BookPDF Available

Decoupling Debunked. Evidence and arguments against green growth as a sole strategy for sustainability. A study edited by the European Environment Bureau EEB

Authors:
  • Looking for a postdoc
  • ZOE. Institute for future-fit economies
  • Sustainable Europe Research Institute SERI Germany

Abstract

Parrique T., Barth J., Briens F., C. Kerschner, Kraus-Polk A., Kuokkanen A., Spangenberg J.H. 2019, 78 pp. Is it possible to enjoy both economic growth and environmental sustainability? This question is a matter of fierce political debate between green growth and post-growth advocates. Over the past decade, green growth clearly dominated policy making with policy agendas at the United Nations, European Union, and in numerous countries building on the assumption that decoupling environmental pressures from gross domestic product (GDP) could allow future economic growth without end. Considering what is at stake, a careful assessment to determine whether the scientific foundations behind this “decoupling hypothesis” are robust or not is needed. This report reviews the empirical and theoretical literature to assess the validity of this hypothesis. The conclusion is both overwhelmingly clear and sobering: not only is there no empirical evidence supporting the existence of a decoupling of economic growth from environmental pressures on anywhere near the scale needed to deal with environmental breakdown, but also, and perhaps more importantly, such decoupling appears unlikely to happen in the future. The validity of the green growth discourse relies on the assumption of an absolute, permanent, global, large and fast enough decoupling of economic growth from all critical environmental pressures. The literature reviewed clearly shows that there is no empirical evidence for such a decoupling currently happening. This is the case for materials, energy, water, greenhouse gases, land, water pollutants, and biodiversity loss for which decoupling is either only relative, and/or observed only temporarily, and/or only locally. In most cases, decoupling is relative. When absolute decoupling occurs, it is observed only during rather short periods of time, concerning only certain resources or forms of impact, for specific locations, and with very small rates of mitigation. There are at least seven reasons to be skeptical about the occurrence of sufficient decoupling in the future: Rising energy expenditures, Rebound effects, Problem shifting, The underestimated impact of services, The limited potential of recycling, Insufficient and inappropriate technological change, and Cost shifting. Each of them taken individually casts doubt on the possibility for sufficient decoupling and, thus, the feasibility of “green growth.” Considered all together, the hypothesis that decoupling will allow economic growth to continue without a rise in environmental pressures appears highly compromised, if not clearly unrealistic. This report highlights the need for a new conceptual toolbox to inform and support the design and evaluation of environmental policies. Policy-makers have to acknowledge the fact that addressing environmental breakdown may require a direct downscaling of economic production and consumption in the wealthiest countries. In other words, we advocate complementing efficiency-oriented policies with sufficiency policies, with a shift in priority and emphasis from the former to the latter even though both have a role to play. From this perspective, it appears urgent for policy-makers to pay more attention to and support the developing diversity of alternatives to green growth.
Decoupling Debunked Evidence and arguments against green growth as a sole strategy for sustainability
We would like to thank our funders.
For the research work, the authors like to thank the following:
Timothée Parrique acknowledges funding received from the Marie Sklodowska Curie Fellowship Action in Excellent
Research (grant agreement n° 675153).
Jonathan Barth acknowledges funding received by the KR Foundation
François Briens thanks the French unemployment allowance scheme which made his contribution possible.
Christian Kerschner acknowledges funding received from the Czech science foundation under the project VE2NEX
(GA CR 16-17978S).
Alejo Kraus-Polk acknowledges funding received from the California Department of Water Resources (grant
agreement n° 4600012167).
For the production, layout, printing and dissemination, the European Environmental Bureau, Deutscher Naturschutzring
and Zoe Institute for future-fit economies like to thank the following:
This report has been produced with the financial assistance of the Federal Ministry
of Sustainability and Tourism from the Republic of Austria. The contents of this
report are the sole responsibility of the authors and can under no circumstances be
taken as reflecting the position of the Austrian Ministry.
This report has been produced with the financial assistance of the European Union. The contents of this report
are the sole responsibility of the authors and can under no circumstances be taken as reflecting the position of
the European Union.
This report has been produced with the financial assistance of the KR foundation. The contents of this report are
the sole responsibility of the authors and can under no circumstances be taken as reflecting the position of the
foundation.
This report was produced and disseminated as part of:
The European-wide project Make Europe Sustainable for All (MESA) is coordinated by the
European Environmental Bureau (EEB) and implemented in 15 European countries by 25
partners. At the core of the project are various campaigns and advocacy activities. This report is
part of the campaign on sustainable consumption and production (2019-2020). The contents of
this report are the sole responsibility of the authors and can under no circumstances be taken as
reflecting the position #SDGS4All https://makeeuropesustainableforall.org
We would like to thank our funders.
For the research work, the authors like to thank the following:
Timothée Parrique acknowledges funding received from the Marie Sklodowska Curie Fellowship Acon in Excellent
Research (grant agreement n° 675153).
Jonathan Barth acknowledges funding received by the KR Foundaon
François Briens thanks the French unemployment allowance scheme which made his contribuon possible.
Chrisan Kerschner acknowledges funding received from the Czech science foundaon under the project VE2NEX
(GA CR 16-17978S).
Alejo Kraus-Polk acknowledges funding received from the California Department of Water Resources (grant
agreement n° 4600012167).
For the producon, layout, prinng and disseminaon, the European Environmental Bureau,
Deutscher Naturschutzring and Zoe Instute for future-t economies like to thank the following:
This report has been produced with the nancial assistance of the Federal
Ministry of Sustainability and Tourism from the Republic of Austria.
The contents of this report are the sole responsibility of the authors and
can under no circumstances be taken as reecng the posion of the
Austrian Ministry.
This report has been produced with the nancial assistance of the European
Union. The contents of this report are the sole responsibility of the authors
and can under no circumstances be taken as reecng the posion of the
European Union.
This report has been produced with the nancial assistance of the KR
foundaon. The contents of this report are the sole responsibility of the
authors and can under no circumstances be taken as reecng the posion
of the foundaon.
This report was produced and disseminated as part of:
The European-wide project Make Europe Sustainable for All (MESA) is coordinated
by the European Environmental Bureau (EEB) and implemented in 15 European
countries by 25 partners. At the core of the project are various campaigns and
advocacy acvies. This report is part of the campaign on sustainable consumpon
and producon (2019-2020). The contents of this report are the sole responsibility of
the authors and can under no circumstances be taken as reecng the posion
#SDGS4All makeeuropesustainableforall.org
We would like to thank our funders.
For the research work, the authors like to thank the following:
Timothée Parrique acknowledges funding received from the Marie Sklodowska Curie Fellowship Action in Excellent
Research (grant agreement n° 675153).
Jonathan Barth acknowledges funding received by the KR Foundation
François Briens thanks the French unemployment allowance scheme which made his contribution possible.
Christian Kerschner acknowledges funding received from the Czech science foundation under the project VE2NEX
(GA CR 16-17978S).
Alejo Kraus-Polk acknowledges funding received from the California Department of Water Resources (grant
agreement n° 4600012167).
For the production, layout, printing and dissemination, the European Environmental Bureau, Deutscher Naturschutzring
and Zoe Institute for future-fit economies like to thank the following:
This report has been produced with the financial assistance of the Federal Ministry
of Sustainability and Tourism from the Republic of Austria. The contents of this
report are the sole responsibility of the authors and can under no circumstances be
taken as reflecting the position of the Austrian Ministry.
This report has been produced with the financial assistance of the European Union. The contents of this report
are the sole responsibility of the authors and can under no circumstances be taken as reflecting the position of
the European Union.
This report has been produced with the financial assistance of the KR foundation. The contents of this report are
the sole responsibility of the authors and can under no circumstances be taken as reflecting the position of the
foundation.
This report was produced and disseminated as part of:
The European-wide project Make Europe Sustainable for All (MESA) is coordinated by the
European Environmental Bureau (EEB) and implemented in 15 European countries by 25
partners. At the core of the project are various campaigns and advocacy activities. This report is
part of the campaign on sustainable consumption and production (2019-2020). The contents of
this report are the sole responsibility of the authors and can under no circumstances be taken as
reflecting the position #SDGS4All https://makeeuropesustainableforall.org
We would like to thank our funders.
For the research work, the authors like to thank the following:
Timothée Parrique acknowledges funding received from the Marie Sklodowska Curie Fellowship Action in Excellent
Research (grant agreement n° 675153).
Jonathan Barth acknowledges funding received by the KR Foundation
François Briens thanks the French unemployment allowance scheme which made his contribution possible.
Christian Kerschner acknowledges funding received from the Czech science foundation under the project VE2NEX
(GA CR 16-17978S).
Alejo Kraus-Polk acknowledges funding received from the California Department of Water Resources (grant
agreement n° 4600012167).
For the production, layout, printing and dissemination, the European Environmental Bureau, Deutscher Naturschutzring
and Zoe Institute for future-fit economies like to thank the following:
This report has been produced with the financial assistance of the Federal Ministry
of Sustainability and Tourism from the Republic of Austria. The contents of this
report are the sole responsibility of the authors and can under no circumstances be
taken as reflecting the position of the Austrian Ministry.
This report has been produced with the financial assistance of the European Union. The contents of this report
are the sole responsibility of the authors and can under no circumstances be taken as reflecting the position of
the European Union.
This report has been produced with the financial assistance of the KR foundation. The contents of this report are
the sole responsibility of the authors and can under no circumstances be taken as reflecting the position of the
foundation.
This report was produced and disseminated as part of:
The European-wide project Make Europe Sustainable for All (MESA) is coordinated by the
European Environmental Bureau (EEB) and implemented in 15 European countries by 25
partners. At the core of the project are various campaigns and advocacy activities. This report is
part of the campaign on sustainable consumption and production (2019-2020). The contents of
this report are the sole responsibility of the authors and can under no circumstances be taken as
reflecting the position #SDGS4All https://makeeuropesustainableforall.org
We would like to thank our funders.
For the research work, the authors like to thank the following:
Timothée Parrique acknowledges funding received from the Marie Sklodowska Curie Fellowship Action in Excellent
Research (grant agreement n° 675153).
Jonathan Barth acknowledges funding received by the KR Foundation
François Briens thanks the French unemployment allowance scheme which made his contribution possible.
Christian Kerschner acknowledges funding received from the Czech science foundation under the project VE2NEX
(GA CR 16-17978S).
Alejo Kraus-Polk acknowledges funding received from the California Department of Water Resources (grant
agreement n° 4600012167).
For the production, layout, printing and dissemination, the European Environmental Bureau, Deutscher Naturschutzring
and Zoe Institute for future-fit economies like to thank the following:
This report has been produced with the financial assistance of the Federal Ministry
of Sustainability and Tourism from the Republic of Austria. The contents of this
report are the sole responsibility of the authors and can under no circumstances be
taken as reflecting the position of the Austrian Ministry.
This report has been produced with the financial assistance of the European Union. The contents of this report
are the sole responsibility of the authors and can under no circumstances be taken as reflecting the position of
the European Union.
This report has been produced with the financial assistance of the KR foundation. The contents of this report are
the sole responsibility of the authors and can under no circumstances be taken as reflecting the position of the
foundation.
This report was produced and disseminated as part of:
The European-wide project Make Europe Sustainable for All (MESA) is coordinated by the
European Environmental Bureau (EEB) and implemented in 15 European countries by 25
partners. At the core of the project are various campaigns and advocacy activities. This report is
part of the campaign on sustainable consumption and production (2019-2020). The contents of
this report are the sole responsibility of the authors and can under no circumstances be taken as
reflecting the position #SDGS4All https://makeeuropesustainableforall.org
Parrique T., Barth J., Briens F., C. Kerschner, Kraus-Polk A., Kuokkanen A., Spangenberg J.H.
Corresponding author at: tparrique@gmail.com
July 2019
Date of publication: July 2019
Authors of the report:
Timothée Parrique, Centre for Studies and Research in International Development (CERDI), University
of Clermont Auvergne, France; Stockholm Resilience Centre (SRC), Stockholm University, Sweden
Jonathan Barth, ZOE.Institute for Future-Fit Economies, Bonn, Germany
François Briens, Independent, Informal Research Centre for Human Emancipation (IRCHE).
Christian Kerschner, Department of Sustainability, Governance, and Methods, MODUL University
Vienna, Austria; Department of Environmental Studies, Masaryk University, Czech Republic
Alejo Kraus-Polk, University of California, Davis, USA
Anna Kuokkanen, Lappeenranta-Lahti University of Technology, Lahti Finland
Joachim H. Spangenberg, Sustainable Europe Research Institute (SERI Germany), Cologne, Germany
The full report should be referenced as follows:
Parrique T., Barth J., Briens F., C. Kerschner, Kraus-Polk A., Kuokkanen A., Spangenberg J.H., 2019.
Decoupling debunked: Evidence and arguments against green growth as a sole strategy for sustainability.
European Environmental Bureau.
Corresponding author: tparrique@gmail.com
Report available online at: eeb.org/library/decoupling-debunked
Report produced for and disseminated by:
The European Environmental Bureau
www.eeb.org
With the assistance of:
Deutscher Naturschutzring
https://www.dnr.de/
With research support from:
Zoe. Institute for Future-Fit Economies
https://zoe-institut.de/en
Acknowledgements
We would like to thank everyone who has contributed to this report in various ways: Sam Bliss, Maria Brück,
Martin Bruckner, Iñigo Capellán Pérez, Mikuláš Černík, Fabrice Flipo, David Font, Jaune Freire, Mario Giampetro,
Stefan Giljum, Christoph Gran, Klaus Hubacek, Theresa Klostermeyer, Anna-Lena Laurich, Nick Meynen, Patrizia
Heidegger, Simon de Muynck, and Hannah Strobel. Special thanks to Kristin Langen for initiating the project.
The responsibility for errors remains with the authors.
Lay-out and printing: Cooperative eco-printery De Wrikker cvba - www.dewrikker.be
Illustrations: Gemma Bowcock (EEB)
Executive summary
Is it possible to enjoy both economic growth and environmental sustainability? This question is a matter
of erce political debate between green growth and post-growth advocates. Over the past decade, green
growth clearly dominated policy making with policy agendas at the United Nations, European Union,
and in numerous countries building on the assumption that decoupling environmental pressures from
gross domestic product (GDP) could allow future economic growth without end.
Considering what is at stake, a careful assessment to determine whether the scientic foundations
behind this “decoupling hypothesis” are robust or not is needed. This report reviews the empirical and
theoretical literature to assess the validity of this hypothesis. The conclusion is both overwhelmingly
clear and sobering: not only is there no empirical evidence supporting the existence of a decoupling
of economic growth from environmental pressures on anywhere near the scale needed to deal with
environmental breakdown, but also, and perhaps more importantly, such decoupling appears unlikely
to happen in the future.
It is urgent to chart the consequences of these ndings in terms of policy-making and prudently move
away from the continuous pursuit of economic growth in high-consumption countries. More precisely,
existing policy strategies aiming to increase eciency have to be complemented by the pursuit of
suciency, that is the direct downscaling of economic production in many sectors and parallel reduction
of consumption that together will enable the good life within the planet’s ecological limits. In the view
of the authors of this report and based on the best available scientic evidence, only such strategies
respect the EU’s ‘precautionary principle,’ the principle that when the stakes are high and the outcomes
uncertain, one should err on the side of caution.
The fact that decoupling on its own, i.e. without addressing the issue of economic growth, has not been
and will not be sucient to reduce environmental pressures to the required extent is not a reason
to oppose decoupling (in the literal sense of separating the environmental pressures curve from the
GDP curve) or the measures that achieve decoupling - on the contrary, without many such measures
the situation would be far worse. It is a reason to have major concerns about the predominant focus
of policymakers on green growth, this focus being based on the awed assumption that sucient
decoupling can be achieved through increased eciency without limiting economic production and
consumption.
4 5
Main ndings:
> Discussing decoupling requires using a rigorous analytical framework. Depending on the
indicators considered to account for economic activities and environmental pressures as well
as the range of their evolution, decoupling can be characterised in dierent ways. It can be
global or local, relative or absolute, territorial- or footprint-based, happen over a short or a long
period of time, and last but not least, it should be put in perspective with relevant environmental
thresholds, political targets and the global socio-economic context, as to assess its adequacy in
magnitude taking into account equity considerations.
> The validity of the green growth discourse relies on the assumption of an absolute, permanent,
global, large and fast enough decoupling of economic growth from all critical environmental
pressures. The literature reviewed clearly shows that there is no empirical evidence for such
a decoupling currently happening. This is the case for materials, energy, water, greenhouse
gases, land, water pollutants, and biodiversity loss for which decoupling is either only relative,
and/or observed only temporarily, and/or only locally. In most cases, decoupling is relative. When
absolute decoupling occurs, it is observed only during rather short periods of time, concerning
only certain resources or forms of impact, for specic locations, and with very small rates of
mitigation.
> There are at least seven reasons to be sceptical about the occurrence of sucient decoupling
in the future. Each of them taken individually casts doubt on the possibility for sucient decoupling
and, thus, the feasibility of “green growth.” Considered all together, the hypothesis that
decoupling will allow economic growth to continue without a rise in environmental
pressures appears highly compromised, if not clearly unrealistic.
1 Rising energy expenditures. When extracting a resource, cheaper options are
generally used rst, the extraction of remaining stocks then becoming a more
resource- and energy-intensive process resulting in a rising total environmental
degradation per unit of resource extracted.
2 Rebound eects. Eciency improvements are often partly or totally compensated by
a reallocation of saved resources and money to either more of the same consumption
(e.g. using a fuel-ecient car more often), or other impactful consumptions (e.g. buying
plane tickets for remote holidays with the money saved from fuel economies). It can
also generate structural changes in the economy that induce higher consumption
(e.g. more fuel-ecient cars reinforce a car-based transport system at the expense
of greener alternatives, such as public transport and cycling).
3 Problem shifting. Technological solutions to one environmental problem can create
new ones and/or exacerbate others. For example, the production of private electric
vehicles puts pressure on lithium, copper, and cobalt resources; the production of
biofuel raises concerns about land use; while nuclear power generation produces
nuclear risks and logistic concerns regarding nuclear waste disposal.
4 The underestimated impact of services. The service economy can only exist on
top of the material economy, not instead of it. Services have a signicant footprint
that often adds to, rather than substitute, that of goods.
5 Limited potential of recycling. Recycling rates are currently low and only slowly
increasing, and recycling processes generally still require a signicant amount of
energy and virgin raw materials. Most importantly, recycling is strictly limited in its
ability to provide resources for an expanding material economy.
6 Insucient and inappropriate technological change. Technological progress is
not targeting the factors of production that matter for ecological sustainability and
not leading to the type of innovations that reduce environmental pressures; it is not
disruptive enough as it fails to displace other undesirable technologies; and it is not
in itself fast enough to enable a sucient decoupling.
7 Cost shifting. What has been observed and termed as decoupling in some local cases
was generally only apparent decoupling resulting mostly from an externalisation of
environmental impact from high-consumption to low-consumption countries enabled
by international trade. Accounting on a footprint basis reveals a much less optimistic
picture and casts further doubt on the possibility of a consistent decoupling in the
future.
