A window of (missed) opportunity? A comprehensive stocktake and energy-system impact assessment of global COVID-19 recovery packages

Original research article A window of (missed) opportunity? A comprehensive stocktake and energy-system impact assessment of global COVID-19 recovery packages Eleftheria Zisarou a , Panagiotis Fragkos a , Dirk-Jan Van De Ven b , Shivika Mittal c,d , Natasha Frilingou e , Cl` audia Rod´ es-Bachs b , Stefanos Tsotras f , Angelos Potiriadis f , Georgios Xexakis f , Konstantinos Koasidis e , Haris Doukas e , Adam Hawkes d , Alexandros Nikas e,* a E3Modelling S.A., Panormou 70-72, 115 23 Athens, Greece b Basque Centre for Climate Change, Edificio Sede 1-1, Parque Científico de UPV/EHU, 48940 Leioa, Spain c Center for International Climate and Environmental Research—Oslo (CICERO), PO Box 1120 Box 1129, Blindern, 0318 Oslo, Norway d Department of Chemical Engineering, Imperial College London, London SW7 2AZ, UK e Energy Policy Unit, School of Electrical and Computer Engineering, National Technical University of Athens, Iroon Politechniou 9, 157 80 Athens, Greece f HOLISTIC P.C., Mesogeion Avenue 507, 153 43 Athens, Greece ARTICLE INFO Keywords: COVID-19 Green recovery packages Energy transition Integrated assessment models Energy affordability ABSTRACT COVID-19 reshaped global economic priorities at the time, with recovery packages offering unprecedented fiscal stimuli aimed at revitalising economies from the impacts of the pandemic and associated policy responses—including lockdowns. While these packages were quickly framed as an opportunity for aligning socioeconomic recovery spending with nearand longer-term climate goals, early assessments of their decarbonisation footprint were constrained by and/or oriented towards optimistic interpretations of the limited information available then. We examine the potential of recovery packages to bridge the mediumand long-term climate ambition gap towards meeting the Paris Agreement goals. To enable such a comprehensive assessment, we first develop an open-access database of global green recovery measures. Second, we explicitly translate these measures as inputs into three Integrated Assessment Models, to assess their implications for energy systems, emissions, and technology development globally. Third, we quantify the missed opportunity in global recovery spending in terms of accelerating the clean energy transition, by exploring a theoretical reallocation of funding from energy affordability measures towards green technologies. Our results suggest that, while the actual synthesis of global recovery funding may not be sufficient to boost climate efforts towards meeting the Paris climate goals with sustained effects post-2030, redirecting part of the funds to low-carbon technologies could accelerate decarbonisation and electrification trends in some sectors. Whether or not the global COVID-19 recovery portfolio is adjusted to better support transition goals, recovery funds alone cannot guarantee a comprehensive and effective transition; this requires complementary systemic reforms, targeted sectoral strategies, and international collaboration. 1. Introduction Soon after COVID-19 shut the world down, nations began announcing economic recovery plans, often including measures for green stimulus—among others. Energyand climate-economy modelling science was quick to respond, by attempting to quantify the environmental and climate impacts of the responses to the pandemic and/or to assess the explicit or hidden potential of recovery efforts to ‘build back better’. Initial analyses provided valuable insights, for example, into emission reductions during lockdowns [1], the potential of promoting * Corresponding author. E-mail addresses: [email protected] (E. Zisarou), [email protected] (P. Fragkos), [email protected] (D.-J. Van De Ven), [email protected] (S. Mittal), [email protected] (N. Frilingou), [email protected] (C. Rod´ es-Bachs), [email protected] (S. Tsotras), [email protected] (A. Potiriadis), [email protected] (G. Xexakis), [email protected] (K. Koasidis), [email protected] (H. Doukas), [email protected] (A. Hawkes), [email protected] (A. Nikas). Contents lists available at ScienceDirect Energy Research & Social Science journal homepage: www.elsevier.com/locate/erss https://doi.org/10.1016/j.erss.2025.104216 Received 30 April 2025; Accepted 4 July 2025 Energy Research & Social Science 127 (2025) 104216 Available online 11 July 2025 2214-6296/© 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/bync-nd/4.0/ ). decarbonisation pathways by means of the anticipated green investments [2], and the role of structural reforms in ensuring resilient recovery at different