors and
differentiates between a variety of fossil, biogenic, nuclear and renewable energy resources (Bauer et al., 2012,
2016, 2017; Klein et al., 2014, 2014; Pietzcker et al., 2014). The model accounts for crucial drivers of energy
system inertia and path dependencies by representing full capacity vintage structure, technological learning of
emergent new technologies, and adjustment costs for rapidly expanding technologies. The learning of
emergent new technologies is typically represented using learning or experience curves, which depict the
relationship between cumulative production or deployment of a technology and its associated costs or
performance. Costs for expanding technologies are adjusted to simulate the decreasing costs and improving
performance of technologies as they are deployed and gain experience in the system) (Pietzcker et al., 2017).

Figure 46: Energy module structure for REMIND

74
The core part of REMIND includes the representation of the energy system via the conversion of primary
energy into secondary energy carriers via specific energy conversion technologies: The conversion chain is
depicted in Figure 46. Around 50 different energy conversion technologies, including fossil fuel based,
renewable energy, bioenergy, nuclear power, energy storage and carbon capture, are included in REMIND. For
example, bioenergy, fossil gas, coal or oil can be transformed into secondary energy carriers such as liquids,
solids, electricity, hydrogen, gas, or district heating. In general, technologies providing a certain secondary
energy type compete linearly against each other, i.e., technology choice follows cost optimization based on
investment costs, fixed and variable operation and maintenance costs, fuel costs, emission costs, efficiencies,
lifetimes, and learning rates. REMIND assumes full substitutability between different technologies producing
one energy type.

A few technologies convert secondary energy into another form of secondary energy, namely the conversion
of electricity to hydrogen via electrolysis and the reconversion via hydrogen turbines, as well as the production
of methanol and methane from hydrogen. In REMIND, technologies are represented as linear transformation
processes that convert one or more inputs into one or more outputs. In- and outputs can be energy, materials,
water, intermediate products or emissions, or labour inputs. The number of in- and outputs is not restricted,
and technologies vary between in- and output characteristics. In the broader system context, technologies and
their deployment interact via various budget constraints, which give rise to competition for resources as well
as the potential to expand feasible production possibilities. A model solution provides a set of activities that is
feasible with all constraints simultaneously. REMIND specifies each technology through several characteristic
parameters:

 Specific investment costs that are constant for most technologies and decrease due to learning-by-doing
for some relatively new technologies

 Cost markups due to financing costs over the construction time

 Fixed yearly operating and maintenance costs in percent of investment costs

 Variable operating costs (per unit of output, excluding fuel costs)

 Conversion efficiency from input to output

 Capacity factor (maximum utilization time per year): this parameter also reflects maintenance periods and
other technological limitations that prevent the continuous operation of the technology

 Average technical lifetime of the conversion technology in years

 If the technology experiences learning-by-doing: the initial learn rate, initial cumulative capacity, and floor
costs that can only be approached asymptotically

REMIND represents all energy technologies as capacity stocks in gigawatt (GW) with full vintage tracking. As
there are no hard constraints on the rate of change in investments, the possibility of investing in different capital
stocks provides high flexibility for technological evolution. However, the model includes cost markups for the
fast upscaling of investments into individual technologies; therefore, a more realistic phasing in and out of
technologies is achieved. The model allows for premature retirement of capital stocks before the end of their
technological lifetime, and the lifetimes differ between various types of technologies. Capital stocks can be
phased out before they reach the end of their technical lifetime by the optimization if the value of their outputs
is lower than the costs of variable inputs, reflecting a situation of asset stranding. This happens predominantly
in “delayed” scenarios, where unanticipated policy changes change the cost or yield assumptions. Furthermore,
capacities of conversion technologies age realistically from an engineering point of view: depreciation rates are
very low in the first half of the lifetime and increase strongly thereafter.

75
REMIND characterizes the exhaustible resources (coal, oil, gas, and uranium) in terms of extraction cost
curves. Fossil resources (e.g., oil, coal, and gas) are further defined by decline rates and adjustment costs (Bauer
et al., 2016b). Extraction costs increase as low-cost deposits become exhausted (Herfindahl, 1967; Rogner,
1997; Aguilera et al., 2009; Bauer et al., 2016a). In REMIND, region specific extraction cost curves relate the
production cost increase to cumulative extraction (Bauer et al., 2016a; Rogner et al., 2012, p. 7). In the model,
the fossil extraction cost input data (see Bauer et al., 2016b for details) are approximated by piecewise linear
functions that are employed for fossil resource extraction curves.

REMIND models resource potentials for non-biomass renewables (hydro, solar, wind, and geothermal) using
region-specific potentials. For each renewable energy type, potentials are classified by different grades,
specified by capacity factors. Superior grades have higher capacity factors, which correspond to more full-load
hours per year. This implies higher energy production for a given installed capacity. Therefore, the grade
structure represents optimal deployment of renewable energy, first using the best sites before turning to sites
with worse conditions.

Figure 47: Energy module technologies of REMIND

The renewable energy potentials of REMIND 35 may appear higher than the potentials used in other models
(Luderer et al., 2014). However, these models typically limit potentials to specific locations that are currently
competitive or close to becoming competitive. The grade structure of REMIND allows for the inclusion of sites
that are less attractive but that may become competitive in the long-term as the costs of technologies and fuels
change.

The model assumes a single electricity market balance that is complemented by equations that implicitly
represent challenges and options related to the temporal and spatial variability of wind and solar power. The
core approach (Pietzcker et al., 2014b) is an aggregated representation of technology- and region-specific wind
and solar PV (variable renewable energy, VRE) integration costs and curtailment rates (i.e., unused surplus
share of VRE electricity generation), which, since 2017, are parameterized with the help of two detailed
electricity production cost models (Scholz et al., 2017; Ueckerdt et al., 2017). Integration costs consist of costs
associated with short-term storage deployment (batteries), long-term hydrogen storage (electrolysis and
hydrogen turbines), transmission and distribution grid expansion and reinforcement, and curtailment of surplus

35 That is, the physical limits of what could be installed disregarding economic considerations. See subsection 2.3 for
details.

76
electricity. These variables are linked to the shares of VRE generation, with higher VRE shares resulting in
higher requirements for storage and grid.

The land-use sector as modelled by the MAgPIE model (see next subsection) is particularly relevant for climate
change mitigation because of its big share of global emissions and its ability to provide the renewable and
comparatively low-emission resource biomass. In REMIND-MAgPIE, biomass is used to produce electricity,
heat, ethanol, diesel, and hydrogen energy sources. Some of the conversion routes are equipped with CCS,
which makes biomass an important source of negative emissions.

2.6 Land and vegetation system

From a climate protection perspective, two aspects of the land-use sector are of particular interest: the supply
of biomass that can be used for energy production (possibly with carbon capture and storage, CCS) and the
total emissions of the land-use sector. REMIND obtains its supply curves for purpose-grown biomass, its data
for land-use emissions, and agricultural production costs from the MAgPIE land-use model. For the NGFS
scenarios, REMIND and MAgPIE are run in an iterative soft-coupled mode (Klein, 2015; Bauer et al., 2020),
where a simultaneous equilibrium of bioenergy and emissions markets is established by an iteration of
simulations in which REMIND provides emissions prices and bioenergy demand to MAgPIE and receives land
use emissions and bioenergy prices from MAgPIE in return (See Figure 48).

The coupling approach between REMIND and MAgPIE is designed to derive scenarios with equilibrated
bioenergy and emissions markets. In equilibrium, bio-energy demand patterns computed by REMIND are
fulfilled in MAgPIE at the same bioenergy and emissions prices that the demand patterns were based on.
Moreover, the emissions in REMIND emerging from pre-defined climate policy assumptions account for the
greenhouse gas emissions from the land-use sector derived in MAgPIE under the emissions pricing and
bioenergy use mandated by the same climate policy. The coupling approach with this iterative process at its
core is explained in Bauer et al., 2014.

Figure 48. Integration of REMIND with MAgPIE

77
MAgPIE36 is a global land use allocation model, which is connected to the grid-based dynamic vegetation model
LPJmL, with a spatial resolution of 0.5°x0.5°. It takes regional economic conditions such as demand for
agricultural commodities, technological development, and production costs as well as spatially explicit data on
biophysical inputs into account. Biophysical inputs, such as agricultural yields, carbon densities and water
availability, are derived from LPJmL. Based on these, the model derives specific land use patterns, yields and
total costs of agricultural production for each grid cell. The objective function of the land use model is to
minimize total cost of production for a given amount of regional food and bioenergy demand under
consideration of biophysical and socioeconomic constraints.

Figure 49. Simplified MAgPIE flowchart of key processes (demand and trade
implementation, data inputs from LPJmL and spatially explicit water shadow prices).

Regional food energy demand is defined for an exogenously given population in 10 food energy categories,
based on regional diets. Future trends in food demand are derived from a cross-country regression analysis,
based on future scenarios on GDP and population growth. Food and feed energy for the demand categories can
be produced by 20 cropping activities and 3 livestock activities. Feed for livestock is produced as a mixture of
crops, crop residuals, processing by-products, green fodder produced on crop land, and pasture. For meeting
the demand, MAgPIE endogenously decides, based on cost-effectiveness, about intensification of agricultural
production, cropland expansion and production relocation (intra-regionally and inter-regionally through
international trade), see Dietrich et al. (2014), Lotze-Campen et al. (2010) and Schmitz et al. (2012).

Variable inputs of production are labour, chemicals, and other capital (all measured in US$). Costs of production
are derived from the Global Trade Analysis Project (GTAP) Database. The model can endogenously decide to

36 Phase IV
NGFS scenarios use version 4.6 of MAgPIE. See https://github.com/magpiemodel/magpie for the last version
and documentation of the model.

78
acquire yield-increasing technological change at additional costs. The costs for technological change for each
economic region are based on its level of agricultural development, measured as agricultural land-use intensity.
These costs grow with further investment in technological change. The use of technological change is either
triggered by a better cost-effectiveness compared to other investments or as a response to resource
constraints, such as land scarcity.

The model LPJmL37 is designed to simulate vegetation composition and distribution as well as stocks and land-
atmosphere exchange flows of carbon and water, for both natural and agricultural ecosystems. Using a
combination of plant physiological relations, generalised empirically established functions and plant trait
parameters, it simulates processes such as photosynthesis, plant growth, maintenance and regeneration
losses, fire disturbance, soil moisture, runoff, evapotranspiration, irrigation, and vegetation structure.

LPJmL is currently the only DGVM (Dynamic Global Vegetation Model) that has dynamic land use fully
incorporated at the global scale and simulates the production of woody and herbaceous short-rotation
bioenergy plantations and the terrestrial hydrology. It differs from other models in the wider field by computing
carbon, nitrogen, and water flows explicitly: most other macro-hydrological models lack the important
vegetation structural and physiological responses that influence the water cycle, while most other vegetation
models lack the advanced consistent water balance of LPJmL or are not global in scale.

2.7 Emissions, abatement costs and Carbon dioxide removal

REMIND simulates emissions from long-lived GHGs (CO2, CH4, and N2O), short-lived GHGs (CO, NOx, and
VOCs), and aerosols (SO2, BC, and OC). REMIND accounts for these emissions with different levels of detail
depending on the types and sources of emissions.

It calculates CO2 emissions from fuel combustion and industrial processes, CH4 emissions from fossil fuel
extraction and residential energy use, and N2O emissions from energy supply based on sources. Fluorinated
gases (F-gases) and emissions from land-use change are included exogenously with different trajectories
depending on the SSP and climate target.

There are mitigation options for CH4, N2O, and CO2 from land-use change, fossil fuel extraction, cement
production, and waste handling that are independent of energy consumption and are calculated in the core of
REMIND. However, there are costs associated with these emission reductions. Therefore, REMIND derives the
mitigation options from marginal abatement cost curves (MACC), which describe the percentage of abated
emissions as a function of the costs (Lucas et al., 2007).

In addition to CCS (carbon capture and storage) with fossil fuels and in the industry sector, four carbon dioxide
removal (CDR) options are available: afforestation and reforestation, bioenergy with CCS (BECCS), direct air
capture with CCS (DACCS), and enhanced weathering of rocks (EW). CO2 emissions from afforestation and
reforestation are derived from the land-use optimization model MAgPIE. The trade-off between land expansion
and yield increases is treated endogenously in the model. BECCS (bioenergy with carbon capture and storage)
is the only CDR technology that provides sizable energy instead of consuming it. The idea of BECCS is to turn
biomass grown on land carbon-negative by capturing the emissions arising during combustion or the refinery
process. BECCS can be used for electricity, hydrogen, gas, or liquid-fuel production with different carbon

37 See https://github.com/PIK-LPJmL/LPJmL for the repository and documentation of the model.

79
capture rates. DACCS (direct air and carbon capture and storage) and enhanced weathering of rocks were
switched off in REMIND for the NGFS scenarios.

2.8 Climate

To translate emissions into changes in atmospheric composition, radiative forcing, and temperature increase,
REMIND is coupled with the MAGICC 6 climate model. Due to numerical complexity, the evaluation of climate
change using MAGICC is performed after running REMIND. Iterative adjustment of emission constraints or
carbon taxes allows for specific temperature or radiative forcing limits to be met in the case of temperature
targets.

MAGICC is a reduced-complexity climate model that calculates atmospheric concentrations of greenhouse

gases and other atmospheric climate drivers, radiative forcing, and global annual-mean surface air
temperature. More details on this model are provided in the dedicated Box: MAGICC: A reduced complexity
Earth system model, and in the corresponding sections of the IAMs modules, as it is also used in post-
processing for the estimation of chronic physical damages.

80
3. Key model inputs

Description of key input variables (e.g., which series, years, sources) and main assumptions
REMIND-MAgPIE uses a range of exogenous data as input to ensure the consistency of scenarios with historic
developments and realistic future projections. Historical data for the year 2005 are used to calibrate most of
the free variables (e.g., primary energy mixes in 2005, secondary energy mixes in 2005, standing capacities in
2005, and trade in all traded goods for 2005). Additional bounds for a select few variables, primarily capacity
(additions), up through 2019 ensure that the 2020 point of departure in current policy cases is proximal to actual
developments. The ability to also run the simulation without these constraints enables important comparisons
of model dynamics from 2005 to 2020 with real world developments.

Technology parameters are projected into the future, generally assuming a certain level of convergence across
regions in the long term.

3.1 Population and GDP

All economic assumptions are taken from the Shared Socioeconomic Pathway 2 (SSP 2), designed to represent
a “middle-of-the-road” future development. Population is a fully exogenous input assumption. Projections of
coherent future demographic and economic developments offer population and labour trajectories from 2005
to 2100 (SSP trajectories; Dellink et al., 2017; KC and Lutz, 2017).

Gross Domestic Product38 is a semi-endogenous output. The model takes the SSP2 GDP trajectories for
calibrating assumptions on exogenous productivity improvement rates in a no-policy reference scenario
(Current policies). GDP trajectories in other scenarios thus reflect the general equilibrium effects of constraints
and distortions by policies (so changes in capital allocation and prices, but without taking potential damages
from climate impacts into account).

3.2 Production function calibration

To align with gross domestic product (GDP) trajectories consistent with the population trajectories from 2005
to 2100, as well as final and useful energy trajectories, REMIND calibrates its production function.

The changes in efficiency parameters over time are tuned such that the baseline scenario meets exogenous
economic growth pathways and final or useful energy pathways in line with the SSPs (O’Neill et al., 2013). The
calibration has to fulfil two constraints: an economic and a technological constraint. The technological
constraint requires the inputs of the CES (constant elasticity of substitution) production function to yield the
desired output. At this stage, there is no economic consideration at all. During a REMIND run, however, the
model will strive to find the most efficient solution in terms of costs. Therefore, the second constraint is an
economic constraint. The derivatives of the CES function, i.e., the marginal increase in income from increasing
the considered input by one unit, must equal the price of that input, i.e., the marginal cost. The economic
constraint defines that the prices are equal to the derivatives. Following Euler’s rule, the technological
constraint determines that, for homogeneous functions of degree 1 (as is the case here), the output is equal to
the sum of the derivatives times the quantity of inputs. Combining both constraints means that the output is

38 This series is provided in the NGFS database with the name GDP|PPP|counterfactual without damage.

81
equal to the sum of inputs valued at their price. Thus, the prices and quantities given exogenously, combined
with the two constraints, are sufficient to determine all the quantities of the CES tree up to the last level with
labour and capital.

3.3 Resource extraction costs and renewable maximum capacities

Figure 50. Extraction marginal costs by fossil resource. Bauer et al. (Energy, 2016)

While biomass resources are given by MAgPIE model, the extraction costs of fossil fuels need to be calibrated,
see Figure 50. More details on the underlying data and method are presented in a separate paper (Bauer et al.,
2016b). In the model, this fossil extraction cost input data is approximated by piecewise linear functions that
are employed for fossil resource extraction curves. For uranium, extraction costs follow a third-order
polynomial parameterization based on data of the Nuclear Energy Agency (NEA); see Bauer et al. (2012a) for
details.

For renewables, maximum production capacity is calibrated by region, see Figure 51. The regionally
aggregated potentials for solar photovoltaics (PV) and concentrated solar power (CSP) used in REMIND were
developed in Pietzcker et al. (2014b) in cooperation with the German Aerospace Center (DLR). To account for
the competition between PV and CSP for the same sites with good irradiation, an additional constraint for the
combined deployment of PV and CSP was introduced in REMIND (Pietzcker et al., 2014b) to ensure that the
model cannot use the available area twice to install both PV and CSP.

82
The regionally aggregated wind potentials were developed based on a number of studies (Hoogwijk, 2004;
Brückl, 2005; Hoogwijk and Graus, 2008; EEA, 2009; Eurek et al., 2017). The technical potentials for combined
on- and offshore wind power amount to 800 EJ yr−1 (half of this amount is at sites with more than 1900 full-
load hours). The total value is approximately half the maximum extractable electric energy from wind over land
area, as estimated in Miller and Kleidon (2016), and about one-fifth of the potential estimated in Lu et al. (2009).
The global potentials of hydropower amount to 50 EJ yr−1. These estimates are based on the technological
potentials provided in the report from WGBU (2003) and the background paper produced for the report
(Horlacher, 2003).

Figure 51. Renewables maximum production by regions. Pietzcker et al. (Applied Energy, 2014)

3.4 Emissions

For each fuel, region, and technology, REMIND applies specific emissions factors, which are calibrated to match
base year GHG inventories (Global Emissions EDGAR v4.2, 2013). Emission factors for CH4 from the residential
sector, and N2O from energy supply are taken from Amous (2000).

Baseline emissions for CH4 fugitive emissions from coal, oil, and gas extraction and processing, are calculated
by source using region- and fuel-specific emission factors. The emission factors for CH4 fugitive emissions are

83
derived using the emissions inventory (Global Emissions EDGAR v4.2, 2013) and the amount of fossil fuel
extracted in each region in REMIND in 2005.

REMIND uses an econometric estimate for CO2 emissions from cement production as well as CH4 and N2O
emissions from waste handling. In both cases, the driver of emissions depends on the development of
population and GDP (as a proxy for waste production) or capital investment (as a proxy for cement production
in infrastructure). REMIND uses exogenous baselines for N2O emissions from transport and industry, and for
CO2, CH4, and N2O emissions from land-use and land-use change based on MAgPIE.

CH4 and N2O emissions from open burning are assumed to remain constant at their 2005 levels.

Emissions of other GHGs (e.g., F-gases and Montreal gases) are exogenous and are taken from the SSP scenario
data set from the IMAGE model (van Vuuren et al., 2017).

For pollutant emissions of SO2, BC, OC, NOx, CO, VOCs, and NH3 related to the combustion of fossil fuels,
REMIND considers time- and region-specific emissions factors coupled to model-endogenous activity data. BC
and OC emissions in 2005 are calibrated to the GAINS model (Klimont et al., 2017; Amann et al., 2011). All other
emissions from fuel combustion in 2005 are calibrated to Global Emissions EDGAR v4.2 (2013).

Emission factors for SO2, BC, and OC are assumed to decline over time according to air pollution policies based
on Klimont et al. (2021). Current near-term policies are enforced in high-income countries, with gradual
strengthening of goals over time and gradual technology (research, development, demonstration, and
deployment). Low-income countries do not fully implement near-term policies but gradually improve over the
century. Emissions from international shipping and aviation and waste of all species are exogenous and are
taken from Fujino et al. (2006). Further, REMIND uses land-use emissions from the MAgPIE model (see Sect.
2.4.1), which, in turn, are based on emission factors from van der Werf et al. (2010).

84
Module 3: IAM – MESSAGE-GLOBIOM

1. Non-technical summary

What is the MESSAGE-GLOBIOM model?

The MESSAGE-GLOBIOM model, or MESSAGE in short, is an Integrated Assessment Model (IAM)
developed by the International Institute for Applied Systems Analysis (IIASA). It combines energy systems,
environmental impacts, and economic analysis to evaluate the long-term implications of energy and climate
policies. Although its name only refers to two of its components, the MESSAGE-GLOBIOM model consists of a
combination of five different models or modules which complement each other and are specialised in different
areas:

 the energy model MESSAGEix39 (Model for Energy Supply Strategy Alternatives and General
Environmental Impact),

 the land-use model GLOBIOM (GLobal BIOsphere Management),

 the air pollution and GHG model GAINS (Greenhouse gas – Air pollution Interactions and Synergies),

 the aggregated macro-economic model MACRO, and

 the simplified climate model MAGICC (Model for the Assessment of Greenhouse-gas Induced Climate
Change).

All models and modules together build the IIASA IAM framework, also referred to as MESSAGEix-GLOBIOM
since the energy model MESSAGE and the land use model GLOBIOM are its most important components. The
five models provide input to and iterate between each other during a typical scenario development cycle.

The MESSAGEix-GLOBIOM model at its core is a technology-detailed energy-engineering optimization model

used for energy planning. Through linkage to macro-economic, land-use and climate models it is capable of
considering important feedback and limitations in these areas outside of the energy system.

What are the key model inputs?

Key model inputs, taken from external sources, relate to GDP and population pathways (taken from the Shared
Socioeconomic Patheway SSP2), energy resource endowments, energy conversion rates, energy end-use,
technological change, fuel blending, add-on technologies, energy demand, modelling policies, macroeconomic
variables, land-use, and emissions.

What are the key model outputs?

Key model outputs comprise regional and country-level variables on emissions, land uses, prices, and quantities
over different scenarios and over a pre-defined horizon common across scenarios.

39 The “ix”stands for integrated assessment modelling with exogenous uncertainties. MESSAGEix is a versatile, dynamic
systems-optimization modelling framework developed by the IIASA Energy, Climate, and Environment (ECE)
Program 1 since the 1980s.

85
What is new in the 2023 edition?
The 2023 edition for Phase IV does not have many changes compared to the 2022 edition used for Phase III.
However, some of the notable differences are:

1. Chemical industry sectors expanded to include ammonia and methanol production processes.
2. A new interpolation scheme for aggregating country-level net-zero trajectories.
3. Updated policy details for the Current Policies and NDC scenarios.
4. Improved method for regionally different carbon price reporting.

2. Overview of model scope and methods

The MESSAGE-GLOBIOM model40 was originally developed to represent global and regional energy
systems41 and can be used as an energy planning tool. The name “MESSAGE” itself refers to the core of the
IIASA IAM framework (see Figure 52) and its main task is to optimise the energy system so that it can satisfy
specified energy demands at the lowest costs. The current version allows for a detailed representation of the
technical-engineering, socioeconomic, and biophysical processes in energy and land-use systems. This is
achieved by linking MESSAGEix to four different models, which have also been developed by IIASA. These four
models are the land-use model GLOBIOM (Global Biosphere Management Model), the air pollution and
greenhouse gas (GHG) model GAINS (Greenhouse gas - Air pollution Interactions and Synergies model), the
aggregated macro-economic model MACRO and the climate model MAGICC (Model for the Assessment of
Greenhouse-gas Induced Climate Change), which complement each other and are specialized in different
areas. The IIASA IAM framework is also referred to as MESSAGEix-GLOBIOM, since the energy model
MESSAGEix and the land-use model GLOBIOM are its central components. Key features of the model include:

 energy system analysis: production, conversion, and consumption across different sectors,

 environmental impacts: enabling the analysis of the environmental consequences of different energy
pathways and policy interventions,

 technological detail: simulating the behaviour and evolution of different energy technologies over time,

 economic analysis: economic implications of different energy and climate policies, including the costs and
benefits of different mitigation strategies, and

 policy assessment: analysing the impacts of policy measures such as carbon pricing, RE subsidies, and
energy efficiency standards.

40 A comprehensive documentation of the model is available at these URLs: https://docs.messageix.org/en/stable/;

https://www.iamcdocumentation.eu/index.php/Model_Documentation_-_MESSAGE-GLOBIOM The source code of
the model is open-source and available at this URL: https://github.com/iiasa/message_ix

41 The energy system analytically traces the process of resource extraction, imports and exports, conversion, transport,
and distribution, up to the provision of energy end-use services such as light, space conditioning, industrial production
processes, and transportation. The energy system in MESSAGEix is represented in Figure 53

86
 Figure 52. Overview of the IIASA IAM Framework. Coloured boxes represent respective specialized disciplinary models which
are integrated for generating internally consistent scenarios. Figure from Riahi et al. (2016).

MESSAGEix42 represents the core of the IIASA IAM framework. Its main task is to optimise the energy
system (see Figure 53), i.e., to satisfy specified energy demands at the lowest costs. This optimisation is
carried out in an iterative setup with MACRO, a single sector macro-economic model, which provides estimates
of the macro-economic demand response that results from energy system and services costs computed by
MESSAGEix. For the six commercial end-use demand categories depicted in MESSAGE (see Energy demand),
based on demand prices, MACRO will adjust useful energy demands, until the two models have reached
equilibrium (see Macro-economy (MACRO)). This iteration reflects price-induced energy efficiency
adjustments that can occur when energy prices change. MESSAGE can represent different energy- and climate-
related policies.

42 Daniel Huppmann, Matthew Gidden, Oliver Fricko, Peter Kolp, Clara Orthofer, Michael Pimmer, Nikolay Kushin,
Adriano Vinca, Alessio Mastrucci, Keywan Riahi, and Volker Krey. The messageix integrated assessment model and
the ix modeling platform (ixmp): an open framework for integrated and cross-cutting analysis of energy, climate, the
environment, and sustainable development. Environmental Modelling & Software, 112:143–156, 2019.
doi:10.1016/j.envsoft.2018.11.012.

87
 Figure 53. Reference Energy System in MESSAGEix

GLOBIOM is a partial-equilibrium model representing the mainland-use sectors, which include agriculture
and forestry. The supply side of the model is built-up from the bottom (spatially explicit land cover, land use,
management systems and economic cost information) to the top (regional commodity markets), as illustrated
in Figure 54. GLOBIOM provides MESSAGEix with information on land use and its implications, including the
availability and cost of bioenergy, and availability and cost of emission mitigation in the AFOLU (Agriculture,
Forestry and Other Land Use) sector. The link between MESSAGEix and GLOBIOM allows researchers to
investigate how land-use changes and biomass supply influence the production of carbon emissions, which in
turn affects energy demand, allowing for an integrated analysis of the energy system and GHG emissions linked
to land use.

88
 Figure 54. Overview of the GLOBIOM model

Air pollution implications of the energy system are accounted for in MESSAGEix by applying technology-
specific air pollution coefficients derived from the GAINS model. The GAINS model is an analytical
framework for assessing future potentials and costs for reducing air pollution impacts on human health and the
environment while simultaneously mitigating climate change through reduced greenhouse gas emissions
(Figure 55). It explores synergies and trade-offs in cost-effective emission control strategies to maximize
benefits across multiple scales. GAINS is calibrated by estimating historic emissions of air pollutants and GHGs
for each country based on data from international energy and industrial statistics, emission inventories and
other data supplied by countries themselves. It assesses emissions over the medium-to-long term, with
projections being specified in five-year intervals until 2050.

Figure 55. Overview of the GAINS model

89
In addition to these two large-scale models, two other key models are also closely related to the use of the
MESSAGE model, namely MACRO and MAGICC. MACRO is a macroeconomic model maximizing the
intertemporal utility function of a single representative producer-consumer in each world region. The
optimization result is a sequence of optimal savings, investment, and consumption decisions. The main
variables of the model are the capital stock, available labour, and energy inputs, which together determine the
total output of an economy according to a nested constant elasticity of substitution (CES) production function.
Thus, by linking the two models it is possible to consistently reflect the influence of energy supply costs, as
calculated by MESSAGE, the mix of production factors considered in MACRO, and the effect of changes in
energy prices on energy service demands. The combined MESSAGEix-MACRO model can generate a consistent
macro-economic response to changes in energy prices and estimate overall economic consequences (on GDP
or consumption) of energy or climate policies. A detailed description of MAGICC is provided separately in the
Box: MAGICC: A reduced complexity Earth system model

Relevant model assumptions for the interpretation of the NGFS scenarios include that the policy scenarios are
a mix of internally consistent and externally imposed constraints on the model. Also, the GDP impacts produced
by the native model, should be interpreted as long-run averages, contrary to the high-resolution data provided
by NiGEM.

90
3. Key model inputs

This section describes key input variables to the MESSAGEix-GLOBIOM framework. In the following section, a
distinction is made between input variables that are common across the models of the framework (i.e.,
MESSAGEix, GLOBIOM, MACRO, MAGICC and GAINS) and input variables that are specific to the different
models. Temporally, the models operate in 5-year or, from 2060 onwards, 10-year steps.

Common input variables

Common input variables are those reflecting socio-economic developments over the projection horizons. More
specifically, exogenous input variables include variables such as GDP and population. These variables are
derived from other analyses and only used as input for the models. More precisely, the main source for socio-
economic assumptions is the database on Shared Socioeconomic Pathways (SSPs) (O’Neill et al., 2015). 43 In
particular, the SSPs reflect five different developments of the world that are characterised by varying levels of
global challenges (see Riahi et al., 2017 for an overview). They include: a world of sustainability-focused growth
and equality (SSP1); a “middle of the road” world where trends broadly follow their historical patterns (SSP2);
a fragmented world of “resurgent nationalism” (SSP3); a world of ever-increasing inequality (SSP4); and a world
of rapid and unconstrained growth in economic output and energy use (SSP5). NGFS scenarios mainly use SSP2
as an input.

In the MESSAGEix scenarios for the NGFS, SSP2 projections of total population and GDP (at purchasing power
parity exchange rates) are the primary drivers of future energy demand. In particular, GDP is common across
scenarios for each region from 2005-2020 and differs depending on the scenarios afterwards. The SSP2 GDP
trajectories are also used for calibrating assumptions on exogenous productivity improvement rates in the
Current Policies scenario. GDP trajectories in other scenarios thus reflect the general equilibrium effects of
constraints and distortions by policies. The other common variable from 2005-2100 is population, which is
further split into rural and urban. Population paths are common across scenarios over the entire projection
horizon, while they differ across regions.

Energy resource endowments (MESSAGEix)

Fossil fuel resources and renewable resource potentials: In MESSAGEix, assumptions on fossil fuel availability and
the underlying extraction cost assumptions are derived from various sources, including global databases such
as The Federal Institute for Geosciences and Natural Resources (BGR) and the U.S. Geological Survey (USGS),
as well as market reports and outlooks provided by different energy institutes and agencies. The availability of
fossil energy resources in different regions 44 is then aligned with the particular storyline of the chosen SSP, i.e.,

43 These pathways cover the range of possible future development of anthropogenic drivers of climate change found in
the literature. The SSP storylines served as the starting point for the development of the quantitative SSP elements.
Each storyline provides a brief narrative of the main characteristics of the future development path of an SSP. The
storylines were identified at the joint IAV and IAM workshop in Boulder, November 2011.

44 Conventional oil and gas are distributed unevenly throughout the world, with only a few regions dominating the reserves.

Nearly half of the reserves of conventional oil is found in Middle East and North Africa, and close to 40% of conventional
gas is found in Russia and the Former Soviet Union states. The situation is somewhat different for unconventional oil of
which North and Latin America potentially possess significantly higher global shares. Unconventional gas in turn is
distributed quite evenly throughout the world, with North America holding most (roughly 25% of global resources). The

91
SSP2 (Rogner, 1997; Riahi et al., 2012) for NGFS scenarios. Specifically, fossil fuel energy resources are aligned
to technical and economic availability of overall resources underlying the SSP2 narrative/pathway. Key
exogenous input variables to achieve this alignment are so called conversion technologies and technological
change which differ across SSPs (e.g., technological change in fossil fuel extraction and conversion
technologies is assumed to be slowest in SSP1). In particular, for SSP2 a continuation of recent trends is
assumed, focusing more on developing extraction technologies for unconventional hydrocarbon resources,
thereby leading to higher potential cumulative oil extraction than in the other SSPs. 45

Nuclear resources: Estimates of available uranium resources in the literature vary considerably, which could
become relevant if advanced nuclear fuel cycles (e.g., the plutonium cycle including fast breeder reactors, the
thorium cycle) are not available. The levels of uranium resources assumed available in the MESSAGE SSP
scenarios are builtupon earlier work developed in the Global Energy Assessment (see Riahi et al., 2012). In the
SSP2 narrative, which underlies NGFS scenarios, the cumulative uranium extracted is assumed at 30 Mt (metric
tonnes) with an approximate price range of 120-230 USD/kg depending on different models of uranium
distribution in the crust and its extraction costs.46

Non-Biomass Renewable Resources: The resources considered are hydro, wind (on-/offshore), solar PV,
concentrating solar power (CSP)47 and geothermal. They are measured in terms of deployment potentials
(EJ/yr.), i.e., in terms of the electricity or heat that can be produced by specific technologies (i.e., from a
secondary energy perspective).48 The estimates used in MESSAGEix are based on different sources, such as the
U.S. National Renewable Energy Laboratory database as described in the Global Energy Assessment (Rogner
et al., 2012). Moreover, resource potentials for solar photovoltaic (PV), concentrating solar power (CSP), and
onshore/offshore wind are further downscaled by region and classified according to resource quality (annual
capacity factor) based on Pietzcker et al. (2014) and Eurek et al. (2017).

Biomass Resources: Bioenergy includes both commercial and non-commercial use. Commercial refers to the
use of bioenergy in, for example, power plants or biofuel refineries, while non-commercial refers to the use of
bioenergy for residential heating and cooking, primarily in rural households of today’s developing countries,
and as such is typically not traded or sold. Bioenergy potentials are derived from the GLOBIOM model and differ

distribution of coal reserves shows the highest geographical diversity. Russia and the former Soviet Union states, Pacific
OECD, North America, and Centrally Planned Asia and China all possess more than 10 ZJ of reserves.

45For assumptions and references on global fossil fuel reserves and resources in the MESSAGE please refer to the tables at
the following link: https://docs.messageix.org/projects/global/en/latest/energy/resource/fossilfuel.html

46 For more details: https://docs.messageix.org/projects/global/en/latest/energy/resource/nuclear.html

47 Unlike CSP which uses the sun’s energy, PV solar panels make use of the sun’s light instead. In other words, photovoltaics

is the direct conversion of light into electricity, while CSP systems produce electric power by converting the sun’s energy
into high-temperature heat using various mirror configurations and this concentrated energy is then used to drive a heat
engine and drive an electric generator.

48 This differs from the technical potentials which instead refer to the flows of energy that could become available as inputs
for technology conversion. So, for example, the technical potential for wind is given as the kinetic energy available for wind
power generation, whereas the deployment potential would only be the electricity that could be generated by the wind
turbines.

92
across SSPs as a result of different levels of competition over land for food and fibre, but ultimately only vary
to a limited degree. The drivers underlying this competition are different land-use developments in the SSPs,
which are determined by agricultural productivity and global demand for food consumption. (Fricko et al.,
2017).

Energy conversion (MESSAGEix)

Energy technologies are characterised by numerical model inputs describing their economic (e.g., investment
costs, fixed and variable operation and maintenance costs), technical (e.g., conversion efficiencies), ecological
(e.g., GHG and air pollutant emissions), and socio-political characteristics.49 Model input data reflecting these
parameters constrains the use of these technologies or, equivalently, determines their omission for some
regions. The specific technologies represented in various parts of the energy conversion sector encompass
“Electricity, Heat, Other conversion” 50 and “Grid, Infrastructure and System Reliability”51,52. Each energy
conversion technology is characterized in MESSAGE by the following data:

 Energy inputs and outputs together with the respective conversion efficiencies.

 Specific investment costs (e.g., per kilowatt, kW) and time of construction as well as distribution of
capital costs over construction time.

 Fixed operating and maintenance costs (per unit of capacity, e.g., per kW).

 Variable operating costs (per unit of output, e.g., per kilowatt-hour, kWh, excluding fuel costs).

49 An example for the socio-political situation in a world region would be the decision by a country or world region to ban
certain types of technologies (e.g., nuclear power plants).

50 Other conversion includes liquid fuel production, gaseous fuel production and hydrogen production.

51 Energy transport and distribution infrastructure is included in MESSAGE at a level relevant to represent the associated
costs as well as transmission and distribution losses. Within individual model regions the capital stock of transmission and
distribution infrastructure and its turnover is modelled for a number of energy carriers (electricity, district heat, natural gas
and hydrogen). For all solid (coal, biomass) and liquid energy carriers (oil products, biofuels, fossil synfuels) a simpler
approach is taken and only transmission and distribution losses and costs are taken into account. Inter-regional energy
transmission infrastructure, such as natural gas pipelines and high voltage electricity grids, are also represented between
geographically adjacent regions. Solid and liquid fuel trade is, similar to the transmission and distribution within regions,
modeled by taking into account distribution losses and costs. A special case are gases that can be traded in liquified form,
i.e., liquified natural gas (LNG) and liquid hydrogen, where liquefaction and re-gasification infrastructure is explicitly
represented in addition to the actual transport process.

52 The global MESSAGE model includes a single annual time period within each modelling year characterized by average
annual load and 11 geographic regions. Seasonal and diurnal load curves and spatial issues such as transmission constraints
or renewable resource heterogeneity are treated in a stylized way in the model.

93
  Plant availability or maximum utilisation time per year.53

 Technical lifetime of the conversion technology in years.

 Year of first commercial availability and last year of commercial availability of the technology.

 Consumption or production of certain materials (e.g., emissions of kg of CO2 or SO2 per produced
kWh).

 Limitations on the (annual) activity and on the installed capacity of a technology.

 Constraints on the rate of growth or decrease of the annually new installed capacity and on the growth
or decrease of the activity of a technology.

 Technical application constraints, e.g., maximum possible shares of wind or solar power in an
electricity network without storage capabilities.

 Inventory upon start up and shutdown, e.g., initial nuclear core needed at the start-up of a nuclear
power plant.

 Lag time between input and output of the technology.

 Minimum unit size, e.g., for nuclear power plants it does not make sense to build plants with a capacity
of a few kilowatts power (optional, not used in current model version).

 Socio-political constraints, e.g., ban of nuclear power plants.

 Inconvenience costs which are specified only for end-use technologies (e.g., cook stoves)

Energy end-use (MESSAGEix)

MESSAGEix distinguishes three energy end-use sectors: transport, residential/commercial (also referred to as
buildings sector), and industry.

Transport sector. The applied MESSAGEix transport sector representation is stylized and essentially includes
fuel switching54 –to account for a key option to reduce emissions; switching depends on fuel-specific relative

53 As an example, electric-sector flexibility in MESSAGE is represented as follows: each generating technology is assigned a
coefficient between -1 and 1 representing (if positive) the fraction of generation from that technology that is considered to
be flexible or (if negative) the additional flexible generation required for each unit of generation from that technology. Load
also has a parameter (a negative one) representing the amount of flexible energy the system requires solely to meet changes
and uncertainty in load.

54 Limitations of switching to alternative fuels may occur, for example as a result of restricted infrastructure availability (e.g.,

rail network) or some energy carriers being unsuitable for certain transport modes (e.g., electrification of aviation). To
reflect these limitations, share constraints of energy carriers (e.g., electricity) and energy carrier groups (e.g., liquid fuels)
are used in the transport sector. In addition, the diffusion speed of alternative fuels is limited to mimic bottlenecks in the

94
efficiencies - and price-elastic demands (via MACRO linkage) as the main responses to energy and climate
policy. According to SSP2, the storyline underlying NGFS scenarios, the electrification rate within transport can
amount up to 50 percent of total transport.55 The following Figure 56 displays a schematic diagram of the
stylized transport sector representation in MESSAGEix.