> This report highlights the need for a new conceptual toolbox to inform and support the design and
evaluation of environmental policies. Policy-makers have to acknowledge the fact that addressing
environmental breakdown may require a direct downscaling of economic production
and consumption in the wealthiest countries. In other words, we advocate complementing
eciency-oriented policies with suciency policies, with a shift in priority and emphasis
from the former to the latter even though both have a role to play. From this perspective, it
appears urgent for policy-makers to pay more attention to and support the developing diversity
of alternatives to green growth.
6 7
Contents
Introduction ...............................................................................................................9
I. What is decoupling? ......................................................................................... 11
1. Relative and absolute decoupling ........................................................................................11
2. The driving variable: Gross Domestic Product ..................................................................12
3. The driven variable: Resources and impacts .....................................................................12
4. Scale: Global or local ...............................................................................................................13
5. Durability: Temporary or permanent ..................................................................................14
6. Magnitude: Sucient or insucient ....................................................................................15
7. Equity in the allocation of decoupling eorts ...................................................................16
Conclusions for Section 1 ............................................................................................................18
II. Is decoupling happening? ............................................................................ 19
1. Resource decoupling ...............................................................................................................20
Materials ................................................................................................................................................... 20
Energy ....................................................................................................................................................... 21
Water ......................................................................................................................................................... 21
2. Impact decoupling ...................................................................................................................24
Greenhouse gases ................................................................................................................................. 24
Land ........................................................................................................................................................... 28
Water pollutants ..................................................................................................................................... 29
Biodiversity loss ...................................................................................................................................... 31
Conclusions for Section 2 ............................................................................................................32
III. Is decoupling likely to happen?................................................................ 33
1. Rising energy expenditure .....................................................................................................33
2. Rebound eects .......................................................................................................................37
3. Problem shifting .......................................................................................................................40
4. The underestimated impact of services .............................................................................42
5. Limited potential of recycling ................................................................................................46
6. Insucient and inappropriate technological change ......................................................49
7. Cost shifting ...............................................................................................................................53
Conclusions for Section 3 ............................................................................................................55
Conclusions: Farewell to green growth ...................................................... 57
Bibliography ............................................................................................................ 60
Appendix ..................................................................................................................72
8 9
Introduction
Is economic growth compatible with ecological sustainability? Almost half a century after
the publication of the Meadows report “Limits to growth” and Sicco Mansholt’s letter to the
President of the European Commission in 1972 in defence of a shift away from the pursuit of
economic growth, the relation between Gross Domestic Product (GDP) and environmental
pressures remains a matter of erce political debate.
The debate has two main sides. Proponents of what has been named “green growth”
argue that technological progress and structural change will enable a decoupling of natural
resources consumption and environmental impacts from economic growth. On the other
hand, advocates of “degrowth” or “post-growth” argue that, because an innite expansion of
the economy is fundamentally at odds with a nite biosphere, the reduction of environmental
pressures requires a downscaling of production and consumption in wealthiest countries,
which is likely to result in a decrease in GDP compared to current levels. On one side,
green growth advocates expect eciency to enable more goods and services at a lower
environmental cost; on the other, degrowth proponents appeal to suciency, arguing that
less goods and services is the surest road to ecological sustainability.
Today, the green growth narrative dominates most political circles. In 2001, the OECD
ocially adopted decoupling as a goal, which later came to play a key role in its strategy
Towards Green Growth (2011).1 It was then followed by the European Commission who, in its
6th Environment Action Programme (Environment 2010: Our Future, Our Choice), announced
its objective to “break the old link between economic growth and environmental damage”
(EU Commission, 2001, p. 3). The commitment of “decoupling growth from resource use”
was repeated in the EU Roadmap to a Resource-Ecient Europe (European Commission,
2011), and in the United Nations Environment Programme’s strategy on green economy
(2011a, p. 18) where green growth was expected to “signicantly reduce environmental
risks and ecological scarcities.”23 Soon after, the World Bank joined the bandwagon with
1 Which they defined as the “breaking of the link between ‘environmental bads’ and ‘economic goods’ ” (OECD, 2002, p. 1).
2 “A key concept for framing the challenges we face in making the transition to a more resource efficient economy is decoupling. As global
economic growth bumps into planetary boundaries, decoupling the creation of economic value from natural resource use and environmental
impacts becomes more urgent” (UNEP, 2011b, pp. 15–16, italics added).
3 “Target 8.4: Improve progressively, through 2030, global resource efficiency in consumption and production and endeavour to decouple eco-
nomic growth from environmental degradation, in accordance with the 10-Year Framework of Programmes on Sustainable Consumption and
Production, with developed countries taking the lead” (italics added).
INTRODUCTION INTRODUCTION
10 11
Inclusive Green Growth: The Pathway to Sustainable Development (2012).4 Since 2012, the
7th Environmental Action Programme guiding the European Commission’s environmental
policy until 2020 Living well, within the limits of our planet (European Commission, 2013) calls
for “an absolute decoupling of economic growth and environmental degradation.” And in
2015, decoupling became a specic target in the Sustainable Development Goals.
Green growth has dominated the discussion and set most of the environmental agenda
based upon the expectation of a decoupling of economic growth and environmental
pressure. A situation with such high stakes calls for a careful assessment to determine
whether the scientic foundations behind the decoupling hypothesis are robust or not.
This is the subject of this report, and as its title clearly indicates, we have found insucient
theoretical and empirical support to warrant the hopes currently placed in decoupling.
The literature on decoupling is abundant. Starting in 2011, the United Nations Environment
Programme (UNEP) has produced a series of report on the topic (UNEP, 2011b, 2014a,
2015). Searching the keywords “decoupling economic growth” on Scopus delivers more than
600 articles, most of them empirical. On such a controversial topic, one would expect wide
divergence in results. Yet, as we will show in the second section of this report, disagreements
within that literature mainly result from slight variations in the way decoupling is dened
and measured. Once these methodological quirks are set aside, ndings converge towards
showing that there is no robust evidence justifying the idea of decoupling as a single or
main policy strategy as it is currently promoted by green growth advocates.
This report is organised in three sections. First, we dene what decoupling means and
specify the dierent forms that it can take. The main point of this section is that behind one
term hides various dierent meanings or situations, some of them more desirable than
others. In the second section, we review the empirical literature on the topic as to assess
whether or not there is evidence of decoupling having occurred in the past. Our nding is
that current scientic knowledge does not support the hypothesis of the type of decoupling
that would be necessary to eectively address climate change and other environmental
crises. In the third section, we discuss how likely is decoupling to occur in the future and
nd that probabilities are too thin to warrant the current central focus placed on the
concept in policy making. In conclusion, the main claim of the report is that green growth,
that is, economic growth that is suciently decoupled from environmental pressures is not
possible and should thus not be the primary objective of environmental policy.
4 For the World Bank (2012): inclusive green growth is “economic growth that is efficient in its use of natural resources, clean in that it minimizes
pollution and environmental impacts, and resilient in that it accounts for natural hazards and the role of environmental management and natural
capital in preventing physical disasters.”
I. What is decoupling?
A constructive discussion requires starting with explicit denitions and clarifying several
terminological and methodological subtleties, having to do with what type of economic
and environmental indicators are considered and how they are statistically correlated; at
which scale, magnitude, and timing decoupling may or may not occur; as well as for what
outcomes in terms of achieving social and environmental targets.
1. Relative and absolute decoupling
Generally speaking, two variables are said coupled if one is driven by the other, making them evolve
in proportion (for instance, more of A means more of B); and they decouple when they cease to
do so. When coupled, both the driven and driving variables move in step, which means that they
evolve over time proportionally. Decoupling refers to a variation over time of the coecient of
proportionality, corresponding to a desynchronization between the two variables tends.
This decoupling can be either relative or absolute (also called weak or strong). Relative
decoupling means that both variables still develop into the same direction but not at the
same speed (a lot of more of A means a little more of B) whereas absolute decoupling
means that the two variables go in opposite directions (more of A and less of B). Assessing
decoupling means estimating the loss of proportionality between one variable towards
another (or more precisely the variable trends) over time.
Relative decoupling, for example between GDP and carbon emissions, refers to a situation
where the emissions per unit of economic output (the coecient of proportionality) declines
but not “fast enough” to compensate for the simultaneous increase in output over the same
period, resulting in an overall increase in total emissions. As a result, although the economy
is relatively less impactful per unit of GDP compared to what it was before, the absolute
volume of emissions has nonetheless increased.
Absolute decoupling is a situation where, to stay with the same example, more GDP coincides
with lower emissions. Relative decoupling becomes absolute decoupling when the growth
INTRODUCTION WHAT IS DECOUPLING?
12 13
rate of the economy is overcompensated by the growth rate of eciency or productivity
having to do with the use of natural resources and the generation of pollutions – a threshold
sometime referred to as the “absolute decoupling point” (Akizu-Gardoki et al., 2018).
When decoupling is absolute, environmental pressure declines without a corresponding
drop in economic activities, or vice versa, economic activities rise without an increase in
environmental pressure.
2. The driving variable: Gross Domestic Product
In the decoupling of economic growth from environmental pressures, the rst term refers to
a measure of market activity, most often Gross Domestic Product (GDP).5 GDP is a measure
of the aggregate market value of all the nal goods and services produced in a country in a
given period of time (often annually), and it is the change of that value that is called economic
growth. Calculating GDP is an intricate process resulting from a number of conventions and
involving a number of subtleties having to do with what to include and exclude and how to
measure it. Since its creation in the 1930s, GDP has been criticised on a number of grounds.
Although this is not the space to review such criticisms, one should still say that the primacy
of this indicator reects a narrow, potentially problematic, framing of prosperity. This being
said, in our context, it matters to take into consideration GDP evolutions in volume or “real
GDP,” that is to say to correct GDP from ination.
3. The driven variable: Resources and impacts
Environmental pressures include all the consequences an economy has on nature. Following
UNEP (2011b), it is possible to distinguish between resource use and environmental impacts.
Resource decoupling is a decoupling of market activity from the volume of resource used
(i.e. extracted from the environment), for example thanks to eciency improvements or
better recycling which both allow for less extraction. It means that the same or a larger
output in monetary terms can be produced with fewer material inputs. The term “resource”
here refers to “natural assets deliberately extracted and modied by human activity for
their utility to create economic value” (UNEP, 2011b, p. 2).6 In this report, we will divide the
natural resources used for economic activities in four categories: materials,7 energy, water,
and land (the latter two dened broadly as to include biodiversity and related ecosystem
services). These resources can be measured using dierent indicators either production-
based (e.g. domestic extraction, primary energy supply, land occupation) or consumption-
based (e.g. material footprint, energy footprint, water footprint, or ecological footprint).
5 There exist other ways of quantifying economic activity, such as total working time or aggregate employment. A small minority of decoupling
studies focus on more encompassing indicators such as the Human Development Index (Akizu-Gardoki et al., 2018); the Index of Sustainable
Economic Welfare (Beça and Santos, 2014); need satisfiers and human well-being (O’Neill et al., 2018). In the report, however, we only focus on
economic growth measured as an increase in GDP for that it is measured as such in the great majority of decoupling studies.
6 The way one accounts for resources matters. For example, including unused extraction of materials (the materials and energy being used, dis-
placed, or damaged in the process of extraction itself) often leads to calculated volumes a few order of magnitude higher than only counting the
inputs to the production process itself. In the case of Chile, for example, the physical trade balance in the year 2003 increases from net exports
of 1 million tons in terms of direct flows to net exports of 634 million tons if calculated including unused extraction materials (Muñoz et al., 2009).
7 Materials can be further broken down into more detailed categories such as, for example, biomass, fossil energy carriers, ores and industrial
minerals, and construction minerals (Fischer-Kowalski et al., 2011: 10).
Impact decoupling refers to a decoupling of GDP from environmental impacts, that is a
decrease in environmental harm per unit of economic output. Environmental impacts
can take various forms such as waste disturbing marine life or pollutants aecting human
and animal health, disturbance of natural processes (e.g. nitrogen, phosphorus, carbon,
and fresh water cycles), or biodiversity loss. There is usually a link between resource use
and environmental impacts; for example, extracting and using more fossil fuels (resource)
generates CO2 emissions contributing to climate change (impact). Although most empirical
studies focus on climate change and greenhouse gas emissions, any deleterious eects
on the biosphere can be taken into consideration as an environmental variable (e.g. light
pollution leading to biodiversity loss, water pollution leading to eutrophication).
In this report, we will refer to overall decoupling for cases where decoupling occurs between
GDP and all selected indicators, including both resource use and environmental impacts.
And we will refer to partial decoupling for cases where one or more environmental indicator
decouples from GDP while coupling remains or intensies for other indicators.
4. Scale: Global or local
Decoupling can be discussed taking into consideration dierent geographical perimeters.
Local decoupling refers to cases where decoupling is observed between variables relative to
a restricted geographical perimeter (e.g. a country or a water basin), while global decoupling
corresponds to decoupling between two variables at the planetary scale (e.g. world GDP
and world greenhouse gas emissions).8
The relevance of using local or global indicators depends on the nature of the environmental
pressure considered and on its causes. For instance, to study local issues, such as the
eutrophication of the Baltic Sea, for which direct causes are located in a rather well dened
geographic area, it makes sense to use local indicators, limited for example to the perimeter
of the watershed. However, global issues like climate change generally call for global
indicators, since greenhouse gases are transboundary pollutants and climate change is a
planetary phenomenon.
In a globalised world, the choice of the boundaries considered for the system under
study matters. Globalisation and the expansion of international trade has led to a spatial
dissociation between places of extraction, production, and consumption, making it more
dicult to determine who is responsible for which impacts. In this context, production-
based (also called territorial) indicators, which relate to geographical areas rather than to
populations, cannot reect responsibilities and are as such insucient. A more comprehensive
approach consists in looking at consumption-based (also called footprint) indicators, in which
embodied impacts from production and end-of-life phases of traded goods and services
are geographically reallocated to nal consumers. Indeed, not accounting for the resources
mobilised and for impacts generated abroad may lead to detecting apparent decoupling at
8 One could even go further and differentiate several local levels: macroeconomic (for instance taking into account the whole national activity),
sectoral (a specific sector of the economy), and microeconomic (single company, city, household). In this report this will not be necessary for that
the majority of empirical studies are either national, regional, or global.
WHAT IS DECOUPLING? WHAT IS DECOUPLING?
14 15
a local level for importing countries which translocate impacting activities abroad. Reversely,
territorial approaches might underestimate decoupling in the case of exporting countries
who host impacting activities intended for the consumption of other nations.
5. Durability: Temporary or permanent
Just like the geographical perimeter, the time period of a decoupling study matters. Indeed,
mitigating environmental pressures in a growing economy not only implies achieving
absolute decoupling from GDP, but also requires maintaining such a decoupling in time
as long as the economy grows. Said dierently, continuous economic growth requires a
permanent absolute decoupling between GDP and environmental pressures. Yet, in the
same way that economic growth and environmental pressures can decouple at one point in
time, they can also recouple later on. As empirical studies often show, decoupling can as well
be temporary, resulting in a further increase of environmental pressures after a temporary
relief. In the literature, this situation is depicted by an N-shaped curve and sometimes
referred to as recoupling or “relinking” (de Bruyn and Opschoor, 1997; Jänicke et al., 1989)
indicating a ‘delinking’ of environmental pressure from economic growth in relation to rising
per capita incomes. The likelihood of such a relationship being persistent is discussed in the
context of a simple macro model of industrial metabolism, and the possibility of ‘relinking’
clearly emerges. Data on specic indicators of environmental pressure (i.e., the throughput
of materials, energy and the volume of transport.
Such pattern can for instance result from a large shift in energy sources. For example, China
moving from coal toward oil and gas and the US increasing the portion of natural gas in their
energy mix caused a temporary levelling of global emissions in 2015 and 2016 reported by
the International Energy Agency. But this decoupling was short-lived: once the shift was
completed and the corresponding decoupling potential spent, emissions recoupled with
economic growth (+1.6% in 2017 and +2.7% in 2018) (Hickel and Kallis, 2019: 8). Another
common example of temporary decoupling is the Global Financial Crisis of 2007-2008 which,
as we will see in detail in Section 2, has momentarily pushed environmental pressures down.
From an ecological sustainability perspective, the necessary type of decoupling is one that
is permanent and not only temporary. Indeed, it makes little sense to cut resource use or
emissions drastically in the short-term only to fall back on a path of increased biophysical
intensity in the longer term. Besides, temporary decoupling only has a marginal eect on
environmental pressures resulting from cumulative impacts, an eect which merely boils
down to a time lag. Findings from decoupling studies should therefore be put in perspective
with the time period considered for what may look like decoupling over a short period
(inverted U-shape curve) might look dierent over a longer period (N-shape curve).
6. Magnitude: Sucient or insucient
A 3% rise in GDP with a 2% drop in total greenhouse gas emissions is by denition absolute
decoupling, but so is a 3% rise in GDP with a 0.02% drop in emissions. Plain to see that the
rst is more desirable if the goal is to mitigate climate change. Our point is the following: the
success of a decoupling strategy should be assessed in relation to specic environmental
targets, and not in terms of abstract decoupling elasticities as often done in the literature.
Once such targets have been dened, one can then speak of decoupling being insucient
or sucient in achieving them – e.g. “absolute decoupling within planetary boundaries” for
Fedrigo-Fazio et al. (2016).
Furthermore, talking about emission or resource productivity measured in emissions/
resource per unit of GDP obscures the fact that most environmental issues are caused by
cumulative, absolute impacts from dierent factors. In reality, not only does this imply that,
to be eective, the required decoupling would have to be covering both resource use and
impacts, in both dimensions being absolute, global, and permanent, but it would also need
to be suciently fast. Long before being exhausted, non-renewable resources get scarce
and can create conicts or exacerbate already existing ones. Adaptation is even more
dicult in the case of ecosystem overload; once overwhelmed i.e. if tipping points have
been passed –, they can collapse or transform into a dierent kind of system (a forest area
becoming savannah, for instance). Both kinds of damage exhaustion and collapse are
often irreversible on a time-scale relevant for humans. Even though it is dicult to measure,
decoupling can be considered suciently fast if the absolute decoupling point is reached
before passing irreversible thresholds of damage such as the nine planetary boundaries
identied by Rockström et al. (2009), Steen et al. (2015) and Steen et al. (2018).9
Climate change provides a good example of a hard deadline for absolute impact decoupling.
With a global carbon budget estimated at 580 GtCO2 that is currently being depleted at the
pace of 42 GtCO2 per year, this leaves only 12 years at current rates of emissions. Reaching
the net zero anthropogenic CO2 by 2040 necessary to limit global warming to 1.5° which a
high level of condence requires an annual reduction of at least 5% of the current emissions,
i.e. a reduction of 8.2 GtCO2 every year. Following this trajectory, the budget will last 20
years and the emissions will be zero at the end of the period with 45% decline in global
emissions by 2030 as an interim target (IPCC, 2018). In light of this constraint, and as we will
show in Section 2, even the decrease of emissions achieved in the most successful national
cases of absolute decoupling are far from being sucient to keep global warming from
passing a critical threshold.
