Structural change and socio-economic disparities in a net zero transition

Abstract

A net zero transition is likely to generate substantial and irreversible economic transformation. High-carbon industries and their related occupations will disappear, while new low-carbon industries and occupations will be created. In the aggregate, the impact of the transition on GDP and employment is commonly projected to be relatively moderate. However, such estimates hide drastic distributional issues that are sectorally and regionally concentrated. We use three sectorally detailed and regionally disaggregated macroeconomic models to explore the possible levels and impacts of structural change in a well below 2°C scenario. In addition to the expected decline in the carbon-intensive industries, we observe secondary impacts, particularly in the services sectors, that vary significantly between models. The risks entailed with structural change involve worsening economic disparity and societal division that could exacerbate existing socioeconomic and political polarisation. Impact assessments of decarbonisation should consider such distributional issues to avoid post-industrial decline and widening socioeconomic inequalities.

KEYWORDS:

• Structural change
• just transition
• climate policy
• multi-sectoral macroeconomic modelling
• post-industrial decline

1. Introduction

The low-carbon transition necessary to achieve the objectives of the Paris Agreement for mitigating climate change is, by design, transformative for the global economy (IPCC, 2022a; Mercure et al., 2021a). The energy supply sector, together with industry, transport, and households must evolve to use new methods and low-carbon technologies and strive to reduce aggregate energy and material use. At the same time, old systems, carbon-intensive manufacturing and fossil-based production processes, durable goods and their value chains must be phased out (Geels et al., 2017).

For the economy and the well-being of workers and citizens, the transition could have impacts more pervasive than could be foreseen by simple aggregate economic models, including some established Integrated Assessment Models (IAMs). While the impacts of the transition on aggregate variables including global GDP and employment are projected to be relatively moderate in many models (IEA, 2021; ILO, 2018; IPCC, 2022a), significant disparities could be created both between and within countries and sectors due to the simultaneous birth and death of industries and occupations (Carley & Konisky, 2020; Fragkos et al., 2021).

The form and shape of decarbonisation policies affect the degree of distributional disparity that their impacts generate (Markkanen & Anger-Kraavi, 2019). Structural changes are by definition unavoidable during a transition from high-carbon (sunset henceforth) industries to low-carbon (sunrise henceforth) industries and activities (Freeman & Louçã, 2001). The degree of mobility of workers, capital and skills between sectors is widely debated but could have large impacts on structural and societal changes (Fallick, 1993; Spencer et al., 2018).

The common assumption in standard economic theory that capital will re-allocate itself from sunset to sunrise activities may be misleading (Mercure, 2022). As a result, economic expectations will, and perhaps already look bleak in regions that are economically focused on activities that lie in the ageing sunset paradigm of fossil-based energy and carbon-intensive industries (Carley & Konisky, 2020). This not only includes industries directly involved with the extraction, transport, and combustion of fossil fuels, but also the manufacturing of internal combustion engines, production of steel, cement and petrochemicals, heavy manufacturing, and plastics industries. These industries are often geographically concentrated (Arthur, 1990; Caldecott et al., 2017; Krugman, 1991). Such regions could therefore be at risk of post-industrial decline and the ability of agents to adapt depends to a large degree on their social and regional mobility. Price, income, and wealth barriers can emerge within countries between regions of high affluence in comparison to regions affected by post-industrial decline, notably in housing markets, hindering mobility from occupations in the sunset regime towards sunrise ones (Lux & Sunega, 2012; Martin et al., 2016).

Meanwhile, the future will naturally look bright in regions that are already oriented towards the new sunrise technologies. This could include not only low-carbon cleantech (e.g. solar photovoltaics, wind turbines), but also, in a broader emergent tech ecosystem, artificial intelligence, automation, biotechnology, batteries, and electric mobility. Early warning signs of division, disparity and discontent have emerged in various parts of the world. This includes the ‘Gilets Jaunes’ protests in France against new taxes on fuels (Hope, 2019). At face value, this could be interpreted to mean that the ‘political acceptability’ of carbon taxes is low (Jewell & Cherp, 2020), yet a more critical analysis of the structure of socio-economic transformation and sectoral interdependencies can help illuminate why such discontent arises even though the predicted aggregate economic and employment impacts are so moderate. What must also be recognised is that, without such a transformation, the physical effects of climate change will drive stark inequalities (Hallegatte & Rozenberg, 2017) and have impacts across the economy (IPCC, 2022b).

In this paper, we use three well-established, highly detailed sectoral macroeconomic models to explore the disparities and structural change that emerge within the economy through a net zero transition. This aims to better understand distributional issues between countries and sectors and to contribute to an emergent literature on structural change, which can help the design of resilient, fair, and socially acceptable climate policies towards achieving a just transition. We build on work by Lefèvre et al. (2022) where a new research agenda for global structural change in the mitigation context was proposed, to investigate the possible ranges, levels and directions of structural change impacts that can be expected in different economic sectors and regions in a net zero transition. We employ models widely used for impact assessment in policy circles that are built upon different theoretical economic bases (demand-led and supply-driven models, see Mercure et al., 2019).

To address these research goals, we begin in Section 2 by drawing from the Great Waves framework to provide a historical perspective on disruptive innovation and the characteristics of economic transformation. We then situate this perspective in the context of the low-carbon transition. Section 3 outlines the methodology and the theoretical differences between the three models used. Section 4 presents model results, exploring the key aspects of structural change focusing on specific sectors and regions. Section 5 discusses the model-based projections of Section 4, extracting insights for future research. Section 6 concludes the research.