Figure 56 . Stylized transport sector representation in MESSAGEix

Residential and commercial sector. The residential and commercial sector in MESSAGEix distinguishes two
demand categories: thermal and specific. Thermal demand, i.e., low temperature heat, can be supplied by a
variety of different energy carriers, while specific demand requires electricity (or a decentralized technology to
convert other energy carriers to electricity).

supply chains, not explicitly represented in MESSAGEix (e.g., non-energy related infrastructure). Both the share as well as
the diffusion constraints are usually parametrized based on transport sector studies that analyse such developments and
their feasibility in much greater detail.

55 The quantitative translation of the storyline elements of SSP1, SSP2 and SSP3 in terms of electrification rate for transport

can be found at https://docs.messageix.org/projects/global/en/latest/energy/enduse/transport.html (see also Fricko et al.,

2017).

95
  The residential and commercial thermal energy demand includes fuel switching as the main
option,56 i.e., different choices about final energy forms to provide thermal energy. In addition to
the alternative energy carriers that serve as input to these thermal energy supply options, their
relative efficiencies also vary. For example, solid fuels such as coal have lower conversion
efficiencies than natural gas, direct electric heating or electric heat pumps. Additional demand
reduction in response to price increases in policy scenarios is included via the fuel switching option
(due to the fuel-specific relative efficiencies) as well as via the linkage with the macro-economic
model MACRO (see Figure 57 below).
 The residential and commercial specific demand (for electricity) can be satisfied either by
electricity from the grid or with decentralized electricity generation options such as fuel cells and
on-site combined heat and power (CHP).57.

56 To reflect limitations of switching to alternative fuels, as a result of limited infrastructure availability (e.g., district heating

network) or some energy carriers being unsuitable for certain applications, share constraints of energy carriers (e.g.,
electricity), and energy carrier groups (e.g., liquid fuels) are used in the residential and commercial sector. In addition, as in
the transport sector, the diffusion speed of alternative fuels is limited to mimic bottlenecks in the supply chains, not
explicitly represented in MESSAGEix (e.g., non-energy related infrastructure).

57 The quantitative translation

of the storyline elements of SSP1, SSP2 and SSP3 in terms of electrification rate within the
residential and commercial sectors can be found at
https://docs.messageix.org/projects/global/en/latest/energy/enduse/transport.html (see also Fricko et al., 2017).

96
 Figure 57 . Schematic diagram of the residential and commercial sector representation in MESSAGEix

Industrial sector. Differently from the two demand sectors above, the industrial sector in MESSAGEix used for
the NGFS scenarios has more detailed representations of sub-sectors, which distinguish direct demands for
industrial materials. We have representations for steel, cement, aluminium, petro-chemical (high-value
chemicals, methanol, ammonia) industries. For the remaining industrial sub-sectors, MESSAGEix receives two
energy demand categories: thermal and specific, linked to MACRO, similarly to the residential and commercial
sectors. Figure 58 and Figure 59 provide schematic representations of the industrial sector in MESSAGEix.

97
Figure 58 . Generic representation of an industry sector modelled in MESSAGEix

98
 Figure 59 . Schematic diagram of the residual industrial sector representation in MESSAGEix

Technological change (MESSAGEix)

Technological change in MESSAGEix is generally treated exogenously. 58 The current cost and performance
parameters, including conversion efficiencies and emission coefficients, are generally derived from the relevant
engineering literature.

58 However, some work endogenization of technological change has been introduced, e.g., the dependence of technology
costs on market structure have been done with MESSAGEix (Leibowicz, 2015).

99
Technological costs vary regionally in all SSPs, reflecting marked differences in engineering and construction
costs across countries observed in the real world. The regional differentiation of technology costs for the initial
modelling periods are based on IEA data (IEA, 2014) with convergence of costs assumed over time driven by
economic development (GDP per capita).59 Estimates for present-day and fully learned-out technology costs
are from the Global Energy Assessment (Riahi et al., 2012) and World Energy Outlook (IEA, 2014).

For technological diffusion, MESSAGEix tracks investments by vintage. In case of shocks (e.g., introduction of
stringent climate policy), it is however possible to prematurely retire existing capital stock such as power plants
or other energy conversion technologies and switch to more suitable alternatives. 60 Also, so called flexible or
soft dynamic constraints have been introduced into MESSAGE (Keppo and Strubegger, 2010). These allow
faster technology diffusion at additional costs and therefore generate additional model flexibility while still
reducing the number of sudden policy reversals and penetration of technologies.61

Fuel blending (MESSAGEix):

Fuel blending is a common practice that allows the shared use of infrastructure by fuels with similar chemical
attributes and thus their combined use at the secondary and final energy level, without requiring the consumer
to adapt the power plant or end-use devices. Fuel blending in the global energy model is modelled for two
distinct blending processes: 62 the blending of natural gas with other synthetic gases and. the blending of light
oil with coal derived synthetic liquids.63

59 Generally, costs start out lower in the developing world and are assumed to converge to those of present-day
industrialized countries as the former becomes richer throughout the century (thus, the cost projections consider both
labour and capital components). This catch-up in costs is assumed to be fastest in SSP1 and slowest in SSP3 (where
differences remain, even in 2100); SSP2 is in between.

60 An important factor in this context that influences technology adoption in MESSAGEix are technology diffusion
constraints. Technology diffusion in MESSAGEix is determined by dynamic constraints that relate the construction of a
technology added or the activity (level of production) of a technology in a period t to construction or the activity in the
previous period t-1 (Messner and Strubegger, 1995).

61 More details on technological diffusion can be found at

https://docs.messageix.org/projects/global/en/latest/energy/tech.html

62 It is important to be able to track the use of blended fuels in the energy model for two reasons. Not all blended fuels can
be used equally within all natural gas applications. For example, hydrogen mixed into the natural gas network is restricted
to use in non-CCS applications only. Secondly, it is essential to keep track of where which of the blended fuels is being used
in order to correctly report emissions and also to potentially restrict the degree to which fuels can be blended for individual
applications. For example, natural gas end-use appliances may only be able to cope with a certain share of hydrogen while
still guaranteeing their safety and longevity. Similarly, for policy analysis, it could be required that a certain minimum share
of a synthetic gas is used sector specifically.

63 For more details refer to https://docs.messageix.org/projects/global/en/latest/energy/fuel_blending.html

100
Add-on technologies (MESSAGEix):
Add-on technologies have a distinct formulation in MESSAGEix.64 The formulation is used to represent two
main types of technical extensions/options for technologies: a) additional modes of operation for a single or
multiple technologies;65 b) depicting emission mitigation options.66 They are defined using the same parameters
as any other technology. What makes a technology an add-on technology, is the fact that their activity is bound
to the activity of one or more other technologies, the so-called parent technology. In particular, a single add-
on technology can be coupled to the activity of multiple parent technologies. Furthermore, multiple add-on
technologies can be linked to the activity of a single parent technology.

Energy demand (MESSAGEix):

Baseline energy service demands are provided exogenously to MESSAGEix, for the NGFS scenarios based on
SSP2 with a COVID update based on Kikstra et al., 2021. These baseline demands are adjusted endogenously
based on energy price changes using the MESSAGEix-MACRO link. There are seven energy service demands
that are provided to MESSAGEix, including:

1. Residential/commercial thermal

2. Residential/commercial specific

3. Industrial thermal

4. Industrial specific

5. Industrial feedstock (non-energy)

6. Transportation

7. Non-commercial biomass.

64 For more details please refer to https://docs.messageix.org/projects/global/en/latest/energy/fuel_blending.html

65 For example, among the electricity generation technologies, a separate technology, known as a pass-out turbine, is
considered an add-on technology. A pass out turbine allows select electricity generation technologies (i.e., parent
technologies) the option to reduce their electricity output in favour of generating electricity and heat. The pass out turbine,
which is a steam turbine in which a certain amount of the pressurized steam is passed out of the turbine for the purpose of
heat production, is restricted to a share of the activity of the selected electricity generation technologies. Technically, this
means that the electricity output of the electricity generation technologies remains unaltered, yet each unit of heat
generated by the pass out turbine, requires a certain electricity input.

66 For example, the possibility to retrofit existing fossil fuel-based energy generation technologies with CCS units. The
separate CCS-retrofit unit is depicted in the model, constrained by the activity of the respective parent technologies. CCS-
retrofits are available for: coal power plants including internal gasification combined cycle plants (IGCC), select gas power
plants, biomass power plants, gas and coal fuel cells as well as for hydrogen and cement production. The share of the total
emissions which can be reduced is limited to the technical feasibility and the combination of which mitigation technologies
are employed.

101
These demands are generated by relating historical country-level GDP per capita to final energy. and using
projections of GDP and population to extrapolate the seven-energy service demands into the future. 67, 68

Modelling policies (MESSAGEix):

MESSAGEix distinguishes between twelve global regions.69 It can account for currently implemented and
planned national policies - such as the nationally determined contributions (NDCs) as agreed upon in the Paris
Agreement - at a lower geographical resolution70, to be able to adequately account for future changes in the
scenario development processes.71 The targets formulated in national policies come in many different flavours.
This applies to the sectors and gases covered by these policies, but it also applies to how the policies are defined
and quantified. In MESSAGEix, four broad categories of policy types related to the different policies embedded
in different scenarios are represented, each of which is translated into a set of constraints: (i) emission targets,
(ii) energy shares, (iii) capacity or generation targets, (iv) macroeconomic targets such as energy-related taxes
and subsidies.72 As the year 2025 is rapidly approaching, the MESSAGE results for orderly 2C and 1.5C scenarios
have been fixed to the NDC trajectory for the year 2025. More stringent mitigation can only happen after 2025
in these scenarios.

Macroeconomic input variables (MACRO):

The main variables of the model are capital stock, available labour force (derived from population projections),
and energy inputs, which together determine the total output of an economy according to a nested constant
elasticity of substitution (CES) production function. The model’s most important driving input variables are the
projected growth rates of total labour, i.e., the combined effect of labour force and labour productivity growth,
and the annual rates of reference energy intensity reduction, i.e., the so-called autonomous energy efficiency
improvement (AEEI) coefficients. In the absence of price changes, energy demands grow at rates that are the

67 The sources for the historical and projected datasets are the following: Historical GDP (PPP) – World Bank (World
Development Indicators, 2012); Historical Population – UN Population Division (World Population Projection, 2010);
Historical Final Energy – International Energy Agency Energy Balances (IEA, 2012); Projected GDP (PPP) – Dellink et al.
(2015), also see Shared Socio-Economic Pathways database (SSP scenarios); Projected Population – KC and Lutz (2014),
also see Shared Socio-Economic Pathways database (SSP scenarios).

68 More details on the techniques and variables used to compute energy demands at regional levels and convergence rates
in final energy intensities across countries and sectors can be found at
https://docs.messageix.org/projects/global/en/latest/energy/demand.html

69 Regions in MESSAGEix-GLOBIOM are, in alphabetical order: China, Eastern Europe, Former Soviet Union, Latin America,

Middle East and North Africa, North America, Other Pacific Asia, Pacific OECD, Rest Centrally Planned Asia, South Asia,
South Asia, Sub-Saharan Africa, Western Europe.

70 National-level NDCs and the detailed policy targets specified are. MESSAGE aggregates those national targets to the
regional level and use it impose bounds in the scenario. Policies which cannot be directly applied as a constraint within
a scenario can be reflected by adjusting MACRO related parameters to reflect improvements on the demand side.

71 National-level NDCs and the detailed policy targets specified are taken into account. MESSAGE aggregates those national

targets to the regional level and uses them to impose bounds within the scenario. Policies which cannot be directly
applied as a constraint within a scenario can further be reflected by adjusting MACRO related parameters to reflect
improvements on the demand side.

72 A more detailed description of (i)-(vi) can be found under

https://docs.messageix.org/projects/global/en/latest/energy/policy.html

102
approximate result of potential GDP growth rates, reduced by the rates of overall energy intensity reduction.
The baseline GDP trajectory is calibrated to an externally provided GDP projection used in all IAMs for the NGFS
scenarios, existing of a combination of IMF short term projections and longer-term SSP2 projections (Dellink
et al., 2017). 73

Land-use input variables (GLOBIOM):

Spatial resolution: In order to enable global bio-physical process modelling of agricultural and forest production,
a comprehensive database has been built (Skalsky et al., 2008), which contains geo-spatial data on soil,
climate/weather, topography, land cover/use, and crop management (e.g., fertilisation, irrigation). The data
were compiled from various sources (FAO, ISRIC, USGS, NASA, CRU UEA, JRC, IFRPI, IFA, WISE, etc.) and
significantly vary with respect to spatial, temporal, and attribute resolutions, thematic relevance, accuracy, and
reliability. Data were harmonized into several common spatial resolution layers as well as country layers. At the
global scale, five altitude classes, seven slope classes, and five soil classes have been included. 74

Crop production: GLOBIOM directly represents production from three major land cover types: cropland,
managed forest, and areas suitable for short rotation tree plantations. Crop production accounts for more than
30 of the globally most important crops. The average yield level for each crop in each country is taken from
FAOSTAT.75

Livestock: GLOBIOM distinguishes between (i) livestock population, (ii) livestock products, (iii) livestock feed,
(iv) grazing forage availability and (vi) livestock dynamics. 76

Forestry: The forestry sector is represented in GLOBIOM with five categories of primary products (pulp logs,
saw logs, biomass for energy, traditional fuel wood, and other industrial logs) which are consumed by industrial
energy, cooking fuel demand, or processed and sold on the market as final products (wood pulp and sawn
wood). These products are supplied from managed forests and short rotation plantations.77

Land use change: Land cover types include cropland, grassland, short rotation plantations, managed forests,
unmanaged forests, other natural land, other agricultural land, wetlands, and not relevant (bare areas, water
bodies, snow and ice, and artificial surfaces). Economic activities are associated with the first four land cover
types.78

73 Cfr. also for more details https://docs.messageix.org/projects/global/en/latest/macro.html

74 For a more detailed description of how spatial resolution is modeled in GLOBIOM please refer to
https://docs.messageix.org/projects/global/en/latest/land_use/spatial.html

75 For more details on yield coefficients and crop management systems please refer to
https://docs.messageix.org/projects/global/en/latest/land_use/crop.html

76 A more detailed description of (i)-(vi) can be found under

https://docs.messageix.org/projects/global/en/latest/land_use/livestock.html

77 For further details on the estimation of harvesting costs and mean annual increments plese refer to
https://docs.messageix.org/projects/global/en/latest/land_use/forest.html

78 Cfr. also https://docs.messageix.org/projects/global/en/latest/land_use/land.html

103
Food demand: Food demand is endogenous in GLOBIOM and depends on population, gross domestic product
(GDP) and own product price. Population and GDP are exogenous variables while prices are endogenous. The
latter are computed via a simple demand system that uses as inputs population GDP per capita and income
elasticities. It is further assumed that food demand in developed countries is more inelastic than in developing
ones. In the latter the value of this elasticity is assumed to decrease with the level of GDP per capita to the price
elasticity of the USA in 2000.79

Land-use emulator: The land-use emulator integrates a set of land-use scenarios into the MESSAGEix energy
system model (RES). Each land-use scenario represents a distinct land-use development pathway for a given
biomass potential and carbon price. The biomass potentials for use in the energy sector are determined by the
biomass price.80 In addition, for each level of biomass potential, different carbon prices reflect the cost of
mitigation for land-use related greenhouse gas (GHG) emissions. The combination of land-use pathways can
therefore be depicted as a trade-off surface, illustrated for SSP2 (Fricko et al., 2017) in the figure below (Figure
60). The figure depicts global biomass potentials and respective GHG emissions at different carbon prices
cumulated from 2010 to 2100.

Figure 60. Land-Use Pathway Trade-Off Surface for SSP2

79 More details
on data sources and the formulas used for computing price elasticities can be found under
https://docs.messageix.org/projects/global/en/latest/land_use/food.html

80 At lower biomass prices, biomass mainly stems from forest residues, for example from sawmills or logging residues. With

increasing prices, land-use will be shifted to make room for fast-rotation tree plantations, purposely grown for use in energy
production which may cause indirectly through increased competition with agricultural land deforestation of today’s forest.
At very high prices, roundwood will be harvested for energy production (for further details see Forestry) competing with
material uses.

104
From this trade-off surface it possible to deduct that when climate policy scenarios are run in MESSAGEix, the
land-use pathways will be chosen such that the optimal balance between the land-use related emission and
biomass use in the energy system is obtained.

Emissions input variables (MAGICC):

MAGICC receives its main inputs, GHG and aerosol emissions, from the MESSAGEix energy system RES (i.e.,
CO2 emissions, non-CO2 GHGs, air pollution) and from GLOBIOM (crop sector emissions, livestock emissions,
land use change emissions).

4. Key model outputs

Description of key output variables/sectors

Output variables are available for the World aggregate and for 12 regions (China, Eastern Europe, Former Soviet
Union, Latin America and the Caribbean, Middle East and North Africa, North America, Other Pacific Asia,
Pacific OECD, Rest Centrally Planned Asia, South Asia, Sub-Saharan Africa, Western Europe). Within these
regions data are further downscaled at country-level. Data are provided in 5-year steps until 2060 and 10-year
steps thereafter until 2110. The output variables available for MESSAGEix-GLOBIOM are denoted by “M” in the
last column of the tables presented in Appendix.

5. What is new in the NGFS Phase IV scenarios?

What is new in the 2023 edition?

In the internal representation of the model, the chemical industry sectors are expanded to include ammonia
and methanol production processes. These variables are currently not specifically reported in the NGFS
scenarios.

A new interpolation scheme for aggregating country-level net-zero trajectories has been used. We now use a
linear reduction of emissions over time rather than an endogenous model-determined timing based on a
cumulative budget in each region.

Policy details and assumptions for the Current Policies and NDC scenarios are updated to include relevant
information up to the cut-off date of Fall 2022.

We apply an improved method for regionally different carbon price reporting based on regionally defined
emission trajectories. Before, the price derived from the global targets was dominant on the regional level, but
now MESSAGEix produces regional prices consistent with mitigation at the regional level.

Scenario implementation: differences from other NGFS models

Net-zero targets
 Translation of national net-zero targets (CO2 and GHG) to regional (R12) level

o Countries without net-zero pledges can keep their emissions amount at the base year in the
region.

105
 o Countries with net-zero pledges collectively set the regional emission bounds, which are
constructed by linearly interpolating aggregated country-level target mitigations in different
target years.

Model period interpretation

 MESSAGEix model year represents the period between the given and the previous model year, e.g.,
2050 is for 2046-2050 (with 5-year interval).

 This also affects the interpretation of net-zero target years in MESSAGE scenarios (i.e., the reported
2050 value is often not yet fully zero, because it also includes the years 2046-2049 which had actual
emissions).

GDP (MER)
 Apply own MER-PPP conversions to the base GDP (PPP) and apply the cumulative growth rates to our
base year GDP

Baseline demand
 Multiply the derived GDP trajectory to the end-use energy intensities extracted from an own COVID-
adjusted scenario (Kikstra et al., 2021)

106
Module 4: IAM – GCAM

1. Non-technical summary

The Global Change Analysis Model (GCAM) is a global market equilibrium model, that combines economic,
energy, land use, and climate systems to analyse the interactions between human activities and global
environmental changes. It is designed to assess the impacts of various policy scenarios and technology options
on energy use, land use change, greenhouse gas emissions and climate change. GCAM is an Integrated
Assessment Model (IAM) applied to the NGFS phase IV scenarios.

Key model inputs into GCAM’s modules:

 Macroeconomy: population; labour productivity growth rate; labour force participation rate; base year
GDP.

 Earth system: atmospheric CO2 concentrations; radiative forcing of emissions; global mean temperature
change; air-land carbon fluxes; air-sea carbon fluxes.

 Land use: historical land use and land cover; vegetation carbon density; soil carbon density.

 Water: crop, electricity, livestock, primary energy, and industry water coefficients; crop and electricity
production; crop, electricity, livestock, and primary energy production; industry output.

 Emissions: emissions and activity data by sector; energy production and consumption; agricultural
production; land use and land use change.

 Marketplace: supply of and demand for all energy commodities; supply of and demand for all agriculture
and land-based commodities; supply of and demand for all water types.

Key model outputs of GCAM for each NGFS phase IV scenario and horizon year:

 Emissions: emissions (CO2 and non-CO2); resource production emissions (CO2 and non-CO2); land use
change emissions; CO2 sequestration.

 Land use: land use and land cover; land use change emissions; change in above and below ground carbon.

 Prices: energy; agriculture and forestry; water; fish.

 Quantity: energy production and consumption; agriculture production and consumption; water
withdrawals; consumption and supply.

Key updates: The 2023 edition of GCAM (version 6.0) used for the NGFS phase IV scenarios, includes a number
of updates: a new residential floorspace expansion model; bio-energy updates; reset of default hotelling rate
for climate stabilization scenarios to 3%; splitting out six detailed industrial sectors from the aggregate industry
sector; updated hydrogen production, distribution, and end-use technologies; a new protected lands definition;
expanded crop commodities; HFC MAC curve fixes; new pollutant emissions controls.

107
2. Overview of model scope and methods

GCAM has been under development for over 40 years. Work began in 1980 with the work first documented
in 1982 in working papers (Edmonds and Reilly, 1982a,b,c) and the first peer-reviewed publications in
1983(Edmonds and Reilly, 1983a,b,c)

Throughout its lifetime, GCAM has evolved in response to the need to address an expanding set of science
and assessment questions. The original question that the model was developed to address was the magnitude
of mid-21st-century global emissions of fossil fuel CO2. Over time GCAM has expanded its scope to include a
wider set of energy producing, transforming, and using technologies, emissions of non-CO2 greenhouse gases,
agriculture and land use, water supplies and demands, and physical Earth systems. GCAM has been used to
produce scenarios for national and international assessments ranging from the very first IPCC scenarios
through the present Shared Socioeconomic Pathways(Calvin et al, 2017). GCAM is increasingly being used in
multi-model, multi-scale analysis, in which it is either soft- or hard-coupled to other models with different
focuses and often greater resolution in key sectors. For example, a range of downscaling tools have been
developed for use with GCAM to be able to obtain land and water outputs at a grid resolution. Similarly, it has
been coupled to a state-of-the-art Earth system model(Collins et al, 2015)

GCAM includes two major computational components: a data system to develop inputs and the GCAM
core. GCAM takes in a set of assumptions and then processes those assumptions to create a full scenario of
prices, energy and other transformations, and commodity and other flows across regions and into the future.
The interactions between these different systems all take place within the GCAM core; that is, they are not
modelled as independent modules, but as one integrated whole.

While the agents in the GCAM model are assumed to act to maximize their own self-interest, the model as
a whole is not performing an optimization calculation. In fact, actors in GCAM can make decisions that
“seemed like a good idea at the time”, but which are not optimal from a larger social perspective and which the
decision maker would not have made had the decision maker known what lay ahead in the future. For example,
the model’s actors do not know about future climate regulations and could install fossil fuel power in the years
preceding the implementation of such policies.

Key scenario assumptions for the GCAM core:

 Macroeconomy: population, labour participation, and labour productivity.

 Energy technology characteristics: e.g., costs, performance, water requirements, GHG and other
emissions coefficients.

 Agricultural technology characteristics: e.g., crop yields, costs, carbon contents, water requirements,
fertiliser requirements.

 Energy and other resources: e.g., capital/extraction costs and availability of fossil fuel resources and
reserves, wind, solar, uranium, groundwater.

 Policies: e.g., wide range of potential regulatory and fiscal policies including emissions constraints, coal
phaseout, renewable portfolio standards, EV targets, fuel efficiency standards, etc.

Key scenario results from the GCAM core:

108
 Energy system: energy demands, flows, technology deployments, international trade, and prices
throughout the energy system.

 Agriculture and land use: prices, supplies, and consumption of all agricultural and forest products, land
use and land use change.

 Water: water demands and supplies for all agricultural, energy, and household uses.

 Emissions: 24 greenhouse gases and short-lived species: CO2, CH4, N2O, halocarbons, carbonaceous
aerosols, reactive gases, sulphur dioxide.

GCAM is an integrated, multi-sector model that explores both human and Earth system dynamics. The role
of models like GCAM is to bring multiple human and physical Earth systems together in one place to shed light
on system interactions and provide scientific insights that would not otherwise be available from the pursuit of
traditional disciplinary scientific research alone. As shown in Figure 61, GCAM is constructed to explore these
interactions in a single computational platform with a sufficiently low computational requirement to allow for
broad explorations of scenarios and uncertainties. Components of GCAM are designed to capture the behavior
of human and physical systems, but they do not necessarily include the most detailed process-scale
representations of its constituent components. On the other hand, model components in principle provide a
faithful representation of the best current scientific understanding of underlying behavior.

Figure 61. GCAM Model Integration

GCAM allows users81 to explore what-if scenarios, quantifying the implications of possible future
conditions. These outputs are conditional forecasts contingent on the validity of input assumptions; they are a
way of analysing the potential impacts of different assumptions about future conditions Figure 62 illustrates
how GCAM reads in external “scenario assumptions” about key drivers (e.g., population, economic activity,

81 In
this context GCAM users are equivalent to GCAM-based NGFS climate policy scenario implementations whose
outputs are provided by the NGFS Scenario Explorer.

109
technology, and policies) and then assesses the implications of these assumptions on key scientific or decision-
relevant outcomes (e.g., commodity prices, energy use, land use, water use, emissions, and concentrations).

Figure 62. Use of scenario assumptions to produce fuller, modelled scenarios

The GCAM core is the component of the model in which economic decisions are made (e.g., land use and
technology choices), and in which dynamics and interactions are modelled within and among different
human and Earth systems. Supplied with input information from the GCAM data system, the GCAM core
is the heart of the dynamic character of GCAM. GCAM takes in a set of assumptions and then processes those
assumptions to create a full scenario of prices, energy and other transformations, and commodity and other
flows across regions and into the future. GCAM represents five different interacting and interconnected
systems. The interactions between these different systems all take place within the GCAM core; that is, they
are not modelled as independent modules, but as one integrated whole. The five systems in the GCAM core are
as follows (see also Figure 63 for a schematic visualization):

 Macroeconomy: This module takes population and labor productivity assumptions as inputs and produces
regional gross domestic product and regional populations as inputs for the other modules. The
macroeconomy sets the scale of economic activity in GCAM.

 Energy systems: The energy system is a detailed representation of the sources of energy supply, modes
of energy transformation, and energy service demands such as passenger and freight transport, industrial
energy use across subsectors, and residential and commercial energy service demands. The module
reports demand for, and supply of, energy forms, as well as emissions of greenhouse gases, aerosols, and
other short-lived species. Energy systems demand bioenergy from agriculture and land systems and water
from water systems.

 Agriculture and land systems: The agriculture and land systems provide information about land use, land
cover, carbon stocks and net emissions, the production of bioenergy, food, fibre, and forest products.
Demands are driven by the size of the population, their income levels, and commodity prices. The module
reports demand for and supply of agricultural and other commodities, land and emissions of greenhouse
gases, aerosols, and other short-lived species. The demand for bioenergy is a derived demand by the
energy sector. Agriculture and land systems demand water from water systems.

 Water systems: The water module provides information about water withdrawals and water consumption
for energy, agriculture, and municipal uses.

110
 Physical Earth system: The physical Earth system in GCAM is modelled using Hector, a physical Earth
system emulator that provides information about the composition of the atmosphere based on emissions
provided by the other modules, ocean acidity, and climate.

Figure 63. Conceptional schematic of GCAM core operation

The exact structure of the model explored in the GCAM core – for example, the number of regions and
technologies – is data driven. In all cases, the GCAM core represents the entire world, but it is constructed with
different levels of resolution for each of these different systems. In release version 6.0 of GCAM (which is used
for the NGFS phase IV scenarios), the energy-economy system operates at 32 regions globally, land is divided
into 384 subregions, and water is tracked for 235 basins worldwide. The Earth system module operates at a
global scale.

The core operating principle for GCAM is that of market equilibrium. Representative agents in GCAM use
information on prices, as well as other information that might be relevant, and make decisions about the
allocation of resources. These representative agents exist throughout the model, representing, for example,
regional electricity sectors, regional refining sectors, regional energy demand sectors, and land users who must
allocate land among competing crops within any given land region. Markets are how these representative
agents interact with one another. Agents indicate their intended supply and/or demand for goods and services
in the markets. GCAM solves for a set of market prices balanced supply and demand in all these markets across
the model. The GCAM solution process is the process of iterating on market prices until this equilibrium is
reached. Markets exist for physical flows such as electricity or agricultural commodities, but they also can exist
for other types of goods and services, for example tradable carbon permits.

111
  In any single model period, GCAM derives a demand for natural gas
starting with all the uses to which natural gas might be put, such as
Explainer box 4 passenger and freight transport, power generation, hydrogen production,
What is an example for heating, cooling, and cooking, fertilizer production, and other industrial
the market equilibrium energy uses.
mechanism in GCAM?
 Those demands depend on the external assumptions about, for example,
electricity generating technology efficiencies, but also on the price of all the
commodities in the model. GCAM then calculates the amount of natural gas
that suppliers would like to supply given their available technology for
extracting resources and the market price. The model gathers this same
information for all the commodities and then adjusts prices so that in every
market during that period supplies of everything from rice to solar power
match demands.

GCAM is a dynamic recursive model, meaning that decision-makers do not know the future when planning
today. After it solves each period, the model then uses the resulting state of the world, including the
consequences of decisions made in that period – such as resource depletion, capital stock retirements and
installations, and changes to the landscape – and then moves to the next time step and performs the same
exercise. For long-lived investments, decision-makers may account for future profit streams, but those
estimates would be based on current prices.

3. Key model inputs

The macroeconomy component of GCAM 6.0 sets the scale of economic activity and associated demands
for model simulations. Assumptions about population and per capita GDP growth for each of the 32
geopolitical regions together determine the gross domestic product (GDP). GDP and population both can drive
the demands for a range of different demands within GCAM. Population and economic activity are used in
GCAM through a one-way transfer of information to other GCAM components (see belowKey model
outputson Key model outputs for explanation on the reported GDP for scenarios, which is different to the one
described here used for the demand determination). For example, neither the price nor quantity of energy nor
the quantity of energy services provided to the economy affect the calculation of the principal model output of
the GCAM macroeconomic system, GDP (due to unidirectionality between the macro and energy modules).
Changes in future per capita GDP and population will affect the final demand for energy, food, and forestry. For
example, increases in population will increase regional consumption proportionally, while changes in per capita
GDP affect consumption through income elasticities. Thus, different assumptions of future GDP and population
growth across different socioeconomic scenarios may play key roles in determining an alternative future. In
addition, regional heterogeneity in future GDP and population growth, leading to heterogeneous regional
demand growth, is also a critical driver to future changes in regional supply, biophysical responses, and trade
patterns. Table 6 shows the inputs for the economic module.

Table 6. Inputs required by the economic module

Name Resolution

Population Region and year

112
 Name Resolution

Labour productivity growth rate Region and year

Labour force participation rate Region and year
Base year GDP Region

The Earth system model (i.e., carbon-cycle climate module) Hector is the default climate model within GCAM
(Hartin et al. ,2015). Users still have the option of running MAGICC in GCAM version 5.1, but this option is no
longer supported beginning with version 6.0 of GCAM. 82 Hector (v2.5.0) runs essentially instantaneously while
still representing the most critical global-scale earth system processes. This model has a three-part main carbon
cycle: a one-pool atmosphere, three-pool land, and four-pool ocean. The model’s terrestrial carbon cycle
includes primary production and respiration fluxes, accommodating arbitrary geographic divisions into, e.g.,
ecological biomes or political units. Hector actively solves the inorganic carbon system in the surface ocean,
directly calculating air-sea fluxes of carbon and ocean acidity. Hector reproduces the global historical trends of
atmospheric CO2, radiative forcing, and surface temperatures. The model simulates all four Representative
Concentration Pathways (RCPs) with equivalent rates of change of key variables over time, consistent with
compared to historical observations, MAGICC, and models from the Coupled Model Intercomparison Project
(CMIP5). Currently the GCAM sectors interact with Hector via emissions. At every time step, emissions from
GCAM are passed to Hector. Hector converts these emissions to concentrations when necessary, and calculates
the associated radiative forcing, as well as the response of the climate system and earth system (e.g.,
temperature, carbon-fluxes, etc.).

Economic land use decisions in GCAM are based on a probabilistic, logit model of land-allocation based on
relative expected profitability of using land for competing purposes. In GCAM, there is a distribution of profit
behind each competing land use within each of the 384 land-use regions. The share of land allocated to any
given use is based on the probability that that use has the highest profit among the competing uses.

The way land types are nested in GCAM, in combination with the logit exponents 83 used, determines the
substitutability of different land types in the model in future periods. Figure 64 shows a nesting diagram of land
with a subregion.

82 This contrasts with MESSAGE-GLOBIOM and REMIND-MAgPIE which use MAGICC as the climate model (see dedicated
box in module 1).
83 GCAM requires the user to specify the logit exponents that determine the substitutability between different leaves and
nodes in the land model.

113
 Figure 64. Agriculture and land use (AgLU) land nesting structure

The following Table 7 shows the inputs relevant for the land-use module.

Table 7. Inputs required by the land allocation module84

Name Resolution

Historical land use and land cover By GLOBE Land Unit (GLU), land type, and year
Vegetation carbon density By GLU and land type
Soil carbon density By GLU and land type
Mature age By GLU and land type
Soil time scale By geopolitical region and land type
Value of unmanaged land By GLU
Profit rate of managed land By GLU
Logit exponents By GLU and land node

In the water module, three distinct sources of fresh water are modelled: renewable water, non-renewable
groundwater, and desalinated water. Renewable water is water that is replenished naturally by surface runoff
and subsurface infiltration and release (groundwater recharge). Non-renewable groundwater is water from
aquifers whose recharge is sufficiently low as to be depletable on a human time scale and which have
replenishment timescales greater than 100 years. Renewable water and non-renewable groundwater are

84 Note that this table differs from Table 29 in that it lists all external inputs to the land module, including information
passed from other modules. This table shows the variables used in the GCAM simulation after processing.

114
separately modelled for each basin. Desalinated water of brackish groundwater and seawater is available as an
additional source of freshwater within each basin and for municipal and industrial end-use demands for water.

Conveyance losses and improvements to water distribution efficiencies are included in the water distribution
sectors. Conveyance losses for irrigated water use has been included and differentiated for each GCAM region.
Conveyance losses/efficiencies for GCAM regions are derived from country level data(From Rohwer et al., 2007)
and are the weighted mean of the original country level data weighted by irrigated harvested area. Water
supplies and demands at each basin are balanced through a market mechanism in which prices for water
(shadow price) are adjusted until water demands are constrained to available supply. The following Table 8
shows the inputs relevant for the water demand module.

Table 8. Inputs required by the water demand module

Name Resolution

Crop water coefficients GLU, GCAM commodity, water type (consumption, withdrawals,
biophysical consumption) and year
Crop production GLU, GCAM commodity, and year
Electricity water coefficients GCAM region, technology, water type (consumption,
withdrawals) and year
Electricity production GCAM region, technology, and year
Livestock water coefficients GCAM region, livestock type, water type (consumption,
withdrawals, biophysical consumption) and year
Livestock production GCAM region, livestock type, and year
Primary energy water GCAM region, fuel, water type (consumption, withdrawals,
coefficients biophysical consumption) and year
Primary energy production GCAM region, fuel, and year
Industry water coefficients GCAM region, water type (consumption, withdrawals, biophysical
consumption) and year
Industry output GCAM region and year
Income and price elasticity By region, demand, and year
GDP per capita By region and year
Population By region and year

GCAM’s Energy Module tracks production of primary energy forms, their transformation into end-use fuels
and electricity, and the production of energy services such as heating, cooling, passenger and freight transport,
and process heat. Figure 65 gives an overview of the GCAM energy system.

GCAM models primary energy production for both depletable and renewable energy forms. GCAM models
depletable resources (oil, unconventional oil, natural gas, coal, and uranium) using graded resource supply
curves. Production of depletable resources occurs out of reserves. Resources are transformed to reserves based
on the cost of finding and bringing resources into production. GCAM’s renewable resources include onshore
wind, offshore wind, solar, geothermal, hydropower, and biomass; some regions are also assigned a “traditional
biomass” resource. In general, the costs of producing electricity from renewable energy forms consist of the

115
sum of the resource costs, the technology costs85, and in some cases, backup-related costs. In the energy
transformation module, the competition between subsectors takes place according to a calibrated logit sharing
function86. Broadly, the energy transformation sectors in GCAM consist of all supply sectors between
the primary energy resources and the final energy demands (i.e., buildings, industry, and transportation).

The main energy transformation sectors are electricity, refining, gas processing, hydrogen production,
and district services. Within the subsectors, there may be multiple competing technologies, where
technologies typically represent either different efficiency levels, and/or the application of carbon dioxide
capture and storage (CCS). Most of the economic activities represented in GCAM present a choice among
several ways to produce the result of the activity. Examples of these choices include choosing between different
fuels or feed stocks, between different technologies, and between transportation modes. In some cases, the
choice is between different uses of a limited resource, such as when land area is allocated to different uses.
Choice in GCAM is based on a single numerical value that orders the alternatives by preference (i.e., a choice
indicator). In practice the choice indicator is either cost or profit rate, though other indicators are possible in
principle. In cases where multiple factors influence a choice, such as passenger transportation (where faster
modes are more desirable), the additional factors are converted into a cost penalty and added to the basic cost
to produce a single indicator that incorporates all the relevant factors. More information in this GCAM technical
document.

Figure 65. GCAM energy system

85 The cost of a technology in any period depends on its exogenously specified non-energy cost, its endogenously
calculated fuel cost, and any cost of emissions, as determined by the climate policy. The first term, non-energy cost,
represents capital, fixed and variable operating and maintenance costs incurred over the lifetime of the equipment
(except for fuel or electricity costs). For electricity technologies, GCAM reads in each of these terms and computes the
levelized cost of energy within the model. For example, the non-energy cost of coal-fired power plant is calculated as
the sum of overnight capital cost (amortized using a capital recovery factor and converted to dollars per unit of energy
output by applying a capacity factor), fixed and variable operations and maintenance costs. The second term, fuel or
electricity cost, depends on the specified efficiency of the technology, which determines the amount of fuel or
electricity required to produce each unit of output, as well as the cost of the fuel or electricity.
86 See the Key model inputsfor explanations.

116
GCAM projects emissions of a suite of greenhouse gases (GHGs) and air pollutants: CO 2, CH4, N2O, CF4, C2F6,
SF6, HFC23, HFC32, HFC43-10mee, HFC125, HFC134a, HFC143a, HFC152a, HFC227ea, HFC236fa, HFC245fa,
HFC365mfc, SO2, BC, OC, CO, VOCs, NOx, NH3. Future emissions are determined by the evolution of drivers
(such as energy consumption, land use, and population), technology mix, and abatement measures. How this
is represented in GCAM varies by emission type. The following Table 9 shows the inputs relevant for the
emissions module.

Table 9. Inputs to the emissions module

Name Resolution

Emissions data by sector for non-CO2 Country, sector, fuel, gas, year
Activity data from GCAM by sector By region, year, sector, fuel
Marginal abatement cost (MAC) assumptions By region, sector, year
Energy production (for emissions driven by production) By region, technology, year
Energy consumption (for emissions driven by By region, technology, year
consumption)
Agricultural production By GLU, technology, year
Land use and land use change By GLU, type, year

GCAM operates by determining a set of prices that ensure supply is equal to demand for all time steps. The
marketplace collects the supplies and demands and uses solver algorithms to determine those prices. Given
a carbon price, the resulting emissions will vary depending on other scenario drivers, such as population, GDP,
resources, and technology. The following Table 10 shows the inputs relevant for the marketplace module.