scales [3]. These early modelling exercises, however, were anchored onto, constrained by, or even fixated on relatively optimistic interpretations of, the limited preliminary information about green recovery packages available at the time. They often assumed unrealistically high levels of clean energy spending [4], overestimated the persistence of temporary behavioural changes such as reduced commuting and travel [5], and/or focused predominantly on large economies that were quick to act [6], such as the European Union (EU) and the United States (USA). Additionally, such studies tended to narrow their scope to a limited set of technologies or sectors, such as renewables or electricity, without comprehensively addressing cross-sectoral impacts or systemic interdependencies [7]. Moreover, they failed to fully reflect the heterogeneity of the recovery plans—even in the EU, despite the high-level orientation of the bloc's Recovery and Resilience Facility (RRF) towards climate action, national implementation strategies varied considerably, with some member states seemingly prioritising infrastructure investments and others emphasising industrial decarbonisation or digitalisation. Finally, early modelling assessments failed to capture the rapid evolution of the post-pandemic economic landscape, in which the inflationary pressures and subsequent energy crisis reshaped policy priorities, with rising energy costs—exacerbated by geopolitical tensions—shifted immediate attention towards energy security and affordability, occasionally delaying or altering the focus on longer-term decarbonisation goals. As a result, while insightful, early findings of these COVID-oriented modelling studies lacked the depth and breadth needed to meaningfully inform the eventual design of recovery strategies or to align them with longer-term climate goals [8]. Our study seeks to address these limitations, in three axes and steps. First, a notable limitation of previous analyses has been the absence of a comprehensive, disaggregated collection of real-world implemented COVID-19 recovery policies with explicit energy-system and emissions implications. Most of the relevant databases available [9–12] track fiscal measures related to recovery efforts but exhibit limitations constraining their applicability for advancing the understanding of green recovery impacts on the low-carbon transition. For example, some do not sufficiently disaggregate fiscal measures directed towards specific green technologies [13], others lack global representation focusing on highincome economies [9,14]—thereby neglecting the potential hidden in diverse recovery spending in developing regions—while others did not expand their focus onto critical social measures, such as energy affordability initiatives and regulatory reforms despite their significant role in shaping energy development pathways. Importantly, few databases [12] are structured explicitly to facilitate integration of recovery packages with energyand climate-economic models, which however is a prerequisite to assessing the complex interactions between recovery funding, technology adoption, and emissions pathways under various policy scenarios, especially in the medium and long run. In this direction, our first contribution is the development of a detailed global database of policy measures and recovery packages, disaggregated by technologies and sectors, in a plug-and-play format tailored to Integrated Assessment Model (IAM) frameworks. This open-access database, available via the data exchange platform IAM PARIS [15], aims to not only serve the purposes of our study but also facilitate future research in this area. Second, after conducting this comprehensive stocktake, we instigate an energy-system impact assessment of COVID-19 recovery packages at the global level (but with regional granularity), drawing on three different IAMs to evaluate the mediumand longer-term implications of all announced green recovery measures worldwide. These efforts are analysed in the context of emissions trajectories and scenarios set by official climate targets and pledges, including the current policies, the latest available versions of Nationally Determined Contributions (NDCs), and the Long-Term Targets (LTTs). Unlike previous studies, our approach captures the full spectrum of announced recovery policies and investment, accounting for regional diversity and sectoral complexity, to explore the interplay between recovery spending and technological innovation on a global scale. Third, we additionally assess the untapped potential of a differentiated recovery portfolio at the global level, by shuffling resources within our database, enabling us to reallocate funds from energy affordability measures [16] towards transformative low-carbon technologies driving the energy transition [17]. We thus attempt to quantify the missed opportunity of post-pandemic stimulus responses to actively accelerate systemic change and bridge the gap between current policy efforts and long-term climate ambition. 