2. Structural change and the low-carbon transition – a framework

2.1. The Great Waves of innovation

Many scholars have noted that economies can be characterised by cycles of varying timeframes. Work on long cycle theory was pioneered by Nikolai Kondratiev and Joseph Schumpeter and focused on the occurrence of economic oscillations of around 40–60 years. Both Kondratiev and Schumpeter attributed this cycle to technological change (Barnett, 1998). This was said to be due to the temporal clustering of innovations and the process of ‘creative destruction’; where new products and technologies displace older systems, methods, and occupations (Schumpeter, 1939).

Freeman and Louçã (2001) identify five technological revolutions or ‘Great Waves’, from the age of the steam engine at the start of the nineteenth century to the more recent IT revolution. Whilst the technologies and industries were different, each technological revolution saw a cycle involving (Perez, 2003) (see Mercure, 2022 for a recent update and detailed analysis):

• The development of a new constellation of connected innovations with revolutionary collective potential

• Product innovation and exponential diffusion of the new constellation, starting the displacement of existing ways of doing things in older incumbent industries

• Process innovation, expansion and consolidation in the new technological regime, completing the displacement of previous ways of doing things

• Saturation, growth decline and possible obsolescence of the new regime as innovation opportunities deplete

In all these technological revolutions, saturation in the mature paradigm coincides with the development of a constellation of sunrise industries, beginning a period of co-existence between the two paradigms. At this time, the opportunities and fortunes of different parts of the economy begin to diverge (Perez, 2003). Process innovation, where entrepreneurs seek ways to achieve the same output at a lower cost, causes unemployment in the sunset industries while accelerated investment and high-profit margins lead to high rates of growth in the sunrise industries. The end of a Great Wave is therefore a turbulent and uneven period and post-industrial decline has been observed and documented in specific sectors and regions towards the end of each of the past five Great Waves (Freeman & Louçã, 2001).

What emerges from the transitions and Great Waves analysis is a product life cycle that corresponds to a sectoral life cycle, and in turn, this generates an economic cycle that is sectorally, and often regionally, significant. Research and development (R&D) investment inversely relates to the age of sectors (Grubb, 2014), going in decreasing order from biotech, artificial intelligence, computers, chemicals, electricity, steel, engines, other metals, agriculture and so on. This matches six successive industrial revolutions in reverse historical order: biotech and artificial intelligence (2010-present), information technology (1970s-2010), petrochemicals and combustion engines (1920–1970s), heavy engineering and electricity (1870s-1910s), the steam engine (1830s-1870s) and textiles (1770s-1830s) (Freeman & Louçã, 2001; Perez, 2003).

A key question is how and where the low-carbon transition fits in this historical perspective. While the above technological revolutions appear to be driven in many cases by the private sector, entrepreneurs and innovation, the low-carbon transition is understood as normatively driven, originating from the need to address climate change (Semieniuk et al., 2021). Yet technological change is often accidental (e.g. Garud & Karnøe, 2003) and the public sector has been influential in developing many technologies (Mazzucato, 2018) and even at least partially driving some past transitions (e.g. the decision of the British military to convert its naval fleet from coal to oil-powered accelerating the pace of the transition towards the internal combustion engine and petrochemicals (Ediger & Bowlus, 2018)). Therefore, it is possible that climate policy is accelerating a transition that might have occurred naturally over time, but not soon enough to mitigate climate change.

Indeed, the transition away from fuel combustion and heat towards electric means of harvesting energy is motivated not only by climate policy, but also by lower wage bills, increased efficiency, and lower operational costs (Way et al., 2022), traded off against higher capital costs (Ondraczek et al., 2015). Moreover, renewable technologies, in particular solar photovoltaics, could be fast approaching a tipping point where their levelized cost is at parity with or below the fossil fuel equivalents (Nijsse et al., 2023). In the Great Waves framing, this could point to a period of co-existence between paradigms. Process innovation, saturation, and overall exhaustion of opportunity characterise the carbon-intensive sectors, while the cleantech sectors are defined by high growth and a jump in productivity over the old paradigm. This raises questions about what we can expect to happen in regions and communities aligned with and dependent on the old carbon-intensive paradigm. Looking at historical transitions can provide some insight.

2.2. Post-industrial decline and the just transition

Deindustrialisation and the associated post-industrial decline have been widely studied, especially in the context of the closure of coal mines, steel, chemical and textile industries in the latter part of the twentieth century in the North of England (Gherhes et al., 2020; Johnstone & Hielscher, 2017; Mah, 2009; Martin et al., 2016; Thorleifsson, 2016), Scotland (Lever, 1991; Walsh et al., 2010), elsewhere in Europe (Walsh et al., 2010) and the Great Lakes region in the US and Canada (Mah, 2009). Deindustrialisation leads not only to loss of income and poverty but also to a loss of identity (Johnstone & Hielscher, 2017; Mah, 2009; Thorleifsson, 2016) as communities typically develop tight-knit cultures around organised industrial work. The loss of those industries has led historically to migration and depopulation, and for those who have not or could not leave, socio-economic decline (Mah, 2009).

Regions affected by deindustrialisation in Europe rank lowest for health and life expectancy in each country (Walsh et al., 2010). The disappearance of once dominant large industries can also leave communities with stymied entrepreneurial capabilities, hindering their recovery in the long run (Gherhes et al., 2020) and reinforcing socio-economic divergence within countries (Gardiner et al., 2013; Martin et al., 2016). Meanwhile, transitions are commonly characterised by strong resistance from incumbent industries and actors (Fouquet, 2016; Geels, 2002; Markard et al., 2020). From the perspective of the actors at the losing end of the transition, the motivating factors are straightforward: loss of affluence, income, employment, purpose, and social status (Baran et al., 2020).