Urgency does not only concern impacts but also resources. The preservation of non-
renewable resources is a matter of intra- and intergenerational equity. Each non-renewable
resource used in one place is a resource that will not be available in another place, and each
non-recyclable resource used today is a resource that will not be available tomorrow. As
for renewable ones, the threshold of sustainable consumption is set by the replenishment
9 To be precise, one should say that the environmental pressure occurring after the decoupling point, even though decreasing, still matters.
Enough resources or carbon budgets (or any other measure of resource and impacts) should be left as to be able to afford the descent from the
peak with still remaining within thresholds of ecosystem stability.
WHAT IS DECOUPLING? WHAT IS DECOUPLING?
16 17
rates of that resource (e.g. avoiding a sh stock being depleted to extinction or the collapse
of soil structure). So when UNEP (2014a, p. 123) concludes their report by arming that
“absolute decoupling of economic growth from resource use is possible,” we want to point
out that it is the magnitude and timing of that decoupling which is at stake more than its
mere statistical existence.
7. Equity in the allocation of decoupling eorts
The last dimension comes on top of the previous one and is about the concept of “shared
but dierentiated responsibilities” that ever since rst agreed at the 1992 United Nations
Conference on Environment and Development in Rio gures in climate agreements.
Decoupling needs to be suciently large in auent countries in order to free the ecological
space necessary for production and consumption in regions where basic needs are unmet.
The fact that there are millions of people in the world who lack access to the means of
satisfying their basic needs puts extra pressures on rich nations to reduce environmental
pressures as much as possible as to give the largest possible leeway to vulnerable communities.
If moving the “global poor” to an income level of US$ 3-8 per day will by itself consume
66% of the available 2°C global carbon budget (Hubacek et al., 2017), then it is imperative
for auent nations to let go of the remaining available climate space. Meyer-Ohlendorf
et al. (2018) calculate that, if the share of carbon budget is derived from 2050 population
numbers as to better account for equity, the current EU target for 2030 would have to
almost double, from 40% reduction of emissions to 71%. Indeed, even if the metabolic
rates of industrial countries would remain stable at 2000 levels (which would already imply
absolute decoupling), the catching up of the rest of the world, using current technology,
would in itself quadruple global emissions by 2050 (Fischer-Kowalski et al., 2011: 29), which
corresponds to levels considered catastrophic in the latest IPCC report (IPCC, 2018).
And again, in world of limited resources, the timing of the peak impact matters as the “safe
operating space” (Steen et al., 2015) may not be large enough for every nation to peak in
a logic of “grow now, clean up later” (Van Alstine and Neumayer, 2010, p. 57). For example,
Storm and Schröder (2018, pp. 20–21) estimate that if China develops along the path of the
production-based Environmental Kuznets Curve they nd for CO2 emissions, they would
exhaust the entirety of the world carbon budget before even reaching the hypothetical
turning point. Decoupling in rich countries can be considered large enough if it compensates
for the increased ecological footprint of poorer nations while still managing to absolutely
and permanently decouple global economic growth from environmental pressures at a
pace that is fast enough to avoid overshooting safe environmental thresholds.10
10 This is a moral, and not a technical, question. Our main point here is that an abstract objective of decoupling is senseless if not connected to
concrete environmental targets, which should themselves be based on moral considerations.
WHAT IS DECOUPLING? WHAT IS DECOUPLING?
18 19
II. Is decoupling happening?
Is decoupling occurring in reality, and if yes, what kind of decoupling is it? The objective
of this section is to assess the validity of the decoupling hypothesis in light of existing
empirical research. To do so, we conduct a comprehensive literature review of a number
of empirical studies having tested the decoupling hypothesis. The discussion that follows is
organised thematically, according to the environmental variables considered: (1) resources
(materials, energy, and water) and (2) impacts (greenhouse gases, land, water pollutants, and
biodiversity loss). In each case, we compare the results reported across studies assessing
them with respect to the dierent dimensions presented in Section 1.
Before diving into the empirical literature, it is worth telling the story of how scientists
came to talk of decoupling in the rst place. In the 1990s, several economists (Grossman
and Krueger, 1995, 1991; Panayotou, 1993; Shak and Bandyopadhyay, 1992) conducted
empirical work that led them to believe that economic growth was negatively correlated
with environmental pressures.11 Environmental impacts would rst grow but then decline
in an inverted bell shaped development that came to be referred to as an Environmental
Kuznets Curve.12 This theory had strong policy implications as it meant that a nation could
grow its way out of an ecological crisis.
This hypothesis of what UNEP (2014a, p. 5) calls a “decoupling through maturation” has
inspired a number of studies in the following decades looking for EKCs for a selection of
environment variables. Today, such assumption of a naturally-occurring decoupling has lost
traction in both scientic and political scenes while it has been recognised that the structural
change of economies leading to decoupling is strongly determined by policies (Smith et al.,
2010; UNEP, 2014a). The way to study decoupling has thus evolved from a semi-natural
phenomenon to something that can be brought into existence via policy intervention.
11 Grossman and Krueger (1991)”publisher”:”National Bureau of Economic Research”,”genre”:”Working Paper”,”source”:”National Bureau of Eco-
nomic Research”,”abstract”:”A reduction in trade barriers generally will affect the environment by expanding the scale of economic activity, by
altering the composition of economic activity, and by bringing about a change in the techniques of production. We present empirical evidence
to assess the relative magnitudes of these three effects as they apply to further trade liberalization in Mexico. In Section 1. we use comparable
measures of three air pollutants in a cross-section of urban areas located in 42 countries to study the relationship between air quality and eco-
nomic growth. We find for two pollutants (sulfur dioxide and \”smoke\” studied air pollutants (sulphur dioxide and other particulates); Shafik
and Bandyopadhyay (1992) focused on water pollution, municipal waste, particulates, sulphur dioxide, deforestation, and carbon emissions; and
Panayotou (1993) considered an array of similar environmental indicators.
12 In 1955, Simon Kuznets elaborated the theory that in the process of expanding economic activity, inequality first increased to a maximum and
then decreased – thus forming an inverted U-shaped curve.
IS DECOUPLING HAPPENING? IS DECOUPLING HAPPENING?
Conclusions for Section 1
As we have shown in this section, decoupling can be dened and measured in dierent
ways. Consequently, carrying a literature review on decoupling calls for a number
of precautions. First, one should be clear about what is being decoupled from what,
specifying the indicators chosen to account for economic activities and environmental
pressures. In particular, one should consider whether these indicators are global or
local and whether they reect territorial (production-based) or footprint (consumption-
based) approaches (scale). Then it matters whether decoupling is studied and discussed
in relative or absolute terms, and over a short or long period of time (durability). Last
but not least, any observed decoupling should be put in perspective with relevant
environmental thresholds and within a broader political context as to assess whether
it manages to reach mitigation targets (magnitude) in a way that is deemed just (equity).
Building on this analytical framework, the next section proposes a review of existing
empirical literature on decoupling.
20 21
In this perspective, we will present a review of recent empirical studies that attempted to
identify decoupling phenomena, adding to existing literature reviews like Li et al., (2007),
Koirala et al., (2011) or Mardani et al., (2019). Although, the literature review we conduct is
one of the most encompassing up to date, it is not systematic and exhaustive, and so the
claims we make about the decoupling literature should be appreciated in regards to the
limited pool of articles under consideration (see full list in appendix). It is also worth noting
that the majority of studies have to do with developed countries (with notable exceptions,
e.g. Wang et al., 2019)this work has made following contributions: (i, and so the claim we
make about decoupling should be understood in that context.13
1. Resource decoupling
Materials
When it comes to aggregate use of materials, the evidence is clear and uncontroversial:
there has been no absolute decoupling of resource use from economic growth. In fact,
the global use of resources is on the rise and global GDP is still tightly coupled with total
resource extraction. (Here and in the remaining of this section, if not otherwise indicated,
the decoupling eects are estimated on the basis of production-based environmental
variables.)
Global material extraction has increased by a factor of 12 in between 1900 and 2015, with a
steady acceleration since the beginning of the 21st century (Krausmann et al., 2018).14 In the
last century, average resource use per capita doubled: a global inhabitant in 2005 required
somewhere between 8.5 (Behrens et al., 2007) and 9.2 tons (Krausmann et al., 2009) of
resources annually, while a hundred years earlier this number was only 4.6 tons (UNEP,
2011b, p. 10).15 Only in the last 40 years, total material use at the global level has tripled
(Schandl et al., 2018). The material footprint of the OECD nations as a whole increased by
almost 50% between 1990 and 2008 in direct relationship with economic activity with every
10% rise in GDP being accompanied with a 6% increase in material footprint (Wiedmann
et al., 2015).16 In the end, the material intensity of GDP per capita has increased by 60%
between 1900 and 2009 (Bithas and Kalimeris, 2018).17
Global material footprint targets are less consensual than carbon targets, and yet an
emerging consensus holds that material consumption needs to be capped to a yearly
13 This is not to say that decoupling is easier in the global South. Nor do we mean that the questions at hand in this report are solely a concern
for the global North; ecological sustainability should be a matter of concern for all. However, we assume that if the global North fails to decouple,
it will be hard to justify why decoupling should be expected to happen in low-income and technologically less advanced countries.
14 Global material extraction increased by 53 per cent between 2002 and 2015, which means that “roughly one third of all materials that have
been extracted since 1900 have been mobilized between 2002 and 2015 only” (Krausmann et al., 2018: 139).
15 Schandl et al. (2018: 4) notes that most of this increase is recent. Indeed, average global material extraction has risen from 7 tones per capita
in 1970 to 10 in 2010.
16 Bithas and Kalimeris (2018) confirms this dependency of the global economy on natural resources. They calculate that the global per capita
consumption of mass resources increased by 78.7 per cent over the last century (1900-2002); this means that a 4.8-fold increase in global in-
come led to a 8.5-fold rise in mass flow. Considering biomass, fossil energy carriers, ores and industrial minerals, and construction minerals,
Krausmann et al. (2018) calculate that global material use increased by a factor of 12 over the 1900-2015 period with a marked shift from the
dominance of renewable biomass towards mineral materials.
17 Same result for Giljum et al. (2014): 93.4% increase in global consumption between 1980 and 2009, which becomes 132% when extended to the
year 2013 (“The Material Flow Analysis Portal,” 2015) Again, that rate picks up at the turn of the century: around 2.5% average increase per year
over the period but 3.4% rise between 2000 and 2009 (Giljum et al., 2014) or 3.85% between 2002 and 2013 (Materialflows.net, 2015).
maximum of 50 billion tons in order to remain ecologically sustainable (Bringezu, 2015;
Dittrich et al., 2012; Hoekstra and Wiedmann, 2014; UNEP, 2014b). In 2009, that number
was already over the threshold at 67.6 billion tons (Giljum et al., 2014).
A surprising fact shown in all studies is that while the world economy had been gradually
dematerialising for a long time, this trend has been reversed in the last two decades. While
in the last century the use of materials was relatively decoupling from GDP at the global level,
the trend has stalled and reverted since the turn of the century. For instance, Krausmann
et al. (2018) show that change in material intensity went from a negative 0.9% per year
between 1945-2002 to a positive yearly 0.4% between 2002 and 2015. Attempting the same
calculation with a dierent method, Bithas and Kalimeris (2018) nd total decreases of
material intensity in the range of 31.9% for 1900-1945 and 48.9% for 1950-2000, but only a
decrease of 0.6% in between 2000 and 2009. Giljum et al. (2014) calls it a re-materialisation,
which is the opposite of decoupling, namely an increase in the material intensity of the
world economy.
From the onset, it seems that rich countries achieve a faster relative decoupling than
others. Yet, this performance wafts away when accounting for cost shifting, i.e. looking
at consumption-based accounts. For example, Wang et al. (2018) compare consumption-
based (material footprint) and production-based measurements (domestic material
consumption) of resource use for the case of six countries, three from the OECD and three
from BRICS. Australia, Japan, India, and the USA do manage to relatively decouple, but only
because they shift their material resource supply abroad. This result is conrmed by both
Bithas and Kalimeris (2018) who report a stagnating material intensity at the global level,
and Wiedmann et al. (2015) who show that using material footprint instead of Domestic
Material Consumption (DMC) cancels an only apparent relative decoupling in the USA, UK,
Japan, the OECD, and EU-27.
One should note that the use of certain materials do decrease along a rising GDP, even
though often only locally – for example aluminium in the USA between 1985 and 2009
(Zhang et al., 2017). But this is counterbalanced by either more of the same material being
extracted elsewhere or other materials whose use rises even faster. For example, global
amounts of extracted iron ore and bauxite have increased faster than global GDP in the
1980-2002 period (Wiedmann et al., 2015).
Energy
The case of energy is less clear cut than the one of materials. Studies diverge on their
results and are dicult to compare because they measure energy consumption dierently
and do so at dierent geographical scales.
Looking at territorial nal energy consumption in the 1971-2004 period, Luzzati and Orsini
(2009) do not nd any evidence of an Environmental Kuznets Curve, neither on a global scale
nor at the level of individual countries. What they nd instead is that the relation between
GDP per capita and energy consumption is stable, both indicators increasing monotonically.
Semeniuk (2018) uses data for 180 countries between 1950 and 2014 and nds that primary
IS DECOUPLING HAPPENING? IS DECOUPLING HAPPENING?
22 23
energy intensity is constant with growth. However, Csereklyei et al. (2016) nd cases of only
relative decoupling between primary energy consumption and GDP for 99 countries over
the 1970-2011 period.
Wu et al. (2018) do nd three cases of absolute decoupling (US, France, and UK) between
2005 and 2015 using production-based approaches (they use decoupling indices which do
not specify how much energy consumption actually decreased) and one case of relative
decoupling in Germany. Wood et al. (2018) nd a relative global decoupling trend for the
period 1995-2011 between nal consumption and GDP. However, it is more common for
authors to nd situations of relative decoupling, mostly at a regional scale: Ward et al. (2016)
in Australia for nal energy consumption, Kovacic et al. (2018) in 14 EU countries (1995-
2013) between energy consumption and hours of labor, Conrad and Cassar (2014) in Malta
(1995-2012), and van Caneghem et al. (2010) in the Flemish industry (1995-2006).
Yet, as for the case of material, a decoupling in one region often hides a recoupling
somewhere else. Moreau and Vuille (2018) test this hypothesis using input-output analysis
for Switzerland between 2000 and 2014. Result: the decrease in territorial nal energy
intensity appears to be compensated by an increase in the energy embodied in imports.
Taking this into account, energy intensity remains roughly the same. In that specic study,
absolute volumes increase both when measured using a territorial approach (+1%, which
is the result of a domestic energy intensity decreasing by 44% being met by an increase in
volume of 45%) and when using a footprint approach (+24.5%), even though the dierence
is signicant. Examining the often-quoted relative decoupling of energy consumption from
economic growth in the UK over the last 15 years, Hardt et al. (2018) show that the majority
of energy intensity improvements is not due to better eciency but instead to oshoring.
The illusion is not only geographical but also sometime sectoral. Using sectoral data for 18
EU countries between 1995 and 2008, Naqvi and Zwickl (2017) nd that even though on
average, relative decoupling occurs in almost all sectors, no country manages to absolutely
decouple nal energy use and GDP in the economy as a whole.
Lastly, that decoupling occurs during a certain period does not guarantee maintaining it
over time. Analysing the Czech Republic, Hungary, Poland, and Slovakia over the 1990-2015
period, Szlavik and Szép, (2017) show that if absolute decoupling occurred at all, it lasted
only during short periods and only in specic places, for example in Poland from 2011 to
2014. This ephemeral breaks in the coupling relation are most often explained by economic
crises and political restructuring, and not by the continuous introduction of ever more
ecient technologies and practices.
Water
Decoupling can be observed on a variety of metrics of water “use,” including water withdrawal
(also called abstraction), which measures the amount of water taken from a natural source
(such as a lake or a river), and water consumption, which measures water used that will not
be returned to its source, and thus not available for reuse.18 The UNEP recently published
18 Decoupling can also be observed between withdrawal and pollution, as well as on per-capita or total use within different economic sectors
that use water, which we will discuss in next part on impact decoupling.
a report entitled Decoupling Economic Growth From Water Use And Water Pollution (UNEP,
2015), which argues that using territorial indicators of water use, many countries have
achieved a relative form of decoupling (UN-Water, 2009), and so has the world as whole
starting in the 1940s (UNEP, 2015, p. 12). Similar production-oriented studies show that the
pace of decoupling signicantly increased after the 1980s, with the global water intensity
of production declining yearly by 1% from 1980 to 2000 (Dobbs et al., 2011). China is a
striking example, with water consumption remaining constant since the 1980s alongside
several decades of two-digit economic growth (Gleick, 2003). Some countries have even
experienced absolute decoupling. This is the case of Australia that has reduced its total
water consumption by 40% over the 2001-2009 period while increasing GDP by over 30%
(Smith, 2011).
As promising as these numbers look, relative decoupling of water and eciency gains were
more than cancelled by the expansion of economic activities, resulting in a net increase in
water consumption. Industrialising countries or regions may indeed reduce overall water
use by decreasing agricultural production. Yet, decreases in agricultural production in one
place require increases elsewhere, and even water-ecient industrialisation often results
in a net increase in industrial water use. Even eciency gains in agriculture may in some
cases generate rebound eects resulting in net increases in water use (Loch and Adamson,
2015; Ward and Pulido-Velazquez, 2008).
A case study from Tianjin City (China), touted as the world’s largest Eco-City and a blueprint
for sustainable urbanisation worldwide (Baeumler et al., 2009), is a perfect example of a
relative decoupling that still results in overall increase in water consumption. According to
recent research by Wang and Li (2018), the city’s industrial water use and rapid economic
growth are still tightly coupled, and perhaps even increasingly so. Data from 2005-2015
indicate that even though the average growth rate of industrial water consumption (+0.18%)
was lower than GDP growth (+15.42%), periods of more rapid economic growth were marked
by a stronger coupling with industrial water consumption.
Just like for materials, it suces to look at global consumption to realise that gains in eciency
are being trumped by increases in volume. On a global level, Wada and Bierkens (2014)
estimate that human water consumption increased more than two-fold (~250%) between
1960-2010, the bulk of which is attributable to the expansion of irrigated agriculture. With
regards to global water withdrawal, the AQUASTAT database of the Food and Agriculture
Organisation (2016) shows a slightly smaller expansion from 2,500 km3/year in 1960 to
nearly 7,000 km3/year in 2010. In parts of Australia and California where absolute decoupling
can be observed, water consumption remains at unsustainable levels, as evidenced by an
increasing number of “anthropogenic droughts” (AghaKouchak et al., 2015; Ashraf et al.,
2017). These should be seen as cases of insucient decoupling.
Another remark has to do with the water embodied in trade. Similarly to the question of
embodied energy, most decoupling studies on water do not account for so-called “virtual
water” (Allan, 1998) which is the water embodied in products (e.g. one kilo of beef requires
around 15,000 litres of water over the full chain of production). Auent countries decrease
their domestic water consumption by importing water-intensive products from abroad,
IS DECOUPLING HAPPENING? IS DECOUPLING HAPPENING?