Table 10. Inputs required by the marketplace

Name Resolution

Supply of all energy commodities Region and year

Demand for all energy commodities Region and year
Supply of all agriculture and land-based commodities Region and year
Demand for all agriculture and land-based
Region and year
commodities
Supply of all water types Basin and year
Demand for water withdrawals and consumption Basin and year

One of GCAM’s uses is to explore the implications of different future policies. There are a number of types of
policies that can be easily modelled in GCAM. There are three primary top-down policy approaches that can be
applied in GCAM to reduce emissions of CO2 or other greenhouse gases: carbon or GHG prices, emissions
constraints, or climate constraints.87 In all cases, GCAM implements the policy approach by placing a price on

87 Besides emissions-related policies GCAM can also integrate energy production as well as land-use policies.

117
emissions. This price then filters down through all the systems in GCAM and alters production and demand. For
example, a price on carbon would put a cost on emitting fossil fuels. This cost would then influence the cost of
producing electricity from fossil-fired power plants that emit CO2, which would then influence their relative cost
compared to other electricity generating technologies and increase the price of electricity. The increased price
of electricity would then make its way to consumers that use electricity, decreasing its competitiveness relative
to other fuels and leading to a decrease in electricity demand. The three policy approaches are described below.
For the NGFS scenarios, mostly emissions constraints are used, with the “Net Zero 2050” scenario using
exogenously calculated GHG prices.

 Carbon or GHG prices: GCAM users can directly specify the price of carbon or GHGs. Given a carbon price,
the resulting emissions will vary depending on other scenario drivers, such as population, GDP, resources,
and technology.

 Emissions constraints: GCAM users can specify the total amount of emissions (CO2 or GHG) as well. GCAM
will then calculate the price of carbon needed to reach the constraint in each period of the constraint.

 Climate constraints: GCAM users can specify a climate variable (e.g., concentration or radiative forcing)
target for a particular year. Users determine whether that target can be exceeded prior to the target year.
GCAM will adjust carbon prices in order to find the least cost path to reaching the target.

In addition to the three primary top-down policy mechanisms that GCAM can model, GCAM can also model
specific sectoral regulatory policies such as renewable portfolio standards, new source performance standards,
and other regulatory policy instruments.

The cost of GHG emissions mitigation is a concept that is not uniquely defined. A wide range of measures are
used in the literature. These include the price of carbon (or as appropriate given the policy) needed to achieve
a desired emission mitigation goal, reduction in gross domestic product (GDP), consumption loss, deadweight
loss (i.e., cost caused by market inefficiency), and equivalent variation. Beyond that is the concept of net cost,
which includes the benefits of emissions mitigation as well as the resource cost of emissions reduction, while
the social cost of carbon is also encountered. GCAM makes no attempt to calculate the benefits, and thus does
not estimate net costs or benefits.

In addition to identifying policy prices as one measure of cost, GCAM employs the “deadweight loss” approach
to measuring welfare loss from emissions mitigation efforts. GCAM employs the deadweight loss approach for
several reasons. First, the deadweight loss approach is numerically straight forward to calculate in GCAM.
Second, the deadweight loss approach provides a computationally tractable method to measuring the change
in welfare, though it is only an approximation. 88 Third, the deadweight loss approach takes advantage of
GCAM’s detailed technological characterisation.

GCAM calculates the cost of emissions mitigation at each GCAM time step.89 For example, in Figure 66 below,
the cost of moving from a reference path without a carbon tax (blue) to the emissions path with a carbon tax
(green) in period T can be calculated simply. Successive scenarios with fixed carbon taxes in period T are run.

88 In principle the equivalent variation is the right approach to measure an individual’s loss in welfare. Equivalent variation
measures the minimum amount of income that would be needed to leave consumers just as happy with the new price
(e.g., carbon tax) as without. However, its calculation requires either knowledge of all of society’s individual
preference functions or the existence of a well-ordered set of social preferences, a requirement that Arrow
(1950) demonstrated to be impossible under ordinary circumstances.
89 Note that calculation of policy costs is currently only supported in GCAM for policies pegged to CO2 prices.

118
The associated emissions are recorded for each carbon tax. The cost is calculated as the area of the purple
triangle, which is the integral of each emissions mitigation step weighted by the carbon tax that was required
to deliver the reduction. The final ton of carbon emissions is the most expensive ton because it is assumed that
for a carbon tax, emissions mitigation occurs with the least expensive tons being reduced first. The final ton of
carbon is simply the carbon tax rate itself. The tax revenue can be calculated as the tax rate times the remaining
emissions, shown in red below.

Figure 66. Carbon emissions paths, costs, and tax revenues

Description of key input variables and main assumptions

GCAM’s demand inputs include information on consumption and prices in the historical period to calibrate
model parameters (see Table 6, Table 7, Table 8). Additional parameters related to income and price
elasticities are needed for modeling future periods. GCAM requires demand data to be globally consistent
with supply data for each of its historical model periods as it solves for market equilibrium in these years as it
does for future years. These inputs are required for each region and historical year. GCAM’s economic inputs
include information on population and income (see Table 9). These inputs are required for each geopolitical
region and historical year. GCAM’s external land inputs include information on land, carbon, other emissions,
and the value of unmanaged land in the historical period (see Table 10). These inputs are required for
each global land unit and historical year. GCAM’s supply inputs include information on production, prices,
technology cost and performance, and other emissions in the historical period to calibrate model parameters
(see Table 30, Table 31, Table 32). In addition, GCAM’s supply modeling requires information on future
technology cost and performance and emissions factors for future periods. GCAM requires that supply data is
globally consistent with demand data for each of its historical model periods as it solves for market equilibrium
in these years as it does for future years. These inputs are required for each region and historical year. GCAM
subdivides the world into 32 geopolitical regions, representing countries or collections of countries (see

Table 33).

119
4. Key model outputs

Description of key output variables/sector

The following tables comprise the general GCAM modular outputs and their respective units. The release
version of GCAM is typically operated in five-year time steps with 2015 as the final calibration year and time
horizon 2100. Figure 67 shows a final overview of GCAM’s model inputs and outputs.

Table 11.GCAM outputs from emissions modelling

Name Resolution Unit

Emissions (CO2) Technology, region, and year MtC/year

Emissions (non-CO2) Technology, region, and year Various
Resource production emissions (CO2) Subresource, region, and year MtC/year
Resource production emissions (non- Subresource, region, and year MtC/year
CO2)
Land use change emissions By GLU and land type MtC/year
Change in above ground carbon By GLU and land type MtC/year
Change in below ground carbon By GLU and land type MtC/year
CO2 sequestration Technology, region, and year MtC/year
The units of non-CO2 emissions vary. Fluorinated gas emissions are reported in Gg of the specific gas per year.
All other emissions are reported in Tg of the specific gas per year (e.g., CH4 emissions are reported in TgCH4/yr.).

Table 12. GCAM outputs from the land model

Name Resolution Unit

Land use and land cover By GLU, land leaf, and year Thousand km²
Land use change emissions By GLU and land leaf MtC/year
Change in above ground carbon By GLU and land leaf MtC/year
Change in below ground carbon By GLU and land leaf MtC/year
Above ground carbon stock By GLU and land leaf MtC
Profit rate By GLU and land leaf 1975$/thousand km²

Table 13. GCAM price outputs

Name Resolution Unit

Price Market and year Various

Food demand prices Region, type and year 2005$/Mcal/day

The price units vary by market. In general, energy-related prices are reported in $1975/GJ, agricultural prices
are in $1975/kg, forestry prices are in $1975/m³, and carbon prices are in $1990/tC.

120
Table 14.GCAM quantity outputs

Name Resolution Unit

Physical Output Technology, region, vintage, and year Various

Resource production Region, resource and year Various
Inputs Technology, input, region, vintage, and Various
year
Supply Market and year Various
Demand Market and year Various
The quantity units vary. In general, energy-related outputs are reported in EJ/yr., agricultural outputs are in
Mt/yr., forestry outputs are in million m³/yr., and water outputs are in km³/yr.

Figure 67. Overview of GCAM inputs and outputs

Table 34 shows a complete list of the sectoral GCAM variables with available output data for all 32 GCAM
regions and implemented NGFS climate policy scenarios as provided by the NGFS Phase IV Scenario Explorer
(indicated as “G” in the IAM column). There, the main variables (variable groups) are also explained briefly.

GCAM utilizes a prescribed (exogenous) GDP trajectory. It currently does not employ an energy-GDP feedback
mechanism. Since the NGFS scenario are representing an estimate of the full economic consequences of
different scenarios, GDP values in non-reference scenarios (so all scenarios except Current Policies) were
replaced with a modified GDP that uses the scenario carbon price and the relationship between the carbon price
and GDP change from the REMIND-MAgPIE model to create a GDP path consistent with the REMIND-MAgPIE
model response to emissions mitigation. However, since the GCAM energy, agriculture and land-use system
produces its own unique carbon prices based on all of the information about energy-agriculture and land-use

121
interactions, the GCAM GDP consistent with transformation pathways is different than the REMIND-MAgPIE
GDP pathway.

The GCAM GDP for scenarios other than the reference scenario was calculated using the following formula:

𝑅𝐸𝑀𝐼𝑁𝐷
𝐺𝐶𝐴𝑀
%∆𝐺𝐷𝑃𝑟𝑒𝑓 (𝑡) 𝐺𝐶𝐴𝑀
𝐺𝐷𝑃𝐺𝐶𝐴𝑀∗ (𝑡) = 𝐺𝐷𝑃𝑟𝑒𝑓 (𝑡) (1 + ( 𝑅𝐸𝑀𝐼𝑁𝐷 ) 𝑃𝐶𝑂2 (𝑡))
𝑃𝐶𝑂2 (𝑡)

Where, the reference scenario, ref, is the Current Policies scenario. GDP is measured in a common currency
using purchasing power parity, PPP. The regional marginal cost of emissions mitigation is measured as the price
%∆𝐺𝐷𝑃𝑅𝐸𝑀𝐼𝑁𝐷
𝑟𝑒𝑓 (𝑡)
of CO2 or PCO2, . We used the REMIND model's regional change in GDP to carbon price ratios, as
𝑃𝑅𝐸𝑀𝐼𝑁𝐷
𝐶𝑂2 (𝑡)
these most closely resembled the macroeconomic effects observed in the GCAM-MACRO model version
currently being developed for future use in the NGFS scenarios. Based on preliminary regional results of this
model, the regional GDP loss ratio between any non-reference scenario and the Current Policy reference case
was capped to 10% for losses, and to 1% for gains.

122
5. What is new in GCAM modeling for NGFS Phase IV scenarios?

1. A new residential floorspace expansion model

2. Bio-energy updates: additional limits to “unsustainable” deployment

3. Default hotelling rate for climate stabilization scenarios is now 3%

4. Split out six detailed industrial sectors from the aggregate industry sector90

5. Updated hydrogen production, distribution, and end-use technologies

6. A new protected lands definition

7. Expanded crop commodities

8. Use spatially explicit soil and vegetation carbon data from Moirai 91

9. HFC MAC curve fixes

10. New pollutant emissions controls

11. Solution improvements, particularly related to water markets

12. Change the XML parser library to RapidXML

13. Add the ability to exit the model early due to solution failure

14. Reduce memory usage (offsets additional memory from expanded crop commodities)

15. GCAM-data: Renv and user modification chunks

90 GCAM 5.3+ from NGFS phase 3 included a preliminary version of the industry split-up. The differences in these industry
module versions and the other differences in this list explain the scenario-independent result changes from phase 3 to
phase 4 for GCAM.

91 The Moirai Land Data System (Moirai LDS) is designed to produce recent historical land data inputs for the AgLU module

of the GCAM data system.

123
Module 5: Chronic physical risks

1. Non-technical summary

Chronic physical risks are associated with long-term shifts in climate patterns, and include risks associated with
long-term increases in temperature, changes in average precipitation patterns, rising sea levels, and ocean
acidification. The impact of these can be reflected in reduced labor and land productivity, capital depreciation,
scarcity of natural resources, forced migrations, increased adaptation costs, etc. 92

The IAM models described in previous sections are capable of calculating policy costs associated with the goals
of different scenarios. However, these models do not estimate the impact of the physical costs associated with
climate change on the economy, including the impacts of chronic physical risks. To fill this gap, the NGFS
scenarios include an ex-post (i.e., computed outside of IAMs) estimate of chronic physical risks.

The approach used for the economic impact estimates from chronic physical risks in this year’s edition of the
NGFS scenarios is the same as the approach used in the 2022 edition. However, the temperature paths for
which the results have been produced have been updated.

The methodology used for the NGFS estimation of chronic physical risk calculates macro-economic impacts
based on damage functions. Damage functions are relationships quantifying the effect of a change in climate-
related variables (e.g., temperature) on economic output. There is a rapidly growing body of empirical research
aimed at developing these functions, as well as other methods for estimating chronic physical risks, with
different approaches generating a wide range of different results (Figure 68). The NGFS methodology uses a
damage function based on mean temperature developed by Kalkuhl and Wenz (2020). While there are several
advantages of this approach (including its global coverage), it is important to recognize that this is an active
research area with large uncertainties.

 The approach used for the economic impact estimates from

chronic physical risks in this year’s edition of the NGFS
Explainer box 5
scenarios is the same as the approach used in the 2022
What is new in the 2023 edition. However, the temperature paths for which the
edition of the NGFS
results have been produced have been updated.
Scenarios ?

92 Note that not all of these impacts are explicitly captured in the NGFS estimates of chronic physical risk.

124
 Figure 68. Global aggregate economic impact estimates by global warming level (annual % global GDP loss relative to GDP
without additional climate change). Source: IPCC AR6 WGII Chapter 16, 2022.

NGFS scenarios account for three uncertainty dimensions related to the chronic risk estimation: (i) IAM
emissions output; (ii) the temperature predicted by the climate module; and lastly (iii) the estimates of the
damage function used. However, the scope of uncertainties on chronic damages is much wider. Regarding the
damage functions, for example, it remains an open question if the damages affect the level or the growth rate
of economic output (i.e., persistence effects). In addition, there are limitations in terms of what is captured by
the damage functions used, with the approach not explicitly accounting for: (i) possible future impacts that are
not reflected in historic relationships between temperature and GDP; (ii) damages from sea level rise, ocean
acidification, and other chronic impacts beyond temperature; and (iii) effects beyond impacts on labor and land
productivity and capital depreciation, including conflict, violence and migration, and biodiversity and
ecosystem impacts.

It is important to remark that, besides the limitations of the damage function methodology presented below,
the estimates of chronic damages are not integrated in IAMs. This means that optimizing agents in these
models are myopic with respect to physical costs and, therefore, the latter are not reflected in the savings
decisions, investment, etc. This limitation contributes to underestimating chronic damages. To address this
issue, an additional run of REMIND with endogenous chronic damages is presented in Scenarios with
integrated transition and physical risks.

125
2. Macro-economic damage estimates

For the NGFS scenarios, we use the results of a recent, state-of-the art econometric estimate by Kalkuhl &
Wenz (2020) to calculate country-level macroeconomic losses. The conceptual model is a stylised Ramsey-type
growth model focusing on aggregate productivity effects Θ(𝑇) and labour productivity growth 𝑔𝐴 (𝑇) on GDP
𝑑 ln 𝐴
growth (𝑔𝑦 ), where T is the global mean temperature change and 𝑔𝐴 ≔ . The equation below shows the
𝑑𝑡
different drivers of growth, with s = savings rate, 𝛿 = capital depreciation rate, L = quantity of labour (equal to
population), 𝑔𝐿 = growth rate of labor, K is capital, Y is GDP. The first term represents the immediate (short-
run) climate effect on the level of productivity, via Θ(𝑇) which captures immediate productivity damages
produced by T diverging from pre-industrial levels. The middle term lists elements related to capital
accumulation and population growth and can be interpreted as transitory effect on the growth rate converging
to zero; the third represents permanent productivity changes driven by T and hence can be interpreted as the
the long-term balanced growth path effect.

Θ′ (𝑇) 𝑌
𝑔𝑦 = 𝑇̇ + Φ (𝑠 − 𝛿 − 𝑔𝐿 ) + (1 − Φ)(𝑔𝐴 (𝑇)) =
Θ(𝑇) 𝐾

Θ′ (𝑇) 𝑌
𝑇̇ + Φ (𝑠 − 𝛿 − 𝑔𝐿 − 𝑔𝐴 (𝑇)) ⏟
+𝑔𝐴 (𝑇) =
Θ(𝑇)
⏟ ⏟ 𝐾 BGP effect
𝐼𝑚𝑚𝑒𝑑𝑖𝑎𝑡𝑒 𝑐𝑙𝑖𝑚𝑎𝑡𝑒 𝑒𝑓𝑓𝑒𝑐𝑡 Ψ(T) ; (ii) transitory effect

𝜕𝐹 𝐾
where Φ = .
𝜕𝐾 𝐹

Based on this framework, Kalkuhl & Wenz use an annual panel approach and specify a regression model linking
temperature change and per capita output growth rate as

2
𝑔𝑖,𝑡 = 𝛼(𝑇𝑖,𝑡 − 𝑇𝑖,𝑡−1 ) + 𝛽𝑇𝑖,𝑡 (𝑇𝑖,𝑡 − 𝑇𝑖,𝑡−1 ) + 𝛾1 𝑇𝑖,𝑡 + 𝛾2 𝑇𝑖,𝑡 + 𝑝𝑖 (𝑡) + 𝛿𝑖 + 𝜇𝑡 + 𝜀𝑖,𝑡

with 𝑝𝑖 (𝑡) controls for slow-moving regional changes affecting growth (like technological or institutional
change), 𝛿𝑖 and 𝜇𝑡 are country- and year-fixed effects. The regression is done on subnational level
(administrative regions), using data from 1900-2014. It should be noted that only 𝛼, 𝛽 𝑎𝑛𝑑 𝛿 capture the impact
on growth related to temperature changes. The coefficients 𝛼 and 𝛽 capture immediate effects of weather
shocks on country level (where Ti is based on temperature downscaling as discussed in the next section), while
𝛾1 and 𝛾2 capture transitory and long-run growth effects, in line with the different terms in the conceptual
model. Note that the approach used in the study by Burke et al. (2015) only captures the latter part, i.e.,
transitory and long-term effects. The empirical analysis finds strong evidence for immediate productivity
effects (𝛼 𝑎𝑛𝑑 𝛽), but not significant evidence for permanent long-run growth reductions (𝛾′𝑠 ). The preferred
model based on various experiments with lag structures, which we use for the calculation of future changes in
the per capita growth rate based on alternative temperature paths, is the one focused on immediate effects:

𝑔̂𝑖,𝑡 = 𝛼1 Δ𝑇𝑖,𝑡 + 𝛼2 Δ𝑇𝑖,𝑡−1 + 𝛽1 𝑇𝑖,𝑡−1 Δ𝑇𝑖,𝑡 + 𝛽2 𝑇𝑖,𝑡−1 Δ𝑇𝑖,𝑡−1

To reflect the uncertainty in these estimates, we also perform calculations at the 95 th confidence interval of the
estimates (reflected in Figure 1 of Kalkuhl & Wenz 2020). We calculate the standard error for 𝑔𝑖,𝑡 , based on

126
variance and co-variance parameters of the coefficients obtained from the authors, and provide as “high
ℎ𝑖𝑔ℎ
damage” estimates based on 𝑔̂𝑖,𝑡 = 𝑔̂𝑖,𝑡 − 1.96𝑆𝐸𝑖,𝑡 93

Parameters are listed in

Table 15 below.

Table 15. Parameter values for damage function from Kalkuhl & Wenz (2020). Values correspond to their specification 5 (Table
4 in the paper, column 5), which reports the results for all parameters under various specifications. The coefficients for delta T
and lag delta T not interacted with T were kept because removing them would risk a biased estimation. Running a regression
on an interaction term between dT and T, the control for T and dT should be kept to avoid an omitted variable bias and to
correctly calculate marginal effects.

𝜶𝟏 𝜶𝟐 𝜷𝟏 𝜷𝟐

Value 0.006410 0.00345 −0.00109 −0.000718

Variance
38.11 · 10-6 26.16 · 10-6 0.288 · 10-6 0.1797 · 10-6
(Var)

Note that these effects capture productivity impacts (e.g., labour and land productivity, capital depreciation)
related to changes in annual temperature. Therefore, non-market effects as well as effects from extreme
events, sea-level rise or indirectly related societal dynamics like migration or conflicts are not included in those
estimates.

Damages are calculated in post-processing using the probabilistic global mean temperature change data. NGFS
scenarios combine the transition pathways of IAMs, the MAGICC climate module 94 and the damage function
by Kalkuhl & Wenz (2020) to provide estimates of chronic damages. The way these pieces fit is shown in Figure
69. It starts with the output of the IAMs for a given scenario. In particular, the projection of emissions is fed into
the MAGICC climate model, which translates this emissions pathway into a global (or regional after
downscaling, see below) temperature path with 90% confidence intervals, representing the first source of
uncertainty. Then, these temperature series are used as input into the damage function, that calculates the
impact on GDP. To reflect the uncertainty related with the methodology, two estimates are presented: one
with the median estimates of the damage function parameters and another with the 95 th percentile. The latter
can serve as a way of gauging the damages that are not captured in the methodology used, such as sea-level
rise and climate-change induced conflicts.

2 2 2 2 2 2 2
93 With 𝑆𝐸𝑖,𝑡 = 𝑉𝑎𝑟(𝛿𝑖,𝑡 ) = Δ𝑇𝑖,𝑡 𝑉𝑎𝑟(𝛼1 ) + Δ𝑇𝑖,𝑡−1 𝑉𝑎𝑟(𝛼2 ) + Δ𝑇𝑖,𝑡 𝑇𝑖,𝑡−1 𝑉𝑎𝑟(𝛽1 ) + Δ𝑇𝑖,𝑡−1 𝑇𝑖,𝑡−1 𝑉𝑎𝑟(𝛽2 ) +
2
2{Δ𝑇𝑖,𝑡 Δ𝑇𝑖,𝑡−1 𝐶𝑜𝑣(𝛼1 𝛼2 ) + Δ𝑇𝑖,𝑡 𝑇𝑖,𝑡−1 𝐶𝑜𝑣(𝛼1 𝛽1 ) + Δ𝑇𝑖,𝑡 Δ𝑇𝑖,𝑡−1 𝑇𝑖,𝑡−1 𝐶𝑜𝑣(𝛼1 𝛽2 ) + Δ𝑇𝑖,𝑡 Δ𝑇𝑖,𝑡−1 𝑇𝑖,𝑡−1 𝐶𝑜𝑣(𝛼2 𝛽1 ) +
2 2
Δ𝑇𝑖,𝑡−1 𝑇𝑖,𝑡−1 𝐶𝑜𝑣(𝛼2 𝛽2 ) + Δ𝑇𝑖,𝑡 Δ𝑇𝑖,𝑡−1 𝑇𝑖,𝑡−1 𝐶𝑜𝑣(𝛽1 𝛽2 )}

94 MAGICC climate module is described in Box: MAGICC: A reduced complexity Earth system model

127
The change in per capita growth rate given by the previous equation is taken into account calculating a
projection of country-level per capita output under climate change following

𝑐𝑙𝑖𝑚 𝑐𝑙𝑖𝑚
𝑦𝑐,𝑡 = 𝑦𝑐,𝑡−1 (1 + 𝑔𝑐,𝑡 + 𝑔̂𝑐,𝑡 )

Where 𝑔𝑐,𝑡 is the unperturbed growth rate in a given country obtained from the downscaled IAM GDP
projections and 𝛿𝑐,𝑡 is the perturbation calculated with the previous equation, depending on country-level
temperature changes. Note that this approach calculates damages compared to present-day conditions, i.e., it
starts with present day GDP (2020), assuming that this already incorporates the effects of past temperature
increases. As the damages are cumulative, this underestimates the overall losses. Furthermore, losses are
underestimated due to the lack of dynamic effects that GDP changes would have, for instance, through the
savings rate or capital accumulation. Results are provided as annual, country-level output change in %, with
losses reported as negative values (e.g. Diagnostics|high/median GDP change|KW panel population-
weighted|GMT AR6 climate diagnostics|Surface Temperature (GSAT)|MAGICCv7.5.3|*.*th Percentile), as well
as net GDP values (e.g. net net GDP|PPP|median/high damage|KW panel population-weighted|GMT AR6
climate diagnostics|Surface Temperature (GSAT)|MAGICCv7.5.3|*.*th Percentile), where median/high
indicates whether the median or the 95th percentile of the damage function estimates are used and |*.*th
Percentile refers to the percentile of the temperature pathway used as input.

3. Chronic damages in post-processing and sources of uncertainty

Figure 69. How post-processing chronic damages are calculated and sources of uncertainty.

4. Temperature downscaling

The global mean temperature pathways provided by the MAGICC postprocessing have to be downscaled to
country-level for the calculation of country-level macroeconomic damages as described in the previous section.
For this we use a statistical downscaling approach based on the multi-model climate data set from Phase 5 of

128
the Coupled Model Intercomparison Project of global climate models (CMIP5, https://esgf-
node.llnl.gov/search/cmip5/). This is aligned with the physical risk data from ISIMIP2b, which are also based on
CMIP5 climate projections.

The country-level mean temperature (in absolute terms) is calculated as

𝑇𝑐,𝑡 = 𝑇̃𝑐,2005 + 𝜅𝑐,𝑡 (𝑇𝑡 − 𝑇2005 )

𝑇̅𝑐,𝑡 −𝑇̅𝑐,2005
with the scaling factor 𝜅𝑐,𝑡 = .
𝑇̅𝑡 −𝑇̅2005

Here, 𝑇𝑡 is the global mean temperature change from the transition scenario as calculated with MAGICC, 𝑇̃𝑐,2005
is the observed 2005 mean temperature of a country calculated from the University of Delaware Air
Temperature and Precipitation v4.01 data set 95The scaling factor 𝜅𝑐,𝑡 is calculated based on gridded mean
temperature anomaly data from CMIP5 (where 𝑇̅𝑐,𝑡 is for a given region and 𝑇̅𝑡 is the global value). Gridded data
are aggregated to the country level using population weights based on SSP2 population data.

5. Scenarios with integrated transition and physical risks

The methodology described in Physical risk modelling approach computes chronic damages outside the IAMs.
This serves as a first approximation of these costs, as the agents in the models optimise policy and allocations
in each scenario without internalising the costs of higher temperatures. Ideally, transition and physical risks
should be modelled together in an integrated framework to capture feedback effects properly. Following the
methodology by Schultes et al. (2021), we provide an additional set of such integrated scenarios for the NGFS
framework. In a nutshell, this approach integrates chronic damages based on the empirical specification by
Kalkuhl & Wenz (2020) into the REMIND-MAgPIE model.

Figure 70. Conceptual framework of scenarios with integrated damages, and (b) comparison of GDP output (dark green) for
non-integrated and integrated runs of REMIND

95 Data can be found here.

129
Overview of the integration
The approach is shown schematically in Figure 70 (left panel). A module that calculates chronic damages based
on Kalkuhl & Wenz (2020) is coupled with the REMIND-MAgPIE core and MAGICC. This module is used to
calculate an additional component in the carbon tax that captures the Social Cost of Carbon (SCC). The
remaining component of the carbon tax is a guardrail tax, which adjusts to the carbon budget of each scenario.
This way, the assumptions of the scenario are satisfied and the economic decisions of agents (savings,
investments, etc.) do respond to physical damages, which are now included in the budget constraint and in turn
on the temperature paths. In other words, with this approach, regions internalize the impact of own emissions
on global temperature.

Iterative approach
The solution is obtained through an iterative approach. The level of the guardrail tax96 is adjusted until the
emissions budget of the scenario is reached and the emissions calculated in the REMIND model are passed to
MAGICC for calculation of global mean temperature change, which is then downscaled to regional
temperature.

Using the temperature pathways, the damage module calculates regional damages based on the approach by
Kalkuhl & Wenz (2020) and computes the associated SCC tax, by solving a global planner problem with Negishi
weights97 that internalize the physical damages. The solution of this problem gives the socially optimal SCC tax
given the output of REMIND. The resulting SSC tax is globally uniform.

This social cost of carbon is internalized in the next iteration of the REMIND model as a component of the
carbon tax, leading to additional mitigation. Damages reduce regional GDP which in turn affects emissions,
capital accumulation and savings dynamics. This iteration continues until a fixed point is reached. 98

Schultes et al. (2021) show that this iterative approach leads to results very close to the solution of the model
with fully endogenous optimization of the SCC tax, which would be computationally very demanding.

Output
The Current Policy scenario with integrated physical risks captures the GDP effect of damages but does not
internalize them for a policy response. The other scenarios combine social costs of carbon and guardrail taxes
as outlined above, on the level of large world regions.

For the rest of scenarios, it is important to notice that the damages are not directly comparable to the ones
reported in Post-Processing of non-integrated runs of REMIND, since the integration leads to additional
dynamic responses. Therefore, the difference in final output between the integrated and the non-integrated
policy runs can be separated into two components, the direct damages, comparable to the post-processed
damages, and the integration costs, which include savings effects and changes in the mitigation strategy in

96 Which follows a a Hotelling form, i.e., rises exponentially with the interest rate. This is a common result for the carbon
tax under a carbon budget.

97 That is, welfare weights that equalize the marginal utility of consumption across regions. This is a common choice in the
literature.

98 See supplementary material of Schultes et al. (2021) for details of the iterative approach.

130
response to the damages. Right panel in Figure 70 provides a guide99 of how to compare GDP counterfactuals
across non-integrated and integrated runs. Notice that only GDP including both direct damages and integration
costs is provided in the integrated run100.

To capture the effect of climate uncertainty in the damage estimate, we select MAGICC6101 configurations at
the median and 95th percentile of the temperature distribution in 2100 from a probabilistic run with 500
outcomes for an RCP2.6 emissions scenario. This climate uncertainty is combined with the damage function
estimates uncertainty to produce two set of output series. “REMIND-MAgPIE 3.2-
4.6IntegratedPhysicalDamages (median)” uses the median estimates of the damage function and the median
of temperature distribution. “REMIND-MAgPIE 3.2-4.6 IntegratedPhysicalDamages (95th-high)” uses the 95th
percentile of both the damage function and the temperature distribution. Table 16 show the output variables
provided by integrated REMIND runs.

Table 16. Output variables for integrated REMIND runs.

Output Variable Description

GDP|MER|Counterfactual without damage GDP net of mitigation costs

GDP|PPP|Counterfactual without damage GDP net of mitigation costs

Macro-Economic Climate Damage|GDP Change Direct and indirect chronic damages (MER)

Policy Cost and Macro-Economic Climate Mitigation costs plus direct and indirect
Damage|GDP Change chronic damages

Policy Cost|GDP Loss Mitigation costs (MER)

GDP|PPP|including chronic physical risk damage GDP net of mitigation costs and indirect
estimate chronic damages (MER, only downscaling)

99 Yellow boxes indicate the differences between GDP series from different runs, represented by dark green boxes.

100 Subtracting Post-processed damages from integrated output and comparing with non-integrated policy run without
damages could serve as an approximation to estimate integration costs.

101 Note that the MAGICC6 version used in the REMIND-MAgPIE framework is different from the version 7.5.3 used to
post-process IAM results, however this just affects the internal damage calculation.

131
Downscaling
To obtain country-level damages for integrated runs we use a pattern-scaling approach, distributing the
regional direct GDP losses and the integration costs to countries using country damages from post-processed
runs as weights. The GDP net of integrated policy costs and chronic physical risk damages is provided on
country level. This is used for the downscaling of further variables (i.e., bringing variables from regional to
country level) of the integrated damage runs. Figure 71 summarizes the differences between non-integrated
and integrated runs in the variables they produced and the downscaling approach.

Figure 71. Non-Integrated REMIND runs vs Integrated REMIND runs

132
Module 6: Acute physical risks

1. Non-technical summary

Acute physical climate risk assessments estimate risk from extreme weather events like floods, heatwaves,
tropical cyclones, and droughts.

Natural catastrophe models for acute risk are based on three main components: hazards or perils, i.e., the
extreme events or physical variable causing the damage; exposure, i.e., a spatial map of the objects exposed to
damage (e.g., assets, infrastructures etc.); and vulnerability, i.e., a function that allows assessing the degree of
damage of the exposed objects. In the NGFS Phase III, acute physical risk was calculated for only two hazards
and at aggregate (world) level only.

In Phase IV, acute physical risk modelling substantially advanced. Firstly, it covers four perils: heatwaves,
tropical cyclones, floods, and droughts. Additionally, estimates are provided at country level, while for each
peril the most relevant transmission channel is used.

The risk projection process generally follows three main steps: (1) estimation of distributions of country-level
impacts, with impacts being in capital stock damages for floods and tropical cyclones, crop yield losses for
droughts, and population impacted for heatwaves, using catastrophe modelling principles and grid level data
across a range of projected temperatures values; (2) projection of these distributions in the future along
temperature paths (expressed in Global Mean Temperature, or GMT in short); and (3) translation of these
shocks into macroeconomic dynamics at country levels, by implementing them as supply and demand shocks
in NiGEM.

2. Differences relative to Phase III

The approach used for acute physical risk characterization in NGFS Phase IV differs from that used for NGFS
Phase III:

Perils covered: The Phase III estimation methodology focuses primarily on floods and tropical cyclones,
whereas the range of perils modelled in Phase IV has been expanded to include droughts and heatwaves.

Method for estimating acute shocks: In Phase III, the representation of acute physical risks was mostly based
on historical shocks (based on EM-DAT) with future multipliers based on productions from the Climate Impact
Explorer. In Phase IV the range of indicators used was extended to cover additional perils and countries to use
an approach based on catastrophe modelling principles.

Macroeconomic modelling: The method for modelling of these shocks in NiGEM has also been updated to
reflect the lack of direct estimates between country GDPs or assets and events like droughts or heatwaves. To
this end, additional channels of transmissions have been used, like yields loss for droughts and labour
productivity for heatwaves. The results using these additional channels also show how much of the physical risk
impacts, which go along various and sometimes very indirect transmission channels (one example is supply
chain disruption, very difficult to model and yet affected by physical risk), is still to be uncovered.

133
Granularity: Whereas world aggregate impacts were reported in Phase III, in Phase IV country-level results are
also provided. The country level results are a great advancement and a significant step toward making these
models available for exercises like stress tests or climate risk analyses, especially if further enriched with further
breakdown like the sectoral one. However, the approach is novel and had to overcome significant data and
model gaps and will benefit from extensive data validation and improvements based on the use.

Hazard
projections Macroeconomic
Hazard indicators Vulnerability
(for range of impact in NiGEM
GMT levels)

Drought Crops: harvested agricultural

SPEI
areas supply

Exposed Labour
Heatwave Wet-bulb CMIP population productivity and
temperature
(ISISMIP) demand

global FLOPROS,
Floods hydrological damage
Capital damage
models (GHM) + function,
ISIMIP ISIMIP2

IBTrACS and Damage

Tropical probabilistic CLIMADA Capital damage
function (EM-
cyclones distirbutions DAT)

The structure of this chapter

The rest of the chapter is structured as follows: the first part covers hazard-modelling, with one section per
hazard. These sections focus on explaining how acute risks can be measured, the exposures and vulnerabilities
quantified, and predictions made on the basis of Global Mean Temperature paths. The last section explains the
implementation in NiGEM, which is also split per hazard.

134
3. Modelling of acute physical risk hazards

3.1 Drought: Yield Exposure to Severe Drought Conditions

Drought conditions, often defined as a long-term lack of precipitation and dry soils, are detrimental to
ecosystems and societally relevant sectors in a variety of ways. For example, it can affect the energy sector
through lack of cooling water and hydropower production, or impact crop yields. The aim of the provided metric
is to estimate crop-land exposure to severe drought conditions under different levels of global warming and a
first order assumption on how those might affect national yields. Several different drought definitions have
been put forward, often related to their temporal evolution (flash drought, mega drought), the underlying
physical or societal causes (meteorological drought, hydrological drought, agricultural drought, socioeconomic
drought102). Out of the several possible indicators for detecting drought conditions, we selected the most
suiting one.

We detect severe drought conditions using the standardized precipitation evaporation index (SPEI) over 12
months, which is a drought indicator based on relevant variables such as precipitation and evaporation. As
evaporation is considered, this indicator is more sensitive to climatic changes compared to, e.g., the
Standardized Precipitation Index (SPI) (Touma et al. 2015). For a detailed description on how the SPEI is
determined we refer to NCAR Climate guide 103. World Bank’s glossary indicates that when SPEI is calculated
for a 12 month period, it may measure the potential impact of drought on ecosystems, crops, and water
resources (like a precipitation deficit or a low remaining soil moisture 104). The duration of a drought event is an
important factor for determining its impact. In addition, the SPEI can be determined on different timescales.
Hence, we use the SPEI-12, which in addition to the current month considers the conditions of eleven preceding
months. Thus, this index considers long-lasting drought conditions on annual scales, which are particularly
detrimental for local food security. Lastly, the SPEI calculation is based on Potential Evapotranspiration (PET)
following the Thornthwaite method, which takes monthly mean temperature as input.

Detecting drought conditions

Drought conditions are determined on a monthly basis. A threshold of 𝑆𝑃𝐸𝐼12 = −3 is used to define drought
conditions for each grid point in a specific month 105. A value of -3 indicates extreme precipitation and
evapotranspiration conditions over an extended time, indicating an exceptionally severe drought, while a value
of zero indicates no drought risk. The resolution is of 0.5 deg. x 0.5 deg. globally which refers to an area of scale
of about 50km x 50km.

102 https://www.ncei.noaa.gov/access/monitoring/dyk/drought-definition

103 https://climatedataguide.ucar.edu/climate-data/standardized-precipitation-evapotranspiration-index-spei

104 https://databank.worldbank.org/metadataglossary/environment-social-and-governance-(esg)-
data/series/EN.CLC.SPEI.XD#:~:text=SPEI%20is%20used%20as%20a,negative%20values%20indicate%20dry%20co
nditions.

105 The SPEI values range from -5 to 5. Smaller values indicate stronger degrees of drought, while the positive values
indicate degree of moisture.

135
Within a year the severity of the ruling drought conditions is determined by counting the total drought months
detected within a year and dividing them by twelve. Thus, a grid-point is considered affected entirely if a
drought is detected throughout all twelve months. Following this approach, if drought conditions of SPEI < -3
are found for one month only, the yields within a grid-point is considered to be affected by 1/12th.

Quantifying effects of drought conditions on national harvests.

Naturally, the impact of a drought on crop yields in a certain region scale with the intensity to which that
particular region is used for agriculture. Therefore, we overlay the global output from the climate models with
a global map that provides the percentage of harvested area per grid point at 0.5° resolution (Figure 72). The
data is based on 2005 estimates from Ray et al. (2015). To determine the annual exposure of harvested area to
drought conditions on a national level we multiply the drought severity determined by the number of affected
months per year with the harvested area for each grid-point within national borders and aggregate all values
for a year. Values are then normalised for each country, thus, yielding a value between zero or one for each
year, where zero means no effect and one refers to a total exposure of drought conditions of the harvested area
of one gridpoint:

(#of months SPEI > −3)𝑖

National annual yield in % (SPEI) = ∑𝑛𝑖 ∙ harvested area share𝑖 ,
12

where 𝑛 is the total number of gridpoints within each country and 𝑖 ∈ {1,2,3, . . . , n} the respective grid-point.

136
 Figure 72. Intensity of agricultural activity as indicated by harvested area based on data provided by Ray et al 2015. Data is
based on 2005 estimates.

Determining future drought conditions and projecting at different levels warming levels
We use the bias-adjusted and down-scaled output from four climate models of the fifth phase of the Coupled
Model Intercomparison Project (CMIP5106): IPSL-CM5A-LR, GFDL-ESM2M, MIROC5, HadGEM2-ES. Following
the approach used in other climate impact studies, four different models are employed to improve the sample
size and provide a better range of possible climate futures.