With a working life of around 40 years (Jarvis et al., 2015), it is highly likely that a substantial number of individuals, young and old, will see their skills become obsolete during a rapid transition before the end of their working lives. Furthermore, since industries are typically geographically clustered (Arthur, 1990), this process leads to regional concentrations of similar or related skills becoming obsolete in large numbers (e.g. coal miners), making it challenging for the local economy to absorb into new occupations required for the sunrise sectors and activities. The challenge in transitions is therefore not just about re-training. For workers unwilling or unable to migrate, without substantial opportunities, a skill excess glut can imply long-term involuntary unemployment and subsequently further persisting social dislocation.

In perhaps one of the most infamous examples of post-industrial decline, the UK coal transition of the 1980s transformed the socio-economic landscape of many former coal mining communities, with long-standing impacts. Despite previously being one of the UK's most important strategic industries of the early twentieth century (Mitchell, 1984), by the 1960s, UK coal mines were facing increasing competition from other fuels that offered economic and environmental advantages. Policymakers adopted a policy of ‘controlled rundown’ of the coal mining industry in response (Ashworth & Pegg, 1986). Following the miners’ strike of 1984–1985, major job losses occurred, leading many workers to take early retirement and others to receive incapacity benefits (Foden et al., 2014). Over 30 years on, former coal mining areas still lag far behind the UK average for metrics including employment and education and have rates above the national average for deprivation, state benefit claimants, and chronic health conditions (Beatty et al., 2019). This kind of persisting social dislocation is not unique to the UK's coal transition and can also be witnessed in Detroit and the Great Lakes region following the partial decline of the automobile industry there (Draus et al., 2010).

For the most part, these historical transitions were not supported by policy to account for disparities and the uneven impacts of structural change and therefore created new and exacerbated existing inequalities. In the current low carbon transition, literature has emerged on the need for a ‘just transition’ (Abraham, 2017; Spencer et al., 2018; Weller, 2019). This is a movement that seeks strategies for climate policy to both address climate change and mitigate present and expected new inequalities (Markkanen & Anger-Kraavi, 2019). Originating from the US labour movement of the late 1990s, the just transition concept has evolved to address issues of equity and justice surrounding the mitigation of climate change, gaining considerable attention through the UN’s Sustainable Development Goals (SDGs) and the Paris Agreement (Bang et al., 2022) as well as national climate transition strategies. Initiatives to promote a just transition have included financial support for the unemployed, regional training schemes, and international transition fund schemes (Krawchenko & Gordon, 2021).

This literature addresses some of the issues that have been hindering progress in rolling out ambitious climate policies, typically related to regional communities predominantly employed in high-carbon industries. For example, due to the expected impacts of European climate policy on the coal-producing country of Poland, the latter has lobbied the European Commission considerably and negotiated with other member states for concessions, otherwise threatening to block further climate policies at the European level (Brauers & Oei, 2020). Similar socio-economic challenges may also be at the root of political polarisation around climate policy in some countries (Kurz et al., 2010; Vona, 2019).

In the context of the low carbon transition, most of the literature on deindustrialisation and social dislocation has focused on the decline of the coal industry (Baran et al., 2020; Burke et al., 2019; Svobodova et al., 2021; Weller, 2019; Young et al., 2023). Communities dependent on the oil and gas industries will soon face similar challenges (Carley & Konisky, 2020) and broader concerns exist surrounding the potential extent of spill-over effects to other areas of the economy (Raimi et al., 2022). These challenges must also be considered alongside the socioeconomic impacts and inequalities that climate change is already causing through natural disasters (Rusca et al., 2023; Smiley et al., 2022), biodiversity loss (Pörtner et al., 2023), disease epidemics (Ebi & Hess, 2020), and more.

However, this is not to say that post-industrial decline is an unavoidable consequence of climate policy and a low-carbon transition. The Ruhr region in Germany transitioned from an economy geared around the coal extraction and steel manufacturing industries to a region dominated by a knowledge-based service-oriented economy. Strategic economic diversification managed by the regional and federal governments as well as trade union involvement and comprehensive retraining schemes led to an economic transition that avoided widespread social dislocation (Arora & Schroeder, 2022; Dahlbeck et al., 2022; Galgóczi, 2014).

3. Methods

Section 2 highlights the need to better anticipate the rise and decline of economic sectors in certain geographical areas. We propose a first step in this direction – using macroeconomic models to explore the possible ranges, levels, and directions of structural change in different economic sectors and regions in a net zero transition. This analysis aims to provide insights that go beyond the conventional assessment of economic impacts at the aggregate level to capture structural change and distributional impacts which must be mitigated in a just transition. We begin this section by first providing an overview of how macroeconomic models can offer insights into structural changes before presenting a description of the three macroeconomic models used in this study.

3.1. Modelling low-carbon transition impacts

Obtaining quantitative insights on the sectoral and regional impacts of a net zero transition using macroeconomic models requires a high level of sectoral and regional disaggregation of industrial output, demand for goods and services, employment, and an adequate temporal resolution. Advanced representations of innovation and technological change dynamics disaggregated across the economy are also required. Furthermore, international trade and sectoral interdependencies are crucial as they provide the necessary interconnections linking agents, products, industries, and regions to one another. Unlike many of the IAMs frequently used for the assessment of climate change mitigation scenarios, the models included in this study (E3ME-FTT, GEM-E3-FIT and Imaclim-R) include many of the features that mean they are useful tools to obtain insights into decarbonisation induced structural change effects.