2. Methods 2.1. Development of a green technology-focused COVID-19 recovery database The COVID-19 pandemic reshaped global economic priorities at the time, with recovery packages offering unprecedented fiscal stimuli aimed at revitalising economies negatively impacted by the pandemic and the confinements forced in response. While these packages may have also presented, and/or been framed as, a significant opportunity for aligning socioeconomic recovery spending with nearand longerterm climate goals, their composition—in terms of the share of spending with direct or indirect implications for energy systems and emissions trajectories—varies widely across countries, sectors, and technologies. To enable an informed, realistic, and relevant model-based assessment of the energy-system and emissions implications of recovery packages, we first put together a comprehensive database comprising the latest available information on fiscal recovery measures relevant to climate action by technology, country, and sector across 105 countries (see Appendix) with such measures announced and/or implemented. The database groups all announced investments in green technologies, household spending, and measures not fitting neatly into predefined categories/sectors yet constituting important chunks of recovery packages related to climate transition (e.g., on critical minerals, methane abatement, recycling, and electricity access). Our methodology included identifying and synthesising data from a diversity of reputable sources [9,11,12,15,18,19], which were thoroughly examined to ensure a broad, representative dataset of green recovery spending around the world. To overcome challenges related to data inconsistencies and gaps, several techniques were employed to extract information and clean data (see Appendix). In particular, while most databases used in the process were available online and free to download (e.g., Global Recovery Observatory [9], Energy Policy Tracker [11], and OECD [19]), the data extraction process involved the use of scrapping methods to gather comprehensive information and data from the International Energy Agency's (IEA) [20] and the Green Recovery Tracker [21]. Despite the comprehensive nature of the databases, we encountered and had to overcome certain challenges in aggregating and classifying the data, including format variations across countries and categories (see Appendix). Also, despite intensive efforts to parse all available information, we still may have missed certain energy-related measures and investments [19,20]. This is because some governments do not sufficiently report on stimulus packages, some recovery strategies have adopted an industry-agnostic approach, while benefits from tax breaks may have been difficult to pin down, quantify, and thus include in the database. The final dataset includes recovery packages related to the energy sector across 105 countries, accounting for 86 % of global population and 94 % of global GDP. This compilation (after removing duplicates) comprises a total of 2109 unique identified policies/measures amounting to USD 2.472 trillion. These measures have been designed for disbursement over a period of multiple years, with some extending implementation planning as far as to 2032. These support measures E. Zisarou et al. Energy Research & Social Science 127 (2025) 104216 2 were categorised into 51 sectors and grouped into 8 broader categories (low-carbon electricity, electricity networks, low-carbon and efficient transport, energy-efficient buildings and industry, fuel and technology innovation, energy affordability, people-centred transition, general/ unclassified energy measures, and other/not explicitly energy-related) within our database (see Appendix), following the IEA's classification system to enhance clarity and usability. This system consists of a broad array of sectors and technologies, including renewable energy sources, low-carbon transportation, and clean energy infrastructure, and was chosen as a highly established framework from a trusted institution that aligns with the focus of our study. This also facilitates the integration of the green recovery packages into the IAMs of our study (see Section 2.2) to assess their mediumand long-term impacts on emissions, technology uptake, and energy system development. The global green