24 25
eectively shifting their water footprint and all its relative environmental issues onto other
countries.
Studies that account for “water footprint” (Hoekstra, 2017)so we must study virtual water
trade and the eects of countries externalizing their water footprint; (2 nd that auent
countries facing water-scarcity tend to reduce local water consumption by importing
virtual water (Oki et al., 2017). In a cross-national study, Wang et al. (2016) conrm that
the decoupling of domestic blue water use and economic growth in high-income nations
occurs through virtual water ows embodied in trade. The same result stands for Feng and
Hubacek (2015) who used a multi-region input-output analysis to understand global virtual
water ows, as well as for other studies that attempt to measure this externalisation of
water footprint (Fulton et al., 2014, 2012; Katz, 2008). Importing water-intensive services,
commodities and energies can create conditions of geopolitical instability. For those
concerned with global water risk and the implications for water justice, the sacrice of one
watershed for the health of another runs counter to the understanding and promise of
global water decoupling.
2. Impact decoupling
Greenhouse gases
The case of carbon dioxide is the most ambiguous of all and requires a detailed discussion.
Most studies do nd patterns of relative decoupling in early industrialised countries and
beyond e.g. 79 countries in Lonhofer and Jorgenson (2017) looking over the 1970-2009
period.19 Some studies even point to cases of absolute decoupling, albeit most often during
short time periods, only in specic locations, and often using production-based (territorial)
indicators. This could be cases for rejoicing, but unfortunately, the magnitude of the
decrease in emissions is negligible. Overall, the reviewed literature converges in saying that
there has never been a global pattern of absolute decoupling of CO2 from economic growth.
But let us look into the details starting with the Environmental Kuznets Curve literature. If
at all, the existence of an EKC for CO2 emission can only be conrmed within single studies
(Azam and Khan, 2016). The three meta-analysis we have screened do not nd any evidence
for decoupling over the 1995-2005 period.20 Out of 588 observations, Li et al. (2007) do not
nd a single case of absolute CO2 decoupling over the 1995-2005 period. What they do nd
is an EKC for more local greenhouse gases (such as SO2, NOx, CO, NO2, and SOx) but at an
income turning point of 37,000 USD per capita, which is seven times larger than the 2000
world average GDP per capita and thus practically unattainable if we aim at staying under the
1.5°C global warming target. Koirala et al. (2011) mobilised around 900 observations from
103 studies for their meta-analysis, and, fail to identify any carbon EKC. The most recent
19 Also Conrad and Cassar (2014) for Malta (1995-2012); Jiang and Li (2017) for several short periods in the US; Marques et al. (2018) for Australia
(1975-2016); Wu et al. (2018) in eight high-income and middle-income countries (1965-2015); and Wood et al. (2018) on a global scale.
20 Regarding a methodological quality of EKC studies Galeotti et al. (2006) show that data sets have negligible impacts on the results. However,
attention should be given to econometric misspecifications. Itkonen (2012) and Wagner (2008) find that a wrong application of methods often
leads to omitted bias and thus to false statements – similar critics have previously been formulated by Stern (2004).
review in date, from Mardani et al. (2019), points in the same direction. After reviewing 175
studies over the 1995-2017 period, they conclude: “While this [decoupling] has happened
in absolute terms in a few countries, the main trend in most developed countries is that
emissions are increasing, or stabilizing at a high level. One can hardly claim that there is
enough empirical evidence to assume that there is an EKC for CO2 emission intensities.”
Absolute decoupling can be spotted only by restricting the scope of observation, that is by
narrowing down either the study period or the geographical perimeter. For example, Chen
et al. (2018) analyse the total emissions of 30 OECD countries between 2001 and 2015. What
they found is that GDP increased by 70.6% over the period with CO2 emissions decreasing
by 3.8%, with most of that drop taking place between 2010 and 2015. The European
Environmental Agency reports a 22% absolute carbon emission reduction between 1990
and 2017, an average of 49 MtCO2e per year (EEA, 2018). Madaleno and Moutinho (2018)
nd evidence for temporary absolute decoupling in the EU-15 for territorial emissions, but
only between 1996 and 1999 (the whole study period was 1995-2014). Similarly, Roinioti
and Koroneos (2017) found two incidences of temporary absolute decoupling, lasting one
and two years respectively for the case of Greece in between 2003 and 2013. Cansino and
Moreno (2018) nd an absolute decoupling eect in Chile, but only for specic years of their
study period (1991-2013).
Cases of absolute decoupling are more likely to be observed looking at geographically
restricted areas and disregarding relations and exchanges with the rest of the world. Focusing
on eco-eciency indicators for industries in Flanders, Van Caneghem et al. (2010) report
an observed absolute decoupling between 1995 and 2006. The study of Azam and Khan
(2016) indicates an absolute decoupling between territorial emissions and GDP happened
in Tanzania and Guatemala using annual a production-based time series data from 1975-
2014. Further evidence is brought forward by Lean and Smyth (2010) for Singapore using
production based measures between 1980 and 2006.
Four remarks on these results. First, if there is absolute decoupling, it remains innitesimal.
For instance, 3.8% in 14 years (Chen et al., 2018) is a meagre performance – that is a compound
annual growth rate of -0.28% per annum, which remains 18 times too slow compared to the
IPCC (2018) 1.5°C target of a yearly 5% decrease. The 8% decrease in emissions between
2007 and 2015 reported by the International Energy Agency is only a yearly abatement of 1%
(IEA, 2016); and the decoupling in the EU reported by the European Environmental Agency
(EEA) would need to be increased 5-fold as to meet a -95% mitigation target for 2050. Other
similarly discouraging rates of absolute decoupling are found by Pilatowska and Wlodarczyk
(2018) in Belgium, Denmark, France, and the UK (1960-2012). In their comparative study, the
strongest eect was in Denmark with minus 1.8% of emissions yearly alongside a 1.16% rise
in GDP. As encouraging as this might look, according to the IPCC (2018), it would need to be
3 times faster and occurring simultaneously in every single country to stay within the 1.5°C
limit to global warming. All of this call for an acceleration of eorts. And yet, studies point to
the opposite: the speed of decoupling in high income countries is decelerating (Fosten et al.,
2012) as the set of easy to implement measures is increasingly depleted. (This is also in line
with current policy impact projections of the EEA, 2018).
IS DECOUPLING HAPPENING? IS DECOUPLING HAPPENING?
26 27
Second, even if decoupling can be spotted over a certain period, it is likely to disappear
if one extends the timeframe of the study. Wang et al. (2018) do observe several periods
in the US where energy related CO2 emissions decline alongside a growing GDP: -1.75%
(2000-2001), -1.61% (2005-2006), and between -2 and -3.31% (2010-2012). If the study had
only looked at these periods, then one would speak of a clear absolute decoupling. Yet,
spread over a longer period (2000 to 2014 in their study), the decrease of emissions is still
absolute but averages 0.006% per year, which is about 833 times too slow compared to
IPCC recommendations. Besides, an important reason for the decline was the switch from
coal to gas, a one-o measure facilitated by the temporary boom of shale oil and gas in the
US, which cannot constitute a permanent trend.
Third, most of these studies only take into account production-based measures. In contrast,
the ones that take a consumption-based perspective nd considerably dierent results. The
most recent long-term climate strategy put forward by the European Commission states that
Europe has managed to successfully decouple greenhouse gas emissions from economic
growth in the past decades (European Commission, 2018).21 However, this includes only
territorial emissions, and not consumption-based emissions including emissions embedded
in international trade. According to van de Lindt et al., (2017), while territorial emissions
declined by 13% during 1990-2010, the carbon footprint in the same period increased by
8%.
Likewise, Jiborn et al. (2018) show that Sweden and the UK (1995-2009) fall o the absolute
decoupling list when carbon leakage is considered (see also results by Hardt et al., 2018
above). What remains is relative decoupling: a rise in GDP (2.9% per year for UK and 3.2%
for Sweden) comes with a smaller rise – but rise nonetheless – in emissions (1.8% per year
for UK and 1.3% for Sweden). Cohen et al. (2018) reach the same result for the UK and
France (1990-2014); if consumption-based greenhouse gases emissions are accounted for
on a footprint basis, absolute decoupling disappears (the exception is Germany due to high
emission exports from the automotive industry). Same case for Singapore, for which Schulz,
(2010)for a comprehensive approach it is suggested to include upstream and downstream
processes of connected socioeconomic systems and the indirect life-cycle related emissions
of imported and exported goods. Singapore is used in this longitudinal study as an example
of an urban scale economy. Accounts for direct emissions are compared with trade
corrected estimates of indirect emissions. Results indicate that direct emissions account
for only about 20% of the overall upstream emissions necessary to sustain the input side of
the economic production process (domestic emissions plus indirect emissions embodied in
imported goods contrasts the result of Lean and Smyth (2010) showing that decoupling is
only relative once indirect trade-related emissions are taken into account.
Even only in terms of relative decoupling, the dierence is important. Cohen et al. (2018)
identify twelve countries in situations of relative decoupling (Brazil, Mexico, Turkey, Korea,
South Africa, Indonesia, India, China, Canada, Japan, Australia, and the USA) while considering
territorial emissions, but only two (UK and France) while measuring greenhouse gases
21 The Sustainable Development Goals reflection paper (European Parliament, 2019) does speak of a consumption-based absolute decoupling
while referring to the analysis of its long-term strategies (European Commission, 2018). Besides that claim, however, we have found no trace of
any supporting evidence in either of these documents.
footprint. Storm and Schröder (2018) analyse data from 61 OECD countries during 1995-
2011 in search for carbon Kuznets curves. What looks like decoupling in production-based
CO2 emissions (with a turning point at 56,000 USD annual per capita income) ceases to be
so when accounting for imported carbon (turning point at 93,000 USD, which is outside of
their sample).
At last, one should take into account the Global Financial Crisis of 2007-2008 and the following
Eurozone Crisis for their consequences on economic activity and therefore emissions. The
rapid decrease of emissions during the crisis is of little surprise. Most studies decomposing
the eects of dierent variables on CO2 emissions (energy consumption, energy intensity,
carbon intensity, GDP) conclude that GDP is one of the biggest drivers of CO2 emissions
(Cansino and Moreno, 2018; Chen et al., 2018; Jiang et al., 2016; Madaleno and Moutinho,
2018; Roinioti and Koroneos, 2017). The review of 175 studies by Mardani et al. (2019) even
points at a bidirectional coupling between GDP and CO2 emissions. Even though a recession
perhaps reduces impacts in the short term (Declercq et al., 2011; Feng et al., 2015; Roinioti
and Koroneos, 2017), it can hardly be considered a policy success in terms of decoupling
for green growth advocates.
To nish, let us scrutinise a specic decoupling study that was widely spread in the media.
In 2016, the World Resource Institute posted an entry on its website titled The Roads to
Decoupling: 21 Countries Are Reducing Carbon Emissions While Growing GDP (Aden, 2016).
To be precise, they show evidence for an absolute decoupling of GDP from territorial
greenhouse gas emissions between 2000 and 2014 in the case of 21 countries. Even if one
takes these results at face value, the decrease in emissions remains too small. Following
their estimation, the fastest decoupling country is Denmark with a 30% cut over the period.
While 30% may seem impressive, it is only a compounded 2.5% yearly decrease, which is
half of what the IPCC recommends. The average reduction for the 21 countries is 15% in
fourteen years (1.15% per year) for a total of 1005 Mt saved over the period. Spread over the
21 country sample, that is 48 Mt per country, and 3.4 Mt per country per year. Let us now
contrast this to the 8.2 GtCO2 yearly reduction recommended by the IPCC (2018). If evenly
distributed over the 195 countries of the United Nations, the necessary mitigation must be
of at least 42 Mt per country per year, which is twelve times the volume of emissions that
has been saved by the most successful cases of decoupling found in the World Resource
Institute study.
This number gets considerably lower when one considers footprint emissions. Evans
and Yeo (2016) redo the calculation with consumption-based indicators. Three countries
(Slovakia, Switzerland, and Ukraine) exit the list. The Danish emissions mitigation eort
shrinks from 30% to 12%. While the average reduction for the 20 countries that achieved
territorial decoupling (we have removed Uzbekistan for which there is no footprint data
available) is of 15.75% in total over the period, the footprint decoupling is only 7.46% (that
is 706.7 Mt of CO2 saved in 14 years) namely a compounded 0.55% yearly drop in emissions.
And again, we should remember that these are the most successful nations in terms of
mitigation, and that the rest of the world remains on a path of increased GDP increased
emissions.
IS DECOUPLING HAPPENING? IS DECOUPLING HAPPENING?
28 29
These numbers should be read carefully as the calculation of footprint emission is only
nascent and extremely complex (Sato, 2014). Considering non-existing data and the level
of sophistication of current models, it is more likely to under-estimate emissions than the
opposite. For example, emissions from aviation and shipping are systematically excluded
from national accounts. In the EU28 (plus Iceland, Norway, Switzerland), CO2 emissions from
aviation alone have been estimated at 151 Mt in 2014; although they have only increased by
5% since 2000, they are expected to rise another 45% until 2035 (EASA-EEA-EUROCONTROL,
2016) Assuming a yearly 150 Mt amounts to 2100 Mt of CO2 emitted over the 2000-2014
period, which is three times all the emissions that were saved through absolute decoupling
in Evans and Yeo (2016) footprint recalculation of the World Resource Institute Study (Aden,
2016).
Land
There are very few empirical studies that have tested the decoupling hypothesis choosing
land measures as environmental variables. And yet, one can nd ample evidence in related
literature that, with growing income, the living space per capita is increasing, and with it the
area of sealed soil. Thus, this section focuses on general relations between GDP and land
use.
In the literature dierent denitions are used to describe land use. Weinzettel et al. (2013)
refers to it as “use of land and ocean area through international supply chains to nal
consumption, modeling agricultural, food, and forestry products”, which is measured either
by land use (gha/capita) or by fraction of global total footprint (%). Another measure is the
Human Appropriation of Net Primary Production (HANPP). The last term is the total carbon
produced annually by plant growth, while the rst term accounts for harvested biomass
and human-induced land use change (Krausmann et al., 2013). Further measures are for
example the ecological footprint (Bagliani et al., 2008; Borucke et al., 2013; Caviglia-Harris et
al., 2009). Other papers only refer to single variables like croplands (Sandström et al., 2017;
Tilman et al., 2011) or forests (Kumar and Aggarwal, 2003).
The existing literature does not provide any indication of an absolute decoupling of economic
activity and land use, only relative ones. Conrad and Cassar (2014) nd evidence for a
relative decoupling of the land area aected by development from GDP in Malta between
1995 and 2012. Globally, the ecological footprint has grown together with economic growth,
showing no signs of decoupling (Bagliani et al., 2008; Caviglia-Harris et al., 2009). Krausmann
et al. (2013) nd that while the human population has grown fourfold and economic output
17-fold, global Human Appropriation of Net Primary Production has only doubled, due to
considerable eciency gains between 1910 and 2005. For dierent measures and regions
these relative trends are also supported by other studies (Conrad and Cassar, 2014; Kastner
et al., 2014; Tilman et al., 2011; Weinzettel et al., 2013), but no absolute decoupling can be
observed.22 Let us take cropland as an example. At the global level, cropland area harvested
for food production increased by 32% from 1963 to 2005 (Kastner et al., 2014), mostly driven
22 A closer look to countries’ ecological footprint and their available biocapacity points out an interesting case of Finland, which ecological
footprint decreased by 6,5% during 2002-2005, while the GDP increased by 9,5% in the same period, whilst also remaining within the limits of
available biocapacity (Mattila, 2012). However this is mainly due to wrong accounting, as Mattila (2012) showed.
by increasing animal calorie demand, being itself strongly inuenced by per capita income
(Tilman et al., 2011). Weinzettel et al. (2013) state that for each doubling of income, the land
footprint increased by 35%.
But not only does income correlate with land use, it also does with the net displacement of
land, which is why footprint indicators are of great importance to understand the relation of
economic activity and land use. When trade is taken into account, high-income countries to
use more biologically productive land per capita than low-income countries (Weinzettel et
al., 2013). EU’s land footprint was 2.5 global hectare (gha) per person compared to a global
average of 1.2 gha per person and total biocapacity of 1.8 gha. For each additional $10.000
income per capita, between 0.1 and 0.4 global hectare per person are displaced outside the
consuming country (Weinzettel et al., 2013), this result being supported by other studies
(Kastner et al., 2014; Yu et al., 2013). In total 60% of land is used for exports (Weinzettel et
al., 2013) whereas high-income countries are the greatest net importers. For example, 33%
of total land use for U.S. consumption purposes is displaced from other countries this
ratio becomes much larger for the EU (more than 50%) and Japan (92%) (Yu et al., 2013). An
average EU citizen in 2004 led to an appropriation of 2.53 gha compared to a global average
of 1.23 gha (Steen-Olsen et al., 2012).
Agricultural production is coupled with environmental pressures and the displacement of
land via international trade means that ecological costs are also displaced (Lambin and
Meyfroidt, 2011; Tukker et al., 2016; Weinzettel et al., 2013; Yu et al., 2013). The EU crop
and livestock imports are a signicant driver of global deforestation over the period 1990–
2008; for example more than 90% of Finland’s impacts on biodiversity occurs elsewhere
via its imports (Sandström et al., 2017). The associated changes in land use are expected
to increase greenhouse gas emissions, about one quarter of which already results from
land use and land use changes (Tilman et al., 2011). Schreinemachers and Tipraqsa (2012)
nd that the use of pesticide, herbicide, and fungicide does not go down as countries reach
higher incomes, and remains strongly associated with crop output. What this shows is that
the relation of economic activity with land use also links to other environmental challenges,
such as biodiversity loss, water scarcity, climate change and energy consumption.
Water pollutants
The aforementioned UNEP report draws on water decoupling or “dewatering” research that
explicitly does not account for water pollution (UNEP, 2015, p. 2). While major advances
have been made in limiting water pollution in industrial and agricultural production, the
contamination of water remains a global issue that contributes to increasing global water
pollution hotspots (Strokal et al., 2019). Most of the global water pollution is from the
production of industrial and agricultural commodities for regional and global trade (Liu et
al., 2017; Mekonnen and Hoekstra, 2016; Vörösmarty et al., 2015; Zhao et al., 2016, 2015).
The concept of return ow, that is the dierence between withdrawal and consumption,
is critical to our understanding of water pollution. Return ow concentrates the pollution
impacts of water-dependent production. Cleaning up return ows can be achieved
IS DECOUPLING HAPPENING? IS DECOUPLING HAPPENING?
30 31
by advances in cleaner production technologies, often prompted by the creation and
enforcement of environmental regulations. These technologies have high costs, which can
prompt the movement of production to areas with fewer or less enforced environmental
regulations related to water pollution. As noted in Schwarzenbach et al.’s (2010) review
of global water pollution and human health, cheap production in emerging economies
continues to be associated with high levels of water pollution. Outsourcing toxic and water-
intensive production can lead to local, regional, and national decoupling of economic growth
from impacts to basin-level water quality, however, on a global scale, the problems of water
quality remain the same or are in some cases exacerbated (van Vliet et al., 2017).