Macroeconomic models also frequently report impacts of climate policy that vary both in terms of the magnitude and even the direction of change. In some models, without accounting for the financial consequences of climate change, stringent climate policy results in a short-run reduction in GDP relative to a counter-factual baseline whereas in other models, the opposite occurs, and climate policy can drive a short-run GDP boost. The main source of this short-run divergence stems from the economic theoretical underpinnings of the models which can be broadly categorised as equilibrium (supply-led) economics and non-equilibrium (demand-led) economics (Mercure et al., 2019). Here, we use models from both paradigms to allow us to consider a wider range of potential impacts and to capture macroeconomic uncertainties.

3.2. Macroeconomic model descriptions

E3ME-FTT is an advanced non-equilibrium macro-econometric model based on a demand-led economic framework. It covers the world in 71 regions with a sectoral disaggregation of 70 sectors of industrial production in the EU and 44 sectors for non-EU regions. Intermediate production in supply chains is represented through a combination of input – output relationships between sectors and bilateral trade relationships between regions. It features both financial and physical energy flows in its econometric representations based on IEA data in 22 fuel user types (Cambridge Econometrics, 2022). FTT (Future Technology Transformation) is a family of evolutionary models of detailed decision-making and S-curve technology diffusion processes. It represents technological decision-making by heterogeneous agents in specific sectors, and the evolution of fleets of technologies from birth to death. Currently, the FTT family covers the power, road transport, residential heating, and steel sectors (Lee et al., 2019; Mercure et al., 2018; Vercoulen et al., 2023).

GEM-E3-FIT is an advanced Computable General Equilibrium (CGE) model, hybrid with bottom-up energy, electricity, and transport modules, ensuring that the economic system in all scenarios remains in general equilibrium. It represents the global economy as a set of 46 interconnected regional and national economies, each composed of 52 production sectors (also connected by input – output and bilateral trade links). Prices determine the interactions between a group of representative firms and a representative household for each regional (or national) economy. Here, prices are also used to minimise production costs and to allow households to maximise their intertemporal welfare whilst conforming to a budget constraint (Capros et al., 2013; Paroussos et al., 2019). The model formulates production technologies in an endogenous manner allowing for a price-driven derivation of all intermediate consumption and the services from capital and labour. In the electricity sector, a bottom-up approach is adopted for the representation of the different power-producing technologies (Fragkos et al., 2021). On the demand side, the model formulates consumer behaviour and distinguishes between durable (equipment) and consumable goods and services and explicitly represents energy-related technologies in buildings and road transport.

Imaclim-R is a multi-sectoral Computable General Equilibrium (CGE) model, hybrid with bottom-up sectoral modules (fossil fuel extraction, electricity, buildings, and transport). It represents the global economy as a set of 12 interconnected regional and national economies, each composed of 19 production sectors (also connected by input – output and trade links). It features consistent input – output accounting of both economic and physical energy flows. The model simulates dynamic trajectories in yearly steps through the recursive and hard-linked succession of static macroeconomic equilibria and bottom-up sectoral modules. It explicitly represents the constraints affecting technical flexibilities and their interplay with macroeconomic trajectories by describing economic patterns in a world with market imperfections, partial use of production factors (labour and capital) and imperfect expectations for investment decisions. Within macroeconomic equilibria, a representative household in each regional economy maximises its utility under both economic and time budget constraints. Productive sectors supply for demand under short-run technical and productive capacity constraints. Between two economic equilibria, bottom-up modules simulate technical adjustments to demand and price changes under imperfect foresight with explicit technologies for the electricity and transport sectors. In the electricity sector, a bottom-up approach is adopted for the representation of the different power-producing technologies. For the demand side, the model represents energy-related technologies in buildings and road transport.

Key model characteristics and differences in theory and assumptions are summarised in Table 1. E3ME-FTT is a demand-driven, non-equilibrium model that assumes that both labour and capital are not fully utilised, whereas GEM-E3-FIT and Imaclim-R are supply-driven, general equilibrium models that assume capital is fully utilised in the baseline scenario. In addition, GEM-E3-FIT and Imaclim-R, in their standard macroeconomic closure¹, assume that investment is exclusively financed by savings and there are no savings unused. Therefore, decarbonisation that requires larger investment volumes will lead to the ‘crowding-out’ of investment in other productive sectors. Meanwhile, E3ME-FTT assumes that financing for projects is not constrained by the financing of other projects elsewhere in the economy, and therefore that investment is not directly constrained by the saving behaviour of agents. Involuntary unemployment is recognised in all three models. In each of these models, climate impacts are not captured. For further comparison of equilibrium and non-equilibrium modelling paradigms, see Mercure et al. (2019). For further comparison of the E3ME-FTT, GEM-E3-FIT, and Imaclim-R models, see Lefèvre et al. (2022).

TABLE 1. Overview of the key macroeconomic model similarities and differences.