 The Coupled Model Intercomparison Project Phase 5 (CMIP5),

managed by the World Climate Research Programme, relies on
Explainer box 6 climate models to analyse Earth’s climate dynamics under
What is CMIp 5 ? different scenarios and current climate conditions. These models
integrate atmospheric, oceanic, terrestrial, and ice processes, giving
insights into climate system responses to external factors like
greenhouse gas emissions. They were a critical component of the 5th
assessment report of the Intergovernmental Panel on Climate
Change.

Annual drought conditions are determined in four general circulation models in historic experiments and under
a high emission scenario (RCP 8.5) to determine samples of exposed crop lands for global mean temperature
(GMT) values between 1 degree to 3.6 degrees above preindustrial levels in 0.1-degree increments. First, we
determine those years at which the 21-year running average GMT reaches a predefined value. For the
respective year and the 20 years surrounding that particular year, drought intensity levels are quantified for
each of the four climate models. Given that we are provided with 21 years per model, and we employ four
different models, we receive a sample size of 84 years per GMT value respectively (Figure 74). The identified
relationships between GMT and impacts allow for a mapping on to the temperature profiles of each NGFS
scenario when running the micromodel (NiGEM). The process of collecting 84 observations around each GMT
step allows us to generate distributions of drought risk per temperature level, rather than point estimates, to
be used in the statistical trials of the macroeconomic model (NiGEM).

106 https://wcrp-cmip.org/cmip-phase-5-cmip5/

137
 Figure 73. Illustration of how global mean Temperature levels are mapped to specific years, which are then used to analyse
drought statistics.

Orange and blue line graphs refer to global mean temperature curves (y-axis) under two distinct scenarios from
2000 until 2100 (x-axis). The grey horizontal lines indicate specific GMT values (1.5 and 2 degrees). The crossing
point of the horizontal lines and the temperature curves mark the year around which the 21 year period (blue
and orange shaded area) are centred and are indicated by vertical arrows.

Figure 74. Increased risk of yield losses from enhanced exposure to drought conditions under future warming levels.

Histograms of percentage of yield affected (x-axis) according to the equation “National annual yield in % SPEI”
based on a sample size of 84 years per GMT value ranging from 1 to 3.6, here shown for Argentina. Colours of
the bar plots refer to different levels of global warming, ranging from green (cool) to brown (warm), hence
resulting in 84 observations (bars) per each colour. A level of one relates to a complete or 100% exposure to
droughts according to the indicator of equation “National annual yield in % SPEI”. Heatwaves are currently
incorporated through population exposure to dangerous levels of humid heat which affect the economy
through labour productivity and consumption. The exposure is measured at a grid-point level at which
population and heatwave changes are quantified. For this reason, regions in which humid heat is projected to
dramatically increase but are not as densely populated will be less impacted. This can be seen, for example,
when comparing North America economies to Mediterranean or Southeast Asia (other direct or secondary
Heatwave impacts e.g., on the energy sector, wildfire probability or on supply chains are not incorporated here).

Suggestions for future improvements

 Note that this indicator still requires further validation and approval from the scientific community
through undergoing peer review processes.
 Drought conditions affect different crop types in different ways. A more accurate estimate of future
crop losses to different warming levels would consider this by taking into account the regional mix of
crop types planted.

138
  A 100% exposure to drought indicator is considered to lead to a 100% yield loss. However, this link
would need to take in account additional factors of protection and adaptation. Other large and long-
term effects, as for example on migration pressures, are not accounted for neither here, nor in the
macroeconomic model (NiGEM).
 The chosen indicator affects non-irrigated crop areas most. A refined indicator would take into account
differences and changes in local water management and irrigation.
 The impacts of drought conditions on yields have a seasonal dependency. The occurrence of a drought
affects yields in a different way if it occurs, for instance, in the sowing season or in the harvesting
season. A more accurate estimate of drought conditions would take into account the exact month in
which it occurs.
 Here, we assume the impact of a drought on yields to scale linearly with the number of months within
a year, however the damage function has likely a more complex character with a more nuanced
dependency based on drought intensity. Future estimates could use a linear dependency on drought
intensity instead of using a threshold-based metric.
 Next to statistical relationships that take into account a single impact driver alone, crop modelling
efforts as, for instance, done within the GGCMI initiative (Jägermeyr et al. 2021), could be used as a
comprehensive source of estimates of crop yields under different emission scenarios.

3.2 Heatwaves: Calculating the nationally based exposure to Heat Stress on different
warming levels

Heatwaves can affect the economic activity in various ways, e.g., they disrupt supply chains by damaging
railways and roads, induce water scarcity and affect labour productivity. For national estimates of exposed
population to dangerous levels of heat stress at different global warming levels we analyse wet bulb
temperature – a measure of humid heat - which is particularly harmful for human health (Hall 2022). Humid
heat affects the body’s cooling capabilities which are based on evaporation of sweat. Once the surrounding air
is saturated with humidity this mechanism stops working. When exposed to such conditions severe health risks
could be the consequence, which can culminate in a total collapse from heatstroke. Wet bulb 107 temperatures
of 35 degrees have been estimated to be fatal, while a value of 32 has been put forward for being fatal when
doing physical labour (Veccellio et al 2020). Below these critical levels, the impacts of a humid heat can still be
very significant but vary with the degree to which the local population is adapted to certain levels of heat stress
characterising the local climatic conditions.

Detecting heat stress events.

Due to locally varying levels of heat stress adaptation, we apply a hybrid approach for detecting humid heat
events for each grid-point. To identify heat stress, a relative threshold is provided by the 84th percentile
calculated from annual maximum values for each grid-point. Assuming a gaussian distribution the 84th
percentile corresponds to one standard deviation which can be considered a good estimate for extreme
conditions. The climatology on which the percentiles are based corresponds to the distribution of humid heat
days for years 1981- 2005 and is determined for each model separately. To that relative threshold, we add an
absolute threshold value of 29.1 degrees, which is considered harmful for outdoor labour irrespective of the
region (Saeed et al. 2021b, Kang et al 2019), we thus apply an upper limit to the quantile-based value and

107 The wet-bulb temperature indicates the lowest level temperature that can be reached at a specific air (dry bulb)
temperature thanks to the effect of water evaporation (i.e. until humidity saturation)

139
consider 29.1 degrees108 as the local threshold wherever the local 84th percentile > 29.1 degrees. Threshold is
hence the min (local 84th percentile, 29.1).

Thus, an event is detected for a grid point if i) a wet-bulb temperature of above the local 84th percentile is
detected or ii) if a wet-bulb temperature is above 29.1 degrees.

Figure 75. Latitudinal dependence of wet-bulb temperatures ranging from North pole (left) to South pole (right) determined by
averaging the 84th percentile values over longitudes and years 1981 – 2019 including ocean and land areas. Highest wet bulb
temperatures are occurring in the tropics around the equator (30°N – 30°S).

Quantifying effects of heat stress on aggregated exposed population

For determining the annually exposed population we use grid-cell based population data from the ISISMIP
project109 for the year 2005 at a resolution of 0.5. x 0.5 degrees, matching the resolution of the climate datasets.
If for a given day the grid-point specific threshold is breached, the population within that specific grid-point is
considered affected. For each year the affected population is then aggregated within national borders. Note
that this approach considers that individuals can be affected multiple times per year as the population within a
grid point is considered for each day within a year separately.

108 This level is considered harmful (while 32 wet-bulb degrees is considered fatal for manual labour) and hence taken as
threshold.

109 https://www.isimip.org/gettingstarted/input-data-bias-adjustment/details/13/

140
Determining future Heat Stress conditions and projecting heatwave distributions under different
warming levels
We use the bias-adjusted and down-scaled output from four climate models of the fifth phase of the coupled
model intercomparison project (CMIP5): IPSL-CM5A-LR, GFDL-ESM2M, MIROC5, HadGEM2-ES. Following the
approach used in other climate impact studies four different models are employed to firstly improve the sample
size and provide a better range of possible climate futures. These models provide relative humidity and
temperature values on a 0.5x0.5-degree (approximately 50km x 50km) grid from which humid heat is calculated
following the approach outlined in (Saeed et al. 2021a).

We analyse data from four climate models at historic conditions (years 1981-2005) and under a high emission
scenario until the end of the century (RCP. 8.5). This pathway was chosen as it provides the largest range of
Global Mean Temperature (GMT) levels, which are then used to sample years with a time slicing approach.

Distribution of exposed population are determined for global mean temperature values between 1 degree to
3.6 degrees in 0.1 degree Celsius increments (compared to preindustrial temperatures). First, we determine
those years at which the 21-year running average global mean temperature reaches a specific GMT value. For
this year and the 20 years (21 in total) surrounding that particular year exposure levels are quantified for each
of the four climate models (84 model-years in total).

This approach provides us with national population exposures for 84 modelled years: for each of the
incremental GMT values we receive 21 years for each of the four models, which then are pooled into one
sample. For each country and each warming level, aggregated population, quantified at the grid-point level, is
fitted with a generalized extreme value distribution (Weibull), a distribution which is particularly well suited for
capturing tails, to provide a continuous distribution (Figure 73) from which events can then be sampled for
further analyses and used in the NiGEM macroeconomic model to estimate GDP losses. While this approach
delivers reasonable results for most countries, geographical location and country size can lead to distributions
with location factor close to zero (i.e., very small countries).

Figure 76. Increased Risk of heat stress exposure under higher global mean temperature levels.

141
 Change in exposed population to dangerous levels of humid heat for in Australia (x-axis) in a given year. Curves
show the Weibull distributions of exposed population to heat stress based on a sample size of 84 years sampled
around the year that exhibits the respective warming level in RCP-8.5.

Suggestions for future improvements

 Note that this indicator still requires validation and approval from the scientific community through
undergoing peer reviewed process.
 Impacts of extremes scale with their magnitude and duration. Here, the duration of a heat-stress event
is considered only implicitly. By assuming that the population is affected each day anew, long duration
events are increasing annual exposure levels each day they last. However, the chosen metric does not
discriminate between days that breach the heatwave thresholds consecutively and those that are
distributed independently within a year. Further, one could argue that a person can only be affected
once. Here, allowing for double counting was chosen to account for the lasting effect of an affected
individual to the labor market, which likely constitutes a simplification of real-world effects.
 Using globally gridded-population data, we do not account for population growth, changes in age
structure or the number of employed citizens or the sectors they are employed in. Outdoor labourers,
active in sectors such as construction or agricultural, are more exposed to the risks of high levels of
heat stress, while the availability of air conditioning and other adaptation strategies might provide
relief for heat stress locally. Differentiating the grid point specific labour structure and conditions
within a country would yield more accurate estimates.
 The approach used delivers reasonable results for most countries but implies that only few events were
captured when the country size is in the same scale as the resolution of the used climate models. In
particular, Island states (e.g., Faroe Islands) exhibit distributions of questionable validity as the climate
models likely don’t resolve the physical mechanisms related to land cover with full accuracy. This likely
leads to an underestimation of heat stress in these countries.
 The societal impacts of heat stress and heat extremes go beyond effects on human health and labor
productivity. Recent heatwaves have severely impacted economies in numerous ways e.g., by
favouring wild-fire conditions, by lowering gauge heights or rivers, thereby disrupting supply chains,
destroying infrastructure such as tarmac roads, and disrupting energy supply (as water temperatures
were too high as to be used for cooling some thermal power plants). Future initiatives within the NGFS
network need to account for these effects to provide a more realistic estimate of future impacts on
global warming on national economies.

3.3 Floods damage modelling

Data for flood detection and protections

Different modelling groups from around the world derived harmonized data on flood depth and flood fraction
at grid point level (150arcsec)110 by running global hydrological models (GHM) and submitted this data to the
ISIMIP project. These GHM’s were driven by four General Circulation Models (GCM) resulting in a number of
GHM/ GCM combination datasets111, that participated in the ISIMIP2 project for the scenarios RCP 2.5, RCP 6.0
and RCP 8.5. This data was used as the basis to calculate average annual economic losses from riverine flood
under different scenarios.

110 ISIMIP, Inter-Sectoral Impact Model Intercomparison Project; www.isimip.org; Database accessible under
https://files.isimip.org/cama-flood

111 GHM’s: GFDL-ESM2S, HADGEM-2-ES, IPSL-CM5A-LR, MIROC5 GCM’s: CLM45,CLM50,CWATM,DBH,H08,JULES-

W1,LPJML,MATSIRO,MPI-HM,PCR-GLOBWB,WATERGAP2

142
A correction for flood protection is taken into account (FLOPROS database, Scussolini et. al, 2016112)
representing the maximum return period (interpretable also as intensity of event) of flood that each country/
region can prevent. The global FLOPROS database consists of several layers: The consolidated or overall layer
is called “Merged layer” which is a combination of three other layers: the “Design layer” combines empirical
data about existing protection infrastructure; the “Policy layer” consists of data on protection standards and
requirements set by policy measures; the “Model layer” is a model output from an observed relationship
between gross domestic product per capita and flood protection. The design layer is considered the most
reliable as it reports information on actual protection standards, while the other two layers are proxies of actual
protection. This threshold procedure113 implies that when the protection level (intended as the maximum return
period) is exceeded, the flood occurs as if there was no initial protection, while below the threshold, no flooding
takes place. For the final assessment, the high-resolution flood depth data from 0.3’ to a 2.5’ resolution (~5 km
× 5 km) is re-aggregate by retaining the maximum flood depth as well as the flooded area fraction, defined as
the fraction of all underlying high-resolution grid cells where the flood depth was greater than zero.

Quantifying flood damages

The damage modelling part closely follows the methodology of Sauer et al. (2021). To derive a local damage
from the annual flood maps and exposure data. The continent-level residential flood depth-damage functions
developed by Huizinga et al. (2017) are applied.

As asset (exposure) layer, a historical gridded Gross Domestic Product dataset (ISIMIP2) is used (Murakami et
al., 2019) but with a fixed exposure set to 2005 and a conversion factor applied to transform the GDP to capital
stock. The exposed assets on the grid level (150 arcmin) based on the flooded fraction obtained from the river
flood model are determined. As a next step, the grid level damage is quantified by multiplying the exposed
assets by the flood fraction and the flood-depth damage function (Figure 72). Then the estimated damage on
the region/country level are calculated by aggregating over all grid cells within a respective region/ country.

Regional aggregation
In line with country level aggregation into regions used by the NIGEM model, the countries Australia, USA +
Canada, China, India, Japan, and Russia are provided separately while all other countries are provided as
regional aggregations according to Table 17.

Table 17. Regional aggregation of countries

Developing
Africa Europe Europe East Asia Latin America Middle East

Egypt Iceland Albania Hong Kong Argentina Afghanistan

112 While the FLOPROS protection layer data has global coverage, it has limitations relative to quality of coverage (in
particular EMDEs) and spatial resolution.

113 Flopros 2015, Scussolini, P., Aerts, J. C. J. H., Jongman, B., Bouwer, L. M., Winsemius, H. C., de Moel, H., and Ward, P.
J.: FLOPROS: an evolving global database of flood protection standards, Nat. Hazards Earth Syst. Sci., 16, 1049–1061,
https://doi.org/10.5194/nhess-16-1049-2016, 2016. Tresholds (return frequencies) are displayed in figure 3 of the
publication available under https://nhess.copernicus.org/articles/16/1049/2016/nhess-16-1049-2016.pdf

143
South Africa Denmark Belarus Taiwan Brazil Algeria

Bosnia and
Angola Norway Herzegovina Indonesia Chile Armenia

Benin Sweden Cyprus South Korea Mexico Azerbaijan

Botswana Finland Kosovo Malaysia Cuba Bahrain

Antigua and
Burkina Faso Switzerland Luxembourg New Zealand Barbuda Djibouti

United
Burundi Kingdom Malta Singapore Aruba Georgia

Cameroon Austria Montenegro Viet Nam Bahamas Iran

Cape Verde Belgium Serbia Bangladesh Barbados Iraq

Central African
Republic France Moldova Bhutan Belize Israel

North Brunei
Chad Germany Macedonia Darussalam Bolivia Jordan

Comoros Ireland Ukraine Cambodia Colombia Kazakhstan

Congo Netherlands Fiji Costa Rica Kuwait

Cote d’Ivoire Croatia Kiribati Dominica Kyrgyzstan

Democratic
Republic of the Lao People's Dominican
Congo Greece Dem. Rep. Republic Lebanon

Equatorial Guinea Italy Maldives Ecuador Libya

Eritrea Portugal Marshall Islands El Salvador Mauritania

Micronesia
(Federated States
Eswatini Spain of) Grenada Morocco

Ethiopia Bulgaria Mongolia Guatemala Oman

144
 Czech
Gabon Republic Myanmar Guyana Pakistan

Gambia Hungary Nauru Haiti Qatar

Ghana Poland Nepal Honduras Saudi Arabia

Guinea Romania Palau Jamaica Somalia

Papua New
Guinea-Bissau Turkey Guinea Nicaragua Sudan

Kenya Estonia Philippines Panama Syria

Lesotho Latvia Samoa Paraguay Tajikistan

Liberia Lithuania Solomon Islands Peru Tunisia

Saint Kitts and

Madagascar Slovakia Sri Lanka Nevis Turkmenistan

United Arab
Malawi Slovenia Thailand Saint Lucia Emirates

Saint Vincent
and the
Mali Timor-Leste Grenadines Uzbekistan

Mauritius Tonga Suriname Yemen

Trinidad and
Mozambique Tuvalu Tobago Palestine

Namibia Vanuatu Uruguay

Niger Venezuela

Nigeria

Rwanda

Sao Tome and

Principe

Senegal

145
 Seychelles

Sierra Leone

South Sudan

Togo

Uganda

United Republic of
Tanzania

Zambia

Zimbabwe

Determining future Floods conditions and projections at different warming levels

To attribute economic losses to warming levels we follow the approach by James et al. 2017 which suggests
that impact indicators can be seen as a function of the Global Mean Temperature (GMT) level. This leads to the
assumption that a given GMT level will, on average, lead to the same change in that indicator even if it is
reached at two different moments in time in two different emission scenarios. This assumption is generally well
justified, and differences are small compared to the widespread changes projected by different models (Herger,
Sanderson and Knutti, 2015).

In each GCM114 simulation corresponding to each RCP scenario, we identify the year for which a certain GMT
level (0.1°C incremented starting with 1°C) is reached. Having identified the year for which a specific GMT level
is reached in a scenario-GCM combination, we average the projected values over a 21-year period centred over
that year in the corresponding GCM (or IM scenario experiment). We then average over all available scenarios
for each GCM (or GCM-IM) combination, before pooling the estimates obtained from all GCMs (or GCM-IM)
combinations, from which we compute their median values for each 0.1°C GMT level increment.

114 Global Climate Models

146
 Figure 77. Modelling chain for flood per country/region losses based on ISIMIP2 data.

The data we are providing as 0.1°C step warming levels, can be used to analyse trends in some regions. While
we see a clear trend towards more flood damage due to higher warming levels in Africa, India and China, there
is an increase visible in Europa and the US + Canada but less significant. For Developing Europe (see Table 17),
the trend shows a decrease that is also reported in other publications. Please note that these results were
derived by aggregating a number of models with global coverage. For damage on a country level, local studies
need to be considered. While Developing Europe show a downwards trend, it is still possible that one or multiple
countries within this regional aggregation can have an upwards trend towards more flood loss with higher
warming levels.

147
 Figure 78. Flood loss (USD) for different warming levels for selected regions.

Suggestions for future improvements

 Provide flood damage estimates on a country level for a set of selected countries, where datasets can
be considered most reliable.
 Compare between different data sources and projections to identify regional trends.
 Conduct research on these models and select the most suitable ones for loss estimation in a certain
region, instead of using the full range of flood models provided within the ISIMIP project

148
  Separate into return frequencies (e.g., 5-year, 10-year, 100-year) for more detailed use by modelling
teams of national banks. The loss data provided currently comes as annual average, and according to
CIE user feedback, changing this would be very useful.
 Consider to also include coastal flooding, and potentially pluvial flooding. Currently only river floods
are considered. Coastal flooding from storm surges compounded with sea level rise should be taken
into account in the future as well (long-term strategy).
 Study the effect of the FLOPROS protection layer (with/without protection).

3.4 Tropical cyclones damage modelling 115

Observations and future conditions by warming levels

Among the costliest natural hazards are tropical cyclones (TCs). TCs, also known as typhoons or hurricanes, are
highly destructive weather phenomena that form over warm tropical oceans, typically between 5° and 30°
latitude North and South. TCs have a devastating impact on many coastal regions in the tropics and subtropics.
For instance, the 2017 TCs Harvey, Irma, and Maria resulted in over 260 billion USD of damages to the United
States (NOAA). Human-induced climate change might have diverse impacts on TCs, including heightened
rainfall and wind speed and an increase in the frequency of extremely intense storms (Knutson et al., 2020). We
estimate future TC risk globally under various Global Mean Temperature (GMT) increases using the CLIMADA
natural catastrophe modelling platform (Aznar-Siguan & Bresch, 2019; Bresch & Aznar-Siguan, 2021). In the
aftermath, we will describe in detail the chosen modelling set-up.

Hazard is given by a probabilistic set of future TC events. This is constructed in two steps. First, a probabilistic
set of historical TCs is built and then, frequencies and intensities of such historic track sets are rescaled
according to expected future changes. The historic track set is built starting from observed tracks given by the
IBTrACS dataset (Knapp et al., 2010). A random walk algorithm is then used to expand such historic set and
generate a larger number of events. This approach is designed to infer a probabilistic distribution of tracks from
a single track and, in so doing, generates a set of probabilistic tracks. The method is described in detail in the
supplementary material of Gettelman et al. (2018). For each of the generated tracks, a wind field map is
estimated by using the parametric wind model proposed in Holland (2008). Figure 79 shows an example of
hazard footprint for tropical cyclone Maria. It is important to highlight that the hazard constructed only
represents extreme winds. Storm surges and coastal flooding associated to TCs are not considered explicitly in
this analysis. Yet, as the storm surge wave is created by the strong winds, they are included implicitly.

149
 Figure 79 Track and wind field of tropical cyclone Maria (2017)

Finally, future probabilistic track sets are derived by rescaling the historic probabilistic wind fields with the
information provided in Knutson et al. (2015). Therein, percentage changes of TCs’ frequency and intensity until
the end of the century (i.e., 2100) and under RCP4.5 are provided. Using linear interpolation, these scaling
factors can be extrapolated for different RCPs, different years and, consequently, different Global Mean
Temperature Levels (GMT). The chosen RCP is 6.0 and tracks are derived every 5 years from 2020 to 2100. The
corresponding GMT levels are then derived by following Table 3 shown in the methodological section of the
previous implementation phase116.

Quantifying Tropical cyclones losses

Exposure is modelled globally at the country level via the LitPop approach proposed by Eberenz et al. (2020).
LitPop disaggregates macro-economic data (e.g., GDP) proportionally to a combination of nightlight intensity
and population data. Vulnerability is modelled using the damage functions provided by Eberenz et al. (2021).

116 https://climate-impact-explorer.climateanalytics.org/methodology/#one-methodology

150
These functions were calibrated using EM-DAT loss data117. They are defined globally and are divided into 9
distinct regions. Figure 80 shows the functions and regions.

Figure 80. The calibrated impact functions as defined by Eberenz et al. (2021)

The probabilistic natural catastrophe model so defined allows generating a large set of synthetic losses. Since
each of these losses have an associated occurrence frequency, output from this model defines the probability
distribution of TC losses. Therefore, by running the model by country and at various time steps under a specific
RCP, countries’ loss distributions at various GMT levels are derived.

Suggestions for future improvements and limitations

The analysis herein presented has several limitations. First, the analysis uses windspeed as a proxy for all losses
and does include storm surge losses only implicitly. Second, due to the global scope of the analysis and the
absence of global asset value datasets, exposure consists of proxy data and may therefore differ from actual
asset values. Also, due to the lack of historical damage data, the same vulnerability functions are used for large
– yet homogenous - geographical areas. While this provides a basic level of regional differentiations, it does not
resolve single country vulnerabilities. More importantly, neither the exposures nor the vulnerabilities are
evolving in time – as no reliable, spatially explicit projection of the sort exists to date at global scale – and thus
we can only report impacts of future climate cyclones on current assets. In addition, the adopted scaling
parameters to simulate TC changes in the future were defined on distinct historical datasets and only for the

117 https://www.emdat.be/

151
end of the century. Finally, in order to derive future tracks for increasing GMT a linear interpolation procedure
is employed, even though future changes in tropical cyclones may not be linear.

4. Modelling of acute risk macroeconomic effects (NiGEM)

Overview
It is common to undertake policy analyses using models of the economy. These usually involve applying a single
shock, such as an increase in energy prices, and evaluating its effects under different policy responses.
Stochastic simulation techniques extend this approach where a variety of shocks are taken at random from a
pre-determined distribution and are repeatedly applied to the model, producing a large variety of possible
outcomes. From this large number of potential outcomes, the moments of the solution of the endogenous
variables can be calculated and used to investigate the degree of uncertainty around projected data values
(forecast or simulation) deriving from the range and distribution of the shocks applied.

Shocks can be estimated in several ways but essentially there are two categories. The first, and most popular
in the past, is Monte Carlo based (MC), where the stochastic shocks are drawn from some assumed parametric
distribution of the errors, usually the multivariate normal (where i ~ (  , ) with mean vector  and variance
matrix ). The second method is the Residual Based (RB) approach, which consists of taking the actual sample
period residuals as the stochastic shocks. In NiGEM, the standard stochastics analysis 118 uses the RB approach
where the stochastic shocks are represented by historical NiGEM equation residuals, leading to a matrix of
shocks of M*T where M is the number of equations in the model and T is the historical data range being used
to run the model. This matrix of potential historical shocks is then drawn randomly from to provide future
shocks. Brown and Mariano (1984) introduced this RB approach using historical residuals.

When analysing projections of acute weather impacts and their likely effects on the economy, we use both MC
and a form of RB with historical equation residuals replaced with future climate projections from Climate
Analytics.

Trade and policy

One characteristic of NiGEM which needs to be addressed in the context of physical risk modelling is the inter-
connected nature of the international economy represented in NiGEM. Via global linkages in NiGEM, if a
country experiences an acute negative shock (of any type), other countries may experience an upswing in their
economy. Only if a global shock of equivalent magnitude is applied would all countries experience a negative
impact and even here, competitiveness, exchange rates etc. would mean the negative effects are not
apportioned equally.

118 Barrell, Ray, Karen Dury and Ian Hurst (2000). ‘International Monetary Policy Coordination: An Evaluation of
Cooperative Strategies using a large Econometric Model.’ National Institute of Economic and Social Research Discussion
Paper 160.

152
When considering future acute impacts, we are dealing with stochastic shocks of differing magnitudes and
frequencies. To facilitate the assessment of the impact of acute climate events on individual countries, we
isolated countries and only consider the impacts domestically which means that the stochastic trials 119 are
executed with trade exogenous (i.e., turned off). This is the same approach followed for chronic physical risk.
In addition, a neutral agent response (governments and central banks) is ensured by also exogenising monetary
and fiscal policy.

Standard economic channels available for shocks in NiGEM require the model to be running with endogenous
global links to correctly adjust to the economic impacts and find a stable solution. The loss of the trade and
asset channels restrict the channels available to shock to only direct supply and demand shocks. To capture the
equivalent impacts of the weather effects on GDP when using demand and supply shocks only, calibration
scenarios were run. These provide the link between the actual economic channel used to model the acute
weather impact (e.g., impact on crop yields) and their equivalent impact on demand and supply in the model,
which is then used in the stochastic trial when trade and monetary policy are discounted.

Example – a commodity price is increased by 1 %. This shock is fed into the standard model providing delta
impacts on both demand and supply. We use the ratio of 1/demand as the multiplier for the stochastic runs
when converting the randomly determined acute impact to an equivalent demand shock in the model. This
process is also used to determine the stochastic supply shock.

Figure 81.Example calibration shock process

Stochastic trial process

NiGEM is executed across all time-periods of the stochastic window (2023-2050) and the resulting output is
saved. A NiGEM trial will contain around 1000 executions to ensure sufficient values are created to remove any
bias in the random selection.

119 Astochastic trial means the final output of n NiGEM simulations run with stochastic impacts where n is normally set in
the region of 1000-5000. Each NiGEM simulation typically represents over 10 million calculations.

153
 Figure 82. NiGEM stochastic execution

The random draw of the shock combines the Global Mean Temperature (GMT) with the climate impact data
provided as input (see the second part of this chapter on hazard modelling) and uses a standard random number
generator to create the resultant shocks. The GMT used is the temperature profile from one of the three
scenarios below:
 Current policies (h_cpol)
 Delayed (d_delfrag)
 Net zero (o_1p5c)

Impact on country regions

NiGEM contains several regions (Africa, Asia, Developing Europe, Latin America & Middle East) that are made
up of an aggregate of constituent countries. The impact in each constituent country is calculated separately
(i.e., differing random shocks within the region) and the aggregate created by summing the constituents.

Model notes and possible refinements

Due to model constraints, the supply shock cannot exceed -70% because a reduction of the trend capacity by
greater than 70% causes instability within NiGEM. This limit is applied to all stochastic trials. Further
experimentation with individual countries could be undertaken to determine the country-specific limits for the
shock size. Alternatively, shocks greater than the maximum could be incorporated into future years to provide
a greater degree of shock persistence than currently considered. However, this approach would have to be
ratified with CA for each acute impact.

The stochastic trials are run using 1000 individual NiGEM runs (approximates to 4e10 model calculations).
Further investigation on the stability of the confidence bounds using increased trial sizes can be undertaken.

The data provided by CA for the stochastic trials is provided in 0.1 C intervals which in turn can produce step
changes in the shocks applied to the model. Investigating into whether the shocks can be applied in a smoother
fashion without affecting the academic integrity of the shocks themselves would remove these step changes.
One possible option is if a weighted average of shocks would be appropriate. For example, currently if the shock
is below GMT at a value of X, the shocks for GMT(X-0.1) are applied. If a reasonably linear progression can be
assumed between the shocks at GMT(X-0.1) and GMT(X), a weighted ratio of the two shocks may be applicable.

154
There are two potential issues with regards to the demand and supply shocks used in the stochastic trials. The
use of proxy demand and supply shocks in the stochastic trials, while providing accurate impacts for GDP, may
not be sufficient to provide the correct size and profile of impacts for other economic variables. Further
investigation on using the resultant GDP from the stochastic trials to create a calibrated shock in the standard
model, which maintains trade and applies the shock to the correct channel, would prove useful here.

Currently, the demand and supply shocks are created, by necessity, with global linkages “On” in the model. The
premise used is that as the shocks are applied equally and globally, the impact of global spillovers is minimised.
However, NiGEM does allow individual country/regional models to be run in isolation (in which there is an
exogenous ‘rest of the world’). Therefore, an investigation on how much trade impacts the size of the demand
and supply shocks, could be conducted for those shocks which can function correctly within this isolated
framework.

Acute weather data

The impact data for four acute weather effects, used as input in the NiGEM runs, are provided as described in
the second section of this chapter

Acute weather Hazard projection data

event

Heatwaves Country level parameters for Weibull distribution for all countries and a range
of GMT.

The Weibull distribution represents the number of people affected by the

heatwave

Cyclones Capital stock damage values in Mn $US, 2017 based on GMT for a range of output
samples along with the probability of that damage occurring in a country.

Floods Annual average capital stock damage values for varying values of GMT for each
country/region.

Drought Projections the loss (%) of agricultural yield due caused by drought for each country
across a range of GMT.

5. Stochastic Implementation

5.1 Cyclones

Cyclones’ impact on the economy is assessed via the channel of asset damages (i.e., capital shock), derived
from the disruptions caused by these type of events.

While capital stock is available for a sub-set of countries in NiGEM, it is primarily used for forecasting and direct
shocks to capital stock can have non-intuitive effects in the simulations. Instead, the more generic investment
premia variables (IPREM) are used by NiGEM to impact housing and business (as well as prices) and can be used

155
to introduce a capital shock. The shock to IPREM will directly relate % damage on assets (from random sample)
to an equivalent premia shock.

1. 1% investment premia (IPREM) shock applied in NiGEM 2023-2050

2. Resultant average productivity, investment (or domestic demand) and employment shock values used
to create equivalent capital stock multipliers.

Figure 83. IPREM shock channels

For the stochastic implementation, the capital stock damage in any timestep is randomly drawn from the range
of data provided for each country for the current scenario GMT. This is then used to create the relevant demand
and supply shocks.

5.2 Floods

Input data
Different from all other acute risks modelled in the NGFS scenarios, flood data are the projected average yearly
damages for all time-periods across a range of GMT. This means that a single shock is required to impact NiGEM
rather than a stochastic trial. Again, absolute damages are transformed into equivalent percentage damages.

Shock implementation
Both flood and cyclone data represent capital stock damages, allowing the use of the same economic channels
(IPREM) as used for Cyclones to implement flood damages. The capital stock percentage damage represents
the flood data provided with the modelling approach described in the previous sections.

5.3 Droughts

Droughts have several channels of impact:

1. Productivity: This is a direct shock to supply based on the % damage to agricultural production.
2. Exports: this links the fall in agricultural production to a country-level fall in total export volumes. The
share of agriculture in an economy is based on UN trade figures.
3. Prices: A fall in supply generates an increase in prices leading to a fall in demand.

156
Note:

 Domestic country level economic impact is the actual % drought shock scaled by UN data relating to
agricultural Gross Value Added (GVA) to total GVA for that country.
 World prices: to overcome the limitation derived from ending trade, country level share of world
agriculture, estimated as part of the physical risk modelling (see previous sections), is used to
determine price increases as a result of droughts. A one-to-one correlation is imposed (e.g., 10%
drought in Africa equals a 10% increase in prices scaled by Africa’s proportion of total agriculture)

Calibration simulations
All three shock channels are again linked to demand and supply shocks, while aggregate drought to
supply/demand shock multipliers are used in the stochastic trial.

1. Productivity: 1% shock to trend capacity (YCAP). Shock provides an equivalent demand multiplier.

Figure 84. Productivity -- demand channels

2. Exports: 1% shock to exports, supply and demand exogenised so all GDP effects come through net
trade impact. The GDP delta is then used as the exogenous impact on demand in the stochastic trial.

Figure 85. Imports adjust to exports and directly impact GDP

3. Prices: 1% shock to world prices of food, beverages, and agriculture to review resultant supply and
demand impacts.

157
 Figure 86. Prices transmission channel

To account for differing country-speed of adjustment, a three-year average is used to create the shock
multipliers for productivity and export scenarios. With prices, an average of all periods in the simulation is used
due to greater volatility (for example the positive impact on commodity exporters in the initial periods of the
scenario, followed by overshoot as trade effects dominate).

Stochastic implementation
The percentage agricultural damage for each timestep is randomly drawn from the range of data available for
the current GMT. This is then converted into equivalent price, demand, and supply shocks.

5.4 Heatwaves

The Weibull distribution explained in the dedicated section represents the size of population affected by a
heatwave. The calibration shock will need to link population changes directly to supply and demand shocks in
NiGEM.

1. 1% population shock applied in NiGEM for one year (2023)

2. Resultant average productivity and demand shock values reviewed120.
a. Supply impact uses the delta for trend capacity (YCAP).
b. Demand impact uses either consumption (C) or domestic demand (DD)
3. Calibration multipliers calculated for each country to provide the link for a 1% population shock
converted to equivalent YCAP & C/DD shocks

120 Applying directly a population shock would require endogenous supply and demand, which is not possible with Trade
set to OFF

158
 Figure 87. Population shock channels

Weibull parameters for each country are selected based on the GMT indicated by the scenario at each timestep.
The parameters are then used (inverse transform) to generate the Weibull distribution value.

1⁄
𝑎
𝑋 = 𝜆(−𝑙𝑛(𝑈))

Where X represents the absolute number of people affected by the heatwave. Scale (λ) and shape (a)
parameters are provided by the models described in the previous section and U is a random number (0<=U<=1)

The value of X is then compared against the average population 1985-2005 to provide the percentage
population shock to be used in NiGEM.

159
Module 7: Country-level downscaling

1. Non-technical summary

What is country-level downscaling?

Downscaling here refers to the process of converting the world-region-level outputs from global integrated
assessment models to the national level. Global IAMs provide projections at the level of world regions, which
may not capture the finer resolution required for regional or local analysis. For the NGFS climate scenarios,
downscaling involves refining the results of global climate models to provide more granular information about
how climate mitigation impacts specific regions, industries, or economic sectors.

Why do we need country-level downscaling?

The goal of the Paris Agreement is to limit long-term global temperature change to well-below 2°C and
pursuing efforts to limit it to 1.5°C. However, as energy and climate policies are not set at the global level, but
by individual countries, these countries have developed and submitted their own plans formulated in Nationally
Determined Contributions (NDCs) and mid-century net zero emissions strategies. Assessments of future
emissions and the effectiveness of climate policies are usually performed with Integrated Assessment Models
(IAMs) at the global and world-region level. However, bringing together insights from IAMs with information at
the country level has remained difficult, as global models usually provide results for a limited number of world
regions.

Several strategies have been developed to overcome these limitations. IAMs have increased regional resolution
and added individual countries as native regions to their models. However, this strategy remains difficult due
to the complexity of the IAMs, catering simultaneously for different modules including energy, economy, and
climate change. Modelling teams such as REMIND (Dietrich et al. , 2023) and MESSAGE (Huppmann et al. ,
2019)are tackling this issue by increasing the spatial heterogeneity. However, running these models for all
countries in the world is still problematic. To solve this issue, downscaling approaches can be used to provide
country-level results. One of the main advantages of applying downscaling techniques is that they do not
require extensive computational time, since they do not increase the spatial resolution of the IAMs themselves.

How does country-level downscaling help to produce NGFS scenarios?

The NGFS Consortium has developed a downscaling methodology that can be used to assess the potential
implications of the NGFS scenarios for 184 countries. To allow for country-level analysis, a subset of key energy-
system-related variables like emissions, primary energy and final energy have been downscaled to country-
level. However, there are also important limitations to consider:

160
  The country level results are derived from and primarily consistent with the regional IAM outputs. If
the IAM scenario does not represent the region well, this will translate into the country level outputs.
 As the country-level results are derived from a standardised methodology that is applied across all
countries, they currently do not incorporate specific policies on a country-by-country or sector-by-
sector basis.
 The downscaling algorithm does not consider technology capacity evolution or investments required
in the electricity grid infrastructure nor backup capacity. Users may need to cross-check these results
with other specific factors and data to ensure the pathway is representative.

2. Overview of method

According to the literature, downscaling approaches should provide results in line with local scale (historical
country-level) data and consistent with the original IAMs results (Van Vuuren et al. , 2010). Criteria used for the
downscaling should be scenario specific, and leading to plausible results, avoiding violation of physical
boundaries (Grubler et al. , 2007). As illustrated in Figure 88, the downscaling tool generates two pathways to
provide results that are both consistent with historical data and IAMs results:121

 Short-term projections that are consistent with both countries historical data and regional IAMs
results.
 Long-term projections that converge to regional IAMs results.

121 The downscaling methodology described in this section was developed by IIASA (International Institute for Applied
Systems Analysis).

161
 Figure 88. Conceptual framework for downscaling variables to the country level

Both pathways are harmonized so that the sum of country level results within a region coincides with the
regional IAM results, where large countries will undertake the biggest adjustments required to match the
regional data. Then a linear interpolation is created to converge from the “short-term” pathway to the “long-
term” pathway between the base year (e.g., 2010) and a future “time of convergence”. The base year is the year
after which model scenarios can start to diverge from historical data. However, historical data information can
be used until more recent available years (hence beyond the base year) as is done for estimating the final energy
demand. Different times of convergence between the short-term to long-term projections are assumed, based
on the type of NGFS policy scenario, to better reflect the underlying scenario storyline. For scenarios
compatible with 1.5°C, a faster convergence is assumed, while convergence is slower for scenarios in line with
current policies or NDCs. Depending on the assumptions on convergence, the downscaling algorithm will
provide a range of energy pathways at the country level.