recovery database follows the format shown in Table 1. ‘Energy Affordability’ measures constitute the lion's share of total funding (35 %, or USD 860 billion). This reflects a key concern in developed and developing countries, to constrain the impacts of the increasing energy prices for consumers, both during and right after COVID-related confinements and during the energy supply crisis that followed (policy/market responses to the Russia-Ukraine conflict). But it also reflects a different concern in developing nations, where maximising and maintaining energy access and affordability for all is a critical aspect of societal equity and economic growth. This category is followed by ‘Low-carbon and Efficient Transport’ (23 %) and ‘Energyefficient buildings & industry’ (16 %), reflecting the importance of enduse sectors for cutting CO 2 emissions in the shortand long-term. Although emphasis on energy affordability and efficiency shows the prioritisation of alleviating societal challenges in the short run (while also enabling long-term sustainable infrastructure investments), ‘Lowcarbon electricity’ has received only about 10 % of total spending, despite being the number one priority in terms of quickly enabling energy transitions towards aligning with climate goals in the long run (Fig. 1). A more comprehensive analysis of the green-technology focused database can be found in the Appendix. 2.2. Model ensemble The study uses three well-established IAMs (GCAM, PROMETHEUS, and TIAM) that have been widely used to assess the environmental, economic, and technology impacts of ambitious climate policies at national and global levels. A brief description of the three models is provided below, while detailed model descriptions can also be found in IAM PARIS [22]. The Global Change Analysis Model (GCAM) [23] is a global IAM that represents human and Earth system dynamics. It explores the behaviour and interactions between the energy system, agriculture and land use, the economy, and climate. The role of GCAM is to bring multiple human and physical Earth systems together to provide scientific insights that would not be available from the exploration of individual scientific research lines. It reads in external “scenario assumptions” about key drivers (e.g., population, economic activity, technology, and policies) and assesses the implications of these assumptions on key scientific or decision-relevant outcomes (e.g., commodity prices, energy use, land use, water use, emissions, and concentrations). PROMETHEUS [24] is a global energy system model covering in detail the complex interactions between energy demand, supply and energy prices at the regional and global level. It aims to assess climate change mitigation pathways and low-emission development strategies and to analyse the energy-system, economic, and emissions implications of a wide spectrum of energy and climate policy measures, differentiated by region and sector. The model provides detailed projections of energy demand, supply, power generation mix, energy-related carbon emissions, energy prices, and investment to the future covering the global energy system. It is thus a fully-fledged energy demand and supply simulation model aiming at addressing energy system analysis, energy price projections, power generation planning, and climate change mitigation policies. The TIMES Integrate Assessment Model (TIAM) [25] is the multiregion, global version of the TIMES energy-system optimisation model, 1 which combines an energy system representation of fifteen different regions with options to mitigate non-CO 2 greenhouse gases as well as non-energy CO 2 mitigation options, such as afforestation. It uses emissions from these sources to calculate temperature changes using a simple climate module. As such, it can be used to explore a variety of questions on how to mitigate climate change through energy systems and transformations, as well as reductions in non-energy CO 2 emissions and non-CO 2 emissions. A comparative overview of models used is presented in Table 2. 