Nitrogen and phosphorus accumulation, the two main macro-nutrients needed for
agricultural production, lead to eutrophication and dead zones in water ecosystems, which
have spread exponentially since the 1960s (Diaz and Rosenberg, 2008). Nitrogen is also
released in the atmosphere, where in reactive form it has higher greenhouse gas eect than
carbon dioxide. N and P fertilizer use rates per unit cropland area increased by approximately
8 times and 3 times, respectively, since the year 1961 (Lu and Tian, 2017). According to Lu
and Tian (2017) fertilizer ratio increased by 0.8g N/g P per decade during 1961-2013, having
human-derived implications on climate change, water quality and ecosystems, food security
and agro-ecosystems at large. Furthermore, recent outlook on fertilizer demand shows that
nitrogen fertilizer demand is still growing even in the rich countries, North America and
Europe (FAO, 2017).
Global biochemical nitrogen and phosphorus ows have transgressed their planetary
boundaries (Steen et al., 2015)introduced in 2009, aimed to dene the environmental
limits within which humanity can safely operate. This approach has proved inuential in
global sustainability policy development. Steen et al. provide an updated and extended
analysis of the PB framework. Of the original nine proposed boundaries, they identify three
(including climate change. This results mainly from the prevalent high-input agriculture and
intensive livestock farming, which lead to atmospheric nitrogen pollution and coastal marine
eutrophication and dead zones (Bouwman et al., 2013). Agricultural nutrient discharge is
the most signicant contributor to groundwater and surface water contamination, much
larger than urban point sources (Billen et al., 2013). A study exploring changes in nitrogen
and phosphorus cycles in agriculture induced by livestock production over 1900-2050
period shows that anthropogenic N and P inputs have grown ve-fold since pre-industrial
times and by 2050 surpluses are expected to further increase by over 20% for N and over
50% for P (Bouwman et al., 2013). Figure X demonstrates that demand for animal-based
products and proteins is likely to grow as countries become wealthier, implying the growing
demand also for nutrient inputs in agriculture.
IS DECOUPLING HAPPENING? IS DECOUPLING HAPPENING?
Biodiversity loss
Biodiversity is dicult to measure,23 but neither individual nor aggregated indicators of
the state of biodiversity showed signicant improvements in their rates of decline, while
all pressure indicators showed increasing trends, with none signicantly decelerating
(Butchart et al., 2012). The last report to date by Intergovernmental Science-Policy Platform
on Biodiversity and Ecosystem Services (IPBES, 2019) has shown that almost all drivers of
biodiversity loss keep increasing, that the dangerous decline of biodiversity is unprecedented,
that the species extinction rates are accelerating, and that the current global response is
insucient. Going in the same direction, the EU 2030 outlook for ecosystem conditions
and services and the Food and Agriculture Organisation report worrying levels of species
decline (EEA, 2018; FAO, 2019, p. 445). Analysing extinction rates in comparison to natural
average background rates since 1500 AD, Ceballos et al. (2015)using extremely conservative
assumptions, whether human activities are causing a mass extinction. First, we use a recent
estimate of a background rate of 2 mammal extinctions per 10,000 species per 100 years
(that is, 2 E/MSY nd that current rate vastly exceeds natural average, and warn of an
impending sixth mass extinction (see also Barnosky et al., 2011).
The empirical literature on EKC relationships between biodiversity and economic growth
is scarce but consistent. The rst meta-analysis used 121 observations gathered from a
set of 25 studies, and 11 environmental indicators, including deforestation (Cavlovic et al.,
2000). The study analysed EKC relationship and estimated hypothetical income turning
points using dierent modelling methods. For deforestation, which is used as a proxy of
biodiversity loss,24 the income turning point was estimated in in the range of 5000-20,000
USD (in 1999 prices).
Koirala et al. (2011) used almost 900 observations from 103 studies for their meta-analysis
and disaggregated environmental quality measures into 12 dierent variables and did
not observe any EKC either for deforestation or landscape/habitat degradation. Dietz and
Adger (2003) do not nd any EKC for deforestation and species richness, a result conrmed
by Mills and Waite (2009)arguing that wealthier countries have the luxury of investing more
heavily in eorts to conserve biodiversity. Under this assumption, we expect a U-shaped
relationship between per capita wealth and proportion of species conserved. We test this
environmental Kuznets curve (EKC. Even stronger is the argument by Asafu-Adjaye (2003)
who nds an inverse relationship between economic growth and species diversity – a result
conrmed by Raymond (2004)continued economic growth eventually leads to superior
environmental quality. This relationship is often described as an ‘Environmental Kuznets
23 Vačkář et al. (2012) provide a comprehensive review of different indices monitoring human impacts on biodiversity. One of the most known
is the Living Planet Index, which shows the change in species abundance and distribution. The other indicators are: Red List Index measuring
changes in extinction risk, collected by IUCN, Marine Trophic Index that is specialized for marine biodiversity, the Natural Capital Index, com-
prised of changes in ecosystem quantity and quality, the Biodiversity Intactness Index, measuring both species richness and population abun-
dance, and Index of Biotic Integrity, which evaluates ecosystems in comparison to a reference state according to various human impacts (Vačkář
et al., 2012). Another index is National Biodiversity Risk Assessment Index (Reyers et al., 2018), but it is not updated regularly.
24 One should be cautious with data inference and interpretation. For instance, per capita income does seem to correlate with state protected
land area, however, rather as part of various socio-economic indicators (social, economic, cultural and natural) than independently on its own
(Dietz and Adger, 2003). In addition, protected areas do not guarantee higher conservation of biodiversity (Bruner et al., 2001). In some previ-
ous studies that have detected EK (Bhattarai and Hammig, 2001), the problem may be in how biodiversity is interpreted. Reforestation through
plantations does not equal to the deforestation of primary rainforests with its accompanying species. Meanwhile, McPherson & Nieswiadomy
(2005) identified an EKC for threatened bird and mammal species (using IUCN data for 113 countries in 2000), and a potential turning point at
around 10,000-15,000$ (1995$ PPP) in per capita income, after which the percent threatened falls. However, the problem with the IUCN data
is that the rate of endangered species or the rate of deforestation may be low in countries which has already experienced much extinction or
deforestation in the past (McPherson and Nieswiadomy, 2005). Hence, they use percent instead of number of species and do a range of other
corrections to the dataset.
32 33
III. Is decoupling likely to happen?
Looking for evidence, we found that the type of decoupling that would be needed to
eectively and equitably mitigate climate change and address other environmental crises
is nowhere to be seen. Yet, lack of empirical support does not suce to fully dismiss the
possibility of decoupling, which some argue could well happen in the future with the right
set of policy changes. The purpose of this section is to assess the validity of this position.
Our claim is the following: adequate (i.e. absolute, permanent, and sucient) decoupling is
extremely unlikely to happen in the near future. We oer seven reasons in defence of that
proposition: (1) rising energy expenditure, (2) rebound eects, (3) problem shifting, (4) the
underestimated impact of services, (5) the limited potential of recycling, (6) insucient and
inappropriate technological change, and (7) cost-shifting. In what follows, we go through
each of these reasons.
1. Rising energy expenditure
The availability of natural resources does not only depend on their absolute quantity (how
much is “out there”) but also on their quality and accessibility (how much eort is required
to extract them). When extracting a resource, cheaper options are generally used rst,
which means that most readily available energy and material resources mobilised by the
economy have already been exploited.25 The extraction of remaining stocks then becomes
a more complex, more technology demanding, more socially disruptive hence generally
more expensive, more resource- and energy-intensive and polluting process resulting in a
rising total environmental degradation per unit of resource extracted. This is the case for
low-concentration metal and mineral depots, tar sands, deep o-shore wells, stocks located
in polar regions or near densely populated cities like shale gas near Paris. These increasing
25 The common-sense idea that easiest and cheapest options are generally used first (the proverbial “reaping the low hanging fruits”) is referred
to in economics as the “law of increasing marginal cost” and, when applied to resources, is sometimes called the “best-first principle.” Such a rule
of thumbs applies widely and can be easily observed in multiple situations: from resource extraction to efficiency gains and pollution abatement.
IS DECOUPLING LIKELY TO HAPPEN? IS DECOUPLING LIKELY TO HAPPEN?
Conclusions for Section 2
In light of the present review, we can safely conclude that there is no empirical evidence
supporting the existence of a decoupling of the type described as necessary in the rst
section of this report that is an absolute, global, permanent, and suciently fast and
large decoupling of environmental pressures (both resources and impacts) from eco-
nomic growth. In the end, our search for robust evidence was unsuccessful, coming
up only with a handful of methodologically peculiar exceptions, most often of relative
decoupling, and if absolute, mainly temporary and restricted in space, only for territo-
rial indicators (that is to say spatially inconsistent), or having to do with specic local,
short-term pollutants. In all cases, the reduction in environmental pressures falls short
of current environmental policy targets. After such an extensive search, it is safe to say
that the type of decoupling acclaimed by green growth advocates is essentially a statisti-
cal gment.
Yet, even though the success of the green growth strategy is nowhere to be seen, this
lack of empirical support does not allow to completely dismiss the decoupling hypothe-
sis. The adequate decoupling of economic activity and environmental pressures remains
theoretically possible if resource productivity grows suciently faster than GDP per-
manently and globally. This might happen, some argue, by increasing the geographical
coverage of emission trading systems (Stiglitz et al., 2017) in combination with phasing
subsidies for fossil fuels (Schwanitz et al., 2014), directing investments into sustainable
infrastructure (Guivarch and Hallegatte, 2011), and a number of other decoupling poli-
cies (Smith et al., 2010; UNEP, 2014a). What is at dispute is the impact of a number of
factors, trends, and phenomena that would enable or prevent such an eciency-driven
decoupling from happening. Putting the decoupling hypothesis in perspective with the
potential impact of those factors is the objective of the nal section of this report.
Curve’ (EKC in a study of 142 countries. Mozumder et al. (2006) rejects the EKC hypothesis
for income and biodiversity risk. Using the same model, Tevie et al. (2011) reach the same
conclusion in their study of 48 American states. Naidoo & Adamowicz (2001), using data
from over 100 countries, investigated the link between numbers of threatened species
and per capita income growth. After dividing species into seven taxonomic groups (plants,
mammals, birds, amphibians, reptiles, shes, invertebrates), they found support for absolute
decoupling only for birds. Meanwhile, for plants, amphibians, reptiles, and invertebrates
the relationship was the opposite, their number of threatened species increased with GDP.
34 35
energetic costs26 of extraction means that more intermediate resources are necessary to
extract the nal resources required for the production of the same quantity of goods and
services, leading to the opposite of decoupling.
The energy expenditure argument is sometimes counteracted by those insisting that energy
only plays a small role in economic activities. And indeed, from a monetary point of view,
the energy sector often accounts for a small fraction of total GDP. Yet, this perspective has
been challenged by a number of scholars (Ayres and Warr, 2009; Georgescu-Roegen, 1971;
Giampietro et al., 2011; Hall and Klitgaard, 2012; Kümmel, 2011). Latest to date, Keen et
al. (2019: 41) argue that energy is not a substitute to labour or capital but precisely what
enables these factors of production to perform useful work “labour without energy is a
corpse, while capital without energy is a sculpture” (Keen et al., 2019, p. 41). Here, common
sense is perhaps more useful than economics: the average speed of a car (GDP growth)
might seem to determine its gasoline consumption (energy use), but no one can reasonably
assume that a car could run without it (Fizaine and Court, 2016, p. 173).
Energy
When it comes to energy resources, the eciency of extraction can be quantied using the
concept of EROI (or EROEI), which stands for Energy Return on Energy Invested. EROI is the
ratio of the quantity of energy obtained from a resource to the quantity of energy that must
be spent to extract it in the rst place.27 It is a measure of net energy output; for instance,
a ratio 1:1 for petroleum would mean that it takes a barrel of oil to extract another barrel
of oil while a ratio of 10:2 would mean that the energy costs of extracting 10 barrels is two
barrels. This concept allows to dierentiate the cost and the surplus of energy (e.g. an EROI
of 50:1 means an energy cost of 2 per cent for an energy surplus of 98 per cent, while one of
5:1 means a cost of 20 per cent for a surplus of 80 per cent). The lower the EROI, the higher
the energy cost or energy expenditure. A declining EROI means that an increasing portion of
energy output must be allocated to obtaining energy, which means an increase in resource
use and impacts.
Several authors make the empirical claim that high levels of energy expenditure are
associated with low economic growth rates, or even that GDP cannot grow over a certain
threshold of relative energy expenditure: 5.5% of total GDP for Murphy and Hall (2011)
looking at the US between 1970-2007; 8-10% for the US and 9-11% for the broader OECD in
Bashmakov (2007); and 11% for Fizaine and Court (2016) looking at the US over the 1850-
2012 period. The logic is simple: if energy expenditures exceed these thresholds, it starts to
act as a limiting factor on employing labour and capital.
26 It should be stressed that there is a difference between the cost and the price of a natural resource. Let us take energy as an example. Where-
as the price denotes the quantity of money that a commodified form of energy commands on the market (e.g. 55€ for a barrel of oil, 0.2€ for
one kWh of electricity), its cost (as used in this section) refers to the real (and not monetary) quantity of energy (e.g. litres of petroleum, cubic
metres of gas, calories of food, kilowatt-hours of electricity, kilos of coal or biomass) that must be spent in order to extract one extra unit of
energy. Another way to put it is that the cost of a natural resource has to do with its extraction and production whereas its price has to do with
its consumption. Because such resource expenditures are usually priced as well, the cost and the price of a natural resource tend to converge
in the long term.
27 Hall et al. (2014)cheap and seemingly limitless fossil energy has allowed most of society to ignore the importance of contributions to the eco-
nomic process from the biophysical world as well as the potential limits to growth. This paper centers on assessing the energy costs of modern
day society and its relation to GDP. Our most important focus is the characteristics of our major energy sources including each fuel’s energy
return on investment (EROI differentiate between four types of EROI. “Standard EROI” is the energy output divided by the sum of the direct and
indirect energy used to generate that output. “Point of Use EROI” adds the costs associated with refining and transporting the fuel. “Extended
EROI” considers the energy required not only to get but also to use a unit of energy. And finally, “societal EROI” is “the overall EROI that might be
derived for all of a nation’s or society’s fuels by summing all gains from fuels and all costs of obtaining them.”
The EROI for fossil fuels is of special interest as it also describes how much greenhouse gas
emissions are generated in a fossil fuel based economy to provide one additional unit of
fossil energy (ton or barrel) – one could even speak of the climate cost of extracting a barrel.
While the carbon intensity of that consumption is xed (e.g. burning one barrel of oil emits
around 120 kg of carbon), a decreasing EROI means an increase in emissions per unit of
primary energy used (the carbon emissions corresponding to the increasing extra energy
burnt to extract that barrel adds up to the 120 kg). According to some estimations, the EROI
for the global production of oil and gas increased from 23:1 in 1992 to 33:1 in 1999 and
declined to about 18:1 in 2005, giving credence to the theory that the eciency gained by
technical improvements is being trumped over time by depletion (Hall et al., 2014)cheap
and seemingly limitless fossil energy has allowed most of society to ignore the importance
of contributions to the economic process from the biophysical world as well as the potential
limits to growth. This paper centers on assessing the energy costs of modern day society
and its relation to GDP. Our most important focus is the characteristics of our major energy
sources including each fuel’s energy return on investment (EROI. Certain authors such as
Morgan (2016) now speak of an “energy sprawl” to describe the necessary expansion of
the infrastructure required to access energy and the growing proportion of GDP that it
will absorb. Accounting for both fossil and renewable energy sources, Capellán-Pérez et al.
(2018) nd that the EROI of the global energy system went from 7:1 in 1995 to 6:1 in 2018.
A prime example of this process of increasing marginal costs concerns the extraction of
dierent types of unconventional oils. Tar sands and oil shale deliver a mean EROI of 4:1
and 7:1 (Lambert et al., 2014)but probably adverse. A major obstacle to examining social
implications of declining EROI is that we do not have adequate empirical understanding of
how EROI is linked, directly or indirectly, to an average citizen′s ability to achieve well-being.
To evaluate the possible linkages between societal well-being and net energy availability, we
compare these preliminary estimates of energy availability: (1. Shale gas is often acclaimed
as an abundant alternative to oil, especially in the United States (Moeller and Murphy, 2016),
but not only is drilling shale wells relatively more expensive in both energetic and nancial
terms, but the rates of decline in production tend to be signicantly faster than traditional
oil wells (Morgan, 2016, p. 63).
Another example is coal. Putting pollution issues to the side for a moment, global reserves
of coal suggests that, in terms of volume, coal is still relatively abundant. Yet, not all forms
of coal are equal in quality. Anthracite, which is the richest coal in terms of energy content,
is increasingly scarce, pushing coal companies to extract bituminous and sub-bituminous
coals of lesser energy density (Kerr, 2009; Morgan, 2016; Schindler and Zittel, 2007).
One could argue that green growth would only run on renewable energies and so that
the EROI of fossil fuels is irrelevant. Even though we will shortly argue that it is not, let us
assume for a moment that a complete replacement of fossil fuel by renewables is possible
materially (nding enough minerals and land to build the energy infrastructure) and
socioeconomically (having renewable energies nding social acceptance and investment
resources to completely replace fossil ones). Even then, according to Murphy and Hall
(2011), the EROI of renewable energies (below 20:1) is still signicantly lower compared to
the high EROIs during the early days of fossil fuels (Hall et al., 2014)cheap and seemingly
IS DECOUPLING LIKELY TO HAPPEN? IS DECOUPLING LIKELY TO HAPPEN?
36 37
limitless fossil energy has allowed most of society to ignore the importance of contributions
to the economic process from the biophysical world as well as the potential limits to growth.
This paper centers on assessing the energy costs of modern day society and its relation to
GDP. Our most important focus is the characteristics of our major energy sources including
each fuel’s energy return on investment (EROI. Capellán-Pérez et al. (2018) simulate what
would happen to average EROI by 2050 should renewable energy sources increase from
15% to 30% (1st scenario) and from 15% to 50% (2nd scenario). In the rst scenario, average
EROI drops from currently 6:1 to 5:1; and down to 3:1 in the second scenario. If energy
expenditures play an important role in the dynamics of economic growth, this means that
renewable energies are fundamentally unable to propel an economy as fast as fossil fuels.
Materials
Similarly, and for the same kind of reasons, the rule of increasing marginal costs or the best-
rst principle applies to material extraction. A series of studies already show how the quality
of ores of essential minerals are declining (e.g. Calvo et al., 2016). Lower ore grades mean
more overburden and environmental damage.
The average concentration of copper in ore/mined material went from 1.8% in 1930 to 0.5%
today (Arnsperger and Bourg, 2017, p. 87), a situation that is common to other minerals.
Lower concentration rates for minerals means that higher volumes of materials need to be
mined and displaced in order to extract the same amount of ore, and with it more energy.
In the rst UNEP decoupling report, Fischer-Kowalski et al. (2011b, p. 25) estimate that, in
average, the extraction of materials today requires to displace three times more matter
than a century ago.