Table 18. Timing of convergence and SSP storyline

NGFS Scenario Convergence SSP122

Delayed transition Slow SSP2

Fragmented World Slow SSP2

Low Energy Demand Fast SSP2

Net Zero 2050 Fast SSP2

Current policies Medium SSP2

122 Shared socioeconomic pathway.

162
 NGFS Scenario Convergence SSP122

NDCs Medium SSP2

Below 2°C Medium SSP2

The definition of slow, medium and fast convergence in terms of the “year of full convergence”, differs
depending on the type of variables, as summarized in Table 19 below. The choice of these years is derived from
experiments with the downscaling method and allows for strong influence of country-level characteristics in
the next few decades across all variations.

Table 19. Timing of convergence

Timing of convergence Final energy Primary and secondary energy

variables variables

Slow 2200 2300

Medium 2150 2250

Fast 2100 2200

Short- and long-term projections are then combined, considering convergence time. Pathways are provided at
the country level c (at time t) by using, e.g., for energy 𝐸𝑁𝑐,𝑡 , a weighted average of these projections (without
violating consistency with the IAMs results), where weights 𝜑 are linearly increasing for the long-term
projections.

(1)
𝐸𝑁𝑐,𝑡 = 𝜑𝑡 𝐸𝑁𝐿𝑜𝑛𝑔𝑐,𝑡 + (1 − 𝜑𝑡 ) 𝐸𝑁𝑆ℎ𝑜𝑟𝑡𝑐,𝑡

Weights will gradually change over time based on the assumption on the timing of convergence tc:

𝑡 − 𝑡𝑐 (2)
𝜑𝑡 =
2010 − 𝑡𝑐

The downscaling algorithm focuses on energy variables such as final energy, secondary energy and primary
energy, and derives the energy-related CO2 emissions from the downscaled energy system characteristics.

3. Key inputs

To downscale these variables, regional input data from IAMs are used. Here, the described downscaling
algorithm uses GDP and population data from baseline scenarios (absent of climate policies) as they are
available in the SSP online database.123 Besides, historical data are used to initialize the country-level variables
at the base year. The IEA Energy Balances 2022 provides energy-related historical data for 183 countries and
regional aggregates (IEA, 2022). In addition, for the electricity sector, power plants information (regarding

123 https://tntcat.iiasa.ac.at/SspDb/dsd?Action=htmlpage&page=welcome

163
remaining economic lifetime of certain power plants and planned capacity additions) around the world are
obtained from the PLATTS database (Platts, 2019). For emissions, PRIMAP is used as historical data source
(Gütschow et al. ,2021). Also, governance indicators at the country level as well as supply-cost curves based on
the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP) for the renewables energy potential
availability are included.

The downscaling tool provides country-level data for final, secondary, and primary energy variables as well as
energy-related CO2 emissions. Final energy variables include energy demand by energy carrier (i.e., electricity,
liquids, gases, solids, heat, hydrogen) and sectors (i.e., transportation, residential and commercial, and
industry). Secondary energy variables include information regarding the fuel mix (e.g., coal, natural gas, oil,
renewables etc.) associated to each energy carrier (e.g., liquids, solids, gases etc.). Primary energy variables
provide information regarding the overall energy mix (including energy transformation losses) by also
differentiating technologies both with and without Carbon Capture and Storage (CCS). For an overview of the
IAM downscaling output variables see Table 36.

Final energy demand can be decomposed into contributing factors by using a Kaya identity approach
(Nakicenovic and Swart, 2000). Final energy consumption (FEN) is decomposed into three contributing
elements: energy intensity (final energy consumption divided by GDP), GDP per capita, and population (POP)
as shown in equation (3).

𝐹𝐸𝑁𝑐,𝑡 𝐺𝐷𝑃𝑐,𝑡 (3)

𝐹𝐸𝑁𝑐,𝑡 = × × 𝑃𝑂𝑃𝑐,𝑡
𝐺𝐷𝑃𝑐,𝑡 𝑃𝑂𝑃𝑐,𝑡

GDP and population projections are used as input to the downscaling tool, as they have been already
downscaled at the country level as part of the SSP framework. As a result, to calculate total final energy
demand, some reasonable assumptions are made about the evolution of the energy intensity over time. Energy
intensity is a metric that allows for comparing how energy is used to produce services and final goods (hence,
GDP) across countries (GEA, 2012). Historical data show that the energy intensity tends to increase in the early
phases of industrialization as traditional (non-commercial) forms of energy are replaced by commercial (and
more efficient) energy. Then, the energy intensity starts to decline again as soon as this transition to
commercial energy is completed – a pattern known as the hill of energy intensity (Grubler et al. ,2012). Apart
from this “peak”, the historical energy intensity is dominated by a general downward trend associated to
increasing income per capita that strengthens improvements in energy efficiency. Although energy intensities
trajectories might differ across individual countries, historical data from 1972-2017 suggest an inverse
relationship between the level of the final energy intensity (defined as final energy consumption divided by the
GDP) and GDP per capita (Figure 89).

164
 Figure 89. Historical energy intensities over GDP per capita across countries, from 1972 to 2017

The literature suggests that energy intensity can still improve by a factor 10 or more in the very long term ‘Ayres
(1989), Gilli et al. (1990), Nakicenovic (1993, 1998), Wall (2006), GEA (2012)). Hence, it is assumed that this
relationship between energy intensity and income per capita will continue in future long-term scenarios, by
using a log-log model, in which the linear regression coefficients are derived after taking the logarithm of both
energy intensity (final energy per unit of GDP) and economic activity (GDP per capita):

𝐹𝐸𝑁𝑐,𝑡 𝐺𝐷𝑃𝑐,𝑡 (4)

log ( ) = 𝛽𝑐 log ( ) + 𝛼𝑐
𝐺𝐷𝑃𝑐,𝑡 𝑃𝑂𝑃𝑐,𝑡

In the above equation, the parameters of the functional form (α and β) are estimated based on:

 Historical data at the country level (historical trend extrapolations for each country).124
 Future regional energy intensity based on IAM results (in this latter case α and β would be the same for
all the countries c).

Long term IAMs projections are based on regionally aggregated IAM results, whereas the short-term
projections 𝐸𝑁𝑆ℎ𝑜𝑟𝑡𝑠,𝑐,𝑡 in each sector s are calculated based on historical trends extrapolations of the energy

124 However, it is also important to evaluate historical data in the context of IAMs results and the future scenario storylines.
IAMs scenarios or SSPs storylines usually envisage increasing GDP per capita over time, whereas historical data show
that in several countries GDP per capita has declined during the period 1980-2010 (including for example Saudi Arabia,
Brunei, Haiti, Venezuela, Zimbabwe etc.). In this case, it might not be entirely appropriate to rely only on historical trend
extrapolations (as future income per capita growth might largely differ from the developments observed in the past).
For this reason (only for countries with declining GDP per capita), an additional data point (with t=2100) is added to the
historical data series, based on long term projections. By doing so, the historical data information (until the most recent
available year) is combined with the energy intensity projections (based on regional IAMs long-term trajectory) in 2100.

165
intensities at the country level. Short-term projections are finally also harmonized to match regional IAMs
results.

For long term projections, it is assumed that the energy intensity path (over GDP per capita) will be the same
across all countries within the same region. Therefore, a relationship between energy intensity 𝐸𝐼𝐿𝑜𝑛𝑔 and
GDP per capita is estimated via regression, based on regional IAMs results.

𝐺𝐷𝑃𝑐,𝑡 (5)
̂ 𝑠,𝑐,𝑡 = 𝑒𝑥𝑝 [ 𝛼𝑐 + 𝛽𝑐 log (
𝐸𝐼𝐿𝑜𝑛𝑔 )]
𝑃𝑂𝑃𝑐,𝑡

Then, the final energy demand (EN) at the country level is calculated by multiplying the energy intensity (EI) by
the GDP projections (available at the country level).

̂ 𝑠,𝑐,𝑡 = 𝐸𝐼𝐿𝑜𝑛𝑔
̂ 𝑠,𝑐,𝑡 × 𝐺𝐷𝑃𝑠,𝑐,𝑡 (6)
𝐸𝑁𝐿𝑜𝑛𝑔

Based on those calculations, countries with the same level of income per capita will have the same level of
energy intensity in a given year. Then, the long-term projections are harmonized to ensure that the sum of
country-level results coincides with the regional IAMs data 𝐸𝑁𝑠,𝑅,𝑡 in a proportional manner.

𝐸𝑁𝑠,𝑅,𝑡 ̂ 𝑠,𝑐,𝑡 (7)

𝐸𝑁𝐿𝑜𝑛𝑔𝑠,𝑐,𝑡 = ⁄∑ ̂ × 𝐸𝑁𝐿𝑜𝑛𝑔
𝑐𝜖𝑅 𝐸𝑁𝐿𝑜𝑛𝑔𝑠,𝑐,𝑡

Using different assumptions on conditional convergence, the downscaling algorithm in principle provides a
range of energy demand pathways at the country level. However, for the NGFS scenarios a mapping of
conditional convergence to each scenario is used (see Table 18), leading to a single projection for each scenario.

The same approach as described here for final energy demand in general is used to downscale final energy
results for individual subsectors (such as electricity, solids, liquids, gas, transportation, industry, residential and
commercial). For example, the subsector electricity of sector final energy is downscaled by considering the
relationship between the share of electricity on total final energy and GDP per capita.

The downscaled results need to be internally consistent. This means that the sum of sub-sectors (such as
industry, transportation, and residential and commercial) needs to be in line with total final energy in each
country. Hence, some adjustments are introduced by using an iterative process: first, the energy carriers are
adjusted in a proportional manner, so that:

 The sum of subsectors coincides with total final energy demand,

 The sum across countries coincides with the regional IAMs results.

These two steps are iterated to obtain results that are in line with the IAMs results and consistent at the sector
level. Note that these adjustments are applied only for short-term projections, which are based on historical
country-level data. Conversely, long-term projections do not need further adjustments as they are entirely
based on regional IAMs results. Finally, the range of projections is calculated based on assumptions on the
timing of conditional convergence.

166
Secondary energy is downscaled by fuel (coal, oil, gas, biomass, nuclear, solar, wind and geothermal energy)
for each energy carrier (e.g., liquids, solids, gases, electricity). For short-term projections, the fuel mix of solids,
liquids and gases is calculated based on historical data at the base year (t=tb).

𝐸𝑁𝑒,𝑐,𝑡=𝑡𝑏,𝑓 (8)
𝐸𝑁𝑆ℎ𝑜𝑟𝑡𝑒,𝑐,𝑡,𝑓 = 𝐸𝑁𝑒,𝑅,𝑡,𝑓 ×
𝐸𝑁𝑒,𝑅,𝑡=𝑡𝑏,𝑓

For the long-term projections the fuel composition f in all countries within a given region R is derived from the
IAM, based on regional IAMs results 𝐸𝑁𝑒,𝑅,𝑡,𝑓 for each energy carrier e, i.e., by the time of the conditional
convergence year (see Table 19), all countries within a region will have the same fuel mix:

𝐸𝑁𝑒,𝑅,𝑡,𝑓 (9)
𝐸𝑁𝐿𝑜𝑛𝑔𝑒,𝑐,𝑡,𝑓 = 𝐸𝑁𝐿𝑜𝑛𝑔𝑒,𝑐,𝑡 ×
∑𝑓 𝐸𝑁𝑒,𝑅,𝑡,𝑓

𝐸𝑁𝑒,𝑐,𝑡=𝑡𝑏,𝑓
𝐸𝑁𝑆ℎ𝑜𝑟𝑡𝑒,𝑐,𝑡,𝑓 = 𝐸𝑁𝑒,𝑅,𝑡,𝑓 × For downscaling the electricity mix EL, a variety of criteria is used to
𝐸𝑁𝑒,𝑅,𝑡=𝑡𝑏,𝑓

calculate the short-term projections (historical data, remaining economic lifetime and planned capacities,
governance, supply cost curves). To this end, a weight for each criterion i is assumed, and the short-term
projections are calculated as a weighted average (where the sum of the criterion weights must equal 1).

̂ 𝑒=𝐸𝐿,𝑐,𝑡,𝑓 = ∑ 𝑤𝑖,𝑓,𝑡 × 𝐸𝑁𝑆ℎ𝑜𝑟𝑡𝑒=𝐸𝐿,𝑐,𝑡,𝑓,𝑖

𝐸𝑁𝑆ℎ𝑜𝑟𝑡 (10)
𝑖

At the base year, electricity generation is initialized by using historical data criteria, and for all other periods,
specific weights are assumed for each fuel. The next steps are as follows:

 Harmonizing the results proportionally to match regional IAMs data for each fuel.
 Updating the results dynamically over time to account for path dependencies, starting from the results
at the base year, and computing the difference in IAM results.
 Allocating this difference to the country level (based on a range of criteria).
 Adjusting the results to match regional IAMs as well as calculating the projections based on the
assumptions on conditional convergence.

The primary energy mix at the country level is calculated by multiplying secondary energy results by using the
same conversion rates used in IAMs.

𝑃𝑟𝑖𝑚𝑎𝑟𝑦𝑒,𝑐,𝑡,𝑓,𝑡𝑐 = 𝑆𝑒𝑐𝑜𝑛𝑑𝑎𝑟𝑦𝑒,𝑐,𝑡,𝑓,𝑡𝑐 × 𝐶𝑜𝑛𝑣𝑒,𝑅,𝑡,𝑓 (11)

Regarding technologies with CCS the same share of CCS versus non-CCS technologies as in regional IAMs
results is applied to the country level.

𝑃𝑟𝑖𝑚𝑎𝑟𝑦_𝐶𝐶𝑆𝑅,𝑡,𝑓,𝑡𝑐 (12)
𝑃𝑟𝑖𝑚𝑎𝑟𝑦_𝐶𝐶𝑆𝑐,𝑡,𝑓,𝑡𝑐 = 𝑃𝑟𝑖𝑚𝑎𝑟𝑦𝑐,𝑡,𝑓,𝑡𝑐 ×
𝑃𝑟𝑖𝑚𝑎𝑟𝑦𝑅,𝑡,𝑓,𝑡𝑐

𝑃𝑟𝑖𝑚𝑎𝑟𝑦_𝑤𝑜_𝐶𝐶𝑆𝑅,𝑡,𝑓,𝑡𝑐 (13)
𝑃𝑟𝑖𝑚𝑎𝑟𝑦_𝑤𝑜_𝐶𝐶𝑆𝑐,𝑡,𝑓,𝑡𝑐 = 𝑃𝑟𝑖𝑚𝑎𝑟𝑦𝑐,𝑡,𝑓,𝑡𝑐 ×
𝑃𝑟𝑖𝑚𝑎𝑟𝑦𝑅,𝑡,𝑓,𝑡𝑐

167
For calculating CCS sequestration from biomass, the same emission factor is applied as used by the various
IAMs. The CCS sequestration from fossils is calculated by using adequate emission factors for oil, gas, and coal.
Finally, total emissions from energy and CCS emissions are harmonised so that the sum of country level results
matches the regional IAMs data.125

To allocate direct land use emissions from IAMs at the country level, a two-step process is
employed. For land use emissions, historical data are available until 2020. As a result, land use
emissions are projected starting from 2020 (t=0) onwards. Firstly, the land use emissions in 2020 are (14)
initialized by using the average 2010-2020 data. Secondly, the change in land use emissions over
time is distributed from regional IAM results to the country level using the following
𝑎𝑏𝑠(ℎ𝑖𝑠𝑡_𝑎𝑣𝑒𝑟𝑎𝑔𝑒_𝑑𝑖𝑟𝑒𝑐𝑡_𝐿𝑈𝑐 )
formulas:𝑠ℎ𝑎𝑟𝑒𝑐 =
𝑎𝑏𝑠(∑𝑐 ℎ𝑖𝑠𝑡_𝑎𝑣𝑒𝑟𝑎𝑔𝑒_𝑑𝑖𝑟𝑒𝑐𝑡_𝐿𝑈𝑐 )

𝑑𝑒𝑙𝑡𝑎_𝑑𝑖𝑟𝑒𝑐𝑡_𝐿𝑈𝑟,𝑡 = 𝐼𝐴𝑀_𝑑𝑖𝑟𝑒𝑐𝑡_𝐿𝑈𝑟,𝑡+1 − 𝐼𝐴𝑀_𝑑𝑖𝑟𝑒𝑐𝑡_𝐿𝑈𝑟,𝑡 (15)

𝑑𝑜𝑤𝑛𝑠𝑐𝑎𝑙𝑒𝑑_𝐿𝑈𝑐,𝑡 = ℎ𝑖𝑠𝑡_𝑎𝑣𝑒𝑟𝑎𝑔𝑒_𝑑𝑖𝑟𝑒𝑐𝑡_𝐿𝑈𝑐,𝑡0 + 𝑑𝑒𝑙𝑡𝑎_𝐷𝑖𝑟𝑒𝑐𝑡_𝐿𝑈𝑟,𝑡 ∗ 𝑠ℎ𝑎𝑟𝑒𝑐 (16)

As land use (LU) emissions can exhibit positive or negative values, the absolute value of historical emissions
shares is employed for allocating regional (r) data to the country level (c), for each time period (t).

IAMs do not account for indirect land use emissions, further contributing to around 5.5 GtCO2 of carbon sinks
globally. Considering the volatile nature of land use, the indirect emissions data are initialized based on the
average values from 2010-2020. The growth rates of the regional adjustment values estimated by the
IMAGE/LPJmL model are then applied to each country belonging to the same region. As these growth rates are
influenced by global mean temperature, a scenario mapping is utilised that aligns the NGFS scenario with the
RCP emissions trajectory reported by Grassi et al. (2021).

Table 20. Mapping NGFS scenarios with RCP emissions trajectories

SSP RCP scenarios NGFS scenarios

SSP2 RCP 1.9 Net Zero 2020, Low Energy Demand

SSP2 RCP 2.6 Delayed transition, Below 2°C

SSP2 RCP 3.4 Fragmented World, NDCs

SSP2 RCP 4.5 Current policies

125 For downscaling non-CO2 emissions, regional IAMs results are linked with the GAINS (Greenhouse Gas – Air Pollution
Interactions and Synergies) country-level results (Höglund-Isaksson et al., 2020, Winiwarter et al., 2018, Purohit et al.,
2020). The GAINS model focuses on cost-effective strategies for greenhouse gas emissions control, emphasizing
improvements in air quality. It provides non-CO2 emissions pathways for baseline and maximum technical abatement
potential scenarios across 96 countries.

168
Scenarios have been mapped based on the global temperature peak, as well as 2100 temperature values from
the NGFS scenarios, in alignment with the SSP/RCP scenarios. Finally, total land use emissions are calculated
as the sum of direct and indirect emissions.

The downscaling method as described so far, does not yet account for policies at the country level that could
influence a country’s emissions. Ideally, in the Current policies and NDC scenarios, the energy system policies
(such as renewable energy targets, or efficiency policies) would be explicitly taken into account in the
downscaling. However, this labour-intensive development is left for future projects. Currently, country-level
policies are incorporated in the downscaling algorithm by introducing an assessment of the NDC emissions
targets at the country level, in order to enhance realism of country-level pathways. Those targets are applied
to total GHG emissions and are introduced as soft constraints, as country-level policies might not be fully
consistent with underlying IAMs results, depending on scenario/storylines considered. In other words, it is
assumed that countries will try to reach their domestic targets, although these might be only partially achieved
(depending on regional policies considered by a given model/scenario). Domestic targets for 2030 are
introduced for all scenarios. The mid-century (2050) strategies are introduced only for the Net Zero 2050, Low
Energy Demand, Fragmented World and Delayed transition scenarios. As described below, policies are
introduced in three steps:

 First, total GHG emissions are computed as the sum of total CO2 emissions, LULUCF (land use, land-
use change and forestry) emissions and total non-CO2 gases based on IPCC AR4 (IPCC 4th Assessment
Report) Global Warming Potentials.
 Second, the gap between current total GHG emissions (without policies) and the emissions targets is
calculated. Then those emissions targets (for 2030 and 2050) are distributed to yearly emissions
targets for all time periods (starting from 2015), assuming that they will gradually tighten over time,
based on a linear interpolation.
 Third, it is assumed that countries will fill the emissions gap by either increasing BECCS (biomass with
CCS) or by replacing fossil fuels with renewables. Here, the assumption is that countries will try to fill
50% of the emissions gap by increasing BECCS. However, the amount of BECCS largely depends on
the type of policy scenario (e.g., BECCS technologies are usually not deployed under a current policy
scenario) and by biomass availability. As a result, it might not be possible to meet 50% of the emission
gap by increasing BECCS. Therefore, it is further assumed that the remaining emission gap (50% or
more) will be met by replacing fossil fuels with renewables.

This approach allows for generating pathways as consistent as possible with country-level NDCs targets and
mid-century net zero strategies. A caveat of this methodology is that final energy is not adjusted to meet the
domestic targets (the downscaling algorithm adjusts all the primary and secondary energy variables, but does
not update the final energy variables), which might create inconsistencies if large adjustments are needed to
achieve those targets.

169
4. Key outputs

After the methodological part, downscaling results and available outputs across the IAMs applied to the NGFS
policy scenarios (i.e., GCAM, MESSAGE-GLOBIOM and REMIND-MAgPIE) are compared. For example, IAM
results were downscaled to the country level and then re-aggregated to the EU27 level for each IAM. The graph
in Figure 90 below compares energy related CO2 emissions for the EU27 regions across different NGFS climate
policy scenarios for all the three models.

Figure 90. Energy related CO2 emissions in the EU27 region across models and scenarios

The graph above shows that projected energy related CO2 emissions depend on the type of model chosen. For
example, under the Current policies scenario, REMIND envisions a faster emissions reduction in the EU27,
compared to MESSAGE and GCAM (results on individual country level might also largely differ). This pattern is
affected by different assumptions regarding final energy demand, as shown in the graph in Figure 91 below.

170
 Figure 91. Final energy in the EU27 region across models and scenarios

In Table 36, the list of output variables that are made available by the downscaling algorithm in the NGFS Phase
IV Scenario Explorer for all individual countries 126 is provided. For convenience, subsectors are named up to a
specific level. For the full trees of IAM- and input-dependent disposable variables see Table 34.

EU27 results have been aggregated based on downscaled results from individual EU countries. For example,
the Figure 92 and Figure 93 below show the results for Germany under a Current policies scenario from the three
IAMs:

GCAM REMIND

MESSAGE

126 For some additional variables, downscaled outputs are available just for specific countries (based on input data).

171
 Figure 92. Germany's downscaled GHG emissions in the Current policies scenario across the three IAMs

GCAM REMIND

MESSAGE

Figure 93. Germany's downscaled primary energy mix in the Current policies scenario across the three IAMs

5. What is new in the 2023 edition?

 Non-CO2 GHG emissions are linked with the GAINS (Greenhouse Gas – Air Pollution Interactions and
Synergies) country-level results (Höglund-Isaksson et al., 2020, Winiwarter et al., 2018). The GAINS
model focuses on cost-effective strategies for greenhouse gas emissions control, emphasizing
improvements in air quality. It provides non-CO2 emissions pathways for baseline and maximum
technical abatement potential scenarios across 96 countries. The country-level results from GAINS are
interpolated and harmonized to be consistent with the regional IAMs pathways.

172
 Accounting for the Grassi issue: IAMs results do not consider indirect emissions from the land use
sector. This leads to a mismatch of around 5.5 GtCO2e when comparing global IAMs results with
aggregated national inventories (Grassi et al., 2021). To solve this issue, indirect emissions are added
from land use to total GHG emissions, by applying regional growth rates – estimated by the
IMAGE/LPJmL model – to the country level data. As these growth rates are influenced by global mean
temperature, a scenario mapping is utilised that aligns the NGFS scenarios with the RCP emissions
trajectory reported by Grassi et al., 2021 (see Table 20).

 Benchmarking of variables to historic data rather than to IAM base year quantities. Downscaled
results are harmonised to match the most recent historical data. This is done by using either offset or
ratio methods, which utilise the difference (ratio) of unharmonised and harmonised results, combined
with convergence methods, and converge to the long-term original results at a given point in time
(Gidden et al., 2018). Regarding the historical data sources, the IEA energy statistics (2022) are used
for primary and secondary energy variables, and PRIMAP (Gütschow et al., 2021) for emissions.

173
Module 8: NiGEM
1. Non-technical summary

The National Institute Global Econometric Model (NiGEM) is a peer-reviewed global econometric model
developed since 1987. NiGEM represents a closed world, where outflows from one country or region are
matched by inflows into other countries and regions. NiGEM consists of individual country models for the major
economies built around the national income identity, and contains the determinants of domestic demand,
trade volumes, prices, current accounts, and asset holdings. Other countries are modelled through regional
blocks (Africa, Middle East, Latin America, Developing Europe, and East Asia), so that the model is fully global
in scope.
The NiGEM model is used for economic forecasting, scenario analysis and stress testing and has been in
continuous development for over 30 years to remain relevant as economic behaviours, structures and theories
have evolved.
The NiGEM climate model is an expanded version of the standard model that introduces channels to model
climate policy instruments through the implementation of energy transition and physical climate shocks.
Further technical details of the NiGEM model are available here.

What are the key model inputs?

For the calibration of climate scenarios, NiGEM takes input from different modules of the NGFS modelling
framework. The Integrated Assessment Models (IAMs) provide data for a new baseline forecast and climate
transition risk scenarios. GDP damages due to chronic physical risk are provided by the Potsdam Institute (PIK)
and the impact of acute physical risks are provided by Climate Analytics. The combination of transition, chronic
and physical risk shocks, when executed as NiGEM scenarios, provide the output for the NGFS macro-economic
variables.

What are the key model outputs?

NiGEM outputs provide the macroeconomic and financial information of the NGFS Scenarios, including
transition, chronic and acute physical risks. Outputs include major macroeconomic and financial variables, like
GDP, Inflation, Unemployment, Consumption, Investment, Exports, Imports, Interest rates.

What is new in the 2023 edition?

Several country model were added (Malaysia, Croatia), or expanded from a reduced country model to a full
country model (Romania, South Africa). Trade matrices and commodity equations updated to 2019 trade
figures (from 2017). The modelling of acute physical risks was expanded (see also Module 6).

2. Introduction to NiGEM

The National Institute of Economic & Social Research (NIESR) has provided policy makers and private sector
organisations around the world with a peer-reviewed global econometric model, the National Institute Global
Econometric Model (NiGEM), since 1987. The model is used for economic forecasting, scenario analysis and
stress testing and has been in continuous development for over 30 years to remain relevant as economic
behaviours, structures and theories have evolved.

NiGEM represents a closed world, where outflows from one country or region are matched by inflows into other
countries and regions. NiGEM is an Econometric model, in that key behavioural equations are econometrically

174
estimated using historical data. This ensures that the dynamics and key elasticities of the model fit the main
characteristics of individual country data.

The NiGEM climate model is an expanded version of the standard model that introduces channels to model
climate policy instruments through the implementation of energy transition and physical climate shocks.
NIESR started developing its climate module in 2018, with an aim of understanding the interactions between
the macroeconomy and climate-related shocks and climate-related policy. Some of this early work was carried
out in collaboration with the Dutch National Bank (Vermeulen et al, 2018) 127. In 2021, NIESR joined the
academic consortium of the Network for Greening the Financial System (NGFS), to contribute to the NGFS
Climate Scenarios (NGFS, 2021)128.

NiGEM v1.23.2 currently forms part of the NGFS Phase IV suite of models, providing the macro-economic
impacts of climate implied by both the IAM transition risks as well as acute and chronic physical risks.

127 Vermeulen, R., Schets, E., Lohuis, M., Kolbl, B., Jansen, D. J., & Heeringa, W. (2018). An energy transition risk stress test
for the financial system of the Netherlands (No. 1607). Netherlands Central Bank, Research Department

128 NGFS (2021). NGFS Climate Scenarios for central banks and supervisors. Available at:
ngfs.net/sites/default/files/media/2021/08/27/ngfs_climate_scenarios_phase2_ june2021.pdf

175
2.1 Country model specification in NIGEM

NiGEM consists of individual country models for the major economies built around the national income identity,
and contains the determinants of domestic demand, trade volumes, prices, current accounts, and asset
holdings. These models also incorporate a well-specified supply-side, which underpins the sustainable growth
rate of each economy in the medium term. Individual country models are linked together through trade in
goods and services and integrated capital markets. So, in NiGEM, a slowdown in a given country, associated
with lower imports, would impact other countries through the effect of lower exports to that economy and
associated shift in asset prices. The overall impact would depend on both the underlying source of the shock
and the policy responses (both in a country where the shock originates and other economies).

Figure 94. NiGEM country model structure

Full country models have a more disaggregated description of domestic demand than reduced country models
and incorporate greater detail on the labour market and the government sector. See 5.1 for country
classification.

2.2 Country coverage

Individual country models are in place for almost all OECD countries. There are also separate models of
Argentina, Brazil, Bulgaria, China, Egypt, Hong Kong, Indonesia, India, Malaysia, Romania, Russian Federation,
South Africa, Singapore, Taiwan and Vietnam. The rest of the world is modelled through regional blocks of
Africa, Middle East, Latin America, Developing Europe, and East Asia, so that the model is fully global in scope.
This ensures that there are no “black holes” in international transactions, as outflows from one country must
be matched by inflows into other countries.

Country models are linked together through trade in goods and services, the influence of trade prices on
domestic inflation, the impacts of exchange rates, and the patterns of asset holdings and associated income
flows.

176
 Figure 95. NiGEM coverage

2.3 Policy environment and key applications

The scenario space in NiGEM, including policy regimes, expectation formation by consumers, firms, wage
setters or financial markets, and other assumptions and judgements can be set by the model user. In standard
simulations, financial markets are normally assumed to look forward and consumers are normally assumed to
be myopic but react to changes in their (forward looking) financial wealth. Monetary policy is set according to
rules, with default parameters calibrated for individual countries. Key applications of NiGEM include:

 The production of economic forecasts for the world economy. NIESR publishes quarterly forecasts produced
with NiGEM, along with a discussion of alternative scenarios around the central forecast and short notes
based on recent model-based research.
 Simulation and analysis tool. Typical simulations involve analysing the effects of changes in monetary or
fiscal policy, or changes in commodity prices such as an oil price shock. The model has a considerable
degree of built-in flexibility, with key assumptions, such as the form of expectation formation in different
markets and the policy rules followed by monetary and fiscal authorities able to be modified.
 The stochastic mode of NiGEM is used to construct error bounds around the central forecast baseline. The
fan charts are based on stochastic shocks drawn from the historical errors on all the key model equations.
Although this mode is not available for the NGFS scenarios, the stochastic work has been extended within
the NGFS to investigate acute damages based on climate damage data provided by Climate Analytics.

177
2.4 Running a NiGEM simulation

An economic forecast normally represents the “most likely” future projections of the macro-economic variables
being considered; in NiGEM this can be viewed as a central baseline scenario. This baseline is conditional on
historical data, economic environment, and forecaster judgement. Conditional assumptions will include
monetary and fiscal policy assumptions, and settings on other key variables such as the oil price and the pace
of technological change.

What we refer to as a “simulation” in NiGEM is an alternative scenario that is assessed relative to a baseline
forecast. The simulation includes one or more changes to the conditional assumptions of the baseline forecast,
within a user-defined policy environment.

Simulation process

 Narrative: What is the source of the shock and your underlying premise for the scenario?
 Channels: How do the shocks propagate within the model and how will they be applied?
 Shocks: Determine the size and profile of the shock, for example is it permanent or temporary?
 Policy: How do agents such as central banks and governments respond, are the shocks anticipated or
unanticipated?

178
3. NiGEM integration into the NGFS scenarios

For the calibration of climate scenarios, NiGEM takes input from different modules of the NGFS modelling
framework. The Integrated Assessment Models (IAMs) provide data for a new baseline forecast and climate
transition risk scenarios. GDP damages due to chronic physical risk are provided by the Potsdam Institute (PIK)
and the impact of acute physical risks are provided by Climate Analytics. The combination of transition, chronic
and physical risk shocks, when executed as NiGEM scenarios, provide the output for the NGFS macro-economic
variables. The various climate scenarios often use the same economic channels in the model, and therefore,
cannot be imposed as a singular shock. Instead, individual scenarios are run as a “stacked” series, which ensures
the output of the stack provides the same output as though all shocks being considered were run
simultaneously. The chart below shows the data links between NiGEM and external NGFS inputs.

• IAM Current • IAM NGFS

policies and scenario
NIESR forecast output

Forecast Transition

Acute Chronic

• CA data • PIK, GDP

impact
(Kalkhul &
Wenz)

Figure 96. External data links with NiGEM

179
3.1 Climate neutral forecast

To ensure NiGEM and the IAMs are using an equivalent starting point for their investigations into climate risk,
particularly in the energy sector, we use a combination of the NIESR v1.23 forecast coupled with IAM data from
the NGFS current policies scenario to create a climate neutral forecast base. Climate neutral refers to the fact
that projected data values do not reflect any climate transition or physical risks.

Figure 97. Creating the climate neutral forecast.

180
3.2 Transition scenarios

The transition scenarios represent two distinct NiGEM simulations : (i) carbon price shock and effects on energy
use, and (ii) carbon revenue recycing.

Figure 98. NiGEM transition risk scenario

The figure below disaggregates for each IAM input and displays the channel pass-through into NiGEM in more
detail.

Figure 99. Transition risk channels from IAM inputs

The carbon price shock uses a combination of energy consumption, carbon tax revenue and useful energy 129
from the IAM scenario output to create the transition risk simulation connected to the application of a carbon

129 The portion of final energy which is actually available after final conversion to the consumer for the respective use. In
final conversion, electricity becomes for instance light, mechanical energy or heat

181
tax. The carbon tax revenue is also used to reflect the budgetary impacts of recycling this revenue into the
economy.

The recycling scenarios use the IAM carbon tax revenue as the basis for the size and profile of the shock to
apply. Orderly scenarios use a recycling option where 50% of the revenue is used for government investment
while the remaining 50% is used to pay off government debt. All other scenarios recycle all revenue through
taxes. The recycling simulations also turn the energy sector in NiGEM off. This is to ensure all energy
movements, including world price of fossil fuels etc., are directly related to the IAM transition shock rather than
because of fiscal stimulus.

The final impacts from transition scenarios are determined by a combination of competing factors across the
two scenarios, as illustrated by the Figure 100 below.

Figure 100.Factors affecting GDP and inflation.

3.3 Chronic physical impacts

The GDP percentage country damages for each scenario are provided directly by the Kalkuhl and Wenz damage
function, as described in section 5.2.Those shocks are implemented in NiGEM to obtain the associated
macrovariables paths..

3.4 Acute physical impacts

Building on the process used for chronic impacts, acute physical impacts use data provided by Climate Analytics
to determine GDP impacts of four differing acute weather events. For acute risk, different channels of
transmissions are assessed depening on the relevant hazard (see below). The different channels of
transmission, e.g. labour force, crops productivity or asset damages, need to be adequately represented in
NiGEM for the macroeconomic impacts to be correctly estimated. The hazard specific approaches are
explained in Stochastic Implementation.

182
  Heatwaves (labour force impact)
 Cyclones (capital stock damages)
 Floods (capital stock impact)
 Drought (productivity, commodity price and export impact)

4. NiGEM NGFS output

NiGEM forecast data is in line with National accounts, so forecast values reference a country’s domestic
currency and base year 130.

𝑠𝑐𝑒𝑛𝑎𝑟𝑖𝑜
𝑥𝑡𝑖𝑚𝑒=𝑡 𝑠𝑐𝑒𝑛𝑎𝑟𝑖𝑜 𝑏𝑎𝑠𝑒
% 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 = ( 𝑏𝑎𝑠𝑒 − 1. ) ∗ 100. 𝐴𝑏𝑠 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 = 𝑥𝑡𝑖𝑚𝑒=𝑡 − 𝑥𝑡𝑖𝑚𝑒=𝑡
𝑥𝑡𝑖𝑚𝑒=𝑡

Database reference Unit NiGEM description

NiGEM|Gross Domestic Product % difference, country base Gross Domestic Product

(GDP) year; local currency (GDP), country base year
NiGEM|Consumption (private) % difference, country base Consumption (private),
year; local currency country base year
NiGEM|Investment (private sector) % difference, country base Investment (private sector),
year; local currency country base year
NiGEM|Gov. consumption % difference, country base Gov. consumption, country
year; local currency base year
NiGEM|Investment (gov.) % difference, country base Investment (gov.), country
year; local currency base year
NiGEM|Domestic demand % difference, country base Domestic demand, country
year; local currency base year
NiGEM|Exports (goods and % difference, country base Exports (goods and services),
services) year; local currency country base year
NiGEM|Imports (goods and % difference, country base Imports (goods and services),
services) year; local currency country base year
NiGEM|Productivity (output per % difference Productivity (output per hour
hour worked); local currency worked)
NiGEM|Unemployment rate; % Abs. difference Unemployment rate
NiGEM|Gross operating surplus % difference, pte Gross operating surplus, pte
corporations; local corporations
currency
NiGEM|Real personal disposable % difference, country base Real personal disposable
income year; local currency income, country base year

130 A base year refers to the base point in time of a time series such as with a GDP deflator to convert GDP at current
market prices into GDP at constant prices.

183
 NiGEM|House prices (residential) % difference, index; House prices (residential),
country base year=100 index
NiGEM|Inflation rate; % Abs. difference Inflation rate
NiGEM|Central bank Intervention Abs. difference Central bank Intervention rate
rate (policy interest rate); % (policy interest rate)
NiGEM|Long term interest rate; % Abs. difference Long term interest rate
NiGEM|Long term real interest Abs. difference Long term real interest rate
rate; %
NiGEM|Nominal exchange rate % difference Exchange rate; local currency
per US$
NiGEM|Effective exchange rate % difference, index; Effective exchange rate, index
2019=100
NiGEM|Equity prices % difference, index; Equity prices, index
2019=100
NiGEM|Energy consumption % difference Energy consumption (total)
(total); MnToe
NiGEM|Quarterly consumption of % difference Quarterly consumption of oil
oil; MnToe
NiGEM|Quarterly consumption of % difference Quarterly consumption of gas
gas; MnToe
NiGEM|Quarterly consumption of % difference Quarterly consumption of coal
coal; MnToe
NiGEM|Quarterly consumption of % difference Quarterly consumption of non-
non-carbon; MnToe carbon
NiGEM|Gross domestic income; % difference Gross domestic income
local currency
NiGEM|Trend output for capacity % difference, country base Trend output for capacity
utilisation year; local currency utilisation, country base year
NiGEM|Oil price; US$ per barrel % difference Oil price
NiGEM|Gas price; US$ per % difference Gas price
barrel(equiv)
NiGEM|Coal price ; US$ per % difference Coal price
barrel(equiv)

Additional notes
 Short-term interest rates: 3-month rates
 Long-term interest rates: a 40 period (10 year) look-ahead average of short-term interest rates
 Inflation: annual rate of change of the consumer expenditure deflator (CED) (YoY growth)
 Nominal exchange rate: country exchange rates are defined in terms of US$ in NIGEM, so positive
delta (change) shows a depreciation, negative delta (change) an appreciation.
 Effective exchange rate: weighted sum of nominal exchange rates, positive delta shows an
appreciation.

184
5. NiGEM Technical references

5.1 Country classification

Full country models Reduced country models

Australia Polanda Africa block Malaysia

Austria Portugal Argentina Mexico
Belgium Romania Brazil Middle East block
b
Canada South Africa Bulgaria New Zealand
China Spain Chile Norway
Czechiaa Sweden Croatia Rep. of Korea
Denmark U.K. Developing Europe Romaniab
Finland USA East Asia block Russian Federation
France Egypt Singapore
Germany Estoniab Slovakiab
Greece Hong Kong Sloveniab
Hungarya India South Africa
Ireland Indonesia Switzerland
Italy Latin America block Taiwan
b
Japan Latvia Turkey
b
Netherlands Lithuania Viet Namb

5.2 Regional country constituents

Based on the IMF's group Sub-Saharan Africa. From this we exclude the
countries modelled individually on NiGEM (South Africa). This group
includes: Angola, Benin, Botswana, Burkina Faso, Burundi, Cabo Verde,
Cameroon, Central African Republic, Chad, Comoros, Democratic
Republic of the Congo, Republic of Congo, Côte d'Ivoire, Equatorial
Africa block
Guinea, Eritrea, Eswatini, Ethiopia, Gabon, The Gambia, Ghana, Guinea,
Guinea-Bissau, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali,
Mauritius, Mozambique, Namibia, Niger, Nigeria, Rwanda, São Tomé and
Príncipe, Senegal, Seychelles, Sierra Leone, South Sudan, Tanzania, Togo,
Uganda, Zambia, and Zimbabwe.
Based on the IMF's group Emerging and Developing Europe. From this we
exclude the countries modelled individually on NiGEM (Bulgaria, Hungary,
Poland, Romania, Russia and Turkey) and we add the advanced European
Developing Europe
economies that are not modelled separately on NiGEM (Iceland,
block
Luxembourg, Malta and Cyprus). This group includes: Albania, Belarus,
Bosnia and Herzegovina, Cyprus, Kosovo, Iceland, Luxembourg, Malta,
Moldova, Montenegro, North Macedonia, Serbia, Ukraine.