2.3. Scenario design Our scenario framework begins with a set of realistic and updated ‘reference’ scenarios (or ‘baselines’, with the two terms used interchangeably hereafter), upon which we assess the impact of recovery packages on emissions pathways. These baselines consist of two core scenarios: the Current Policies (CP-EI) and the combination of Nationally Determined Contributions (NDCs) and Long-Term Targets (LTTs) (NDC-LTT), altogether representing different levels of policy ambition. The first (CP-EI) reflects a continuation of current trends with climate policies remaining at their currently implemented or officially announced levels. The second (NDC-LTT) assumes that all countries achieve the targets documented in both their post-Glasgow version of their NDCs and LTTs. This scenario reflects the level of near-term climate ambition (by 2030), as included in the country NDCs, while LTTs show the level of countries' long-term ambition which is not necessarily embodied in policy agendas or legislative policy measures. This baseline NDC-LTT scenario therefore reflects national long-term targets (i.e., predominantly net-zero emissions around or after midcentury) on top of and linearly extrapolated from currently pledged NDC targets (by 2030) for each model region [26]. All three IAMs (see Section 2.2) were coordinated with the use of a harmonisation protocol [27], by aligning key model assumptions with the latest official projections in terms of socioeconomic assumptions, technoeconomics, fossil fuel prices, and policies: •Socioeconomic assumptions (population and GDP): we used data from the Europop [28] and UN [29] datasets for population as well as from the IMF [30] short-term outlook and SSP2 long-term trends for GDP [31]. •Technoeconomics (power sector): We used the technology cost assumptions from the IEA's World Energy Outlook until 2050, where relevant [32]. •Fossil fuel prices: We used the price projections from the IEA's World Energy Outlook 2024 and historical data from the World Bank, where relevant [33,34]. •Policies: A database of collected policies and targets for G20 countries (and all EU member states) served as a guideline for scenario modelling; post-Glasgow NDCs and LTTs were used, with NDC targets based on the direct interpretation of countries' unconditional NDC pledges or the less ambitious range for pledges where NDC targets are given in ranges. The green recovery packages collected in the database (discussed in Section 2.1) are integrated in a ‘Green Recovery’ scenario (−GR), either as direct investment or subsidies to low-carbon technologies by region/ country. In this scenario, the climate policy intensity remains fixed to 1 TIMES is a modelling platform for local, national or multi-regional energy systems, which provides a technology-rich basis for estimating how energy system operations will evolve over a long-term, multiple-period time horizon. E. Zisarou et al. Energy Research & Social Science 127 (2025) 104216 3 the baseline levels, meaning that carbon prices in both CP-EI-GR and NDC-LTT-GR are drawn directly from the respective baselines in all regions, so that any change to the emissions trajectory traces back to the integration of the recovery packages. In this scenario, about USD 1.3 trillion in recovery funding is allocated to green technologies that directly contribute to emissions reductions, such as renewable energy deployment, clean mobility, and energy efficiency enhancements. Investments under energy affordability, people-centred transitions, and general (unclassified) measures, collectively amounting to about USD 1.2 trillion, are excluded, as these interventions are not directly related with climate efforts (see Appendix for details into what these encompass). The global synthesis of total recovery packages, discussed in Section 2.1, begged the question of whether an alternative strategy to the substantial focus on energy affordability could instead accelerate decarbonisation efforts even further, if the relevant spending on energy affordability is redirected towards green technologies, prompting considerations of whether such investments represent a missed opportunity to advance the transition to cleaner and more sustainable energy sources. We thus develop another set of scenarios, which—additionally to the -GR scenarios—assume that recovery spending originally designated for energy affordability measures would have been allocated towards green technologies, representing this “Lost Opportunity” (−GRLO). In other words, the hypothetical -GR-LO scenarios are intended to quantify the missed opportunity for climate action owing to governments' funding prioritisation, by redistributing the USD 860-billion budget for energy affordability between low-carbon electricity and transport (equally), as these two sectors are the second and third priorities of most recovery plans (see Fig. 1), while also providing clear opportunities for fast and cost-efficient emissions reductions [35]. In essence, these scenarios are intended to capture whether the potential reallocation of funds could further accelerate decarbonisation efforts and help achieve more ambitious climate targets rather than enabling the continued use of fossil fuels. Table 3 summarises all scenarios used in the study. Table 1 Gathered Recovery Funding in Billion