This is particularly problematic when it comes to green technologies (Calvo et al., 2016;
Valero et al., 2018). Indeed, the mineral intensity of renewable energies is higher than the
one for fossil fuels – 1kWh of renewable energy requires 10 times more metals than 1kWh
of fossil energy (Arnsperger and Bourg, 2017, p. 87). Add increasing production into this,
and the following vicious circle emerges: more energy will be necessary to extract more
minerals which are needed to build more energy infrastructure, part of which is needed
to provide the additional energy required to extract more minerals and so on and so on.
Renewable energies can mitigate some environmental impacts but they cannot trump
resource scarcity.
What is often forgotten is that this increasing resource scarcity also translated into an ever
further expansion of the so-called commodity frontier (Moore, 2000), that is advancements
into previously untouched pristine areas, often at the cost of indigenous communities and
ecosystems’ health. Current examples include the extraction of tar sand in Alberta, Canada,
oil in the Peruvian rain forest, or, most famously, in a national park in Ecuador. While
these involves fossil fuels, the reach for the minerals required to build renewable energy
infrastructure poses similar threat to socio- and biodiversity.
2. Rebound eects
Improving resource eciency is probably the most common argument put forward in
defence of decoupling. However, every action that responds to savings in resources is prone
to rebound eects, that is a dierence between the projected and the realised environmental
savings from an eciency improvement. Such a phenomenon was hinted at already in
the 18th century by Stanley Jevons in The Coal Question (1865, pp. 140–142): “It is wholly a
confusion of ideas to suppose that the economical use of fuel is equivalent to a diminished
consumption. […] Whatever, therefore, conduces to increase the eciency of coal, and to
diminish the cost of its use, directly tends to augment the value of the steam-engine, and
to enlarge the eld of its operations” – hence the rebound eect often qualied as “Jevons
Paradox” (Giampietro and Mayumi, 1998; Jevons, 1865).
This idea that eciency changes would rebound into more consumption gained ground
in the eld of energy economics in the context of the oil crises of the 1970s, most notably
with the work of Khazzoom (1980) and Brookes (1990) – later referred to as the “Khazzoom-
Brookes postulate” (Saunders, 1992). After more than 40 years of research, the literature
has expanded to encompass a variety of causes and eects.28 In order to account for
overall decoupling, the concept we nd most relevant is the “environmental rebound
eect” (originally used by Goedkoop et al., 1999, and then by others such as Murray, 2013;
Spielmann et al., 2008; and Takahashi et al., 2004), which goes beyond energy issues to
encompass a wider range of environmental concerns.29
Several types of rebound eects
Rebound eects come in many shades depending whether eciency leads to an increase of
consumption of the same product or service (direct rebound eect), whether freed resources
are allocated elsewhere (indirect rebound eect), or whether consumption is induced by
structural changes in the economy as a whole (structural rebound eect). These eects, alone
or together, are then either partial or total depending on the magnitude of their impact on
resource use.
28 Here are a few examples that shows the wide span of the concept: time rebound effects (Jalas, 2002), socio-psychological or mental rebound
effect (de Haan et al., 2006; Girod and de Haan, 2009; Santarius and Soland, 2018), international rebound effects (Bergh, 2017).
29 For a general framework for the study of environmental rebound, see Font Vivanco et al. (2016).
IS DECOUPLING LIKELY TO HAPPEN? IS DECOUPLING LIKELY TO HAPPEN?
Energy and material are crucial for the functioning of an economy, and even more so for
one that is growing. Just like a living organism, an economy requires energy and material
not only to grow, but also only to maintain its current size. All available evidence points
towards increasing costs of extraction for both energy sources and materials. If economic
growth requires more energy and material, and it takes increasingly more energy and
material to extract energy and material, then rising energy expenditure acts as a limit
to growth and constitutes a barrier to decoupling. In order to argue that decoupling is
possible, one must show how to deal with the increasing marginal cost of energy and
material extraction.
38 39
First order: direct rebound eects
Direct or 1st order rebound eects refer to cases where the eciency gain is reinvested as
additional consumption of the same product or service. This is especially true for normal
goods for which a decrease in the cost of use perceived by users translates into a higher
consumption. For instance, driving a more fuel-ecient car more often, faster, or over
longer distances; the petrol that was saved in eciency by the car rebounded into more
usage of the car. Direct rebound eects can also occur in production, for example when
the acquisition of a more energy-ecient machine motivates additional production (output
eect).
Second order: indirect rebound eects
Indirect or 2nd order rebound eects refer to cases where resources freed by an eciency
or suciency improvement are re-allocated to another type of consumption (re-spending
eect). For example, driving a more fuel-ecient vehicle (eciency) or deciding to use it
less often (sobriety) could save money (income eect), which can then be spent on impactful
products or services (e.g. a far-away holiday trip by plane) or invested on problematic
nancial products (e.g. related to fossil fuel extraction). For producers, prots resulting
from productivity gains can be reinvested into expanding production capacity (re-investment
eect).
What Wallenborn (2018) call “structural rebound eect” is a good example of such indirect
rebound.30 It is structural because it has to do with economic structures such as markets,
ownership, and money. In a globalised economy where money can be used to buy almost
anything (one then speaks of general-purpose money), all purchasing power is a potential
polluting power. Even if euros are spent on green products, and even if the sellers of these
products spend these euros in a sustainable way, at some point down the chain, these
euros are likely to be used in a polluting manner. Even euros not spent will cause resource
consumption and pollution when re-lent by the bank to nance new investments. The only
way to avoid this eect would be to change the structure of the economic system itself
(decommodication, localisation, special-purpose monies like complementary currencies,
etc.).
Third order: economy-wide rebound eects
Eciency in resource use can also rebound at the macro level (economy-wide or
macroeconomic rebound eect). For instance, eciency gains in internal combustion engines
have help made private car transportation eective and aordable, and resulted in a wide
diusion of this technology. This generalisation of private car transport has in turn driven
the spatial conguration of cities and territories, resulting in extensive spatial congurations
which now rely on, and even require, the use of private cars. This wide scale modication
of the system of needs now results in a dramatically higher energy consumption from the
30 In the words of Jevons’s himself writing in the The Coal Question (1865): “[…] In fact, there is hardly a single use of fuel in which a little care, in-
genuity, or expenditure of capital may not make a considerable saving. But no one must suppose that coal thus saved is spared – it is only saved
from one use to be employed in others, and the profits gained soon lead to extended employment in many new forms. The several branches of
industry are closely interdependent, and the progress of any one leads to the progress of nearly all” (Jevons, 1865: 136 cited in Missemer, 2012,
p. 99).
IS DECOUPLING LIKELY TO HAPPEN? IS DECOUPLING LIKELY TO HAPPEN?
transport sector. In other words, more fuel-ecient cars reinforce the hegemony of cars at
the expense of more sustainable modes of transportation like trains and bikes. Resource
eciency can also lead to a restructuring of the economy around nature-intensive activities
(composition eect). For example, abandoned mining activities can be resumed after the
development of new ecient techniques makes it economically protable again, as it is
currently the case for gold mining where lower grade ores (including the former overburden)
are now reprocessed.
Partial and total rebound
Depending on its magnitude, a rebound eect can result in either an overall decrease (partial
rebound) or increase in resource use (total rebound, also known as overshoot or back-re). In
the rst case, the savings are larger than the extra rebounded consumption (e.g. a heater
consumes 50% less and rebounds in being used 1.5 times more, which means there are
still 25% net savings). In the case of total rebound, however, the rebounded consumption
is larger than the savings and savings are totally oset (e.g. if the money saved by using a
car consuming 30% less energy per km is used to pay for a holiday trip by plane where it
pays for much more energy than in the case of gasoline which unlike kerosene is heavily
taxed).31 In relation to decoupling, this means that a rebound eect can either slow down
the expected rate of decoupling (partial rebound) or reverse it altogether (total rebound).
Empirical evidence of rebound
Indirect and structural rebound being highly complex, most empirical research focuses on
direct rebound eects, which are easier to measure. In their review of energy use rebounds,
Ackerman and Stanton (2013, pp. 120–121) conclude that evidence for total direct rebound
eects is rare: “estimates of 10 to 30 percent seem common […] actual evidence of rebound
eects of 100 percent or more appears to be non-existent.” Same conclusions for surveys
conducted by Greening et al. (2000) and Sorrell (2007) who nd a diverse range of rebounds,
sometime low like in the case of lighting (up to 15%), moderate like in the case for aviation
(19%), or very high like in the case for motorised transport (up to 96%).32 Galvin (2014) reports
a rebound for household energy conservation in the range of 0-50% for older EU member
states between 2000 and 2011 – certain countries, notably Eastern European countries, as
well as Finland and Denmark, shows situations of total rebound. Grafton et al. (2018) show
that higher use of ecient technology rarely reduces water consumption. Kyba et al. (2017)
reports a situation of backre in the case of LED technology for outdoor lighting. Antal and
van den Bergh (2014) estimate the re-spending rebound for saving energy from gasoline to
range between 45 and 60% for larges economies such as Russian, China, and India.
31 In the literature, and following Ehrhardt-Martinez and Laitner (2010), what we call partial and total rebound are often referred to as “typical
rebound” and “back-fire.” The authors (ibid. 7-77) also add a third category: a “negative rebound” for situations where actual energy savings are
higher than expected (e.g. “a family that installs a new energy-efficient hot water heater may be motivated to find other ways to save energy by
taking shorter showers, washing clothes in cold water, or by limiting dishwasher use to full loads”; negative rebound, better example, direct cau-
sality: isolating walls reduces heating demand, making existing heating installations oversized. This in turn requires installing new and smaller
boilers, which are more efficient, so energy demand sinks again. or on the producer side if the price of a new machine is greater than the saving
in operating cost it allows). To avoid confusion, others prefer to speak of a “super-conservation” effects (Saunders, 2005) or “amplifying” and
“leverage” effects (Spielmann et al., 2008).
32 For all figures given, readers should be aware that the methodology used influences the results. For instance, studies using Life Cycle Analysis
together with the concept of environmental rebound effect find a higher likelihood of backfire. This is the case for Font Vivanco et al. (2016)
looking at electric cars.
40 41
Magee and Devezas (2017) examine numerous statistical sources to estimate the use of
69 dierent materials from 1960 to 2010, arguing that the Jevons paradox applies to just
about every substance. Out of their sample, they nd only 6 cases of absolute decline. Four
of these materials asbestos, beryllium, mercury, and thallium have been phased out
deliberately by legal restrictions because of toxicity issues. The other two are wool, which
has declined without decreasing the global populations of domestic sheep or other wool-
producing animals, and tellurium, a byproduct of rening copper whose use in solar panel
manufacturing means its overall consumption is likely rising again.
Empirical studies of macroeconomic rebound eects are scarcer than their micro
counterparts. In his review of the literature, van den Bergh (2017, p. 4) concludes that “the
majority of economy-wide studies suggest overall rebound is above 50% and possibly much
higher.” In a survey of computable general equilibrium studies, Dimitropolous (2007) nds
three cases of total rebound, three others above 50%, one in the range of 30-50%, and one
around 15%. Even though rebound eects of the 2ndor 3rdorder are the most determining
ones, these remain the most dicult to study empirically.
Example 1: renewable energy
Renewable energy is often depicted as clean and unlimited, but it is far from being free of
environmental pressures. Renewable energies and eciency-enhancing ICT technologies
reduce carbon emissions but exacerbate land use (e.g. solar farms and biomass/biofuels),
and water conicts in the case of hydropower (Capellán-Pérez et al., 2017; Havlík et al., 2011;
Scheidel and Sorman, 2012; Yang et al., 2012). They increase metal demand and the local
conicts associated with their extraction (Ali, 2014; Chancerel et al., 2015; Kleijn et al., 2011;
Vidal et al., 2013), and, in the case of photovoltaic infrastructure, generate environmental
pollution and emissions of greenhouse gases (Andersen, 2013; Hernandez et al., 2014;
Zehner, 2012). The extraction of rare earth minerals, which are essential for many green
technologies including wind mills, causes enormous environmental damage, for example in
China (Pitron and Védrine, 2018).
Let us take three more examples among many. The production of batteries for electric cars
puts pressure on the extraction of lithium, cobalt, nickel, and manganese (Bednik, 2016, p.
101; Valero et al., 2018). The expansion of biomass for biofuels can encroach on protected
areas and lead to an increase of monocultures, negatively impacting biodiversity and its
conservation (IPBES, 2019), a good example being deforestation in the Indonesian rainforest
for palm-oil plantation (Koh and Wilcove, 2008; Margono et al., 2012). And hydropower
produces methane emissions when algae growth is catalysed by the silt trapped by the
dam, sometime generating more greenhouse gas emissions than a fossil-fuel-red plant
(Deemer et al., 2016).
Example 2: Nuclear energy
Nuclear energy is a good case in point. Being relatively carbon-neutral,33 it is considered the
principal factor that allowed countries like France, Sweden, United Kingdom and Germany
to reduce their energy-related carbon emissions. Nuclear energy, however, requires the
extraction of uranium as fuel as well as titanium, cobalt, tantalum, zirconium, hafnium,
indium, silver, selenium, and lithium for construction materials (Sersiron, 2018, p. 165). A
shift to nuclear power means intensifying the coupling of economic activity with various
materials, starting with uranium.34 Mining and transporting these materials is itself a source
of environmental pressures, for example in terms of water pollution or biodiversity loss
through land change (Conde and Kallis, 2012). Furthermore, nuclear energy involves a
dierent set of social-ecological hazards linked with the storage of toxic waste as well as the
risks of nuclear accidents and nuclear weapon proliferation. In sum: nuclear electrication
shifts the coupling from one impact (CO2 emissions from fossil fuel) to other impacts (e.g.
biodiversity loss, water pollution, and other impacts related to mining and transport, toxic
waste) and resource use (e.g. uranium scarcity).
33 This remains a matter of controversy, as it is difficult to calculate the carbon footprint of the entire life-cycle of a nuclear plant, including
indefinite waste storage and potential clean-up operations after accidents.
34 If only for the case of uranium, currently identified reserves – 7.6 million tonnes commercially recoverable at less than 260 US$/kgU in 2015
(OECD, 2016) –, would only allow 13 years of electricity production at current demand (Brown et al., 2018, p. 840).
IS DECOUPLING LIKELY TO HAPPEN? IS DECOUPLING LIKELY TO HAPPEN?
The rebound eect argument minimises the plausibility of the decoupling hypothesis.
Thus rebound eects must be taken into account while considering decoupling scenarios
as they might make rates of resource use more or less sensitive to the introduction of
resource-saving technologies and suciency-driven behavioural changes. The point is not
to argue against those, which may still have positive overall impacts, as long as rebound
eects remains limited, especially if anticipated by decision makers and counterbalanced
with proactive policies. But it remains very risky to rely exclusively on sectoral and technical
improvements. Rather, what is necessary is an in-depth and systemic consideration and
anticipation of potential rebound eects in the design of sustainability policies.
3. Problem shifting
An additional argument to be considered alongside rebound eects is that eorts to solve
one environmental problem can create new ones and/or exacerbate others. In other words,
decoupling of one environmental factor can occur at the expense of the (re-)coupling of
another one. As Ward (2017) points out to illustrate this argument, the world decoupled
GDP growth from build-up of horse manure in city streets and from whale oil, but only by
substituting it by alternative uses of nature. In what follows, we consider the example of
climate change mitigation and show how four dierent sources of energy often considered
as solutions for green growth merely change the form that the environmental burden takes,
often with unintended spill-over eects.
42 43
Example 3: Natural gas
The switch from coal to natural gas is a good example of shifting problems from one
greenhouse gas to another. The World Resource Institute (2016) reports a 6% fall in measured
US greenhouse gases emissions between 2000 and 2014, which alongside a 28% increase in
GDP appears to be a temporary absolute decoupling. This corresponds to a large shift away
from coal to natural gas (Feng et al., 2015), which was lauded by public authorities for its
ecological benets.35 The problem is that the extraction of natural gas emits methane, a gas
roughly 28 times more potent at heat-trapping than CO2 over a century (IPCC, 2013) which
easily escapes into the air before it can be captured in a pipeline. Turner et al. (2016) nds
that US methane emissions increased by more than 30% over the 2002-2014 period, which
more than cancels the drop in CO2. Same results for Howarth et al. (2011) who show that if
more than 3% of the methane from shale-drilling operations leaked into the atmosphere,
this would make shale gas more climate disruptive than coal (the leaks they report are
in the range of 3.6 to 7.9 per cent).36 The problem of methane leakages goes beyond the
relatively new phenomenon of shale gas extraction and concern convention gas operations
as well, especially the ones with faulty infrastructure.
4. The underestimated impact of services
Another hope for the decoupling of growth and environmental pressures lies in the
tertiarisation of the economy, that is the shift from extractive industries (agriculture and
mining) and manufacturing to services. This was already one of the explanation proposed
by the scholars who rst described the Environmental Kuznets Curve: “economic growth
brings about structural change that shifts the center of gravity of the economy from low-
polluting agriculture to high-polluting industry and eventually back to low-polluting services”
(Panayotou et al., 2000). Indeed, the service sector as such is much (only considering direct
consumption) less nature intensive than the primary and secondary one, and so if economic
growth is mostly driven by the expansion of economic activities where the product is mostly
information (e.g. nance, insurance, education), then raw materials and energy consumption
as well as environmental harms can be expected to decrease.37 We challenge the possibility
for such dematerialisation-through-services on several grounds.
35 Closing President Trump’s speech justifying the withdrawal from the Paris Agreement on June 1st, 2017, Scott Pruitt, then administrator of the
Environmental Protection Agency, announced: “before the Paris Accord was ever signed, America had reduced its CO2 footprint to levels from
the early 1990s. In fact, between the years 2000 and 2014, the United States reduced its carbon emissions by 18-plus percent.”
36 This leaking issue is not unique to fracking. It also happens because of ancient infrastructure or in the case of open mines where methane is
not actively captured.
37
Relative and absolute tertiarisation
For tertiarisation to contribute to decoupling, it must translate into an absolute, and not
only relative, decrease of the volume of industrial activities. A situation where the volume of
services grows without a corresponding and simultaneous shrinking of other sectors may
indeed be called a “relative” tertiarisation of the economy (the share of industrial activities in
the whole economy decreases while its volume still increases), but one that actually results
in higher environmental pressures.
With the impacts from the primary and secondary sector constant, a growing tertiary sector
adds to the pressures, even though it lowers the average energy intensity per euro. In
reality, this situation seems to be the rule rather than the exception.38 The development
of new types of services adds-up to other polluting activities instead of substituting to
them: consumers buy a Netix account with, and not instead, of a computer, and workers
can produce services if they are nourished, transported, and housed, not instead of food,
vehicles, and homes. Immaterial products require a material infrastructure. Software
requires hardware, a massage parlour requires a heated room, and the platform on which
we are writing these very words requires a computer along with all the material equipment
and energy necessary to make the Internet run. Services cannot be generated without raw
material extraction, energy provision, and infrastructure building, all of which are tightly
coupled with environmental pressures. The expansion of the service sector can hardly be
decoupled because it is part of an economy that grows as an integrated whole.