	E3ME-FTT	GEM-E3-FIT	Imaclim-R
Model type	Macroeconometric (E3ME), hybrid with bottom-up evolutionary modules (FTT)	CGE model, hybrid with bottom-up modules	CGE model, hybrid with bottom-up modules
Macroeconomic theory branch	Non-equilibrium (Demand driven)	Equilibrium (Supply driven)	Equilibrium (Supply driven)
Technological change	Endogenous (power, road transport, domestic heat, and steel sectors)	Endogenous (power, road transport, heating)	Endogenous (power, road transport) with high inertia
Labour market representation	Imperfect and flexible labour markets	Imperfect and flexible labour markets, 5 skill levels are represented	Imperfect labour markets with limited flexibility
Investment & Finance	No crowding out, no fixed limit on finance	Crowding-out, finance is constrained	Crowding-out, finance is constrained
Sector coverage	44 (70 in the EU)	52	19
Regional coverage	71 countries and regions	46 countries and regions	12 countries and regions

3.3. Study design

To explore possible ranges, levels, and directions of structural change impacts in a low-carbon transition, we develop baseline and global net zero emission scenarios for the macroeconomic models mentioned above. The baseline only includes currently implemented energy and climate policies without intensification in the future (see Lefèvre et al. (2022) for more detail on structural change in a baseline scenario). Net zero scenarios show the impact of ambitious climate and energy policies on economic structural change up to 2050.

The scenarios are compliant with an end of century carbon budget of between 600–700 GtCO₂ and reach global net zero emissions between 2060–2065. We consider this to be in line with the Paris Agreement target to limit warming to well below 2°C, with similar budgets featured in such scenarios in IPCC’s AR6 Working Group III (IPCC, 2022a). In all three models, the net zero scenario includes a combination of market-based (e.g. carbon pricing) and regulatory (e.g. clean technology support) policies. Table S1 (supplementary material) shows the policy packages in each of the net zero scenarios. The amount of cumulative global negative emissions is limited to below 50 GtCO₂ in 2050. Between 2020–2050, the average annual global carbon intensity reductions in the net zero scenarios are between 5.8 and 7.5 percent. All scenarios include the short-term socio-economic impacts of the COVID-19 pandemic. We develop a consistent aggregation of economic sectors (described in Table S2 – supplementary material) and include results for major fossil fuel exporter and importer economies.

4. Results

We begin this section by describing and contrasting the trajectories the three models describe at the economy-wide level in a net zero scenario. In the second part of this section, we show the impact of a net zero transition on disaggregated metrics including sectoral output and employment.

4.1. Regional socio-economic impacts of a net zero transition

At the aggregate level, we find considerable differences in the economic output projections between the models that illustrate alternative possible storylines of the net zero transition. These differences are linked to different visions and economic theories embedded in the models. We note three key differences: (i) the flexibility of the economy, socio-technical, labour, and energy systems in the short to medium run, (ii) the availability of capital for low carbon investments and the resulting stimulus effect in sunrise sectors, and (iii) alternative visions and databases for modelling fossil fuel markets and the distribution of declining production between countries.

4.1.1. Aggregate level differences between regions and models

While a net zero transition will inevitably generate different impacts in different parts of the world, we identify four broad country groups (Table 2) according to (i) the degree to which the models report similar outcomes and (ii) the aggregate GDP impacts of a net zero transition presented in Figure 1. Group 1 represents regions where the model results are similar, with all models projecting some of the most significant negative impacts on GDP in a net zero transition. Countries in Group 1 are major fossil fuel exporter economies with a high carbon intensity (Russia, Saudi Arabia). In Group 2, GDP impacts are less severe. The trajectories described by the models are still fairly similar, where negative GDP impacts are still mostly reported. Like the regions in Group 1, Group 2 consists of fossil fuel exporters, though these countries have a more moderate carbon intensity and are more economically diversified than those in Group 1.

Figure 1. Discounted cumulative GDP differences from baseline (3% discount rate) in 2050 for each study region in E3ME-FTT, GEM-E3-FIT, and Imaclim-R.

TABLE 2. Description of four common country groups derived from economic carbon intensity, fossil fuel exporter/importer status, and the size of the aggregate GDP changes in a net zero transition.

Group	Degree of model similarity	GDP impacts	Countries	Fossil fuel Importer or exporter	Economic Carbon intensity
1	High	Largest negative impacts	Russia, Saudi Arabia	Exporter	High
2	High/medium	Small – moderate positive or negative impacts	Brazil, Canada, US	Exporter	Medium
3	Low/medium	Smallest positive or negative impacts	EU, Japan	Importer	Low
4	Low	Largest positive or negative impacts	China, India	Importer	High

The transition trajectories described by the models begin to diverge across models in Group 3. E3ME-FTT, the simulation-based non-equilibrium model, describes positive GDP impacts while the equilibrium models continue to project net costs. Overall, GDP impacts (positive or negative) are the most limited in Group 3 which describes developed Global North fossil fuel importing regions with a low carbon intensity (EU, Japan). Group 4 exhibits the lowest level of model result similarity with strong positive GDP impacts in E3ME-FTT in contrast to notable losses in the other models. This group represents emerging economies that are fossil fuel importers with a higher carbon intensity (China, India).

Altogether, as Figure 1 shows, and in line with other models featured in IPCC AR6, a net zero scenario in each of the models generates mostly small to moderate impacts on aggregate GDP (IPCC, 2022a). The models largely agree that fossil fuel exporter economies will witness a negative GDP impact in a net zero transition. This is mainly driven by the loss of exports due to lower global demand for fossil fuels. As a result, countries in Group 1 fare the worst as they are more dependent on fossil fuel rents than the exporters in Group 2, who have more diversified exports. Furthermore, the models also mostly agree that impacts, whether positive or negative, are particularly limited in economies with lower carbon intensities (Groups 2 and 3). The picture is more nuanced in fossil fuel importer economies, where E3ME-FTT projects that a net zero transition has the potential to increase GDP.