185
 Based on the IMF's group Emerging and Developing Asia. From this we
exclude the countries modelled individually on NiGEM (China, India,
Indonesia, and Viet Nam). This group includes: Bangladesh, Bhutan,
East Asia block Brunei Darussalam, Cambodia, Fiji, Kiribati, Lao P.D.R., Maldives, Marshall
Islands, Micronesia, Mongolia, Myanmar, Nauru, Nepal, Palau, Papua New
Guinea, Philippines, Samoa, Solomon Islands, Sri Lanka, Thailand, Timor-
Leste, Tonga, Tuvalu, Vanuatu.
Based on the IMF's group Latin America and the Caribbean. From this we
exclude the countries modelled individually on NiGEM (Argentina, Brazil,
Chile, and Mexico). This group includes: Antigua and Barbuda, Aruba, The
Bahamas, Barbados, Belize, Bolivia, Colombia, Costa Rica, Dominica,
Latin America block
Dominican Republic, Ecuador, El Salvador, Grenada, Guatemala, Guyana,
Haiti, Honduras, Jamaica, Nicaragua, Panama, Paraguay, Peru, St. Kitts
and Nevis, St. Lucia, St. Vincent and the Grenadines, Suriname, Trinidad
and Tobago, Uruguay, and Venezuela.
Based on the IMF's group Middle East and Central Asia. To this we add the
advanced Middle East economies that are not modelled separately on
NiGEM (Israel). This group includes: Afghanistan, Algeria, Armenia,
Azerbaijan, Bahrain, Djibouti, Georgia, Iran, Iraq, Israel, Jordan,
Middle East block
Kazakhstan, Kuwait, Kyrgyz Republic, Lebanon, Libya, Mauritania,
Morocco, Oman, Pakistan, Qatar, Saudi Arabia, Somalia, Sudan, Syria,
Tajikistan, Tunisia, Turkmenistan, United Arab Emirates, Uzbekistan,
West Bank and Gaza, and Yemen.

5.3 Theoretical foundation

NiGEM is based on a broadly New Keynesian structure with many of the characteristics of dynamic stochastic
general equilibrium (DSGE) models, individual country models are grounded in textbook macroeconomic
foundations, with features such as sticky prices, rational or model-consistent expectations, endogenous
monetary policy based on a Taylor rule or other standard specifications, and long-run fiscal solvency. The
structure of NiGEM is designed to correspond to macroeconomic policy needs.

From a theoretical perspective, NiGEM can be classed among global general equilibrium macroeconomic
models, which are fundamentally grounded in Walrasian general equilibrium theory. It therefore strikes a
balance between theoretical underpinnings that guide economies towards long-run market clearing equilibria,
and data-driven individual country characteristics that fit the main characteristics of real-world data outturns.

5.4 Where does NiGEM sit within the spectrum of macroeconomic models?

Blanchard (2018) distinguishes five different classes of general equilibrium macroeconomic models:
foundational models, dynamic stochastic general equilibrium (DSGE) models, policy models, toy models and
forecasting models. Blanchard posits that each class of model is best suited to a specific purpose: foundation
models are designed to make a deep theoretical point; DSGE models are designed to explore the macro
implications of distortions; policy models are best suited to study the dynamic effects of specific shocks; toy
models present the essence of an answer from a more complicated model; and forecasting models are designed
for short-term forecasting. Under this framework, NiGEM best falls into the category of policy models, as it is:
“aimed at analysing actual macroeconomic policy issues”. Models in this class should fit the main characteristics
of the data, including dynamics, and allow for policy analysis and counterfactuals. In terms of general
methodological approach, it can be described as incorporating micro-founded long-run relationships – sharing

186
some properties of standard DSGE models – but with more flexible lag structures that are fitted to the data.
This combination ensures that NiGEM is useful for both policy analysis and forecasting.

5.5 Model usage

A key feature of the model is its flexibility, which allows users to define the scenario space, including policy
regimes, expectation formation by consumers, firms, wage setters or financial markets, and other assumptions
and judgements. Financial markets are normally assumed to look forward and consumers are normally
assumed to be myopic but react to changes in their (forward looking) financial wealth. However, both of these
default settings can be modified. Monetary policy is set according to rules, with default parameters calibrated
for individual countries. However, these feedback rules can also be changed, and their parameters adjusted.
Hence, to describe the results of a given scenario, rather than using a phrase such as 'the NiGEM simulation
results suggest…' a better description would be 'under these assumptions, the NiGEM simulation results
suggest…'. This is different from many other models, and it explains the widespread use of the NiGEM for policy
analysis.

187
6. IAM input variables

6.1 Climate neutral forecast baseline

Table 21. IAM inputs for NiGEM baseline

IAM variables input into Variable Units NiGEM Processing for use in NiGEM
NiGEM Description Suffix

Primary Energy|Coal Energy EJ/yr COLC Level import

consumption  Exajoules to Million
tonnes of oil equivalent
 Annual to quarterly
Primary Energy|Gas Energy EJ/yr GASC Level import
consumption  Exajoules to Million
tonnes of oil equivalent
 Annual to quarterly
Primary Energy|Oil Energy EJ/yr OILC Level import
consumption  Exajoules to Million
tonnes of oil equivalent
 Annual to quarterly
Primary Energy|Biomass Energy EJ/yr RNWC Level import
consumption  Exajoules to Million
Primary
tonnes of oil equivalent
Energy|Geothermal
 Non-carbon = summation
Primary Energy|Hydro
 Annual to quarterly
Primary Energy|Solar
Primary Energy|Wind
Primary Energy|Nuclear
GDP|PPP/Trend capacity GDP/YCAP billion Y Growth rate import
US$2010/yr  Annual to quarterly
To prevent additional inflationary
impacts from supply/demand
imbalances, growth rates set
equal to IAM GDP
Population Population million POPT Level import
 Millions to 1000’s

188
6.2 Carbon price

Table 22. IAM carbon tax inputs

IAM variables input into NiGEM Variable Units NiGEM Processing for use in
Description suffix NiGEM

Primary Energy|Coal Energy EJ/yr COLC Level import:

consumption  Exajoules to
Million tonnes of
oil equivalent
 Annual to
quarterly

Primary Energy|Gas Energy EJ/yr GASC Level import:

consumption  Exajoules to
Million tonnes of
oil equivalent
 Annual to
quarterly

Primary Energy|Oil Energy EJ/yr OILC Level import:

consumption  Exajoules to
Million tonnes of
oil equivalent
 Annual to
quarterly

Primary Energy|Biomass Energy EJ/yr RNWC Level import:

consumption  Exajoules to
Primary Energy|Geothermal
Million tonnes of
Primary Energy|Hydro
oil equivalent
Primary Energy|Solar
 Annual to
Primary Energy|Wind quarterly
Primary Energy|Nuclear  Non-carbon =
summation

189
 Price|Carbon Carbon price US$2010/t CBTAX Level import
CO2  Constant to
current prices
using NiGEM US
GDP deflator
(NIESR).

Deprecated since phase iii

as the carbon revenue is
now provided directly from
the IAMs to account for
CDR & CSS
Useful Energy|Industry Useful Energy EJ/yr OIVOL Multiplicative residual
import
Useful Energy|Residential and
 Delta calculated
Commercial
(w.r.t. current
Useful Energy|Transportation
policies)
 Annual to
Electricity quarterly
Gases
Heat
Hydrogen
Liquids
Solids
Revenue|Government|Tax|Carbon Carbon Revenue billion ETAX Level import
US$2010/yr  Constant to
current prices
using NiGEM US
GDP deflator.
 PPP (2019) used to
convert to local
currency
 Annual to
quarterly

6.3 Revenue recycling

Table 23.IAM recycling input

IAM variables input into NiGEM Variable Units NiGEM Processing for
Description Suffix use in NiGEM

Revenue|Government|Tax|Carbon Carbon Revenue billion ETAX Automated in

S$2010/yr NiGEM using

190
 NiGEM c-tax
output file

6.4 Chronic physical damage

Table 24. Damage function input for chronic physical risk

IAM variables input Variable Units NiGEM Processing for use in NiGEM
into NiGEM Description Suffix

Diagnostics | high GDP Chronic billion YDMG Level import

GDP change | KW damage (%) US$2010/yr  Annual to quarterly
panel population-  95th percentile for Current policies
weighted and NDCs
 50th percentile for all other
scenarios

191
Appendix

Table 25. GCAM external inputs used for demand of energy

Name Description Type

Historical demand for Demand for energy in the historical period; used for External data
energy initialisation/calibration of GCAM
Historical demand for Demand for floorspace in the historical period; used for External data
floorspace initialisation/calibration of GCAM
Price elasticity of demand Elasticity determining how demand responds to changes Assumption
in price
Value of time in transit Factor multiplied by the wage rate to determine the Assumption
multiplier value of time in transit, used in the transportation
module
Cost Cost of production Assumption
Default input-output Default amount of input required per unit of output Assumption
coefficients produced; can be overwritten by region-specific
information derived from historical data
Default efficiencies Default amount of output produced per unit of input; can Assumption
be overwritten by region-specific information derived
from historical data
CO2 capture rates Fraction of CO2 captured in CCS technologies Assumption
Retirement rules For vintaged technologies, GCAM requires the user to Assumption
specify the lifetime, and the parameters required for
phased and profit-based shutdown
Logit exponents GCAM requires the user to specify the logit exponents Assumption
that determine the substitutability between
technologies
Share weight These rules dictate how share weights (GCAM’s Assumption
interpolation rules calibration parameter) are specified in future years
Fuel preference elasticity Elasticity dictating how share weights change with GDP Assumption
per capita
Residential floorspace Estimated parameters for residential floorspace demand Analysis/assumption
parameters
Satiation levels Assumed satiation values for commercial floorspace and Assumption
building energy services
Income elasticity of Elasticity determining how demand responds to changes Assumption
demand in per capita output for industry and cement

192
 Name Description Type

Energy intensities Energy intensity for energy-for-water processes External data

(desalination, abstraction, treatment, distribution,
wastewater treatment)
Desalinated water Water produced through desalination, used to estimate External data
production energy-for-water
Shares of wastewater Shares of wastewater treated, used to estimate energy- External data
treated for-water
Non-renewable Electricity inputs to groundwater production External data
groundwater supply
curves – electricity inputs
Historical non- Historical emissions of non-CO2 External data
CO2 emissions

Table 26. GCAM external inputs used for demand of water

Name Description Type

Agriculture water coefficients Water coefficients for agricultural commodities, External data set
including blue (irrigation) and green (rain) water,
includes data for a single year circa 2000
Industrial manufacturing Water coefficients for industrial manufacturing for External data set
water coefficients 1995
Livestock water coefficients Water coefficients for drinking and the servicing of Mekonnen, M.M.
livestock commodities, includes data for the period and A.Y. Hoekstra
1996-2005 (2010). Volume 2:
Appendices
Electricity cooling system Historical shares of cooling system types associated UCS and Schakel
shares with power plants aggregated to GCAM3 regions Inventories
Electricity water coefficients Water withdrawal and consumption coefficients for External data set
power plants and cooling system types
Primary energy water Water coefficients for the consumption of water during Maheu, A. (2009)
coefficients the process of mining primary energy fuel sources
Municipal water withdrawals Water withdrawal values for municipalities include FAO Aquastat
data, as reported, from 1987 to 2017
Municipal water use efficiency Water efficiency values for municipalities Shiklomanov, I.A.
(2000)

193
 Name Description Type

Municipal water cost Price per unit of water delivered to municipalities International
Benchmarking
Network for Water
and Sanitation
Utilities (IBNET)

Table 27. GCAM external inputs used for demand of food, feed, and forestry

Name Description Type

Historical demand for Demand for agricultural commodities in the historical period; External data
crops used for initialization/calibration of GCAM
Historical demand for Demand for livestock commodities in the historical period; used External data
livestock for initialisation/calibration of GCAM
Historical demand for Demand for forest products in the historical period; used for
External data
forest initialisation/calibration of GCAM
Income and price Income and price elasticity of demand (for non-food, non-feed
Assumption
elasticity demand)
Food demand Set of 11 parameters required for the food demand model
External data
parameters
Logit exponents Share parameters dictating substitution between different
Assumption
commodities

Table 28. GCAM external inputs used for economics

Name Description Type

Population Population by country and year, used for 1700-1900 External data
set
Population Population by country and year, used for 1950-2015 External data
set
Population Population by country and year, used for 2015-2100 External data
set
GDP Historical GDP used for most countries for GDP prior to 2015 External data
set
GDP Historical GDP used for remaining countries for GDP prior to 2015 External data
set
GDP growth rate Near-term growth rate of GDP (2015-2024) External data
set

194
 Name Description Type

GDP GDP by country and year, used for 2025-2100 External data
set

Table 29. GCAM external inputs used by the land model

Name Description Type

Historical land use and land Land area by region, land type and year. Land cover data is External data
cover provided beginning in 1700 in order to spin-up the carbon set
cycle within GCAM. Crop-specific harvested area is used to
downscale FAO data to a subnational level; however, this
data is only available for a single year. Similarly, the division
between irrigated and rainfed land is only available for a
single year only.
Historical harvested area Harvested land area by country, crop, and year External data
set
Historical cropland cover Arable land, temporary crops, and temporary fallow land External data
area by country and year set
Terrestrial carbon Inputs include potential vegetation and soil carbon density External data
information (i.e., carbon density if the land grew to equilibrium), and set
mature age for vegetation carbon. Note that vegetation
carbon contents for crops are calculated from crop yields. All
other carbon parameters are external inputs.
Soil time scale Inputs include the number of years for soil carbon changes to Assumption
occur. Note that this is not the time to equilibrium, which is
much longer.
Value of unmanaged land GCAM requires profit rates for all land types in the historical External data
period for calibration. Managed land profit is calculated in set
the supply model. For unmanaged land, however, the value
is input into the model.

195
 Name Description Type

Share parameters GCAM requires the user to specify the logit exponents that Assumption
determine the substitutability between different leaves and
nodes in the land model. These parameters were chosen to
produce land supply elasticities comparable to those found in
the literature, although it should be noted that there is not a
transformation between logit exponents and supply
elasticities for all land types.
Parameters to introduce a For land types that do not exist in the historical period, GCAM Assumption
new land type requires parameters to introduce these land types in the
future. Specifically, GCAM needs to know how that land type
will compete with other land types in its nest if it were to have
equal profit.

Table 30. GCAM external inputs used for supply of energy

Name Description Type

Historical supply of Supply of energy in the historical period; used for External data
energy initialization/calibration of GCAM
CO2 capture rates Fraction of CO2 captured in CCS technologies Assumption
Retirement rules For vintaged technologies, GCAM requires the user to specify Assumption
the lifetime, and the parameters required for phased and profit-
based shutdown
Logit exponents GCAM requires the user to specify the logit exponents that
Assumption
determine the substitutability between technologies
Share weight These rules dictate how share weights (GCAM’s calibration
Assumption
interpolation rules parameter) are specified in future years
Cost of conversion Cost of production for conversion technologies
External data
technologies
Capital cost Overnight capital cost of electricity generation technologies External data
Fixed O&M costs Fixed operating and maintenance (O&M) costs for electricity
External data
generation technologies
Variable O&M costs Variable operating and maintenance (O&M) costs for electricity
External data
generation technologies

196
 Name Description Type

Capacity factor Ratio of generation to capacity for electricity generation

Assumption
technologies
Fixed charge rate Factor used to levelise capital cost Assumption
Default efficiencies Default amount of output produced per unit of input; can be
overwritten by region-specific information derived from Assumption
historical data
Default input-output Default amount of input required per unit of output produced;
coefficients can be overwritten by region-specific information derived from Assumption
historical data
Resource supply curves Mapping between cost and resource extraction. Resource
extraction is cumulative for depletable resources and annual for External data
renewable resources
Historical non- Historical emissions of non-CO2
External data
CO2 emissions
CO2 emissions Default carbon content of fuels
External data
coefficients
Historical CO2 emissions Historical emissions of CO2 External data

Table 31. GCAM external inputs used for supply of water

Name Description Type

Surface water supply curves Xanthos derived total maximum runoff values, Exogenous
(cost and availability) combined with accessible water calculation to Data
determine water available at very low price and the level
of accessible water for cost-curve inflection

Groundwater supply curves Amount of groundwater available in each basin at Turner et al.
(cost and availability) increasingly high graded levels (2019)

197
 Name Description Type

Desalination cost Cost of desalinated water within a basin which is Exogenous

available at high cost and available once the price of Data
water within a basin surpasses a certain threshold

Table 32. GCAM external inputs used for supply of food, feed, and forestry

Name Description Type

Historical country-level Production of agricultural commodities by country in External data

production of crops the historical period; used for initialisation/calibration
of GCAM
Historical country-level harvested Harvested area for agricultural commodities by country External data
area for crops in the historical period; used for
initialisation/calibration of GCAM
Historical sub-national Production of agricultural commodities by water basin External data
production of crops in a single year; used for initialisation/calibration of
GCAM
Historical sub-national harvested Harvested area of agricultural commodities by water External data
area of crops basin in a single year; used for initialisation/calibration
of GCAM
Historical production of livestock Production of livestock commodities in the historical External data
period; used for initialisation/calibration of GCAM
Livestock feed coefficients Livestock feed input, animal output, and meat output External data
by systems
Historical cost of production Historical cost of crop production in the USA External data
Historical prices Historical prices of agriculture and livestock External data
commodities; used for initialisation/calibration of
GCAM
Agriculture productivity growth Projected yields through 2050 for agricultural External data
commodities
Logit exponents Share parameters dictating substitution between Assumption
different feed options for livestock
Historical non-CO2 emissions Historical emissions of non-CO2 External data

198
Table 33. Mapping from GCAM region to country

GCAM Region Countries

Africa_Eastern Burundi, Comoros, Djibouti, Eritrea, Ethiopia, Kenya, Madagascar, Mauritius,

Reunion, Rwanda, Sudan, Somalia, Uganda
Africa_Northern Algeria, Egypt, Western Sahara, Libya, Morocco, Tunisia
Africa_Southern Angola, Botswana, Lesotho, Mozambique, Malawi, Namibia, Swaziland,
Tanzania, Zambia, Zimbabwe
Africa_Western Benin, Burkina Faso, Central African Republic, Cote d’Ivoire, Cameroon,
Democratic Republic of the Congo, Congo, Cape Verde, Gabon, Ghana, Guinea,
Gambia, Guinea-Bissau, Equatorial Guinea, Liberia, Mali, Mauritania, Niger,
Nigeria, Senegal, Sierra Leone, Sao Tome and Principe, Chad, Togo
Argentina Argentina
Australia_NZ Australia, New Zealand
Brazil Brazil
Canada Canada
Central America and the Aruba, Anguilla, Netherlands Antilles, Antigua & Barbuda, Bahamas, Belize,
Caribbean Bermuda, Barbados, Costa Rica, Cuba, Cayman Islands, Dominica, Dominican
Republic, Guadeloupe, Grenada, Guatemala, Honduras, Haiti, Jamaica, Saint
Kitts and Nevis, Saint Lucia, Montserrat, Martinique, Nicaragua, Panama, El
Salvador, Trinidad and Tobago, Saint Vincent and the Grenadines
Central Asia Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyzstan, Mongolia, Tajikistan,
Turkmenistan, Uzbekistan
China China
Colombia Colombia
EU-12 Bulgaria, Cyprus, Czech Republic, Estonia, Hungary, Lithuania, Latvia, Malta,
Poland, Romania, Slovakia, Slovenia
EU-15 Andorra, Austria, Belgium, Denmark, Finland, France, Germany, Greece,
Greenland, Ireland, Italy, Luxembourg, Monaco, Netherlands, Portugal, Sweden,
Spain, United Kingdom
Europe_Eastern Belarus, Moldova, Ukraine
European Free Trade Iceland, Norway, Switzerland
Association
Europe_Non_EU Albania, Bosnia and Herzegovina, Croatia, Macedonia, Montenegro, Serbia,
Turkey
India India
Indonesia Indonesia
Japan Japan

199
GCAM Region Countries

Mexico Mexico
Middle East United Arab Emirates, Bahrain, Iran, Iraq, Israel, Jordan, Kuwait, Lebanon, Oman,
Palestine, Qatar, Saudi Arabia, Syria, Yemen
Pakistan Pakistan
Russia Russia
South Africa South Africa
South America_Northern French Guiana, Guyana, Suriname, Venezuela
South America_Southern Bolivia, Chile, Ecuador, Peru, Paraguay, Uruguay
South Asia Afghanistan, Bangladesh, Bhutan, Sri Lanka, Maldives, Nepal
Southeast Asia American Samoa, Brunei Darussalam, Cocos (Keeling) Islands, Cook Islands,
Christmas Island, Fiji, Federated States of Micronesia, Guam, Cambodia, Kiribati,
Lao Peoples Democratic Republic, Marshall Islands, Myanmar, Northern Mariana
Islands, Malaysia, Mayotte, New Caledonia, Norfolk Island, Niue, Nauru, Pacific
Islands Trust Territory, Pitcairn Islands, Philippines, Palau, Papua New Guinea,
Democratic People’s Republic of Korea, French Polynesia, Singapore, Solomon
Islands, Seychelles, Thailand, Tokelau, Timor Leste, Tonga, Tuvalu, Viet Nam,
Vanuatu, Samoa
South Korea South Korea
Taiwan Taiwan
USA United States

200
Table 34. NGFS Phase IV Scenario Explorer IAM sector output

Sector Subsector Variable IAM131

Agricultural Demand Agricultural Demand (TOTAL) G M R

Agricultural Demand Crops Crops (TOTAL) G M R
Agricultural Demand Crops Energy (TOTAL) G M
Agricultural Demand Crops Feed G M
Agricultural Demand Crops Food M R
Agricultural Demand Crops Other G M R
Agricultural Demand Livestock Livestock (TOTAL) G M R
Agricultural Demand Livestock Food M R
Agricultural Demand Livestock Other G M R
Agricultural Agricultural Production (TOTAL) M R
Production
Agricultural Energy Energy (TOTAL) M R
Production
Agricultural Energy Crops (TOTAL) M R
Production
Agricultural Energy Residues R
Production
Agricultural Non-Energy Non-Energy (TOTAL) M R
Production
Agricultural Non-Energy Crops (TOTAL) M R
Production
Agricultural Non-Energy Livestock (TOTAL) M R
Production
Capacity Additions Electricity Biomass (TOTAL) G M R
Capacity Additions Electricity Biomass w/ CCS G M R
Capacity Additions Electricity Biomass w/o CCS G M R
Capacity Additions Electricity Coal (TOTAL) G M R
Capacity Additions Electricity Coal w/ CCS G M R
Capacity Additions Electricity Coal w/o CCS G M R
Capacity Additions Electricity Gas (TOTAL) G M R
Capacity Additions Electricity Gas w/ CCS G M R
Capacity Additions Electricity Gas w/o CCS G M R
Capacity Additions Electricity Geothermal G M R
Capacity Additions Electricity Hydro R
Capacity Additions Electricity Nuclear G M R
Capacity Additions Electricity Oil (TOTAL) G M R
Capacity Additions Electricity Oil w/o CCS G M

131 G = GCAM, M = MESSAGE-GLOBIOM, R = REMIND-MAgPIE

201
Sector Subsector Variable IAM131

Capacity Additions Electricity Solar (TOTAL) G M R

Capacity Additions Electricity Solar CSP G M R
Capacity Additions Electricity Solar PV G M R
Capacity Additions Electricity Wind (TOTAL) G M R
Capacity Additions Electricity Wind Offshore G M R
Capacity Additions Electricity Wind Onshore G M R
Capacity Electricity Electricity (TOTAL) G M R
Capacity Electricity Biomass (TOTAL) G M R
Capacity Electricity Biomass w/ CCS G M R
Capacity Electricity Biomass w/o CCS G M R
Capacity Electricity Coal (TOTAL) G M R
Capacity Electricity Coal w/ CCS G M R
Capacity Electricity Coal w/o CCS G M R
Capacity Electricity Gas (TOTAL) G M R
Capacity Electricity Gas w/ CCS G M R
Capacity Electricity Gas w/o CCS G M R
Capacity Electricity Geothermal G M R
Capacity Electricity Hydro G M R
Capacity Electricity Nuclear G M R
Capacity Electricity Oil (TOTAL) G M R
Capacity Electricity Oil w/o CCS G M R
Capacity Electricity Other R
Capacity Electricity Solar (TOTAL) G M R
Capacity Electricity Solar CSP G M R
Capacity Electricity Solar PV G M R
Capacity Electricity Storage R
Capacity Electricity Wind (TOTAL) G M R
Capacity Electricity Wind Offshore G M R
Capacity Electricity Wind Onshore G M R
Capacity Gases Gases (TOTAL) M R
Capacity Gases Biomass (TOTAL) M R
Capacity Gases Biomass w/o CCS M
Capacity Gases Coal(TOTAL) M R
Capacity Gases Coal w/o CCS M
Capacity Hydrogen Hydrogen (TOTAL) M R
Capacity Hydrogen Biomass (TOTAL) M R
Capacity Hydrogen Biomass w/ CCS M R
Capacity Hydrogen Biomass w/o CCS M R
Capacity Hydrogen Coal (TOTAL) M R

202
Sector Subsector Variable IAM131

Capacity Hydrogen Coal w/ CCS M R

Capacity Hydrogen Coal w/o CCS M R
Capacity Hydrogen Electricity (TOTAL) M R
Capacity Hydrogen Gas (TOTAL) M R
Capacity Hydrogen Gas w/ CCS M R
Capacity Hydrogen Gas w/o CCS M R
Capacity Liquids Liquids (TOTAL) M R
Capacity Liquids Biomass (TOTAL) M R
Capacity Liquids Biomass w/ CCS M
Capacity Liquids Biomass w/o CCS M
Capacity Liquids Coal (TOTAL) M R
Capacity Liquids Coal w/ CCS M
Capacity Liquids Coal w/o CCS M
Capacity Liquids Gas (TOTAL) M
Capacity Liquids Gas w/ CCS M
Capacity Liquids Gas w/o CCS M
Capacity Liquids Oil (TOTAL) M R
Capital Cost Electricity Biomass w/ CCS G R
Capital Cost Electricity Biomass w/o CCS G R
Capital Cost Electricity Coal w/ CCS G R
Capital Cost Electricity Coal w/o CCS G R
Capital Cost Electricity Gas w/ CCS G R
Capital Cost Electricity Gas w/o CCS G R
Capital Cost Electricity Geothermal G M R
Capital Cost Electricity Nuclear G M R
Capital Cost Electricity Solar CSP G R
Capital Cost Electricity Solar PV G M R
Capital Cost Electricity Wind Offshore G M R
Capital Cost Electricity Wind Onshore G M R
Carbon Sequestration CCS CCS (TOTAL) G M R
Carbon Sequestration CCS Biomass (TOTAL) G M R
Carbon Sequestration CCS Biomass Energy Demand Industry G R
Carbon Sequestration CCS Biomass Energy Supply (TOTAL) G M R
Carbon Sequestration CCS Biomass Energy Supply Electricity G M R
Carbon Sequestration CCS Biomass Energy Supply Hydrogen G M R
Carbon Sequestration CCS Biomass Energy Supply Liquids G M R
Carbon Sequestration CCS Fossil (TOTAL) G M R
Carbon Sequestration CCS Fossil Energy Demand Industry G
Carbon Sequestration CCS Fossil Energy Supply (TOTAL) G M R

203
Sector Subsector Variable IAM131

Carbon Sequestration CCS Fossil Energy Supply Electricity G M R

Carbon Sequestration CCS Fossil Energy Supply Hydrogen G M R
Carbon Sequestration CCS Fossil Energy Supply Liquids M R
Carbon Sequestration CCS Industrial Processes G M R
Carbon Sequestration Land Use Land Use (TOTAL) M R
Carbon Sequestration Land Use Afforestation M R
Emissions BC BC (TOTAL) G M R
Emissions BC AFOLU G M R
Emissions BC Energy (TOTAL) G M R
Emissions BC Energy Demand Industry G M R
Emissions BC Energy Demand Residential and Commercial G M R
Emissions BC Energy Demand Transportation G M R
Emissions BC Energy Supply G M R
Emissions C2F6 G R
Emissions CF4 G M R
Emissions CH4 CH4 (TOTAL) G M R
Emissions CH4 AFOLU G M R
Emissions CH4 Energy (TOTAL) G M
Emissions CH4 Energy Demand Industry G M
Emissions CH4 Energy Demand Residential and Commercial G M
Emissions CH4 Energy Demand Transportation G M
Emissions CH4 Energy Supply G M R
Emissions CH4 Other G
Emissions CO CO (TOTAL) G M R
Emissions CO AFOLU G M R
Emissions CO Energy (TOTAL) G M R
Emissions CO Energy Demand Industry G M R
Emissions CO Energy Demand Residential and Commercial G M R
Emissions CO Energy Demand Transportation G M R
Emissions CO Energy Supply G M R
Emissions CO2 CO2 (TOTAL) G M R
Emissions CO2 AFOLU G M R
Emissions CO2 Energy (TOTAL) G M R
Emissions CO2 Energy and Industrial Processes (TOTAL) M R
Emissions CO2 Energy Demand (TOTAL) G M R
Emissions CO2 Energy Demand Industry (TOTAL) G M R
Emissions CO2 Energy Demand Industry Cement G M R
Emissions CO2 Energy Demand Industry Chemicals (TOTAL) G M R
Emissions CO2 Energy Demand Industry Chemicals Ammonia G

204
Sector Subsector Variable IAM131

Emissions CO2 Energy Demand Industry Non-ferrous metals G M

Emissions CO2 Energy Demand Industry Other G R
Emissions CO2 Energy Demand Industry Steel G M R
Emissions CO2 Energy Demand Residential and Commercial G M R
(TOTAL)
Emissions CO2 Energy Demand Residential and Commercial G
(Commercial)
Emissions CO2 Energy Demand Residential and Commercial G
(Residential)
Emissions CO2 Energy Demand Transportation (TOTAL) G M R
Emissions CO2 Energy Demand Transportation Freight G
Emissions CO2 Energy Demand Transportation Passenger G
Emissions CO2 Energy Demand
Transportation|Aviation|Passenger
Emissions CO2 Energy Demand Maritime|Freight
Emissions CO2 Energy Demand Transportation Rail Freight R
Emissions CO2 Energy Demand Transportation Rail Passenger R
Emissions CO2 Energy Demand Transportation Road R
Emissions CO2 Energy Demand Transportation Road Freight R
Energy Demand Transportation Road
Emissions CO2
Passenger R
Energy Demand Transportation Road
Emissions CO2
Passenger Bus R
Energy Demand Transportation Road
Emissions CO2
Passenger LDV R
Emissions CO2 Energy Demand Transportation Rail Freight R
Emissions CO2 Energy Supply (TOTAL) G M R
Emissions CO2 Energy Supply Electricity G M R
Emissions CO2 Energy Supply Gases G M R
Emissions CO2 Energy Supply Heat G M R
Emissions CO2 Energy Supply Liquids G M R
Emissions CO2 Energy Supply Other Sector R
Emissions CO2 Energy Supply Solids G M R
Emissions CO2 Energy and Industrial Processes G
Emissions CO2 Industrial Processes G M R
Emissions F-Gases G M R
Emissions HFC HFC (TOTAL) G M R
Emissions HFC HFC125 G M
Emissions HFC HFC134a G M
Emissions HFC HFC143a G M
Emissions HFC HFC227ea G M
Emissions HFC HFC23 G M
Emissions HFC HFC245fa G M

205
Sector Subsector Variable IAM131

Emissions HFC HFC32 G M

Emissions Kyoto Gases Kyoto Gases (TOTAL) G M R
Emissions Kyoto Gases AFOLU G M R
Emissions Kyoto Gases Cement G
Emissions Kyoto Gases Chemicals G
Emissions Kyoto Gases Electricity G
Emissions Kyoto Gases Industry G R
Emissions Kyoto Gases Other G
Emissions Kyoto Gases Other Energy Supply G
Emissions Kyoto Gases Other Industry G
Emissions Kyoto Gases Residential and Commercial G R
Emissions Kyoto Gases Steel G
Emissions Kyoto Gases Supply G R
Emissions Kyoto Gases Transportation G R
Emissions N2O N2O (TOTAL) G M R
Emissions N2O AFOLU G M R
Emissions N2O Energy G M
Emissions N2O Other G
Emissions NH3 NH3 (TOTAL) G M R
Emissions NH3 AFOLU G M R
Emissions NH3 Energy (TOTAL) G M R
Emissions NH3 Energy Demand Industry G M R
Emissions NH3 Energy Demand Residential and Commercial G M R
Emissions NH3 Energy Demand Transportation G M R
Emissions NH3 Energy Supply G M R
Emissions NH3 Other R
Emissions NH3 Waste R
Emissions NOx NOx (TOTAL) G M R
Emissions NOx AFOLU G M R
Emissions NOx Energy (TOTAL) G M R
Emissions NOx Energy Demand Industry G M R
Emissions NOx Energy Demand Residential and Commercial G M R
Emissions NOx Energy Demand Transportation G M R
Emissions NOx Energy Supply G M R
Emissions NOx Other R
Emissions NOx Waste R
Emissions OC OC (TOTAL) G M R
Emissions OC AFOLU G M R
Emissions OC Energy (TOTAL) G M R

206
Sector Subsector Variable IAM131

Emissions OC Energy Demand Industry G M R

Emissions OC Energy Demand Residential and Commercial G M R
Emissions OC Energy Demand Transportation G M R
Emissions OC Energy Supply G M R
Emissions OC Other R
Emissions OC Waste R
Emissions PFC G
Emissions SF6 G M R
Emissions Sulfur Sulfur (TOTAL) G M R
Emissions Sulfur AFOLU G M R
Emissions Sulfur Energy (TOTAL) G M R
Emissions Sulfur Energy Demand Industry G M R
Emissions Sulfur Energy Demand Residential and Commercial G M R
Emissions Sulfur Energy Demand Transportation G M R
Emissions Sulfur Energy Supply G M R
Emissions Sulfur Other R
Emissions Sulfur Waste R
Emissions VOC VOC (TOTAL) G M R
Emissions VOC AFOLU G M R
Emissions VOC Energy (TOTAL) G M R
Emissions VOC Energy Demand Industry G M R
Emissions VOC Energy Demand Residential and Commercial G M R
Emissions VOC Energy Demand Transportation G M R
Emissions VOC Energy Supply G M R
Emissions VOC Other R
Emissions VOC Waste R
Fertilizer Use Nitrogen Nitrogen (TOTAL) M
Fertilizer Use Phosphorus Phosphorus (TOTAL) M
Energy Service Residential and Commercial Floor Space G
Energy Service Residential and Commercial Residential Floor Space G
Energy Service Transportation Aviation G
Energy Service Transportation Freight (TOTAL) G
Energy Service Transportation Freight International Shipping G
Energy Service Transportation Freight Road G
Energy Service Transportation Passenger (TOTAL) G
Energy Service Transportation Passenger Aviation G
Energy Service Transportation Passenger Bicycling and Walking G
Energy Service Transportation Rail G
Energy Service Transportation Road G

207
Sector Subsector Variable IAM131

Final Energy Final Energy (TOTAL) G M R

Final Energy Electricity G M R
Final Energy Gases Gases (TOTAL) G M R
Final Energy Heat G M R
Final Energy Hydrogen G M R
Final Energy Industry Industry (TOTAL) G M R
Final Energy Industry Cement (TOTAL) G M R
Final Energy Industry Cement Electricity G M R
Final Energy Industry Cement Gases M R
Final Energy Industry Cement Hydrogen M R
Final Energy Industry Cement Liquids M R
Final Energy Industry Cement Liquids Bioenergy M
Final Energy Industry Cement Liquids Fossil M
Final Energy Industry Cement Solids M R
Final Energy Industry Cement Solids Bioenergy M
Final Energy Industry Cement Solids Fossil M
Final Energy Industry Chemicals High value chemicals M
Final Energy Industry Chemicals High value chemicals Electricity M
Final Energy Industry Chemicals High value chemicals Gases M
Final Energy Industry Chemicals High value chemicals Heat M
Final Energy Industry Chemicals High value chemicals Hydrogen M
Final Energy Industry Chemicals High value chemicals Liquids M
Final Energy Industry Chemicals High value chemicals Liquids M
Bioenergy
Final Energy Industry Chemicals High value chemicals Liquids Fossil M
Final Energy Industry Chemicals High value chemicals Solids M
Final Energy Industry Chemicals High value chemicals Solids M
Bioenergy
Final Energy Industry Chemicals High value chemicals Solids Fossil M
Final Energy Industry Chemicals (TOTAL) G R
Final Energy Industry Chemicals Ammonia (TOTAL) G
Final Energy Industry Chemicals Ammonia Gases G R
Final Energy Industry Chemicals Ammonia Hydrogen G R
Final Energy Industry Chemicals Ammonia Liquids G R
Final Energy Industry Chemicals Ammonia Solids (TOTAL) G R
Final Energy Industry Chemicals Ammonia Solids Fossil G
Final Energy Industry Chemicals Electricity G
Final Energy Industry Chemicals Gases G
Final Energy Industry Chemicals Heat G
Final Energy Industry Chemicals Hydrogen G

208
Sector Subsector Variable IAM131

Final Energy Industry Chemicals Liquids G

Final Energy Industry Chemicals Solids (TOTAL) G
Final Energy Industry Chemicals Solids Bioenergy G
Final Energy Industry Chemicals Solids Fossil G
Final Energy Industry Electricity (TOTAL) G M R
Final Energy Industry Electricity Share R
Final Energy Industry Gases (TOTAL) G M R
Final Energy Industry Gases Bioenergy R
Final Energy Industry Heat G M R
Final Energy Industry Hydrogen G M R
Final Energy Industry Liquids (TOTAL) G M R
Final Energy Industry Liquids Bioenergy M
Final Energy Industry Liquids Oil M
Final Energy Industry Non-ferrous metals (TOTAL) G M
Final Energy Industry Non-ferrous metals Electricity G M
Final Energy Industry Non-ferrous metals Gases G M
Final Energy Industry Non-ferrous metals Liquids G M
Final Energy Industry Non-ferrous metals Solids (TOTAL) G M
Final Energy Industry Non-ferrous metals Solids Bioenergy G M
Final Energy Industry Non-ferrous metals Solids Fossil G M
Final Energy Industry Other (TOTAL) M
Final Energy Industry Other Electicity R
Final Energy Industry Other Gases R
Final Energy Industry Other Heat R
Final Energy Industry Other Hydrogen R
Final Energy Industry Other Liquids R
Final Energy Industry Other Solids R
Final Energy Industry Solids (TOTAL) G M R
Final Energy Industry Solids Biomass G M R
Final Energy Industry Solids Coal G M R
Final Energy Industry Steel (TOTAL) G M R
Final Energy Industry Steel Electricity G M R
Final Energy Industry Steel Gases G M R
Final Energy Industry Steel Hydrogen G M R
Final Energy Industry Steel Liquids G M R
Final Energy Industry Steel Solids (TOTAL) G M R
Final Energy Industry Steel Solids Bioenergy G M
Final Energy Industry Steel Solids Fossil G M
Final Energy Liquids Liquids (TOTAL) G M R