To the question “do societies with a larger service sector actually dematerialise?” Fix (2019)
answers an unequivocal “no.” Looking at 217 countries over the 1991-2017 period, he
concludes that “the evidence indicates that a service transition does not lead to absolute
carbon dematerialisation” (ibid. 4). Similarly, Suh (2006)increasing to 0.83 kg when supply-
chain-induced emissions are taken into account. Services produce less than 5% of total U.S.
GHG emissions directly, and their direct GHG emission intensities per dollar output are
much less (0.04 kg CO2 equiv/$ calculates that in 2004 in the United States, $1 spent on
seemingly material-free services requires 25 cents of output from manufacturing, utility, and
transportation service sectors. In Denmark, Jespersen (1999) nds that, if one includes all
indirect uses of energy, the service sector is actually as energy intense as the manufacturing
one. In Spain, Alcántara and Padilla (2009)demand volume, feed-back, internal and spill over
components nd the service sector responsible for the lion share of increases in emissions,
and this because of its reliance on other, polluting economic activities.
Additionally, workers in the service sectors receive wages, which are used for purchasing
material items produced in the manufacturing sectors. If the value of a dematerialised
good increases, it means that the purchasing power of those who sell that good increases
too (potential re-spending rebound) and that customers may work longer hours in order
to aord it (potential re-investment rebound), both having resources implications. So
the direct ecological intensity of a company specialised in internet advertisement may
be relatively low, but because it provides its employee with high-salary, and additionally
38 We should also say that situations where tertiarisation in one country occurs at the expense of (re)industrialisation in another is equally prob-
lematic for that it only shifts the environmental burden somewhere else (we will treat this point at length in Reason 7: cost shifting).
IS DECOUPLING LIKELY TO HAPPEN? IS DECOUPLING LIKELY TO HAPPEN?
What has been shown above for greenhouse gases emissions can be repeated for various
other environmental issues. The point is that piecemeal solutions are likely to fall short in
addressing a complex, systematic environmental crisis with many interdependent factors
at play. Substituting one problem like climate change for another such as biodiversity loss
cannot be considered problem solving. In order to argue that decoupling is possible, one
must show that a decoupling in one type of environmental pressure will not translate into
signicantly increasing another type of pressure.
44 45
because the advertising that it produces fosters the consumption of material or energy
intensive products and services such as cars, clothes, technological gadgets, and far-away
holiday travels, its indirect ecological intensity is higher than it seems.
From an environmental perspective, not all services are equally desirable and so certain
forms of tertiarisation are more desirable than others. Services in one sector do often
spill over in more consumption or production in another. Think of nancial and marketing
activities whose purpose is to boost sales of manufactured products and investment in
extractive industries. But also IT services and software development, which allows for-prot
enterprises to engage in planned obsolescence, or more generally to faster upgrades in
hardware. Or also of those services that rely on material and impactful tools, for example
being chair lifted up a ski slope or sky diving o a plane. In contrast, the expansion of yoga
clubs, couple therapists, and climbing centres may be less intensive on nature, even though
not necessarily so (see Services have a footprint too just below).
Not much tertiarisation left to do
Tertiarisation only provides a partial decoupling, and, importantly, one that has already
occurred in most OECD countries. In these economies, the share of services in GDP is
often already high, which is problematic because it is precisely those countries which have
the highest ecological footprint per capita and thus should reduce their impact the most.
Countries that have already reached a high degree of tertiarisation (more than 70% of value
added is generated in the service sector) retain a small industrial part that is increasingly
dicult to compress.
That is because certain sectors simply cannot be dematerialised. This is the case for
agriculture, transport, and housing construction, which, are often in the top sectors in terms
of emissions and used materials. Cement is a good example. Representing 5% of global
greenhouse gas emissions, its production implies both high levels of process emissions
and energy consumption, as well as important amount of increasingly scarce marine sand
(Rubenstein, 2012; The Pembina Institute, 2014). Although constructions can substitute
other materials to cement, it is dicult to imagine how services could possibly oer
adequate substitutes to most industrial production with regards to elementary needs such
as food, shelter, or mobility (the service of having a pizza home delivered requires roads,
a vehicle, and, not least, a pizza made from material ingredients) Hence, dematerialisation
only concerns a limited fraction of the global economy, leaving most of environmental
pressures unsolved.
Services have a footprint too
Even if services are less nature-intensive than industrial goods, they still have material
requirements and environmental repercussions, and so cannot be expected to fuel a
biophysically unbounded process of value creation. In one of their decoupling report, UNEP
(2014a, p. 70) nd a linear relation between expenditure in services and emissions of CO2 in
the direction of more services, more emissions.
Gadrey (2008) points to three factors explaining such correlation. Services require people to
travel, either from provider to customer (e.g. mail delivery) or the opposite (e.g. commuting
to school) which is made possible by material infrastructure, vehicles, and energy uses.
Then they are often anchored in specic material spaces (university building, train station,
airport, hospital, oces), whose construction, operation, and maintenance requires
materials and energy. They also rely on material tools, which production and use is far from
being environmentally-neutral. (ICT, computers, credit card readers, screens and displays,
cooling infrastructure in data centres).
In terms of materials, the making of information and communication technology products
such as computers, mobile telephones, LED screens, batteries, and solar cells require scarce
metals like gallium, indium, cobalt, platinum, in addition to rare minerals. An expansion
of services means more transactions using more devices, which require more minerals
whose extraction involves environmental impacts. Not only these material requirements
imply signicant environmental impact (from their mining) but their limited availability and
recyclability (cf. Reason 5) also put absolute limits to the growth of material-based services.
And even if it is common to observe a decline in the quantity of material products needed
to manufacture equipment, these eciency gain are being trumped by growth in volume of
equipment and intensity of usage (cf. Reason 2), often having to do with decreasing life-time
due to planned obsolescence (cf. Reason 5).
Services require energy, not only to build the material infrastructure they rely on, but also
to simply run. Not only for end-user equipment (laptops, smartphones, routers) but also
for the infrastructure, such as data centres and access networks (the wiring and antennas
that carry data). Malmodin et al. (2010) calculate that ICT used 3.9% of global electricity
in 2007, accounting for 1.3% of global greenhouse gas emissions. Numbers are similar in
other studies; for instance, the information and technologies sector produced 2% of global
CO2 emissions in 2007 (830 MtCO2e), half of it accounting for computers and devices and the
other half for data centres and telecoms (The Climate Group, 2008). Starting from Malmodin
et al.’s (2010) 3.9% of global electricity used by ICT, Van Heddeghem et al. (2014)there is also
a growing attention to the electricity consumption associated with ICT equipment. In this
paper we assess how ICT electricity consumption in the use phase has evolved from 2007 to
2012 based on three main ICT categories: communication networks, personal computers,
and data centers. We provide a detailed description of how we calculate the electricity use
and evolution in these three categories. Our estimates show that the yearly growth of all
three individual ICT categories (10%, 5%, and 4%, respectively nd that it went up to 4.6% by
2012. Forecasting to 2030, Andrae and Edler (2015) estimate that ICT could consume up to
51% of global electricity, contributing up to 23% of global greenhouse gas emissions.
In itself, the Internet accounts for between 1.5 and 2% of the world’s energy consumption
(CEET, 2013). Only considering the users’ side, the 100 most visited French website require
8.3 GWh or the energy consumption equivalent of 3,077 households (WEA, 2014). Energy
consumption resulting from Bitcoin emits an annual 69 mtCO2 and, if more broadly used,
could alone produce enough emissions to push warming above 2°C within less than three
decades (Mora et al., 2018). Carr (2006) estimates the energy consumption of a Second Life
IS DECOUPLING LIKELY TO HAPPEN? IS DECOUPLING LIKELY TO HAPPEN?
46 47
avatar to be around 1,752 kWh per year, which he compares to a world average for humans
of 2,436 kWh. Looking at the ecological cost of music in the US, Devine and Brennan (2019)
discovers that, even though music has become almost completely digital, it is, in terms of
greenhouse gases, more polluting than it has ever been: from 140 million kg in 1977 to 157
in 2000 and between 200 and 350 in 2016.
Because of prevalence of fossil sources in the current energy mix of countries hosting data
centres, ICT ends up with a heavy contribution in terms of emissions. The Greenpeace
report “How Clean is Your Cloud?” (2012) nds that, for example, 39.4% of the electricity
used by Facebook servers is generated by coal plants, while it is 49.7% for Apple. This
energy consumption adds up to an already high level of energy demand, exacerbating the
environmental impact of the energy sector. And perhaps yes, this climate impact would
disappear should all services run on renewable energy, but, assuming that this is even
possible (cf. Reason 1), then it would still generate an array of environmental issues (cf.
Reason 3).
5. Limited potential of recycling
Recycling is a common strategy advocated for decoupling often associated to the idea of
a circular economy. The idea is that resource decoupling could be possible if all materials
required for the production of new products were extracted from the old products that
have been thrown away and not from nature. The traditional linear process of production
would then be turned into a “closed-loop” (Stahel and Reday-Mulvey, 1981), “zero waste”
(Palmer, 2005), “cradle et cradle” (McDonough and Braungart, 2010) economy. Of course,
closing the loop between waste and extraction via recycling is a sensible goal, and in theory,
one would want any economy to be as circular as possible. What we are about to argue is
that there are limits to this circularity and that these limits are quickly reached in a fast-
growing economy.
Recycling itself requires new materials and energy
Perpetual motion machines do not exist in reality. Even though signicant gains can be
expected from better recycling, the process of recycling itself necessitates energy and,
most of the time, new materials, which would then also need to be recycled at some point,
requiring the use of additional new material, and this ad innitum (Georgescu-Roegen, e.g.
1971, p. 132, spoke of an “innite regress”). This means that because of unescapable laws
of nature (here the entropy law), the technically feasible recycling rates are always below
the theoretically possible ones. On top of that, the economically justiable rates are often
signicantly below what is technically possible for that the marginal cost tends to increase
the more a process approaches its theoretical maximum (cf. Reason 1).
Since materials inevitably degrade through time (2nd law of entropy), they can only be
recycled into the same products for a limited number of times before they have to be used
to produce other products with lower grade requirements. Put another way, sooner or
later, any recycling is necessarily downcycling. For instance, plastic bottles can be recycled
into plastic bre for clothing but not back into plastic bottles, and they can nally end up in
the noise protection walls along motorways. Paper cellulose bres can only endure 3 to 6
cycles, for which they need to be mixed with new bres, and until they become too fragile
to be used for paper before being used for cardboard and later as housing isolation and
nally as biofuel. Just like for energy, this wearing down of materials sets absolute limits on
how circular any economy can be.
Giampietro (2019) proposes another way of thinking about it. In a way, nature already
recycles all materials for free, albeit too slowly for current rates of extraction. Arguing that
materials and energy will then be recycled within the economy, and not outside of it, comes
with an energy price tag. As always, production requires labour, tools, and energy, except
that this time, what is being produced is recycling services. Put another way, it is a use of
primary energy and material to recycle waste, that is secondary energy and material. In a
world where the economy is relatively small compared to its environment and where the
ows of primary energy and materials are larger than the secondary ows, an economy
can indeed be circular. Yet, when the scale of the second matches the ones of the rst,
circularity is compromised. As the author puts it: “what really matters in relation to the
potential of recycling is the size of the required input ows and the waste ows generated
by the economy (technosphere) compared to the size of the primary sources and primary
sinks made available by ecological processes (biosphere)” (ibid. 149). If economic growth
means an increase in size of the economy compared to its environment, then it means that
growing economies will sooner or later reach the limits of circularity.
Recycling rates are far from 100%
Of course, one can argue that this entropy argument is irrelevant to a situation where rates
of recycling are low and that simply increasing those rates to match the pace of increase
of resource use will be enough to achieve absolute decoupling. But here comes a practical
consideration: How likely is it for recycling rates to increase that much?
IS DECOUPLING LIKELY TO HAPPEN? IS DECOUPLING LIKELY TO HAPPEN?
The so-called “service economy” carries a heavier biophysical backpack than one would
think. In the countries with the most urgent mitigation imperatives, the service sector has
already been developed to its maximum without the benets of absolutely decreasing
environmental pressures. Services have a footprint, that even though lower than manu-
factured products, is often only added on top of the environmental pressure pile without
much substitution occurring. This is because the service economy can only exist on top
of the material economy, not instead of it. Moreover, services such as advertising or -
nancial products do sometime actively foster more polluting production, which results in
an overall rise in environmental pressure. Again, we are not arguing against services; on
the contrary, it is crucial to replace jobs in resource-intensive sectors with more labour-
intensive work. Rather, the point we make is that directly reducing output in the problem-
atic sectors would be more eective than developing activities around them hoping that
substitution would somehow occur.
48 49
Let us rst assume that recycling does not require extra energy and that all materials can
be recycled perfectly. In 2005, 62Gt/yr material have been processed, generating 41Gt of
outputs, (19Gt biomass for feed, food and fodder, 12Gt fossil fuels, 4.5Gt mined ores) (UNEP,
2011). At the same time, only 4Gt of material have been recycled. This is not surprising for
that certain materials that are currently used cannot be recycled. For example, fossil fuels
and biomass burnt for energy.39 One fth of total resources used worldwide are fossil fuels,
and almost half are energy carriers. The 98% of fossil fuels that are burnt as a source of
energy along with the biomass consumed for feed, food, and fodder cannot be re-used or
recycled. Of course, shifting to a 100% renewable energy provision would solve this problem
(although perhaps at the cost of creating others, cf. Reason 2), but we are still far from this
situation.
Another problem is that many modern products are too complex to be recycled.
Miniaturisation can save material but renders the recovery of materials more dicult
and when this is technically feasible (which is not always the case), more costly and thus
less economically interesting. Reuter et al. (2018) study the recyclability of one of the most
modular smartphone (Fairphone 2) and nd that the best possible recycling scenario would
only recover about 30% of the materials. Most problematically, this is also the case for
technology to harvest and store renewable energy. UNEP (2011) estimated that less than
1% of specialty metals are recycled.
A third point is that improvements in recycling are often more than cancelled out by rises
in rates of replacement (sometime fuelled by planned obsolescence). Indeed, if rates of
recycling are increasing at a slower pace than the reduction of products’ average lifetime
(i.e. the rate of product replacement), then resource use is set to increase. If the ability to
recycle is slower than the will to produce, then virgin resources will have to be used.
There is not enough waste to recycle
This last argument is a matter of basic arithmetic. Just for now, let us still assume that rates
of recycling would increase signicantly faster than their current trends (while still relaxing
the assumption that recycling in itself requires energy and new materials). Yet, even this
would in itself not be a guarantee to maintain the growing economy’s throughput, since in
an economy with increasing resource use, the amount of used material that can be recycled
will always be smaller than the material needed for growth. As the economy keeps on
expanding, more materials will be required than the ones available from previous periods
of time, and so the materials available for recycling within this economy will not suce. This
would be like a snake trying to make a larger skin out of the scraps of its previous, smaller
skin.
As shown by Grosse (2010), in an economy where material consumption increases, recycling
can only delay resource depletion. The author takes the example of steel, the best-recycled
material worldwide. At a current 62% recycling rate and with a yearly rise in consumption of
39 This is also the case for dispersive uses that divert materials from recycling circuits. (e.g. scarce metals used in ink and painting pigments,
additives in glass and plastic).
3.5%, recycling is only delaying depletion by 12 years. If we keep consumption rates steady,
even increasing recycling rates to 90% would only add an extra 7 years before depletion.
Arnsperger and Bourg (2017, p. 73) apply the Grosse (2010) calculation to copper. They
assume that the residence time of copper in the economy is of 40 years and that 60% of it
can be recycled with current technologies. Out of the 6 million tons of copper used in 1975,
this means that 4 millions could have been recovered by 2015. However, consumption of
copper has grown to 16 million in the last forty years and so, despite recycling, 12 million
tons of virgin copper must still be extracted. In this case, even with assuming an illusory
100% recycling rate, the extraction would have more than doubled during the period.
What exacerbates the limited availability of products to be recycled is the fact that a
signicant portion of all resources used end up in infrastructure, often for quite some time.
De Decker (2018) proposes a simple back-of-the-envelope calculation. In 2005, the world
used 62Gt of natural resources: 4Gt for disposable products lasting less than one year and
26Gt in buildings, infrastructure, and consumer goods lasting more than one year. The
same year, 9Gt of resources were disposed of in the process of production. The author
concludes that the total quantity of materials available for recycling at the start of a second
year of production is 13Gt (4Gt of disposable products + 9Gt of surplus resources), of which
only a third could be eectively recycled. Plain to see that this number is not only short of
what would be needed just to produce the same as in the previous year (62Gt), but even
more so for a growing economy.
6. Insucient and inappropriate technological change
The debate on the likeliness of future decoupling is, at its very core, a debate on the potential
of technological innovation. Decoupling may have not occurred yet, and economic growth
may seem biophysically constrained, either because of rising costs of extraction (cf. Reason
1), unforeseen problem shifting (cf. Reason 3), material infrastructure (cf. Reason 4), or
limited recycling (cf. Reason 5), but the green growth discourse develops on the assumption
that future innovations soon to come would do away with that. In our opinion, this
hypothetical argument has several shortcomings having to do with the purpose, unintended
IS DECOUPLING LIKELY TO HAPPEN? IS DECOUPLING LIKELY TO HAPPEN?
An innitely growing circular economy is an arithmetical impossibility, and a contradic-
tion in terms. Recycling is itself limited in its ability to provide resources for an expanding
material economy. In the end, our point is not to question the usefulness or relevance of
recycling, which could on the contrary play a crucial role in a non-growing economy, but
merely to point to the fact that hopes of decoupling based on recycling are misinformed.
The reality is that recycling rates are currently low and only slowly increasing, that recy-
cling processes generally still require a signicant amount of energy and virgin raw mate-
rials, and that it is mathematically impossible for recycling to match rates of replacement
in a context of increasing consumption.
50 51
consequences, and pace of technological change. Simply put: technological progress is (1)
not targeting the factors of production that matter for ecological sustainability and not
leading to the type of innovations that reduce environmental pressures; (2) it is not disruptive
enough as it fails to displace other undesirable technologies; and (3) it is not in itself fast
enough to enable a decoupling that is absolute, global, permanent, large and fast enough.
Essentially we are not arguing against innovation in itself. Our point is that technological
innovation is most often ambivalent when it comes to addressing environmental issues, and
that the potential of future technological innovations is most likely too limited, and in any
case uncertain. Relying on the belief that technological innovation will bring all necessary
solutions to environmental problems appears as an extremely risky and unreasonable bet.
Not leading to relevant innovations
Innovation is not in and of itself a good thing for ecological sustainability. The desirable
type of innovation is eco-innovation or one that results “in a reduction of environmental risk,
pollution and other negative impacts of resources use compared to relevant alternatives”
(Kemp and Pearson, 2008). But this is only one type among several. In general, rms have
an incentive to innovate so as to economise on the most expensive factors of production in
order to maximise prots. Because labour and capital are usually relatively more expensive
than natural resources, it is likely that more technological progress will continue to be
directed towards labour- and capital-saving innovations, with limited benets, if any, for
resource productivity and a potential rise in absolute impacts due to more production. But
decoupling will not occur if technological innovations contribute to saving labour and capital
while leaving resource use and environmental degradation unchanged.