4.1.2. Drivers of aggregate level model result difference

The difference in the trajectory of aggregate GDP between E3ME-FTT and the two other models described in Section 4.1.1 is largely due to the absence of crowding-out of low-carbon investments on other investments. There is no limit on the availability of the money supply or finance in E3ME-FTT. The model also has a flexible supply that can more readily adapt to the demand and evolution of technical systems and the capital stock than the other models.

Furthermore, a stimulus effect is generated from the investment required to develop low-carbon technologies and sectors. Fossil fuel importer countries are key winners of the transition in E3ME-FTT as decarbonisation facilitates improved trade balances. This effect benefits most countries worldwide since fossil fuel production is concentrated in a limited number of countries. The reverse effect occurs in fossil fuel exporting countries in E3ME-FTT, where the loss of activity amplifies through drastic losses of investment, with scrapped capital not assumed to be recycled towards other activities. In addition, E3ME-FTT adopts a representation of fossil fuel markets based on a detailed extraction costs dataset (Mercure et al., 2021b) that penalises the high-cost producers (USA, Canada) in a more pronounced way than the other models.

While the largest differences between models can be explained by the contrasting macro-innovation theory representation, the two equilibrium models also often project GDP impacts of differing magnitude. Both GEM-E3-FIT and Imaclim-R project negative GDP impacts in all regions in a net zero scenario, but such impacts are almost always more negative in Imaclim-R. This is due to the higher flexibility of technical systems as embedded in the CES (constant elasticity of substitution) production functions and in the bottom-up modules to represent energy and transport systems in GEM-E3-FIT (Fragkos & Fragkiadakis, 2022). This means that the energy system in Imaclim-R is characterised by greater inertia. Additionally, a more flexible labour market in GEM-E3-FIT limits short to medium-run transition costs.

4.2. Structural change in a net zero transition

Whilst the economic impacts of a net zero transition might well appear relatively moderate for most countries at the aggregate level, we find substantial sectoral variation regarding the direction and magnitude of change in both output and employment. This reflects the potential for climate policy to induce large-scale structural change, with uneven impacts between sectors and regions (Lefèvre et al., 2022). The sectoral impact of a net zero transition on output is shown in Figure 2, expressed as the percentage difference from baseline. Despite the different theoretical underpinnings of the models, a common narrative of structural change in a net zero transition emerges, where the models report similar sectoral trajectories.

Figure 2. Contributions of a range of sectors to the relative aggregate output difference from baseline in a net zero scenario for E3ME-FTT, GEM-E3-FIT, and Imaclim-R.

4.2.1. A common storyline of structural change

First, in line with the aim of the net zero scenarios, the fossil fuel sector’s share of total GDP falls rapidly in every region. This contraction in contribution to overall GDP is particularly considerable in the fossil fuel exporter economies in Group 1 that are currently less well economically diversified. Figure 3 also shows that employment in the fossil fuel sector falls in every region in all models in the net zero scenarios, consistent with the sectoral output. Again, the most substantial losses are projected in large fossil fuel-producing regions, where a greater portion of the total workforce is currently employed in the fossil fuel sector. Furthermore, the fossil fuel trade balance (exports-imports) generally worsens in fossil fuel exporter countries as their hydrocarbon exports decrease (Figure S1 – supplementary material). Simultaneously, the fossil fuel trade balance generally improves in importer countries, as decreased demand for fossil fuels reduces their import bill.

Figure 3. Contributions of a range of sectors to the relative aggregate employment difference from baseline in a net zero scenario for E3ME-FTT, GEM-E3-FIT, and Imaclim-R.

Meanwhile, the electricity sector witnesses an increase (albeit limited) in its share of total GDP in almost all regions as electrification of the road transport and heating sectors accelerates as these sectors decarbonise. This is also reflected in an increase in electricity sector employment in all regions. The energy-intensive sector's share of total GDP also increases across most regions. In E3ME-FTT and GEM-E3-FIT, this is largely driven by an increased demand for critical materials on account of a greater share of solar photovoltaics in electricity systems.

4.2.2. Net zero transition impacts beyond the fossil fuel sector

A key feature in the process of decline in high-carbon industries is the impact on the intermediate production of goods and services that occurs across those value chains, which goes far beyond the fossil fuel sector itself. This includes output losses in manufacturing and construction for projects that do not happen (relative to a counterfactual baseline scenario). This also includes vehicles and machinery that are never produced, fuel that is never distributed, steel for pipelines that is never ordered, and so on. Therefore, downward pressure in many sectors develops, particularly amongst fossil fuel exporters due to the more significant loss of activity in the fossil fuel sector. At the same time, rapid growth occurs in the renewable energy sectors driven by net zero policy which also has knock-on effects in other areas of the economy though these are less pronounced in GEM-E3-FIT and Imaclim-R due to the crowing-out of investment assumed in CGE models.

Figure 2 shows that in many countries a substantial part of this contraction of GDP stems from the services sectors, which provide a rough barometer of the general economic situation. This is because the services account for the largest share of GDP globally and they generally become the end recipients of income generated in most other sectors as salaried workers spend their wages (e.g. on food, clothing, etc.). In fossil fuel importer economies, however, the models show different trajectories for the services sector, due to the different ways they portray the creation of (or competition for) financial resources for the investment required for decarbonisation. For fossil fuel importer countries in Groups 3 and 4, E3ME-FTT generally projects increased service sector output and employment in a net zero scenario whilst GEM-E3-FIT and Imaclim-R typically project a somewhat limited contraction.