209
Sector Subsector Variable IAM131

Final Energy Other Sector Other Sector (TOTAL) R

Final Energy Other Sector Electicity R
Final Energy Other Sector Gases R
Final Energy Other Sector Heat R
Final Energy Other Sector Hydrogen R
Final Energy Other Sector Liquids R
Final Energy Non-Energy Use Non-Energy Use (TOTAL) G M
Final Energy Non-Energy Use Biomass G M
Final Energy Non-Energy Use Coal G M
Final Energy Non-Energy Use Gas G M
Final Energy Non-Energy Use Oil G M
Final Energy Residential and Commercial Residential and Commercial (TOTAL) G M R
Final Energy Residential and Commercial Commercial Cooling G
Final Energy Residential and Commercial Commercial Electricity G
Final Energy Residential and Commercial Commercial Gases G
Final Energy Residential and Commercial Commercial Heat G
Final Energy Residential and Commercial Commercial Heating Space G
Final Energy Residential and Commercial Commercial Hydrogen G
Final Energy Residential and Commercial Commercial Liquids G
Final Energy Residential and Commercial Commercial Solids (TOTAL) G
Final Energy Residential and Commercial Commercial Solids Biomass G
Final Energy Residential and Commercial Commercial Solids Coal G
Final Energy Residential and Commercial Cooling G
Final Energy Residential and Commercial Electricity G M R
Final Energy Residential and Commercial Gases (TOTAL) G M R
Final Energy Residential and Commercial Gases Biomass R
Final Energy Residential and Commercial Gases Natural gas R
Final Energy Residential and Commercial Heat G M R
Final Energy Residential and Commercial Heating Space G
Final Energy Residential and Commercial Hydrogen G M R
Final Energy Residential and Commercial Liquids (TOTAL) G M R
Final Energy Residential and Commercial Liquids Biomass R
Final Energy Residential and Commercial Liquids Oil R
Final Energy Residential and Commercial Residential Cooling G
Final Energy Residential and Commercial Residential Electricity G R
Final Energy Residential and Commercial Residential Gases G R
Final Energy Residential and Commercial Gases Biomass R
Final Energy Residential and Commercial Gases Natural Gas R
Final Energy Residential and Commercial Residential Heat G R

210
Sector Subsector Variable IAM131

Final Energy Residential and Commercial Residential Heating Space G

Final Energy Residential and Commercial Residential Hydrogen G R
Final Energy Residential and Commercial Residential Liquids G R
Final Energy Residential and Commercial Solids (TOTAL) G M R
Final Energy Residential and Commercial Solids Biomass (TOTAL) G M R
Final Energy Residential and Commercial Solids Biomass Traditional G M R
Final Energy Residential and Commercial Solids Coal G M R
Final Energy Solids Solids (TOTAL) G M R
Final Energy Solids Biomass (TOTAL) G M R
Final Energy Solids Biomass Traditional G M R
Final Energy Solids Coal G M R
Final Energy Transportation Transportation (TOTAL) G M R
Final Energy Transportation Aviation (TOTAL) G
Final Energy Transportation Aviation Passenger G R
Final Energy Transportation Electricity G M R
Final Energy Transportation Freight (TOTAL) G R
Final Energy Transportation Freight Electricity G R
Final Energy Transportation Freight Gases G R
Final Energy Transportation Freight Hydrogen G R
Final Energy Transportation Freight Liquids G R
Final Energy Transportation Freight Other G
Final Energy Transportation Gases (TOTAL) G M R
Final Energy Transportation Gases Bioenergy R
Final Energy Transportation Gases Fossil R
Final Energy Transportation Hydrogen G M R
Final Energy Transportation Liquids (TOTAL) G M R
Final Energy Transportation Liquids Bioenergy R
Final Energy Transportation Liquids Coal R
Final Energy Transportation Liquids Natural Gas R
Final Energy Transportation Liquids Oil R
Final Energy Transportation Maritime (TOTAL) G R
Final Energy Transportation Maritime Freight G R
Final Energy Transportation Other G M R
Final Energy Transportation Passenger (TOTAL) G R
Final Energy Transportation Passenger Electricity G R
Final Energy Transportation Passenger Gases G R
Final Energy Transportation Passenger Hydrogen G R
Final Energy Transportation Passenger Liquids G R
Final Energy Transportation Rail (TOTAL) G R

211
Sector Subsector Variable IAM131

Final Energy Transportation Rail Freight G R

Final Energy Transportation Rail Passenger G R
Final Energy Transportation Road Freight (TOTAL) G R
Final Energy Transportation Road Freight Electric G R
Final Energy Transportation Road Freight FC G R
Final Energy Transportation Road Freight ICE G R
Final Energy Transportation Road Passenger (TOTAL) G R
Final Energy Transportation Road Passenger 2W&3W G
Final Energy Transportation Road Passenger Bus G R
Final Energy Transportation Road Passenger LDV G R
Food Demand Food Demand (TOTAL) M R
Food Demand Crops (TOTAL) M R
Food Demand Livestock (TOTAL) M R
Forestry Demand Roundwood Roundwood (TOTAL) M
Forestry Demand Roundwood Industrial Roundwood M
Forestry Demand Roundwood Wood Fuel M
Forestry Production Roundwood Roundwood (TOTAL) M
Forestry Production Roundwood Industrial Roundwood M
Forestry Production Roundwood Wood Fuel M
GDP MER G R
GDP PPP G R
GDP MER Counterfactual without damage M
GDP PPP Counterfactual without damage M
GDP PPP including high chronic physical risk M
damage estimate
GDP PPP including medium chronic physical risk M
damage estimate
Investment Energy Supply Energy Supply (TOTAL) R
Investment Energy Supply CO2 Transport and Storage G M R
Investment Energy Supply Electricity (TOTAL) G M R
Investment Energy Supply Electricity Biomass (TOTAL) G M R
Investment Energy Supply Electricity Biomass w/ CCS G M R
Investment Energy Supply Electricity Biomass w/o CCS G M R
Investment Energy Supply Electricity Coal (TOTAL) G M R
Investment Energy Supply Electricity Coal w/ CCS G M R
Investment Energy Supply Electricity Coal w/o CCS G M R
Investment Energy Supply Electricity Gas (TOTAL) G M R
Investment Energy Supply Electricity Gas w/ CCS G M R
Investment Energy Supply Electricity Gas w/o CCS G M R
Investment Energy Supply Electricity Geothermal G M R

212
Sector Subsector Variable IAM131

Investment Energy Supply Electricity Hydro G M R

Investment Energy Supply Electricity Non-Biomass Renewables G M R
Investment Energy Supply Electricity Nuclear G M R
Investment Energy Supply Electricity Oil (TOTAL) G M R
Investment Energy Supply Electricity Oil w/o CCS G M R
Investment Energy Supply Electricity Solar G M R
Investment Energy Supply Electricity Transmission and Distribution G M R
Investment Energy Supply Electricity Wind G M
Investment Energy Supply Extraction Coal G M
Investment Energy Supply Extraction Fossil G M
Investment Energy Supply Extraction Gas G M
Investment Energy Supply Extraction Oil G M
Investment Energy Supply Extraction Uranium M
Investment Energy Supply Heat M R
Investment Energy Supply Hydrogen (TOTAL) M R
Investment Energy Supply Hydrogen Fossil M R
Investment Energy Supply Hydrogen Other M R
Investment Energy Supply Hydrogen Biomass R
Investment Energy Supply Hydrogen Electricity R
Investment Energy Supply Hydrogen Renewable M R
Investment Energy Supply Liquids (TOTAL) M R
Investment Energy Supply Liquids Biomass M R
Investment Energy Supply Liquids Coal and Gas M R
Investment Energy Supply Liquids Oil M R
Investment Energy Supply Other M R
Land Cover Land Cover (TOTAL) G M R
Land Cover Built-up Area G R
Land Cover Cropland Cropland (TOTAL) G M R
Land Cover Cropland Cereals M R
Land Cover Cropland Cropland Energy Crops G M R
Land Cover Forest Forest (TOTAL) G M R
Land Cover Forest Afforestation and Reforestation M r
Land Cover Forest Managed G M R
Land Cover Forest Natural Forest G M R
Land Cover Forest Secondary R
Land Cover Other Land G M R
Land Cover Pasture G M R
Population Population (TOTAL) G M R
Population Population Rural M

213
Sector Subsector Variable IAM131

Population Population Urban M

Price Agriculture Corn Index G R
Price Agriculture Non-Energy Crops Index G M R
Price Agriculture Soybean Index G R
Price Agriculture Wheat Index G R
Price Carbon Carbon (TOTAL) G M R
Price Carbon Demand Industry G R
Price Carbon Demand Residential and Commercial G R
Price Carbon Demand Transportation G R
Price Carbon Supply G R
Price Final Energy Industry Electricity G R
Price Final Energy Industry Gases (TOTAL) G R
Price Final Energy Industry Gases Bioenergy R
Price Final Energy Industry Gases Fossil R
Price Final Energy Industry Hydrogen R
Price Final Energy Industry Liquids (TOTAL) G R
Price Final Energy Industry Liquids Bioenergy R
Price Final Energy Industry Liquids Fossil synfuel R
Price Final Energy Industry Solids Bioenergy R
Price Final Energy Industry Solids Coal R
Price Final Energy Industry Electricity G
Price Final Energy Transportation Electricity G R
Price Final Energy Transportation Gases G R
Price Final Energy Transportation Liquids (TOTAL) G R
Price Final Energy Transportation Liquids Fossil synfuel R
Price Final Energy Transportation Hydrogen R
Price Final Energy Residential and Commercial Residential M R
Electricity
Price Final Energy Residential and Commercial Residential M R
Electricity Index
Price Final Energy Residential and Commercial Residential Gases M R
Natural Gas
Price Final Energy Residential and Commercial Residential Gases M R
Natural Gas Index
Price Final Energy Residential and Commercial Residential Liquids M R
Biomass
Price Final Energy Residential and Commercial Residential Liquids M R
Biomass Index
Price Final Energy Residential and Commercial Residential Liquids M R
Oil
Price Final Energy Residential and Commercial Residential Liquids M R
Oil Index

214
Sector Subsector Variable IAM131

Price Final Energy Residential and Commercial Residential Solids R

Biomass
Price Final Energy Residential and Commercial Residential Solids R
Biomass Index
Price Final Energy Residential and Commercial Residential Solids R
Coal
Price Final Energy Residential and Commercial Residential Solids R
Coal Index
Price Industry Cement M
Price Primary Energy Biomass (TOTAL) G M R
Price Primary Energy Biomass Index G M R
Price Primary Energy Coal (TOTAL) G M R
Price Primary Energy Coal Index G M R
Price Primary Energy Gas (TOTAL) G M R
Price Primary Energy Gas Index G M R
Price Primary Energy Oil (TOTAL) G M R
Price Primary Energy Oil Index G M R
Price Secondary Energy Electricity (TOTAL) G M R
Price Secondary Energy Electricity Index G M R
Price Secondary Energy Gases Natural Gas (TOTAL) G M R
Price Secondary Energy Gases Natural Gas Index G M R
Price Secondary Energy Hydrogen (TOTAL) G M R
Price Secondary Energy Liquids (TOTAL) G M R
Price Secondary Energy Liquids Biomass (TOTAL) G M R
Price Secondary Energy Liquids Biomass Index G M R
Price Secondary Energy Liquids Oil (TOTAL) G M R
Price Secondary Energy Liquids Oil Index G M R
Price Secondary Energy Solids Coal (TOTAL) G M R
Price Secondary Energy Solids Coal Index G M R
Primary Energy Primary Energy (TOTAL) G M R
Primary Energy Biomass Biomass (TOTAL) G M R
Primary Energy Biomass 1st Generation M R
Primary Energy Biomass Electricity (TOTAL) M R
Primary Energy Biomass Electricity w/CCS M R
Primary Energy Biomass Electricity w/o CCS M R
Primary Energy Biomass Energy Crops M R
Primary Energy Biomass Residues M R
Primary Energy Biomass Modern G M R
Primary Energy Biomass Traditional G M R
Primary Energy Biomass Gases R

215
Sector Subsector Variable IAM131

Primary Energy Biomass Heat R

Primary Energy Biomass Hydrogen R
Primary Energy Biomass Liquids R
Primary Energy Biomass Solids R
Primary Energy Coal Coal (TOTAL) G M R
Primary Energy Coal Coal Electricity M R
Primary Energy Coal Coal Electricity w/CCS M R
Primary Energy Coal Coal Electricity w/o CCS M R
Primary Energy Coal w/ CCS G M R
Primary Energy Coal w/o CCS G M R
Primary Energy Coal Gases R
Primary Energy Coal Heat R
Primary Energy Coal Hydrogen R
Primary Energy Coal Liquids R
Primary Energy Coal Solids R
Primary Energy Fossil Fossil (TOTAL) G M R
Primary Energy Fossil w/ CCS G M R
Primary Energy Fossil w/o CCS G M R
Primary Energy Gas Gas (TOTAL) G M R
Primary Energy Gas Gas Electricity G M R
Primary Energy Gas Gas Electricity w/CCS G M R
Primary Energy Gas Gas Electricity w/o CCS G M R
Primary Energy Gas w/ CCS G M R
Primary Energy Gas w/o CCS G M R
Primary Energy Gas Gases R
Primary Energy Gas Heat R
Primary Energy Gas Hydrogen R
Primary Energy Gas Liquids R
Primary Energy Gas Solids R
Primary Energy Geothermal G R
Primary Energy Hydro G R
Primary Energy Non-Biomass Renewables G M R
Primary Energy Non-Biomass Renewables Geothermal M R
Primary Energy Non-Biomass Renewables Hydro M R
Primary Energy Non-Biomass Renewables Ocean M
Primary Energy Non-Biomass Renewables Solar M R
Primary Energy Non-Biomass Renewables Wind M R
Primary Energy Nuclear G M R
Primary Energy Oil Oil (TOTAL) G M R

216
Sector Subsector Variable IAM131

Primary Energy Oil Oil Electricity M

Primary Energy Oil Oil Electricity w/ CCS M
Primary Energy Oil Oil Electricity w/o CCS M
Primary Energy Oil w/ CCS G
Primary Energy Oil w/o CCS G M R
Primary Energy Solar G R
Primary Energy Wind G R
Production Cement G M R
Production Chemicals M
Production Non-ferrous metals G M
Production Steel G M R
Revenue Government Tax Carbon (TOTAL) G M R
Revenue Government Tax Carbon Demand Industry G M R
Revenue Government Tax Carbon Demand Residential and G M R
Commercial
Revenue Government Tax Carbon Demand Transportation G M R
Revenue Government Tax Carbon Supply G M R
Secondary Energy Secondary Energy (TOTAL) G R
Secondary Energy Electricity Electricity (TOTAL) G M R
Secondary Energy Electricity Biomass (TOTAL) G M R
Secondary Energy Electricity Biomass w/ CCS G M R
Secondary Energy Electricity Biomass w/o CCS G M R
Secondary Energy Electricity Coal (TOTAL) G M R
Secondary Energy Electricity Coal w/ CCS G M R
Secondary Energy Electricity Coal w/o CCS G M R
Secondary Energy Electricity Fossil M
Secondary Energy Electricity Fossil w/ CCS M
Secondary Energy Electricity Fossil w/o CCS M
Secondary Energy Electricity Gas (TOTAL) G M R
Secondary Energy Electricity Gas w/ CCS G M R
Secondary Energy Electricity Gas w/o CCS G M R
Secondary Energy Electricity Geothermal G M R
Secondary Energy Electricity Hydro G M R
Secondary Energy Electricity Non-Biomass Renewables G M R
Secondary Energy Electricity Nuclear G M R
Secondary Energy Electricity Oil (TOTAL) G M R
Secondary Energy Electricity Oil w/o CCS G M R
Secondary Energy Electricity Solar (TOTAL) G M R
Secondary Energy Electricity Solar CSP G M R

217
Sector Subsector Variable IAM131

Secondary Energy Electricity Solar PV G M R

Secondary Energy Electricity Wind (TOTAL) G M R
Secondary Energy Electricity Wind Offshore G M R
Secondary Energy Electricity Wind Onshore G M R
Secondary Energy Gases Gases (TOTAL) G M R
Secondary Energy Gases Biomass G M R
Secondary Energy Gases Coal G M R
Secondary Energy Gases Natural Gas G M R
Secondary Energy Gases Other M R
Secondary Energy Heat Heat (TOTAL) M R
Secondary Energy Heat Heat Biomass M R
Secondary Energy Heat Heat Coal M R
Secondary Energy Heat Heat Gas M R
Secondary Energy Heat Heat Geothermal M R
Secondary Energy Heat Heat Oil M
Secondary Energy Heat Heat Other M
Secondary Energy Hydrogen Hydrogen (TOTAL) G M R
Secondary Energy Hydrogen Biomass (TOTAL) G M R
Secondary Energy Hydrogen Biomass w/ CCS G M R
Secondary Energy Hydrogen Biomass w/o CCS G M R
Secondary Energy Hydrogen Coal (TOTAL) G M R
Secondary Energy Hydrogen Coal w/o CCS G M R
Secondary Energy Hydrogen Coal w/o CCS G M R
Secondary Energy Hydrogen Electricity G M R
Secondary Energy Hydrogen Fossil (TOTAL) G M R
Secondary Energy Hydrogen Fossil w/ CCS G M R
Secondary Energy Hydrogen Fossil w/o CCS G M R
Secondary Energy Hydrogen Gas (TOTAL) G M R
Secondary Energy Hydrogen Gas w/ CCS G M R
Secondary Energy Hydrogen Gas w/o CCS G M R
Secondary Energy Liquids Liquids (TOTAL) G M R
Secondary Energy Liquids Biomass (TOTAL) G M R
Secondary Energy Liquids Biomass w/ CCS G M R
Secondary Energy Liquids Biomass w/o CCS G M R
Secondary Energy Liquids Coal (TOTAL) G M R
Secondary Energy Liquids Coal w/o CCS G M R
Secondary Energy Liquids Fossil (TOTAL) G M R
Secondary Energy Liquids Fossil w/o CCS G M R
Secondary Energy Liquids Gas (TOTAL) G M R

218
 Sector Subsector Variable IAM131

Secondary Energy Liquids Gas w/ CCS R

Secondary Energy Liquids Gas w/o CCS R
Secondary Energy Liquids Oil G M R
Secondary Energy Solids Solids (TOTAL) G M R
Secondary Energy Solids Biomass G M R
Secondary Energy Solids Coal G M R
Trade Primary Energy Biomass Volume M R
Trade Primary Energy Coal Volume M R
Trade Primary Energy Gas Volume M R
Trade Primary Energy Oil Volume M R
Water Consumption (TOTAL) M
Water Consumption Irrigation M R
Water Withdrawal Irrigation M
Yield Cereal M R
Yield Oil crops M R
Yield Sugar crops M R

Table 35. Mapping of IAM regions and downscaled countries

ISO MESSAGEix-GLOBIOM 1.1- GCAM 6.0 NGFS REMIND-MAgPIE 3.2-4.6

R12

ABW Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
AFG South Asia South Asia Other Asia
AGO Sub-Saharan Africa Africa_Southern Sub-Saharan Africa
AIA Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
ALA Western Europe EU 28
ALB Eastern Europe Europe_Non_EU Non-EU28 Europe
AND Western Europe EU-15 Non-EU28 Europe
ANT Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
ARE Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
ARG Latin America and the Caribbean Argentina Latin America and the Caribbean
ARM Former Soviet Union Central Asia Countries from the Reforming
Economies of the Former Soviet
Union
ASM Other Pacific Asia Southeast Asia Other Asia
ATA Latin America and the Caribbean
ATF Other Asia
ATG Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
AUS Pacific OECD Australia_NZ Canada, NZ, Australia
AUT Western Europe EU-15 EU 28
AZE Former Soviet Union Central Asia Countries from the Reforming
Economies of the Former Soviet
Union

219
ISO MESSAGEix-GLOBIOM 1.1- GCAM 6.0 NGFS REMIND-MAgPIE 3.2-4.6
R12

BDI Sub-Saharan Africa Africa_Eastern Sub-Saharan Africa

BEL Western Europe EU-15 EU 28
BEN Sub-Saharan Africa Africa_Western Sub-Saharan Africa
BES Latin America and the Caribbean
BFA Sub-Saharan Africa Africa_Western Sub-Saharan Africa
BGD South Asia South Asia Other Asia
BGR Eastern Europe EU-12 EU 28
BHR Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
BHS Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
BIH Eastern Europe Europe_Non_EU Non-EU28 Europe
BLM Western Europe Latin America and the Caribbean
BLR Former Soviet Union Europe_Eastern Countries from the Reforming
Economies of the Former Soviet
Union
BLZ Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
BMU Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
BOL Latin America and the Caribbean South America_Southern Latin America and the Caribbean
BRA Latin America and the Caribbean Brazil Latin America and the Caribbean
BRB Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
BRN Other Pacific Asia Southeast Asia Other Asia
BTN South Asia South Asia Other Asia
BVT Latin America and the Caribbean
BWA Sub-Saharan Africa Africa_Southern Sub-Saharan Africa
CAF Sub-Saharan Africa Africa_Western Sub-Saharan Africa
CAN North America Canada Canada, NZ, Australia
CCK Other Pacific Asia Southeast Asia Other Asia
CHE Western Europe European Free Trade Association Non-EU28 Europe
CHL Latin America and the Caribbean South America_Southern Latin America and the Caribbean
CHN China China China
CIV Sub-Saharan Africa Africa_Western Sub-Saharan Africa
CMR Sub-Saharan Africa Africa_Western Sub-Saharan Africa
COD Sub-Saharan Africa Africa_Western Sub-Saharan Africa
COG Sub-Saharan Africa Africa_Western Sub-Saharan Africa
COK Other Pacific Asia Southeast Asia Other Asia
COL Latin America and the Caribbean Colombia Latin America and the Caribbean
COM Sub-Saharan Africa Africa_Eastern Sub-Saharan Africa
CPV Sub-Saharan Africa Africa_Western Sub-Saharan Africa
CRI Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
CUB Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
CUW Central America and Caribbean Latin America and the Caribbean
CXR Pacific OECD Southeast Asia Other Asia
CYM Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean

220
ISO MESSAGEix-GLOBIOM 1.1- GCAM 6.0 NGFS REMIND-MAgPIE 3.2-4.6
R12

CYP Western Europe EU-12 EU 28

CZE Eastern Europe EU-12 EU 28
DEU Western Europe EU-15 EU 28
DJI Sub-Saharan Africa Africa_Eastern Sub-Saharan Africa
DMA Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
DNK Western Europe EU-15 EU 28
DOM Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
DZA Middle East and North Africa Africa_Northern Middle East, North Africa, Central
Asia
ECU Latin America and the Caribbean South America_Southern Latin America and the Caribbean
EGY Middle East and North Africa Africa_Northern Middle East, North Africa, Central
Asia
ERI Sub-Saharan Africa Africa_Eastern Sub-Saharan Africa
ESH Sub-Saharan Africa Africa_Northern Middle East, North Africa, Central
Asia
ESP Western Europe EU-15 EU 28
EST Eastern Europe EU-12 EU 28
ETH Sub-Saharan Africa Africa_Eastern Sub-Saharan Africa
FIN Western Europe EU-15 EU 28
FJI Other Pacific Asia Southeast Asia Other Asia
FLK Latin America and the Caribbean EU-15 Latin America and the Caribbean
FRA Western Europe EU-15 EU 28
FRO Western Europe EU-15 EU 28
FSM Other Pacific Asia Southeast Asia Other Asia
GAB Sub-Saharan Africa Africa_Western Sub-Saharan Africa
GBR Western Europe EU-15 EU 28
GEO Former Soviet Union Central Asia Countries from the Reforming
Economies of the Former Soviet
Union
GGY Western Europe EU 28
GHA Sub-Saharan Africa Africa_Western Sub-Saharan Africa
GIB Western Europe EU-15 EU 28
GIN Sub-Saharan Africa Africa_Western Sub-Saharan Africa
GLP Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
GMB Sub-Saharan Africa Africa_Western Sub-Saharan Africa
GNB Sub-Saharan Africa Africa_Western Sub-Saharan Africa
GNQ Sub Saharan Africa Africa_Western Sub-Saharan Africa
GRC Western Europe EU-15 EU 28
GRD Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
GRL Western Europe EU-15 Non-EU28 Europe
GTM Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
GUF Latin America and the Caribbean South America_Northern Latin America and the Caribbean
GUM North America Southeast Asia Other Asia
GUY Latin America and the Caribbean South America_Northern Latin America and the Caribbean

221
ISO MESSAGEix-GLOBIOM 1.1- GCAM 6.0 NGFS REMIND-MAgPIE 3.2-4.6
R12

HKG China China China

HMD Pacific OECD Canada, NZ, Australia
HND Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
HRV Eastern Europe Europe_Non_EU EU 28
HTI Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
HUN Eastern Europe EU-12 EU 28
IDN Other Pacific Asia Indonesia Other Asia
IMN Western Europe EU-15 EU 28
IND South Asia India India
IOT Western Europe Other Asia
IRL Western Europe EU-15 EU 28
IRN Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
IRQ Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
ISL Western Europe European Free Trade Association Non-EU28 Europe
ISR Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
ITA Western Europe EU-15 EU 28
JAM Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
JEY Western Europe EU 28
JOR Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
JPN Pacific OECD Japan Japan
KAZ Former Soviet Union Central Asia Countries from the Reforming
Economies of the Former Soviet
Union
KEN Sub-Saharan Africa Africa_Eastern Sub-Saharan Africa
KGZ Former Soviet Union Central Asia Countries from the Reforming
Economies of the Former Soviet
Union
KHM Rest Centrally Planned Asia Southeast Asia Other Asia
KIR Other Pacific Asia Southeast Asia Other Asia
KNA Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
KOR Other Pacific Asia South Korea Other Asia
KWT Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
LAO Rest Centrally Planned Asia Southeast Asia Other Asia
LBN Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
LBR Sub-Saharan Africa Africa_Western Sub-Saharan Africa
LBY Middle East and North Africa Africa_Northern Middle East, North Africa, Central
Asia
LCA Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
LIE Western Europe European Free Trade Association Non-EU28 Europe
LKA South Asia South Asia Other Asia
LSO Sub-Saharan Africa Africa_Southern Sub-Saharan Africa

222
ISO MESSAGEix-GLOBIOM 1.1- GCAM 6.0 NGFS REMIND-MAgPIE 3.2-4.6
R12

LTU Eastern Europe EU-12 EU 28

LUX Western Europe EU-15 EU 28
LVA Eastern Europe EU-12 EU 28
MAC Rest Centrally Planned Asia China China
MAF Western Europe Latin America and the Caribbean
MAR Middle East and North Africa Africa_Northern Middle East, North Africa, Central
Asia
MCO Western Europe EU-15 Non-EU28 Europe
MDA Former Soviet Union Europe_Eastern Countries from the Reforming
Economies of the Former Soviet
Union
MDG Sub-Saharan Africa Africa_Eastern Sub-Saharan Africa
MDV South Asia South Asia Other Asia
MEX Latin America and the Caribbean Mexico Latin America and the Caribbean
MHL Other Pacific Asia Southeast Asia Other Asia
MKD Eastern Europe Europe_Non_EU Non-EU28 Europe
MLI Sub-Saharan Africa Africa_Western Sub-Saharan Africa
MLT Western Europe EU-12 EU 28
MMR Other Pacific Asia Southeast Asia Other Asia
MNE Eastern Europe Europe_Non_EU Non-EU28 Europe
MNG Rest Centrally Planned Asia Central Asia Other Asia
MNP Other Pacific Asia Southeast Asia Other Asia
MOZ Sub-Saharan Africa Africa_Southern Sub-Saharan Africa
MRT Sub-Saharan Africa Africa_Western Sub-Saharan Africa
MSR Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
MTQ Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
MUS Sub-Saharan Africa Africa_Eastern Sub-Saharan Africa
MWI Sub-Saharan Africa Africa_Southern Sub-Saharan Africa
MYS Other Pacific Asia Southeast Asia Other Asia
MYT Western Europe Southeast Asia Sub-Saharan Africa
NAM Sub-Saharan Africa Africa_Southern Sub-Saharan Africa
NCL Other Pacific Asia Southeast Asia Other Asia
NER Sub-Saharan Africa Africa_Western Sub-Saharan Africa
NFK Western Europe Southeast Asia Other Asia
NGA Sub-Saharan Africa Africa_Western Sub-Saharan Africa
NIC Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
NIU Other Pacific Asia Southeast Asia Other Asia
NLD Western Europe EU-15 EU 28
NOR Western Europe European Free Trade Association Non-EU28 Europe
NPL South Asia South Asia Other Asia
NRU Other Pacific Asia Southeast Asia Other Asia
NZL Pacific OECD Australia_NZ Canada, NZ, Australia
OMN Middle East and North Africa Middle East Middle East, North Africa, Central
Asia

223
ISO MESSAGEix-GLOBIOM 1.1- GCAM 6.0 NGFS REMIND-MAgPIE 3.2-4.6
R12

PAK South Asia Pakistan Other Asia

PAN Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
PCN Western Europe Southeast Asia Other Asia
PER Latin America and the Caribbean South America_Southern Latin America and the Caribbean
PHL Other Pacific Asia Southeast Asia Other Asia
PLW Other Pacific Asia Southeast Asia Other Asia
PNG Other Pacific Asia Southeast Asia Other Asia
POL Eastern Europe EU-12 EU 28
PRI North America USA Latin America and the Caribbean
PRK Rest Centrally Planned Asia Southeast Asia Other Asia
PRT Western Europe EU-15 EU 28
PRY Latin America and the Caribbean South America_Southern Latin America and the Caribbean
PSE Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
PYF Other Pacific Asia Southeast Asia Other Asia
QAT Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
REU Western Europe Africa_Eastern Sub-Saharan Africa
ROU Eastern Europe EU-12 EU 28
RUS Former Soviet Union Russia Countries from the Reforming
Economies of the Former Soviet
Union
RWA Sub-Saharan Africa Africa_Eastern Sub-Saharan Africa
SAU Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
SDN Middle East and North Africa Africa_Eastern Middle East, North Africa, Central
Asia
SEN Sub-Saharan Africa Africa_Western Sub-Saharan Africa
SGP Other Pacific Asia Southeast Asia Other Asia
SGS Western Europe Latin America and the Caribbean
SHN Western Europe EU-15 Sub-Saharan Africa
SJM Western Europe European Free Trade Association Non-EU28 Europe
SLB Other Pacific Asia Southeast Asia Other Asia
SLE Sub-Saharan Africa Africa_Western Sub-Saharan Africa
SLV Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
SMR Western Europe EU-15 Non-EU28 Europe
SOM Sub-Saharan Africa Africa_Eastern Sub-Saharan Africa
SPM Western Europe EU-15 Canada, NZ, Australia
SRB Eastern Europe Europe_Non_EU Non-EU28 Europe
SSD Middle East and North Africa Africa_Eastern Sub-Saharan Africa
STP Sub-Saharan Africa Africa_Western Sub-Saharan Africa
SUR Latin America and the Caribbean South America_Northern Latin America and the Caribbean
SVK Eastern Europe EU-12 EU 28
SVN Eastern Europe EU-12 EU 28
SWE Western Europe EU-15 EU 28

224
ISO MESSAGEix-GLOBIOM 1.1- GCAM 6.0 NGFS REMIND-MAgPIE 3.2-4.6
R12

SWZ Sub-Saharan Africa Africa_Southern Sub-Saharan Africa

SXM Central America and Caribbean Latin America and the Caribbean
SYC Sub-Saharan Africa Southeast Asia Sub-Saharan Africa
SYR Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
TCA Latin America and the Caribbean EU-15 Latin America and the Caribbean
TCD Sub-Saharan Africa Africa_Western Sub-Saharan Africa
TGO Sub-Saharan Africa Africa_Western Sub-Saharan Africa
THA Other Pacific Asia Southeast Asia Other Asia
TJK Former Soviet Union Central Asia Countries from the Reforming
Economies of the Former Soviet
Union
TKL Pacific OECD Southeast Asia Other Asia
TKM Former Soviet Union Central Asia Countries from the Reforming
Economies of the Former Soviet
Union
TLS Other Pacific Asia Southeast Asia Other Asia
TON Other Pacific Asia Southeast Asia Other Asia
TTO Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
TUN Middle East and North Africa Africa_Northern Middle East, North Africa, Central
Asia
TUR Western Europe Europe_Non_EU Non-EU28 Europe
TUV Other Pacific Asia Southeast Asia Other Asia
TWN Other Pacific Asia Taiwan China
TZA Sub-Saharan Africa Africa_Southern Sub-Saharan Africa
UGA Sub-Saharan Africa Africa_Eastern Sub-Saharan Africa
UKR Former Soviet Union Europe_Eastern Countries from the Reforming
Economies of the Former Soviet
Union
UMI North America Other Asia
URY Latin America and the Caribbean South America_Southern Latin America and the Caribbean
USA North America USA United States of America
UZB Former Soviet Union Central Asia Countries from the Reforming
Economies of the Former Soviet
Union
VAT Western Europe EU-15 Non-EU28 Europe
VCT Latin America and the Caribbean Central America and Caribbean Latin America and the Caribbean
VEN Latin America and the Caribbean South America_Northern Latin America and the Caribbean
VGB Western Europe EU-15 Latin America and the Caribbean
VIR North America USA Latin America and the Caribbean
VNM Rest Centrally Planned Asia Southeast Asia Other Asia
VUT Other Pacific Asia Southeast Asia Other Asia
WLF Western Europe EU-15 Other Asia
WSM Other Pacific Asia Southeast Asia Other Asia
YEM Middle East and North Africa Middle East Middle East, North Africa, Central
Asia
ZAF Sub-Saharan Africa South Africa Sub-Saharan Africa

225
 ISO MESSAGEix-GLOBIOM 1.1- GCAM 6.0 NGFS REMIND-MAgPIE 3.2-4.6
R12

ZMB Sub-Saharan Africa Africa_Southern Sub-Saharan Africa

ZWE Sub-Saharan Africa Africa_Southern Sub-Saharan Africa

Table 36. NGFS Phase IV Scenario Explorer downscaling output

Sector Subsector Variable IAM132

Carbon Sequestration CCS Biomass G M R

Carbon Sequestration CCS Fossil G M R
Emissions CO2 Energy G M R
Final Energy Final Energy G M R
Final Energy Electricity G M R
Final Energy Gases G M R
Final Energy Heat G M R
Final Energy Hydrogen G M R
Final Energy Liquids G M R
Final Energy Solids G M R
Final Energy Industry Electricity G M R
Final Energy Industry Gases G M R
Final Energy Industry Heat G M R
Final Energy Industry Hydrogen G M R
Final Energy Industry Liquids G M R
Final Energy Industry Solids G M R
Final Energy Residential and Commercial Electricity G M R
Final Energy Residential and Commercial Gases G M R
Final Energy Residential and Commercial Heat G M R
Final Energy Residential and Commercial Liquids G M R
Final Energy Residential and Commercial Solids G M R
Final Energy Transportation Electricity G M R
Final Energy Transportation Gases G M R
Final Energy Transportation Hydrogen G M R
Final Energy Transportation Liquids G M R
Primary Energy Biomass G M R
Primary Energy Coal G M R
Primary Energy Coal w/ CCS G M R
Primary Energy Coal w/o CCS G M R
Primary Energy Fossil G M R

132 G = GCAM, M = MESSAGE-GLOBIOM, R = REMIND-MAgPIE

226
 Sector Subsector Variable IAM132

Primary Energy Fossil w/ CCS G M R

Primary Energy Fossil w/o CCS G M R
Primary Energy Gas G M R
Primary Energy Gas w/ CCS G M R
Primary Energy Gas w/o CCS G M R
Primary Energy Geothermal G R
Primary Energy Hydro G R
Primary Energy Nuclear G M R
Primary Energy Oil G M R
Primary Energy Oil w/ CCS G
Primary Energy Oil w/o CCS G M R
Primary Energy Solar G R
Primary Energy Wind G R
Secondary Energy Electricity Biomass G M R
Secondary Energy Electricity Coal G M R
Secondary Energy Electricity Gas G M R
Secondary Energy Electricity Geothermal G M R
Secondary Energy Electricity Hydro G M R
Secondary Energy Electricity Nuclear G M R
Secondary Energy Electricity Oil G M R
Secondary Energy Electricity Solar G M R
Secondary Energy Electricity Wind G M R
Secondary Energy Gases Biomass G M R
Secondary Energy Gases Coal G M R
Secondary Energy Gases Natural Gas G M R
Secondary Energy Liquids Biomass G M R
Secondary Energy Liquids Coal G M R
Secondary Energy Liquids Oil G M R
Secondary Energy Solids Biomass G M R
Secondary Energy Solids Coal G M R

Table 37. Acronyms and meanings

Acronym Term

AFOLU Agriculture, forestry and other land use

AgLU Agriculture and land use
BC Black carbon
BECCS Bioenergy with Carbone capture storage
C2 F 6 Hexafluoroethane

227
Acronym Term

CCS Carbon capture and storage

CCS Carbone capture storage
CES Constant Elasticity of Substitution
CF4 Tetrafluoromethane
CH4 Methane
CMIP5 Coupled Model Intercomparison Project – phase 5
CO Carbon monoxide
CO2 Carbon dioxide
DACCS Direct air capture with Carbone capture storage
EJ Exajoule
ETS Emissions trading system
ETS Emissions trading system
EW Enhanced weathering of rocks
EW Enhanced weathering of rocks
Gg Gigagram
GHG Greenhouse gases
GJ Gigajoule
GLU GLOBE Land Unit
HDV Heavy-duty vehicles
HDV Heavy-duty vehicles
HFC Hydrofluorocarbon
HFC125 Pentafluoroethane
HFC134a 1,1,1,2-Tetrafluoroethane
HFC143a 1,1,1-Trifluoroethane
HFC152a 1,1-Difluoroethane
HFC227ea 1,1,1,2,3,3,3-Heptafluoropropane
HFC23 Fluorophore
HFC236fa 1,1,1,3,3,3-Hexafluoropropane
HFC245fa 1,1,1,3,3-Pentafluorpropan
HFC32 Difluoromethane
HFC365mfc 1,1,1,3,3-Pentafluorobutane
HFC43-10mee 1,1,1,2,3,4,4,5,5,5-Decafluoropentane
IPCC Intergovernmental Panel on Climate Change
Kyoto gases Basket of CO2, CH4, N2O, HFC, PFC, SF6
LDV Light-duty vehicles
MAC Marginal abatement cost

228
 Acronym Term

MCal Million calories

MER Market exchange rate
MtC Million tonnes carbon
N2 O Nitrous oxide
NH3 Ammonia
NOx Nitrogen oxides
PFC Perfluorocarbon
PPP Purchasing power parity
RCP Representative Concentration Pathways
SSP Socioeconomic Development Pathways
TC technological change
Tg Teragram
VOC Volatile organic compounds
VRE Variable renewable energy

Table 38. Regional net-zero targets implemented in the 3 IAMs.

Country Net-zero year GCAM MESSAGE-GLOBIOM REMIND-MAgPIE

Argentina 2050 GHG GHG GHG (as LAM)

Australia 2050 GHG (as AUS_NZ) GHG GHG (as CAZ)

Brazil 2050 GHG GHG GHG (as LAM)

Canada 2050 GHG GHG GHG (as CAZ)

China 2060 GHG GHG GHG

Colombia 2050 GHG CO2‡ GHG (as LAM)

EU+UK 2050 GHG (for total EU12 GHG GHG

and EU15)
India 2070 CO2 GHG CO2

Indonesia 2060 GHG

Japan 2050 GHG GHG GHG

New Zealand 2050 GHG (as AUS_NZ) GHG GHG (as CAZ)

Russia 2060 GHG GHG (as REF)

South Africa 2050 GHG GHG

South Korea 2050 GHG CO2‡

USA 2050 GHG GHG GHG

229
† In GCAM, these country targets are implemented as one rest of world (ROW) constraint, and results show that
all net-zero targets are met (or very close to be met in the case of India).