Another issue is that technologies do not only solve environmental problems but also tend
to create new ones. Assuming that resource productivity becomes a priority over labour
and capital productivity, there is still nothing preventing technological innovations from
creating more damage. For example, research into processes of extractions can lead to
better ways to locate resources (imaging technologies and data analytics), to extract
them (horizontal drilling, hydraulic fracturing, and automated drilling operations), and to
transport them (Arctic shipping routes). These innovations may target resource use but
with a result opposite to the objective of decoupling, that is more extraction. And this is not
even considering unintended side-eects, which often accompany the development of new
technologies (Grunwald, 2018).
Not disruptive enough
Another problem has to do with the replacement of harmful technologies. Indeed, it is not
enough for new technologies to emerge (innovation), they must also come to replace the
old ones in a process of “exnovation” (Kimberly, 1981). What is required is a “push and
pull strategy” (Rockström et al., 2017): pushing environmentally-friendly technologies into
society and pulling harmful ones, like fossil-based infrastructure out of it.
First, in reality, such a process is slow and dicult to trigger. Most polluting infrastructures
(power plants, buildings and city structures, transport systems) require large investments,
which then creates inertia and lock-in (Antal and van den Bergh, 2014, p. 3). Let us for
instance consider the energy, buildings, and transport sectors, which account for the large
majority of world energy consumption and greenhouse gas emissions: initial lifetime for a
nuclear or a coal power plant is about 40 years. Buildings can last at least as much. Average
lifetime for a car is 12-15 years, and this is about what it takes for an innovation to spread
in the vehicle eet. The wide availability of petrol refuelling stations gives an infrastructural
advantage to petrol-based cars, whereas this is the opposite situation for electric, gas,
or hydrogen vehicles that would require dierent and new supporting infrastructures.
Building a highway or a nuclear plant is a commitment to emit for at least as long as these
infrastructures will last – Davis and Socolow (2014) speaks of “committed emissions.”
Energy is a good case in point: using more renewable energy is not the same as using less
fossil fuels. The history of energy use is not one of substitutions but rather of successive
additions of new sources of energy. As new energy sources are discovered, developed, and
deployed, the old sources do not decline; instead, total energy use grows with additional
layers on the energy mix cake (see gure below). York (2012) nds that each unit of energy
use from non-fossil fuel sources displaced less than one-quarter of a unit of its fossil-fuel
counterpart, showing empirical support for the claim that expanding renewable energies
is far from enough to curb fossil fuel consumption. The relative part of coal in the global
energy mix has been reduced since the advent of petroleum but this occurred in spite of an
absolute growth in the use of coal (Krausmann et al., 2009).
Moreover, even if the decision to substitute renewables to all fossil energies was enacted,
it is doubtful whether this process can happen fast enough – or even at all, taking material
requirements into consideration. In a recent study, the International Renewable Energy
Association (IRENA, 2018) estimates that a continued GDP growth compatible with a 2°C
warming target would require the addition of 12,200 GW of solar and wind capacity by 2050.
This means increasing renewable capacity addition rates by b 2.3 to 4.6 times. Because the
study assumes a parallel decrease in energy intensity of 2.8% per year (double the historical
rate), and because it aims for the 2°C target (and not the more ambitious 1.5°C), one might
consider that the speed of renewable energy development would need to be even higher:
for instance, Garrett (2012) calculates that one would need to build one nuclear power plant
per day (or equivalent in renewables) in order to decarbonise an energy demand steadily
growing at current rates.
This pattern observed with energy whereby new technologies supplement rather than
replace existing ones, can be observed in many other sectors as well. Computers have not
brought about the paperless oce because computers and papers came to complement
each other (York, 2006). The rise of synthetic rubber, whose production was established
during World War II, did not stop natural rubber production and consumption from
increasing steadily throughout the 20th century (Cornish, 2001). Likewise, the explosion of
synthetic bers like polyester and nylon has not displaced natural ber production. While
yearly world production of synthetic bers has grown from less than 2 Mt in 1950 to above
60 Mt today, the production of natural bers has more than tripled, from under 10 to roughly
30 Mt, with annual variations due to climactic conditions(The Fiber Year, 2016). Additional
consumption largely surpassed substitution.
IS DECOUPLING LIKELY TO HAPPEN? IS DECOUPLING LIKELY TO HAPPEN?
52 53
Not fast enough
In light of the past decades of technological change, the rate of improvement that is needed
for high-income, high-footprint economies to absolutely decouple appears disproportionate
in contrast to past and present rates of technical progress.
Let us consider the example of carbon emissions. Jackson, (2016, pp. 96–100) considers several
simple hypothetical decoupling scenarios. The rst, baseline scenario runs as follow: extending
the trend of global annual per capita economic growth of 1.3% in parallel of 0.8% of expected
annual population growth and with the average annual decline of carbon intensity of 0.6% that
has been observed since 1990, would result in carbon emissions growing by 1.5% per year (1.3%
+ 0.8% – 0.6% = 1.5). In order to achieve a 90% emission reduction in 2050 compared to current
levels with the same GDP and demographic hypotheses, the emission intensity would need
to decline at an average rate 8% per year until 2050 – reducing the average carbon content of
economic output to 20 gCO2/$, that is to say 1/26 of what it is today (497 gCO2/$). In comparison,
the carbon intensity of the global economy fell from about 760 grams of CO2 per dollar in 1965
to just under 500 g/CO2/$ in 2015, that is to say an annual decline of only 1%.
Many more ambitious scenarios can be imagined,40 but the message is already clear: relying
only on technology to mitigate climate change implies extreme rates of eco-innovation
improvements, which current trends are very far from matching, and which, to our
knowledge, have never been witnessed in the history of our specie. Such an acceleration of
technological progress appears highly unlikely, especially when considering the following
elements.
First, global carbon intensity improvement has been slowing down since the turn of the
century, from an average yearly 1.28% between 1960 and 2000 to 0% between 2000 and
2014 (Hickel and Kallis, 2019, pp. 8–9). Narrowing the scope to high-income OECD countries
only, where most innovations are developed, the improvement rate of CO2 intensity still
declines from 1.91% (1970-2000) to 1.61% (2000-2014), which is a long way from matching
appropriate levels to curb emissions to a 2°C target, let alone to 1.5°C.
This empirical observation is nothing like a surprise with regards to theory. Technological
innovation is limited as a long-term solution to sustainability issues because it itself exhibits
diminishing returns (cf. Reason 1). Tracking the number of utility patents per inventor in the
United States over the 1970-2005 period, Strumsky et al. (2010) provides evidence that the
productivity of invention declines over time, including in the sectors such as solar and wind
power as well as information technologies (which are often acclaimed for their innovative
potentials). “Early work […] solves questions that are inexpensive but broadly applicable.
[Then] questions that are increasingly narrow and intractable. Research grows increasingly
complex and costly […]” (ibid. 506). Looking at total factor productivity changes from 1750
to 2015, Bonaiuti (2018) argues that humanity has entered an overall phase of decreasing
40 Since in the aforementioned baseline scenario, the carbon budget ends up being fully used by 2025, the author calculates in a second scenario
the requirement for a 95% reduction holding all else equal. The rate of improvement rises to a 10.4% reduction in carbon intensity year on year,
but the carbon budget still runs out by the end of the 2020s. In order to avoid this, a third scenario sets the target year to 2035 instead of 2050,
and the necessary speed of technological change becomes 13% for a 90% reduction and 15% for a 95% reduction. In scenario 4, low-income
countries are expected to match the income of the richer ones (with a 2% expansion in rich countries, it will take a rate of growth of 7.6% in poor
ones for both levels of income to converge) Under those conditions, the carbon intensity must be less than 2 gCO2/$ to achieve a 95% reduction,
almost 1/250 of what it is today. Meeting these targets by 2035 requires a reduction of carbon intensity to average an annual 18%, 100 times
faster than the current rate of change.
marginal returns to innovation.
To sum up, technology is no panacea. It is indeed impossible to predict what the future holds in
terms of innovations over the long term. Yet the point is that reasons to be sceptical about the
potential for technological change to foster the type of decoupling we described as necessary are
multiple and serious. First, many technologies that could have severed part of the link between
GDP and environmental pressures have been here for several decades now with only minimal
eects. More importantly, all innovations do not go in the direction of more ecological sustainabil-
ity. In a capitalist and growth-oriented economy, innovation is most often strongly dependent on
prot-making opportunities, hence partly oriented to this aim. In such a context, most innovations
may result in GDP increase, but only few of them might help mitigate environmental pressures.
Future technological changes may perhaps bring some additional improvements, provided these
are not cancelled by rebound eects (cf. Reason 2), and provided they do not result in problem
shifting (cf. Reason 3). Past and current paces of technological evolutions are clearly at odds with
the urgent and radical changes that the environmental crises call for, and declining marginal rates
of improvement (cf. Reason 1) gives little reason for optimism about the future.
7. Cost shifting
The absolute decoupling shown in early-industrialised nations is only apparent if those countries
outsource their biophysically-intensive production somewhere else. This leakage eect41 – also
sometime called “decoupling through burden shifting” (UNEP, 2014a) or “virtual decoupling”
(Moreau and Vuille, 2018)– can be either intentional or conjectural (Peters, 2008). It is intentional
or direct when the geographical shift in production results from an obvious choice to relocate to
jurisdictions with less stringent environmental regulations – this is referred to as the “pollution
heaven hypothesis.” It is conjectural or indirect when the eect is attributed to a broader set of
factors (e.g. dierences in cost of labour, industrial capacity, access to resources, or technology).
Based on this premise, globalisation would cause polluting activities to concentrate in the least
regulated most often low-income countries. Put another way, trade would enable the decoupling
of certain regions at the expense of an intensication of environmental pressures elsewhere;
or in other words, would allow high-consumption countries to externalise the environmental
costs of production to low-consumption countries (one then speak of “embodied” impacts, e.g.
embodied emissions, embodied energy).
Empirical evidence of environmental cost shifting
The empirical literature on the embodied environmental pressure in trade is consistent.
Reviewing embodied carbon studies, Sato (2014) identied a large and growing volume of
embodied carbon emissions in international trade, which accounted in 2006 for around one
fourth of global emissions. Looking at 113 countries, Peters et al., (2011) nd that the net
41 Because mostly focusing on carbon, this phenomenon is referred to as “carbon leakage” in the empirical literature. The term “leakage” depoliti-
cises the process and so we prefer, following Kapp (1950) and the school of world-system analysis (most notably Hornborg, e.g. 1998)neoclassical
economic ideology has dispelled all possible criteria for assessing a market transaction as unequal or unfair. One way to assess the occurrence
of unequal exchange may be to look at the direction of net flows of energy and materials (concrete, productive potential, to call it a process of
environmental cost shifting whereby richer nations systematically impose the environmental cost of their consumption onto poorer countries.
IS DECOUPLING LIKELY TO HAPPEN? IS DECOUPLING LIKELY TO HAPPEN?
54 55
Conclusions for Section 3
In this section we have oered a number of reasons to be sceptical about decoupling: (1) Rising
energy expenditures, (2) rebound eects, (3) problem shifting, (4) the underestimated impact
of services, (5) the limited potential of recycling in a growing economy, (6) insucient and
inappropriate technological change, and (7) cost shifting. Each of them taken individually casts
doubt on the possibility for decoupling and thus the feasibility of “green growth.” Considered
all together, the decoupling hypothesis appears highly compromised, if not clearly unrealistic. It
is urgent to draw the consequences in terms of policy making, and following the precautionary
principle, to move away from the continuous pursuit of economic growth in high-consumption
countries, in particular in the EU. Following the arguments we have discussed in this section, the
burden of proof rests on decoupling advocates. Unless adequate and convincing demonstrations
are brought against each and all of the above-mentioned arguments, the concept of decoupling
remains an act of pure belief with little relevance for policy making.
emission transfers via international trade from low-income to high-income countries has
quadrupled between 1990 and 2008.
This does not only concern emissions but also resources. In between 1997 and 2001, 16%
of the global water footprint was embodied in global trade (Hoekstra and Chapagain, 2007).
Raw material embodied in international trade accounted for 30% of the global material
consumption increase during the 1990-2010 period, “this eect being due to the growing
contribution of less material-ecient economies to global production” (Plank et al., 2018,
p. 19). Likewise, Schandl et al. (2018, p. 8) report that global material eciency is declining
because of a “large shift of economic activity from very material-ecient economies, such
as Japan, the Republic of Korea, and Europe, to the currently much less material-ecient
economies of China, India, and Southeast Asia.”
For example, a 2011 OECD report claimed that Germany, Canada, Italy, and Japan had
achieved an absolute decoupling of greenhouse gases emissions since the 1980. Even
though, as pointed out by Bednik (2016, p. 107) the authors of the report pinpoint that “parts”
of this decoupling is due to the exportation of manufacturing activities in emerging and
developing countries (OECD, 2011, pp. 15–16). The dierence between the gross emissions
(measured with a production approach) and net emissions (measured with a consumption
approach) was indeed of 27.7% for Germany and 24.7% for Italy in 2004, and as high as 44%
for France (Laurent, 2012).
More generally, Davis and Caldeira (2010) estimate the dierence between production and
consumption emissions to be around 30% in rich countries. When compared to the rates
of supposedly absolute decoupling announced in certain studies, the sole factor of cost-
shifting is enough to explain the observation.42
Why cost shifting happens?
What is observed empirically nds its theoretical explanation in world-system analysis and
dependency theory (Amin, 1976; Emmanuel, 1972; Wallerstein, 1974). Building on such
tradition, Hornborg (1998, p. 38) calls this process “ecologically unequal exchange”: “a
relation of exchange, even when it has been entered voluntarily, can generate a systematic
deterioration of one party’s resources, independence, and development potential.” From
this particular perspective, the world can be divided into core countries, semi-periphery
countries, and periphery countries, with the former having more power to import wealth
from and export illth to others.
Emmanuel (1972) showed how dierences in price of labour between nations lead to net
transfer of embodied labour from the poorest to the richest. What is relevant for decoupling is
that the same mechanism is at work but with material, energy, and pollutions. If it is cheaper
to produce what is most polluting elsewhere, and as a consequence there will be a net transfer
of environmental burden from the global North to the global South. In decoupling terms, this
would mean that core countries nd themselves in a situation of ecological decit with their
periphery.
42 In their study of embodied emissions in British imports, (Druckman et al., 2008, p. 594) conclude that “any progress towards the U.K.’s carbon
reduction targets (visible under a production perspective) disappears completely when viewed from a consumption perspective.”
Decoupling in certain regions of the world would be a “local illusion” (Hornborg, 2016, p. 115)
or “geographical illusion” (Fischer-Kowalski and Amann, 2001) that is enabled by a process of
“environmental load displacement” (Muradian et al., 2001) or “cost shifting” (Kapp, 1950) from
one locality to another or from the present to the future. Following this line of thinking, Hornborg
(2001, p. 33) invites us to “think of the world as a system, in which one country’s environmental
problems may be the ip side of another country’s growth.” This is especially relevant when it
comes to technological change. Hornborg (2019, p. 15) argue that modern technology “should
be understood not simply as an index of ingenuity, but as a social strategy of appropriation (of
labour and land)” or “a strategy of displacement (of work and environmental loads).” A vacuum-
cleaner may save time in cleaning the house, but it does so at the expense of someone having
to spend time and energy building the vacuum, and a lot of more people having to extract the
materials necessary for it.
It would be irrelevant to celebrate decoupling in one country if this one is achieved at the expense
of coupling in another one, especially if the latter one is poorer than the former. There are strong
theoretical reasons to believe that the few cases of local decoupling that are celebrated (which
remain exceptions) are in fact mostly a displacement of environmental pressures elsewhere, as
we have shown in Section 2. If that is so, it means that ecological sustainability can only be achieved
via a downscaling of polluting production. This reason is perhaps the most problematic of all. As
long as individuals, rms, and nations stay engaged in cost-competition, there will be incentives to
swipe ecological costs under the rug, with the lightening of footprints remaining a mere statistical
trick.
56 57
Conclusions:
Farewell to green growth
This report has sought to make a number of points. To begin with, scientic studies and
political discussions about decoupling must be precise as to how they dene the term (is
it relative or absolute, dealing with resource use or impacts, global or local, and temporary or
permanent?) and how it relates to existing environmental thresholds and political targets:
Is it sucient to achieve the target? Does it account for a fair distribution of burdens and
benets?
In the second section, we have reviewed the empirical decoupling literature searching for
evidence of the type of decoupling that would justify green growth as a political strategy.
Our nding is clear: the decoupling literature is a haystack without a needle. Of all the
studies reviewed, we have found no trace that would warrant the hopes currently invested
into the decoupling strategy. Overall, the idea that green growth can eectively address the
ongoing environmental crises is insuciently supported by empirical foundations.
Here, it is important to note that decoupling is neither a new nor a never-tried strategy; it
has been the main sustainability plan, at least for the OECD and the European Commission,
since 2001, and a key feature of many member states’ environmental and industrial policies
since the 1990s. Decoupling is not an innovative strategy but rather the continuation of
what has been done in the European Union in the last decades. The meagre achievements
of the decoupling strategy until now reported in Section 2 cast serious doubt as to whether
prospects for the short- to medium-term future are better. Considering the last two
decades as a trial period, one must confront the fact that decoupling has failed to deliver
the ecological sustainability it promised
At last, we claimed that there were several reasons to be sceptical about the occurrence
of decoupling in the future. (1) Rising energy expenditure, (2) rebound eects, (3) problem
shifting, (4) the underestimated impact of services, (5) limited potential of recycling, (6)
insucient and inappropriate technological progress, and (7) cost shifting can, each
CONCLUSIONS: FAREWELL TO GREEN GROWTH CONCLUSIONS: FAREWELL TO GREEN GROWTH
58 59
individually, and even more all together, compromise or even dismiss the possibility of
“green growth.” The insight here is not that eciency improvements are unnecessary (and in
that sense, we support most of the decoupling-targeted policies advocated by UNEP in their
2014a report), but instead that it is theoretically and empirically unrealistic to expect those
to absolutely, globally, and permanently delink a constantly growing economic metabolism
from its biophysical base. Given the historical correlation of Gross Domestic Product and
environmental pressure as well as the required technological improvements needed for a
suciently large and fast reduction in resource use and environmental degradation, relying
on decoupling alone to solve environmental problems appears to be an extremely risky and
irresponsible bet. Framing issues of social ecological justice with the concept of decoupling
is like trying to cut a tree with a spoon: it is likely to be a long attempt and most likely to fail
in the end.
As Daly (1977, p. 115) already argued forty years ago, the bet we are facing is similar to
Pascal’s Wager. Either we hope that somehow these seven problems will solve themselves,
continue growth-as-usual and risk a social and environmental collapse; or we acknowledge
that decoupling is likely to fail with irreversible consequences on the environment, and