This is, again, a direct consequence of the assumptions around economic equilibrium and crowding-out effects, where a stimulus effect from building activity related to low-carbon technologies in E3ME-FTT drives an increase in demand. Meanwhile for GEM-E3-FIT and Imaclim-R, the cost of the transition reduces the demand for services as financial resources are limited and additional investment requirements for decarbonisation puts stress on the capital markets (‘crowding-out’ effect). This cancels investment in other productive sectors and increases the price of capital in the entire economy (Fragkos & Paroussos, 2018). These changes in output are directly reflected in employment projections of the models (see Figure 3).

There are similar contrasting outcomes in the construction sector, where in E3ME-FTT, output and employment increase relative to a baseline scenario. This is due to climate policy that induces large-scale investment in low-carbon capital in the form of renewable infrastructure. In GEM-E3-FIT and Imaclim-R, the impact of a net zero transition on the construction sector is slightly negative for most regions, as the negative impacts of GDP losses counterbalance the increased demand for construction to build the new low-carbon infrastructure (e.g. renewable energy plants, enhanced renovation of buildings, recharging infrastructure etc.).

As indicated in Section 4.1.1, the model results are the most different for China and India, where E3ME-FTT projects output and employment gains across several sectors while the two equilibrium models project almost an economy-wide decline relative to a baseline scenario. While the models agree that China and India can both improve their trade balances in a net zero transition by reducing their imports of fossil fuels, the direction and magnitude of change in other sectors are less clear. For instance, the services sector in E3ME-FTT grows both in terms of output and employment whilst the opposite occurs in GEM-E3-FIT and Imaclim-R. Once again, this is due to decreased economic growth driven by climate policy under the strict closure rule of General Equilibrium models.

5. Discussion

Using three sectorally and regionally disaggregated macroeconomic models of different theoretical foundations, we show that the transition to a low-carbon economy will generate substantial uneven impacts on economic output (Figure 2), employment (Figure 3), and trade (Figure S1) both within and between economies. The results described in Section 4 show that, in a net zero transition, some sectoral trends are more certain than others. Most notably, large-scale unemployment will emerge in the fossil fuel sector, especially in fossil fuel exporting regions. Employment losses could be witnessed in other areas of the economy, through supply chain linkages with the sunset sectors.

There is still significant uncertainty regarding the degree of losses in the service sectors in particular, as the three models present often contrasting directions of change due to their different theoretical positions concerning the creation of investment capital and their degree of optimism regarding the flexibility of sociotechnical systems. It is not possible at this stage to claim with any degree of certainty which trajectory the future will be closer to. In fact, since investment capital is neither limitless (as in E3ME-FTT) nor fully crowded-out (as in GEM-E3-FIT and Imaclim-R), the future trajectory of the economy could lie between the boundaries delimited by these models (Pollitt & Mercure, 2018). It is also important to highlight that because these models do not consider the physical impacts of climate change and their economic consequences, further inaction on climate change will lead to additional socioeconomic impacts and inequalities that are not represented here. Therefore, delaying or avoiding all policy to support the transition on the grounds of justice is ill considered.

5.1. Limits of aggregate metrics

An important insight of this study is that structural change is much larger in magnitude than the aggregate impacts of a net zero transition. For an ordinary household, emerging economic pressures will depend starkly upon which sector(s) and region its working family members live and work in, in contrast with any definition of a representative agent (or average person) as commonly used in economic and integrated assessment modelling. The decarbonisation impacts on financial, economic, social, health and standard of living metrics for certain households could be extreme, even if ‘average’ or ‘median’ impacts across households are moderate or zero. Additionally, these disparities could be particularly important in post-growth and degrowth scenarios given that these scenarios not only project more significant aggregate GDP impacts (due in part to less ambitious requirements for technological change) (Keyßer & Lenzen, 2021; Nieto et al., 2020), but also require a differentiated down – or upscaling of different sectors depending on ecological and social importance (Hickel et al., 2021).

The frequently used aggregate metric of ‘consumption loss’ or ‘GDP loss’ as used in standard policy assessments could therefore be misleading, as it could be interpreted as the amount of consumption or income lost by all households in a region because of climate policy. In contrast, we show here that moderate aggregate consumption loss can mean important benefits for some households, simultaneous to the end of entire livelihoods and communities for others (e.g. employees in regions that rely significantly on the fossil fuel industry), even within the same regions. Those differences in turn can fuel societal inequality, socioeconomic division, and political polarisation.

Regardless of whether stringent climate policies adequate to achieve ambitious emission reduction targets are developed and implemented, technology is constantly changing in a transition towards higher energy efficiency, renewable energy sources, and increasing electrification in the present economic trajectory (Nijsse et al., 2023; Way et al., 2022). With ambitious climate policies, where the aim of a just transition is ignored, the accelerated rate of structural change could destabilise many parts of the economy from the incumbent supply chains, labour markets, and financial markets (Semieniuk et al., 2021). Destabilisation could also be significant in alternative pathways looking at low or degrowth transitions (Hickel et al., 2021) or in scenarios with limited mitigation and strong climate change impacts across the economy. This rapid evolution of the socio-technical regime must be recognised as it could become the dominant source of socio-economic transformation for the next few decades, driving key socio-economic indicators and even political processes.

5.2. Post-industrial decline in a net zero transition

Focusing only on the aggregate impacts of climate policy could risk contributing to a transition that is unjust and is in danger of creating stranded capital, labour, skills (Mercure et al., 2021c), and even entire regions (Spencer et al., 2018) that become precluded from many of the opportunities presented by the sunrise sectors. As highlighted in Section 2.2, similar industry tends to cluster due to an agglomeration effect (Arthur, 1990). This means that whole cities or even countries may have formed and grown around certain industrial activities. For example, Fort McMurray, an urban service area in Northern Canada, might not have existed without the development of the tar sands industry. Similarly, Rotterdam could have been a less prominent city without the development of its petrochemical industry, which powers mobility and many other industrial activities upstream of the Rhine River in Germany and elsewhere.