‡ MESSAGE-GLOBIOM applies the net-zero GHG constraint for all countries/regions in its model
implementation. Thus, for those countries with CO2-only targets, we assume an approximation of a 10-year lag
for changing CO2 targets with GHG targets. E.g., for Colombia, the model sets GHG net-zero target by 2060
(instead of CO2 net-zero by 2050).

230
Bibliography

Aguilera, R. F., Eggert, R. G., Lagos C. C., G., and best worlds, Climatic Change, 114, 145168,
Tilton, J. E.: Depletion and the Future Availability https://doi.org/10.1007/s10584-011-0129-2, 2012b.
of Petroleum Resources, Energ. J., 30, 141–174,
https://doi.org/10.5547/ISSN0195-6574EJ-Vol30- Bauer, N., Brecha, R. J., and Luderer, G.:
No1-6, 2009. Economics of nuclear power and climate change
mitigation policies, P. Natl. Acad. Sci. USA, 109,
Amann, M., Bertok, I., Borken-Kleefeld, J., 16805–16810,
Cofala, J., Heyes, C., Höglund-Isaksson, L., https://doi.org/10.1073/pnas.1201264109, 2012a.
Klimont, Z., Nguyen, B., Posch, M., Rafaj, P.,
Sandler, R., Schöpp, W., Wagner, F., and Bauer, N., Hilaire, J., Brecha, R. J., Edmonds, J.,
Winiwarter, W.: Cost-effective control of air Jiang, K., Kriegler, E., Rogner, H.-H., and Sferra,
quality and greenhouse gases in Europe: Modeling F.: Assessing global fossil fuel availability in a
and policy applications, Environ. Modell. Softw., scenario framework, Energy, 111, 580–592,
26, 1489–1501, https://doi.org/10.1016/j.energy.2016.05.088,
https://doi.org/10.1016/j.envsoft.2011.07.012, 2016a.
2011.
Bauer, N., Klein, D., Humpenöder, F., Kriegler,
Amous, S.: Non-CO2 emissions from stationary E., Luderer, G., Popp, A., and Strefler, J.: Bio-
combustion. Background paper to “Good Practice energy and CO2 emission reductions: an
Guidance and Uncertainty Management in integrated land-use and energy sector perspective,
National Greenhouse Gas Inventories,” IPCC, 2000. Climatic Change, 163, 1675–1693,
https://doi.org/10.1007/s10584020-02895-z, 2020.
Arrow, K.J. 1950. “A difficulty in the concept of
social welfare”, Journal of Political Economy 58(4):
Bauer, N., Mouratiadou, I., Luderer, G.,
328–346.
Baumstark, L., Brecha, R. J., Edenhofer, O., and
Aznar-Siguan, G., & Bresch, D. N. (2019). Kriegler, E.: Global fossil energy markets and
CLIMADA v1: A global weather and climate risk climate change mitigation – an analysis with
assessment platform. Geoscientific Model REMIND, Climatic Change, 136, 69–82,
Development, 12(7), 3085–3097. https://doi.org/10.1007/s10584-0130901-6, 2016b.
https://doi.org/10.5194/gmd-12-3085-2019
Bauer, N., Rose, S. K., Fujimori, S., Vuuren, D. P.
Baranović, Ivana, Busies, Iulia, Coussens, van, Weyant, J., Wise, M., Cui, Y., Daioglou, V.,
Wouter, Grill, Michael and Hempell, Hannah S., Gidden, M. J., Kato, E., Kitous, A., Leblanc, F.,
(2021), “The challenge of capturing climate risks in Sands, R., Sano, F., Strefler, J., Tsutsui, J.,
the banking regulatory framework: is there a need Bibas, R., Fricko, O., Hasegawa, T., Klein, D.,
for a macroprudential response?”, ECB Kurosawa, A., Mima, S., and Muratori, M.: Global
Macroprudential Bulletin, 15, issue, number 1. energy sector emission reductions and bioenergy
use: overview of the bioenergy demand phase of
Barro, R. J. and Sala-i-Martin, X.: Economic the EMF-33 model comparison, Climatic Change,
growth, 2nd Edn., MIT Press, Cambridge, 163, 1553–1568, https://doi.org/10.1007/s10584-
Massachusetts, 2004. 018-2226-y, 2018.

Bauer, N., Baumstark, L., and Leimbach, M.: The Bertram, C., Luderer, G., Popp, A., Minx, J. C.,
REMIND-R model: the role of renewables in the Lamb, W. F., Stevanović, M., ... & Kriegler, E.
low-carbon transformation – first-best vs. second- (2018). Targeted policies can compensate most of

231
the increased sustainability risks in 1.5 C mitigation Formulation and functionality.” Geoscientific
scenarios. Environmental Research Letters, 13(6), Model Development 8, no. 7: 2203-2219.
064038. https://doi.org/10.1088/1748-9326/aac3ec
Dellink, R., Chateau, J., Lanzi, E., Magné, B.
(2017) Long-term economic growth projections in
Bertram, C., Riahi, K., Hilaire, J., Bosetti, V.,
the Shared Socioeconomic Pathways, Global
Drouet, L., Fricko, O., ... & Luderer, G. (2021). Environmental Change, Volume 42, 200-214.
Energy system developments and investments in https://doi.org/10.1016/j.gloenvcha.2015.06.004.
the decisive decade for the Paris Agreement
goals. Environmental Research Letters, 16(7), Dietrich, J. P., Schmitz, C., Lotze-Campen, H.,
074020. Popp, A., & Müller, C. (2014) Forecasting
technological change in agriculture—An
Bosetti, V., Massetti, E., and Tavoni,M.: endogenous implementation in a global land use
TheWITCHModel, Structure, Baseline, Solutions, model. Technological Forecasting and Social
2007, FEEM Working Paper No. 10.2007, CMCC Change, 81(Supplement C), 236–249.
Research Paper No. 7, 2007. https://doi.org/10.1016/j.techfore.2013.02.003.

Bresch, D. N., & Aznar-Siguan, G. (2021). Dietrich, J., L. Baumstark, S. Wirth, A.

CLIMADA v1.4.1: Towards a globally consistent Giannousakis, R. Rodrigues, B. Bodirsky, U.
adaptation options appraisal tool. Geoscientific Kreidenweis, D. Klein and P. Führlich. 2023.
Model Development, 14(1), 351–363. madrat: May All Data be Reproducible and
https://doi.org/10.5194/gmd-14-351-2021 Transparent, MADRaT.

Brückl, O.: Global Technical Potential of Electricity Drouet, L., Bosetti, V., Padoan, S. A., Aleluia
Production from Wind Energy, Internal Final Reis, L., Bertram, C., Dalla Longa, F., ... &
Report for BMW AG Study, Institute for Energy Tavoni, M. (2021). Net zero-emission pathways
Economy and Application Technology (IfE) reduce the physical and economic risks of climate
Germany, Technische Universität München, change. Nature Climate Change, 11(12), 1070-
Munich, Germany, 2005. 1076.
Calvin, K., B. Bond-Lamberty, L. Clarke, J.
Edmonds, J. Eom, C. Hartin, S. Kim, P. Kyle, R. Eberenz, S., Lüthi, S., & Bresch, D. N. (2021).
Link, R. Moss, H. McJeon, P. Patel, S. Smith, S. Regional tropical cyclone impact functions for
Waldhoff and M. Wise. 2017. “The SSP4: A world globally consistent risk assessments. Natural
of deepening inequality”, Global Environmental Hazards and Earth System Sciences, 21(1), 393–
Change 42: 284-296. 415. https://doi.org/10.5194/nhess-21-393-2021.

Carney, M., Villeroy de Galhau, F. and Elderson,

Eberenz, S., Stocker, D., Röösli, T., & Bresch, D.
F. (2019), “Open letter on climate-related financial
N. (2020). Asset exposure data for global physical
risks”, Bank of England, April.
risk assessment. Earth System Science Data, 12(2),
Article 2. https://doi.org/10.5194/essd-12-817-2020
Clarke et al., 2014; Forster et al., 2018; Jia et al.,
2019; Rogelj, Shindell, et al., 2018; UNEP, 2018, Edmonds, J. and J. Reilly. 1982a. “Global energy
IPCC 2022a and CO2 to the year 2050”, IEA/ORAU Working
Paper Contribution: 82-6.
Collins, W.D., A.P. Craig, J.E. Truesdale, A.V. Di
Vittorio, A.D. Jones, B. Bond-Lamberty, K.V. Edmonds, J. and J. Reilly. 1982b. “Global energy
Calvin, J.A. Edmonds, A.M. Thomson, B. Bond- production and use to the year 2050”, IEA/ORAU
Lamberty, P. Patel, S.H. Kim, P.E. Thornton, J. Working Paper Contribution: 82-7.
Mao, X. Shi, L.P. Chini and G.C. Hurtt. 2015. “The
integrated Earth system model version 1: Edmonds, J. and J. Reilly. 1982c. “An introduction
to the use of the IEA/ORAU, long-term, global

232
energy model”, IEA/ORAU Working Paper quantification of the shared socioeconomic
Contribution: 82-9. pathway 2: a middle-of-the-road scenario for the
21st century. Global Environmental Change,
Edmonds, J. and J. Reilly. 1983a. “Global energy 42:251–267, 2017.
and CO2 to the year 2050,” The Energy Journal 4(3):
21-47.
FSB-NGFS (2022), “Climate Scenario Analysis by
Jurisdictions: initial findings and lessons”,
Edmonds, J. and J. Reilly. 1983b. “A long-term,
November.
global, energy-economic model of carbon dioxide
release from fossil fuel use”, Energy Economics 5(2):
Fujino, J., Nair, R., Kainuma, M., Masui, T., and
74-88.
Matsuoka, Y.: Multi-gas mitigation analysis on
stabilization scenarios using AIM global model,
Edmonds, J. and J. Reilly. 1983c. “Global energy
Energ. J., in: MultiGreenhouse Gas Mitigation and
production and use to the year 2050,” Energy 8(6):
Climate Policy, 343–354,
419-32.
https://doi.org/10.5547/ISSN0195-6574-EJ-
Edmonds, J., J. Reilly, J.R. Trabalka and D.E. VolSI2006-NoSI317, 2006.
Reichle. 1984. “An analysis of possible future
atmospheric retention of fossil fuel CO2”, DOE TR- Gettelman, A., Bresch, D. N., Chen, C. C.,
013, National Technical Information Service, U.S. Truesdale, J. E., & Bacmeister, J. T. (2018).
Department of Commerce, Springfield Virginia Projections of future tropical cyclone damage with
22161. a high-resolution global climate model. Climatic
Change, 146(3–4), Article 3–4.
Eurek, K., P. Sullivan, M. Gleason, D. Hettinger, https://doi.org/10.1007/s10584-017-1902-7
D.M. Heimiller, and A. Lopez. An improved global
wind resource estimate for integrated assessment Gidden, M. J., Fujimori, S., van den Berg, M.,
models. Energy Economics, 64:552–567, 2017. Klein, D., Smith, S. J., van Vuuren, D. P., & Riahi,
K. (2018). A methodology and implementation of
Eurek, K., Sullivan, P., Gleason, M., Hettinger, automated emissions harmonization for use in
D., Heimiller, D., and Lopez, A.: An improved Integrated Assessment Models. Environmental
global wind resource estimate for integrated Modelling & Software, 105, 187-200.
assessment models, Energ. Econ., 64, 552–567,
https://doi.org/10.1016/j.eneco.2016.11.015, 2017. Global Emissions EDGAR v7:
https://edgar.jrc.ec.europa.eu/
Fricko, O., Havlik, P., Rogelj, J., Klimont, Z.,
Gusti, M., Johnson, N., Kolp, P., Strubegger, M., Grassi, G., Stehfest, E., Rogelj, J., Van Vuuren,
Valin, H., Amann, M., Ermolieva, T., Forsell, N., D., Cescatti, A., House, J., ... & Popp, A. (2021).
Herrero, M. Heyes, C., Kindermann, G., Krey, V., Critical adjustment of land mitigation pathways for
McCollum, D. L., Obersteiner, M., Pachauri, S., assessing countries’ climate progress. Nature
Rao, S., Schmid, E., Schoepp, W., and Riahia, K.: Climate Change, 11(5), 425-434.
The marker quantification of the Shared Gütschow, J., Jeffery, L., Gieseke, R., Gebel, R.,
Socioeconomic Pathway 2: A middle-of-the-road Stevens, D., Krapp, M., & Rocha, M. The PRIMAP-
scenario for the 21st century, Global Environ. hist national historical emissions time series (1750-
Change, 42, 2 2019) (v2. 3.1, updated September 2021).

Fricko, O., P. Havlik, J. Rogelj, Z. Klimont, M. Hall et al., Spatial analysis of outdoor wet bulb
Gusti, N. Johnson, P. Kolp, M. Strubegger, H. globe temperature under RCP4.5 and RCP8.5
Valin, M. Amann, T. Ermolieva, N. Forsell, M. scenarios for 2041–2080 across a range of
Herrero, C. Heyes, G. Kindermann, V. Krey, D. L. temperate to hot climates, Weather and Climate
McCollum, M. Obersteiner, S. Pachauri, S. Rao, Extremes 35 (2022) 100420.
E. Schmid, W. Schoepp, and K. Riahi. The marker

233
Hartin, C.A., P. Patel, A. Schwarber, R.P. Link
and B.P. Bond-Lamberty. 2015. “A simple object- Huizinga et al. (2017): Huizinga, J., De Moel, H.
oriented and open-source model for scientific and and Szewczyk, W., Global flood depth-damage
policy analyses of the global climate system – functions: Methodology and the database with
Hector v1.0”, Geosci. Model Dev., 8, 939-955. guidelines, EUR 28552 EN, Publications Office of
the European Union, Luxembourg, 2017, ISBN 978-
Heckscher, E. F., Ohlin, B., Flam, H., and
92-79-67781-6, doi:10.2760/16510, JRC105688.
Flanders, M. J.: Heckscher-Ohlin trade theory, MIT
Press, Cambridge, Massachusetts, 234 pp., 1991.
Huppmann D, Gidden M, Fricko O, Kolp P,
Orthofer C, Pimmer M, et al. The MESSAGEix
Herfindahl, O. C.: Depletion and Economic
Integrated Assessment Model and the ix modeling
Theory, in: Extractive Resources and Taxation,
platform (ixmp): An open framework for integrated
edited by: Gaffney, M., University of Wisconsin
and cross-cutting analysis of energy, climate, the
Press, Madison, Wisconsin, 1967.
environment, and sustainable development.
Environ Model Softw 2019.
Herger, N., Sanderson, B. M., & Knutti, R. (2015).
https://doi.org/10.1016/j.envsoft.2018.11.012.
Improved pattern scaling approaches for the use in
climate impact studies. Geophysical Research
Letters, 42, 3486-3494. International Institute for Applied Systems
doi:10.1002/2015GL063569 Analysis, TWI2050 - The World in 2050 (2018).
Transformations to Achieve the Sustainable
Herger, Sanderson and Knutti (2015).
Development Goals. Report prepared by the
World in 2050 initiative. (IIASA), Laxenburg,
Höglund-Isaksson, L., Gómez-Sanabria, A., Austria. www.twi2050.org
Klimont, Z., Rafaj, P., & Schöpp, W. (2020).
Technical potentials and costs for reducing global
anthropogenic methane emissions in the 2050 IEA, I. 2022. World Energy Statistics and Balances,
timeframe–results from the GAINS model. IEA.
Environmental Research Communications, 2(2),
025004 IEA (2023), World Energy Outlook 2023, IEA, Paris
https://www.iea.org/reports/world-energy-
Holland, G. (2008). A Revised Hurricane Pressure– outlook-2023, License: CC BY 4.0 (report); CC BY
Wind Model. Monthly Weather Review, 136(9), NC SA 4.0 (Annex A)
3432–3445.
https://doi.org/10.1175/2008MWR2395.1 IEA. World energy model - investment costs.
Report, International Energy Agency (IEA), 2014.
Hoogwijk, M. and Graus, W.: Global potential of URL:
renewable energy sources: a literature assessment, http://www.worldenergyoutlook.org/media/weow
Ecofys, REN21 Renewable Energy Policy Network ebsite/2014/weio/WEIO2014PGAssumptions.xlsx.
for the 21st Century, 2008. International Energy Agency. Energy Balances.
Technical Report, International Energy Agency,
Hoogwijk, M.: On the global and regional potential 2012.
of renewable energy sources, PhD Thesis,
Universiteit Utrecht, Faculteit Scheikunde, Jägermeyr J, Müller C, Ruane AC, et al. Climate
Utrecht, 2004. impacts on global agriculture emerge earlier in new
generation of climate and crop models. Nat Food.
Horlacher, H.-B.: Globale Potenziale der 2021;2(November). doi:10.1038/s43016-021-
Wasserkraft. Externe Expertise für das WBGU- 00400-y
Hauptgutachten 2003 “Welt im Wandel: Kalkuhl, M., & Wenz, L. (2020) The Impact of
Energiewende zur Nachhaltigkeit,” WBGU, Climate Conditions on Economic Production.
Heidelberg, Germany, 2003. Evidence from a Global Panel of Regions. Journal

234
of Environmental Economics and Management, Klein, D., Luderer, G., Kriegler, E., Strefler, J.,
103, 102360. Bauer, N., Leimbach, M., Popp, A., Dietrich, J.
https://doi.org/10.1016/j.jeem.2020.102360 P., Humpenöder, F., LotzeCampen, H., and
Edenhofer, O.: The value of bioenergy in low
Kang, S., Pal, J. S., & Eltahir, E. A. B. (2019). stabilization scenarios: an assessment using
Future heat stress during Muslim pilgrimage (Hajj) REMIND-MAgPIE, Climatic Change, 123, 705–718,
projected to exceed “extreme danger” levels. https://doi.org/10.1007/s10584013-0940-z, 2014b.
Geophysical Research Letters, 46, 10094– 10100.
https://doi.org/10.1029/2019GL083686 Klein, D.: Bioenergy markets in a climate
constrained world, Doctoral Thesis, Fakultät VI –
KC, S. and W. Lutz. The human core of the shared Planen Bauen Umwelt, Technische Universität
socioeconomic pathways: Population scenarios by Berlin, Berlin, Germany,
age, sex and level of education for all countries to https://doi.org/10.14279/depositonce-4307, 2015.
2100. Global Environmental Change, 2014.
Klimont, Z., Hoglund, L., Heyes, C., Rafaj, P.,
KC, S., & Lutz, W. (2017) The human core of the Schoepp, W., Cofala, J., Borken-Kleefeld, J.,
shared socioeconomic pathways: Population Purohit, P., Kupiainen, K., Winiwarter, W.,
scenarios by age, sex and level of education for all Amann, M., Zhao, B., Wang, S. X., Bertok, I.,
countries to 2100. Global Environmental Change, and Sander, R.: Global scenarios of air pollutants
42,181–192. and methane: 1990–2050, in preparation, 2021.
https://doi.org/10.1016/j.gloenvcha.2014.06.004
Klimont, Z., Kupiainen, K., Heyes, C., Purohit,
Keppo, I. and M. Strubegger. Short term P., Cofala, J., Rafaj, P., Borken-Kleefeld, J., and
decisions for long term problems–The effect of Schöpp, W.: Global anthropogenic emissions of
foresight on model based energy systems analysis. particulate matter including black carbon, Atmos.
Energy, 35(5):2033–2042, 2010. Chem. Phys., 17, 8681–8723,
https://doi.org/10.5194/acp-178681-2017, 2017.
Kikstra JS, Nicholls ZR, Smith CJ, Lewis J,
Lamboll RD, Byers E, Sandstad M, Meinshausen Knapp, K. R., Kruk, M. C., Levinson, D. H.,
M, Gidden MJ, Rogelj J, Kriegler E. The IPCC Sixth Diamond, H. J., & Neumann, C. J. (2010). The
Assessment Report WGIII climate assessment of International Best Track Archive for Climate
mitigation pathways: from emissions to global Stewardship (IBTrACS): Unifying Tropical Cyclone
temperatures. Geoscientific Model Development. Data. Bulletin of the American Meteorological
2022 Dec 20;15(24):9075-109. Society, 91(3), Article 3.
https://doi.org/10.1175/2009BAMS2755.1
Kikstra, J.S., Vinca, A., Lovat, F. et al. Climate
mitigation scenarios with persistent COVID-19- Knutson, T. R., Sirutis, J. J., Zhao, M., Tuleya, R.
related energy demand changes. Nat Energy 6, E., Bender, M., Vecchi, G. A., Villarini, G., &
1114–1123 (2021). https://doi.org/10.1038/s41560- Chavas, D. (2015). Global Projections of Intense
021-00904-8 Tropical Cyclone Activity for the Late Twenty-First
Century from Dynamical Downscaling of
Klein, D., Humpenöder, F., Bauer, N., Dietrich, J. CMIP5/RCP4.5 Scenarios. Journal of Climate,
P., Popp, A., Bodirsky, B. L., Bonsch, M., and 28(18), 7203–7224. https://doi.org/10.1175/JCLI-D-
Lotze-Campen, H.: The lobal economic long-term 15-0129.1
potential of modern biomass in a climate-
constrained world, Environ. Res. Lett., 9, 074017, Knutson, T., Camargo, S. J., Chan, J. C. L.,
https://doi.org/10.1088/1748-9326/9/7/074017, Emanuel, K., Ho, C.-H., Kossin, J., Mohapatra,
2014a. M., Satoh, M., Sugi, M., Walsh, K., & Wu, L.
(2020). Tropical Cyclones and Climate Change

235
Assessment: Part II: Projected Response to https://doi.org/10.5547/ISSN0195-6574-EJ-Vol31-
Anthropogenic Warming. Bulletin of the American NoSI-5, 2010b.
Meteorological Society, 101(3), Article 3.
https://doi.org/10.1175/BAMS-D-18-0194.1 Lotze-Campen, H., Müller, C., Bondeau, A.,
Rost, S., Popp, A., and Lucht, W.: Global food
Kriegler et al., 2013, 2014, 2015; Luderer et al., demand, productivity growth, and the scarcity of
2013; Riahi et al., 2015; Tavoni et al., 2013 land and water resources: a spatially explicit
Kuramochi, T., Ramírez, A., Turkenburg, W., mathematical programming approach, Agr. Econ.,
and Faaij, A.: Comparative assessment of CO2 39, 325–338, https://doi.org/10.1111/j.1574-
capture technologies for carbonintensive industrial 0862.2008.00336.x, 2008.
processes, Prog. Energ. Combust., 38, 87112,
https://doi.org/10.1016/j.pecs.2011.05.001, 2012. Lotze-Campen, H., Popp, A., Beringer, T.,
Müller, C., Bondeau, A., Rost, S., & Lucht, W.
Lange, S., Volkholz, J., Geiger, T., Zhao, F., (2010) Scenarios of global bioenergy production:
Vega, I., Veldkamp, T., ... & Frieler, K. (2020). The trade-offs between agricultural expansion,
Projecting exposure to extreme climate impact intensification and trade. Ecological Modelling,
events across six event categories and three spatial 221(18), 2188–2196.
scales. Earth's Future, 8(12). https://doi.org/10.1016/j.ecolmodel.2009.10.002
https://doi.org/10.1029/2020EF001616
Lu, X., McElroy, M. B., and Kiviluoma, J.: Global
Leibowicz. B. D., Growth and competition in potential for wind-generated electricity, P. Natl.
renewable energy industries: insights from an Acad. Sci. USA, 106, 1093310938,
integrated assessment model with strategic firms. https://doi.org/10.1073/pnas.0904101106, 2009.
Energy Economics, 52, Part A:13 – 25, 2015.
doi:10.1016/j.eneco.2015.09.010. Lucas, P. L., van Vuuren, D. P., Olivier, J. G. J.,
and den Elzen, M. G. J.: Long-term reduction
Leimbach, M. and Bauer, N.: Capital markets and potential of nonCO2 greenhouse gases, Environ.
the costs of climate policies, Environ. Econ. Policy Sci. Policy, 10, 85–103,
Stud., https://doi.org/10.1007/s10018-021-00327-5, https://doi.org/10.1016/j.envsci.2006.10.007, 2007.
online first, 2021.
Lucas, R. E.: Why Doesn’t Capital Flow from Rich
Leimbach, M. and Giannousakis, A.: Burden to Poor Countries?, Am. Econ. Rev., 80, 92–96,
sharing of climate change mitigation: global and 1990.
regional challenges under shared socio-economic Luderer, G., Krey, V., Calvin, K., Merrick, J.,
pathways, Climatic Change, 155, 273291, Mima, S., Pietzcker, R., Vliet, J. V., and Wada,
https://doi.org/10.1007/s10584-019-02469-8, 2019. K.: The role of renewable energy in climate
stabilization: results from the EMF27 scenarios,
Leimbach, M., Bauer, N., Baumstark, L., and Climatic Change, 123, 427–441,
Edenhofer, O.: Mitigation Costs in a Globalized https://doi.org/10.1007/s10584013-0924-z, 2014.
World: Climate Policy Analysis with REMIND-R,
Environ. Model Assess., 15, 155–173, Luderer, G., Pietzcker, R. C., Kriegler, E., Haller,
https://doi.org/10.1007/s10666-009-9204-8, M., and Bauer, N.: Asia’s role in mitigating climate
2010a. change: A technology and sector specific analysis
with ReMIND-R, Energ. Econ., 34, S378–S390,
Leimbach, M., Bauer, N., Baumstark, L., Luken, https://doi.org/10.1016/j.eneco.2012.07.022, 2012.
M., and Edenhofer, O.: Technological Change and
Maheu, A. 2009. Energy choices and their impacts
International Trade – Insights from REMIND-R,
on demand for water resources: An assessment of
Energ. J., 31, 109–136,
current and projected water consumption in global
energy production. Unisféra.

236
Malte Meinshausen, SCB Raper, and TML limits, P. Natl. Acad. Sci. USA, 113, 13570,
Wigley. Emulating coupled atmosphere-ocean and https://doi.org/10.1073/pnas.1602253113, 2016.
carbon cycle models with a simpler model,
MAGICC6–Part 1: Model description and
calibration. Atmospheric Chemistry and Physics, Murakami, D.; Yamagata, Y. Estimation of
11(4):1417–1456, 2011. Gridded Population and GDP Scenarios with
Spatially Explicit Statistical Downscaling.
Manne, A., Mendelsohn, R., and Richels, R.: A Sustainability 2019, 11, 2106.
Model for Evaluating Regional and Global Effects https://doi.org/10.3390/su11072106
of GHG Reduction Policies, Energ. Policy, 23, 17–
34, 1995. Nakićenović, N., A. Grübler, A. Inaba, S.
Messner, S. Nilsson, Y. Nishimura, H.-H. Rogner,
McCollum, D.L., Zhou, W., Bertram, C. et al. A. Schäfer, L. Schrattenholzer and M.
Energy investment needs for fulfilling the Paris Strubegger. 1993. “Long-term strategies for
Agreement and achieving the Sustainable mitigating global warming”, Energy 18, 401.
Development Goals. Nat Energy 3, 589–599 (2018).
https://doi.org/10.1038/s41560-018-0179-z Nakicenovic, N., N. Nakićenović, A. Grübler, A.
Meinshausen, M., Nicholls, Z. R., Lewis, J., Grübler, A. McDonald and A.T. McDonald. 1998.
Gidden, M. J., Vogel, E., Freund, M., ... & Wang, Global Energy Perspectives. Cambridge University
R. H. (2020). The shared socio-economic pathway Press.
(SSP) greenhouse gas concentrations and their
extensions to 2500. Geoscientific Model Network for Greening the Financial System
Development, 13(8), 3571-3605. (2018), “First progress report”, Banque de France,
October.
Meinshausen, M., Raper, S. C., & Wigley, T. M.
(2011). Emulating coupled atmosphere-ocean NGFS (2023), “NGFS Survey on Climate Scenarios:
and carbon cycle models with a simpler model, key findings”, June
MAGICC6–Part 1: Model description and
calibration. Atmospheric Chemistry and NGFS (2023), “NGFS Survey on Climate Scenarios:
Physics, 11(4), 1417-1456. key findings”, June.

Mekonnen, M.M. and A.Y. Hoekstra. 2010. The Nicholls Z, Meinshausen M, Lewis J, Smith CJ,
green, blue and grey water footprint of farm Forster PM, Fuglestvedt JS, Rogelj J, Kikstra JS,
animals and animal products. Riahi K, Byers E. Changes in IPCC scenario
assessment emulators between SR1. 5 and AR6
Messner, S., and L. Schrattenholzer, MESSAGE–
unraveled. Geophysical Research Letters. 2022 Oct
MACRO: linking an energy supply model with a
28;49(20):e2022GL099788.
macroeconomic module and solving it iteratively,
Energy, Volume 25, Issue 3, 2000, Pages 267-282,
Nicholls Z, Meinshausen M, Lewis J, Corradi MR,
ISSN 0360-5442, https://doi.org/10.1016/S0360-
Dorheim K, Gasser T, Gieseke R, Hope AP, Leach
5442(99)00063-8.
NJ, McBride LA, Quilcaille Y. Reduced complexity
Model Intercomparison Project Phase 2:
Messner, S., and M. Strubegger. User's Guide for
Synthesizing Earth system knowledge for
MESSAGE III. 1995. URL:
probabilistic climate projections. Earth's Future.
http://pure.iiasa.ac.at/id/eprint/4527/1/WP-95-
2021 Jun;9(6):e2020EF001900
069.pdf.
NOAA,National Centers for Environmental
Miller, L. M. and Kleidon, A.: Wind speed
Information (NCEI) U.S. Billion-Dollar Weather
reductions by largescale wind turbine deployments
and Climate Disasters (2023).
lower turbine efficiencies and set low generation

237
https://www.ncei.noaa.gov/access/billions/, DOI: and concentrating solar power. Applied Energy,
10.25921/stkw-7w73 135:704–720, 2014.

Nordhaus, W. D. and Yang, Z.: A Regional Platts, S.G. 2019. Platts-World Electric Power
Dynamic GeneralEquilibrium Model of Alternative Plants Database.
Climate-Change Strategies, Am. Econ. Rev., 86,
741–765, 1996. Popp .A, K. Calvin, S. Fujimori, P. Havlik, F.
Humpenöder, E. Stehfest, B. Leon Bodirsky, J. P.
O’Neill, B. C., Kriegler, E., Riahi, K., Ebi, K. L., Dietrich, J. C. Doelmann, M. Gusti, T. Hasegawa, P.
Hallegatte, S., Carter, T. R., Mathur, R., and Kyle, M. Obersteiner, A. Tabeau, K. Takahashi, H.
Vuuren, D. P.: A new scenario framework for Valin, S. Waldhoff, I. Weindl, M. Wise, E. Kriegler,
climate change research: the concept of shared H. Lotze-Campen, O. Fricko, K. Riahi, D. P. van
socioeconomic pathways, Climatic Change, 122, Vuuren,.: Land-use futures in the shared socio-
387400, https://doi.org/10.1007/s10584-013-0905- economic pathways, 2017.
2, 2013. https://doi.org/10.1016/j.gloenvcha.2016.10.002

O’Neill, B.C., E. Kriegler, K. L. Ebi, E. Kemp- R. Skalsky, Z. Tarasovicova, J. Balkovic, E.

Benedict, K. Riahi, D. S Rothman, B. J van Schmid, M. Fuchs, E. Moltchanova, G.
Ruijven, D. P. van Vuuren, J. Birkmann, and K. Kindermann, and P. Scholtz. Geo-bene global
Kok. The roads ahead: narratives for shared database for bio-physical modeling v.1.0.
socioeconomic pathways describing world futures concepts, methodologies and data.technical
in the 21st century. Global Environmental Change, report. Report, IIASA, accessed 13.03.09 2008.
2015. URL: http://www.geo-bene.eu/?q=node/1734S.

Pauliuk, S., Milford, R. L., Müller, D. B., and Ray DK et al.. Climate variation explains a third of
Allwood, J. M.: The Steel Scrap Age, Environ. Sci. global crop yield variability. Nat Communications
Technol., 47, 3448–3454, 2015;6:5989. doi:10.1038/ncomms6989
https://doi.org/10.1021/es303149z, 2013. Recoveries, M. D. (2021). World economic outlook.
International Monetary Fund.
Pietzcker, R. C., Longden, T., Chen, W., Fu, S.,
Reilly, J.M., J.A. Edmonds, R.H. Gardner and
Kriegler, E., Kyle, P., and Luderer, G.: Long-term
A.L. Brenkert. 1987. “Uncertainty analysis of the
transport energy demand and climate policy:
IEA/ORAU CO2 emissions model,” The Energy
Alternative visions on transport decarbonization in
Journal 8(3): 1-29.
energy-economy models, Energy, 64, 95–108,
https://doi.org/10.1016/j.energy.2013.08.059, Riahi, K., D. P. van Vuuren, E. Kriegler, J.
2014a. Edmonds, B. O’Neill, S. Fujimori, N. Bauer, K.
Calvin, R. Dellink, O. Fricko, W. Lutz, A. Popp, J.
Pietzcker, R. C., Stetter, D., Manger, S., and Crespo Cuaresma, S. KC, M. Leimbach, L. Jiang,
Luderer, G.: Using the sun to decarbonize the T. Kram, S. Rao, J. Emmerling, K. Ebi, T.
power sector: The economic potential of Hasegawa, P. Havlik, F. Humpenoder, L. Aleluia
photovoltaics and concentrating solar power, Appl. Da Silva, S. Smith, E. Stehfest, V. Bosetti, J.
Energ., 135, 704–720, Eom, D. Gernaat, T. Masui, J. Rogelj, J. Strefler,
https://doi.org/10.1016/j.apenergy.2014.08.011, L. Drouet, V. Krey, G. Luderer, M. Harmsen, K.
2014b. Takahashi, L. Baumstark, J. Doelman, M.
Kainuma, Z. Klimont, G. Marangoni, H. Lotze-
Pietzcker, R.C. D. Stetter, S. Manger, and G. Campen, M. Obersteiner, A. Tabeau, and M.
Luderer. Using the sun to decarbonize the power Tavoni. The Shared Socioeconomic Pathways and
sector: the economic potential of photovoltaics their Energy, Land Use, and Greenhouse Gas
Emissions Implications. Global Environmental

238
Change, 42:153–168, 2017. URL: regional disaggregation. Nature Communication
http://pure.iiasa.ac.at/13280/doi:10.1016/j.gloenvc 12, 2128 (2021). https://doi.org/10.1038/s41467-
ha.2016.05.009. 021-22153-9

Riahi, K., F. Dentener, D. Gielen, A. Grubler, J. Schmitz, C., Biewald, A., Lotze-Campen, H.,
Jewell, Z. Klimont, V. Krey, D. McCollum, S. Popp, A., Dietrich, J. P., Bodirsky, B., Krause,
Pachauri, S. Rao, B. van Ruijven, D. P. van M., & Weindl, I. (2012) Trading more food:
Vuuren, and C. Wilson. Chapter 17 - Energy Implications for land use, greenhouse gas
Pathways for Sustainable Development. In Global emissions, and the food system. Global
Energy Assessment - Toward a Sustainable Future, Environmental Change, 22(1), 189–209.
pages 1203–1306. Cambridge University Press, https://doi.org/10.1016/j.gloenvcha.2011.09.013
Cambridge, UK and New York, NY, USA and the
International Institute for Applied Systems Scholz, Y., Gils, H. C., and Pietzcker, R. C.:
Analysis, Laxenburg, Austria, 2012. URL: Application of a highdetail energy system model to
http://www.globalenergyassessment.org. derive power sector characteristics at high wind
and solar shares, Energ. Econ., 64, 568–582, 2017.
Rogner, H.-H., Aguilera, R. F., Bertani, R.,
Bhattacharya, S. C., Dusseault, M. B., Gagnon, Scussolini, P., Aerts, J. C. J. H., Jongman, B.,
L., Haberl, H., Hoogwijk, M., Johnson, A., Bouwer, L. M., Winsemius, H. C., de Moel, H.,
Rogner, M. L., Wagner, H., and Yakushev, V.: and Ward, P. J.: FLOPROS: an evolving global
Chapter 7 – Energy Resources and Potentials, in: database of flood protection standards, Nat.
Global Energy Assessment – Toward a Sustainable Hazards Earth Syst. Sci., 16, 1049–1061,
Future, Cambridge University Press, Cambridge, https://doi.org/10.5194/nhess-16-1049-2016, 2016.
UK and New York, NY, USA and the International
Shiklomanov, I.A. 2000. “Appraisal and
Institute for Applied Systems Analysis, Laxenburg,
Assessment of World Water Resources”, Water
Austria, 423–512, 2012.
International 25: 11-32.

Rogner, H.-H.: An assessment of world Touma et al. A multi-model and multi-index

hydrocarbon ressources, Annu. Rev. Energ. Env., evaluation of drought characteristics in the 21st
22, 217–262, century, Journal of Hydrology, Volume 526, 2015,
https://doi.org/10.1146/annurev.energy.22.1.217, Pages 196-207
1997.
Rohwer, J., D. Gerten and W. Lucht. Turner S.W.D., M. Hejazi, C. Yonkofski, S. Kim
and P. Kyle. 2019. “Influence of groundwater
2007. Development of functional irrigation types
extraction costs and resource depletion limits on
for improved global crop modelling”. PIK Report
simulated global nonrenewable water withdrawals
No. 104, Potsdam Institute for Climate Impact
over the 21st century”. Earth’s Future.
Research.

Saeed et al. Deadly Heat Stress to Become Ueckerdt, F., Pietzcker, R., Scholz, Y., Stetter,
Commonplace Across South Asia Already at 1.5°C D., Giannousakis, A., and Luderer, G.:
of Global Warming Geophysical Research letters, Decarbonizing global power supply under region-
Volume48, (2021a) e2020GL091191 specific consideration of challenges and options of
integrating variable renewables in the REMIND
Saeed et al. From Paris to Makkah: heat stress model, Energ. Econ., 64, 665–684,
risks for Muslim pilgrims at 1.5 °C and 2 °C (2021b) https://doi.org/10.1016/j.eneco.2016.05.012, 2017.
Environ. Res. Lett. 16 024037 UN Population Division. World Population
Projection. Technical Report, UN, 2010.
Sauer, I.J., Reese, R., Otto, C. et al. Climate
signals in river flood damages emerge under sound

239
Van der Werf, G. R., Randerson, J. T., Giglio, L., Veccellio et al. Utility of the Heat Index in defining
Collatz, G. J., Mu, M., Kasibhatla, P. S., Morton, the upper limits of thermal balance during light
D. C., DeFries, R. S., Jin, Y., and van Leeuwen, T. physical activity (PSU HEAT Project) (2022)
T.: Global fire emissions and the contribution of International Journal of Biometeorology 66(2) DOI:
deforestation, savanna, forest, agricultural, and 10.1007/s00484-022-02316-z
peat fires (1997–2009), Atmos. Chem. Phys., 10,
11707–11735, https://doi.org/10.5194/acp-10- Warszawski, L., et al. The Inter-Sectoral Impact
11707-2010, 2010. Model Intercomparison Project (ISI-MIP): Project
framework. PNAS, 111, 9, 3228-3232, 2013.
Van Vuuren, D. P., Stehfest, E., Gernaat, D. E. H.
J., Doelman, J. C., van den Berg, M., Harmsen, WGBU: Welt im Wandel: Energiewende zur
M., de Boer, H. S., Bouwman, L. F., Daioglou, V., Nachhaltigkeit (Wissenschaftlicher Beirat der
Edelenbosch, O. Y., Girod, B., Kram, T., Bundesregierung Globale Umweltveränderung),
Lassaletta, L., Lucas, P. L., van Meijl, H., Müller, https://doi.org/10.1007/978-3-642-55861-0, 2003.
C., van Ruijven, B. J., van der Sluis, S., and
Tabeau, A.: Energy, land-use and greenhouse gas Winiwarter, W., Höglund-Isaksson, L., Klimont,
emissions trajectories under a green growth Z., Schöpp, W., & Amann, M. (2018). Technical
paradigm, Global Environ. Chang., 42, 237–250, opportunities to reduce global anthropogenic
https://doi.org/10.1016/j.gloenvcha.2016.05.008, emissions of nitrous oxide. Environmental research
2017. letters, 13(1), 014011.

240