This concentration of industry will lead to regionally concentrated output and employment losses in a net zero transition. In such regions, output losses are unlikely to be confined to the high-carbon industries and they will instead likely impact the service sectors too (Raimi et al., 2022) as households have less disposable income to spend on food, retail, and other goods. Just as regions aligned with the sunset technological paradigms in previous transitions were (Freeman & Louçã, 2001), regions currently highly dependent on carbon-intensive activities could be at serious risk of post-industrial decline in a net zero transition. Left unmitigated, this could mean such areas become left behind by the transition, potentially unable to take advantage of the economic opportunities offered by the sunrise cleantech industries.

5.3. Implications for climate policy and future work

Addressing post-industrial decline after the fact is challenging, as evidenced by the persistence of poor socioeconomic outcomes in many former industrial areas (Beatty et al., 2019; Draus et al., 2010). Where programmes have been successful, they have been at high cost (Arora & Schroeder, 2022). While lessons can be learned from previous examples of economic transitions, adequate policy frameworks should be highly region and sector-dependent (Dahlbeck et al., 2022). Vocational training and relocation schemes for displaced workers in sunset industries and related supply chains are likely to be important tools to address disparities. Such schemes can be (at least partly) funded by increasing carbon tax revenues.

However, redistributive policy alone is unlikely to be adequate to deal with post-industrial decline, especially considering that fossil fuel exporter economies in particular will be faced with a reduced fiscal space (as shown by Figure S1 – supplementary material) from which to fund transition support initiatives due to losses of exports. Instead, transformative economic policy could be more effective to enable former fossil fuel dependent regions to capture the opportunities of the transition. This could include economic diversification schemes that utilise existing regional economic capabilities to develop new export opportunities (Mealy & Teytelboym, 2022).

International mechanisms can play a role too, as demonstrated by the recent Just Energy Transition Partnerships, where the European Union and other countries in the Global North financially support the acceleration of energy system decarbonisation in the Global South. The partnerships explicitly acknowledge calls for a just transition, placing focus on supporting affected communities in this transition (European Commission, 2021). Such partnerships are currently supporting transition efforts in Indonesia, Senegal, South Africa, and Vietnam (Simpson et al., 2023).

Research to further investigate the regional disparities both between and within countries is also needed to help inform future discussions on structural change induced by ambitious climate policy. Coupled with this, advanced modelling tools that offer a high sectoral and regional disaggregation of outputs embedded in economy-wide frameworks with endogenous trade impacts are required to identify risks and opportunities for national and subnational regions and sectors and to support effective decision-making and design of efficient and just industrial, labour, climate, and economic policy. Finally, given the call to explore alternative pathways with different growth assumptions, e.g. degrowth and post-growth scenarios (Li et al., 2023), the same methodology could be applied in future research to investigate potential levels of structural change and disparities in such disruptive development pathways (Lefèvre, 2023).

6. Conclusion

Analysis of the economic impacts of climate policy in a sustainability transition must evolve to examine both regional and sectoral disparities caused by the transformation away from fossil fuels and related carbon – and energy-intensive industries. This paper aims to kickstart an investigation of structural change in a low-carbon transition by comparing the disaggregated outputs of three leading sectorally disaggregated macroeconomic models of contrasting economic schools of thinking. We find that structural change impacts, when measured on a sectoral and regional basis, are much stronger and more drastic (either positive or negative) than the impacts observed at the aggregate economy-wide level.

Given that aggregate level impacts are typically moderate, this suggests that socio-economic impact assessments of decarbonisation could usefully move their focus away from aggregate metrics such as GDP or consumption and towards detailed sectoral and regional analyses. However, assessment outcomes, particularly the impact a net zero transition will have on the services sectors, also vary depending on which model is used under which theoretical framework. This suggests that attention must be given to better identify the sources of investment and growth available to economic sectors and geographical regions. We call for the development of modelling tools and analytical practices that can explore, with ever greater detail, the regional and sectoral impacts and societal transformations induced by climate policy and innovation.

CRediT authorship contribution statement

Cormac Lynch: Conceptualization, Methodology, Software, Formal analysis, Writing – original draft.

Yeliz Simsek: Conceptualization, Methodology, Software, Formal analysis, Writing – original draft.

Jean-Francois Mercure: Conceptualization, Methodology, Software, Formal analysis, Writing – original draft, supervision, project administration, funding acquisition.

Panagiotis Fragkos: Conceptualization, Methodology, Writing – review & editing, project administration, funding acquisition.

Julien Lefevre: Conceptualization, Methodology, Writing – review & editing, project administration, funding acquisition.

Thomas Le Gallic: Conceptualization, Methodology, Software, Writing – review & editing.

Kostas Fragkiadakis: Conceptualization, Methodology.

Leonidas Paroussos: Conceptualization, Methodology.

Dimitris Fragkiadakis: Conceptualization, Methodology.

Florian Leblanc: Conceptualization, Methodology, Software, Writing – review & editing.

Femke Nijsse: Conceptualization, Methodology, Writing – review & editing.

Disclosure statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Additional information

Funding

The research leading to these results was supported by funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 821124 (NAVIGATE) and under grant agreement No 101022622 (ECEMF).

Notes

1 GEM-E3-FIT can adopt a financial mechanism that relaxes this constraint. See Paroussos et al. (2019) for more detail.