USD per region and classified sector. Electricity networks Energy affordability Energy-efficient buildings & industry Fuel & technology innovation General Low-carbon and efficient transport Low-carbon electricity Other Peoplecentred transitions Total Australia/ New Zealand 1.6 5.3 3.5 0 0.5 2.1 29.7 0.9 0.04 43.9 Canada 0 0.9 8.2 1.5 1.9 15.8 0.06 0.02 0.07 28.6 China 0 0.1 0 0 0 2.9 28.6 0 0 31.7 EU 18.7 592 132.4 10.4 31 206.9 58.9 11.6 3.9 1066.3 India 13.2 31.3 0 0.29 0 23.9 0.9 0.8 0 70.7 Japan 0 60.4 0.7 0.04 0.1 1.3 0.3 18.2 0 81.3 Korea 0 4.6 8.1 3.1 0 1 7.9 0 0 34.9 Russia 0.3 0 0 0.3 0 2.6 0 34.5 2.7 40.5 UK 0 41.8 14.1 0.6 0.5 13.6 4.2 0.3 0 75.3 USA 50 0.3 33.5 1 28.2 179.9 232.7 0 16.3 542.3 Other countries 1 4 123.6 21.3 14 4.3 116.1 34 139 0.06 456.7 Total 87.8 860.3 221.8 31.23 66.5 566.1 397.26 205.32 23.07 2472.2 1 The ‘Other countries’ refers to nations grouped under broader regional or unspecified categories in the source databases. Fig. 1. Global allocation of green recovery funding by major sector. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) E. Zisarou et al. Energy Research & Social Science 127 (2025) 104216 4 3. Results 3.1. Impacts on emissions Starting with the current policy baseline, all three models agree on the potential of recovery spending to accelerate emissions reductions, although we observe different behaviours depending on the modelling approach. GCAM reflects the dynamic of climate policies tending to be more effective when implemented as part of a comprehensive mix rather than in isolation [36]: the introduction of recovery funding on top of currently implemented policies would lead to reduced emissions by 65 MtCO 2 in 2030 (CP-EI-GR), while the funding reallocation of energy affordability funding towards decarbonisation-related interventions to a reduction of 306 MtCO 2 (CP-EI-GR-LO), in 2030 compared to the baseline. Despite this early trend, the long-term sustainability of emissions reductions would depend on a broader mix of policies and continued investments in innovation and infrastructure, hinting at more moderate longer-term reductions (48–150 MtCO 2 in 2040 and 43–95 MtCO 2 in 2050, with the lower end in the -GR and the higher in the -GR-LO scenarios). PROMETHEUS, in line with previous studies like the IEA [37], the OECD [38] and others [39,40], suggests that recovery funding (CP_EI_GR) could have a markedly more pronounced impact on emissions reductions due to the enhanced focus on green recovery policies and large-scale infrastructure (−1530 MtCO 2 in 2030 and up to −1950MtCO 2 in 2050, compared to the baseline). Including additional funding (CP-EI-GR-LO) would further accelerate the deployment of renewable technologies, such as wind and solar PV, eventually contributing to bolder emissions reductions (−2460 MtCO 2 by 2030). In this case of current policy baselines, TIAM results lie in between, projecting a decrease of 1161 MtCO 2 in 2030, 1396 in 2040, and 1145 in 2050 in the recovery scenario (CP-EI-GR) compared to the baseline; interestingly, the insertion of additional green recovery funding (CP-EIGR-LO) leads to a doubling of emissions reductions by 2040 compared to (CP-EI-GR). In 2050, annual emissions reductions reach 2252 MtCO 2 below baseline levels, highlighting the pivotal role of sustained lowcarbon investment in accelerating deep decarbonisation. The introduction of targeted green funding on top of a baseline that is compatible with the quantified NDC and LTT targets suggests that financial support alone does not significantly alter long-term emission trajectories without complementary structural shifts, as it is the stringency of climate policy the main driver for decarbonisation. This time, GCAM shows even smaller gains, while the significant added value of recovery funds remains in PROMETHEUS, albeit to a lesser extent; this is in part induced by the accelerated learning dynamics integrated in the model. TIAM instead points to a synergistic effect between green recovery spending and NDC-LTT-compatible pathways—were the further greening of the recovery packages (-LO) would result to about 8000 MtCO 2 below baseline levels in 2030, the rate of the decline though, slows over time, reaching 7384 MtCO 2 in 2040 and 2918 MtCO 2 in 2050 compared to the NDC-LTT baseline. The model-based analysis shows that recovery funding can facilitate the clean energy transition, but its long-term impact is limited, as the pace of decarbonisation is more heavily influenced by factors such as climate policy ambition, global infrastructure development, market dynamics, and the pace of cleantech roll-out (Fig. 2), rather than the availability of investment stimulus alone. The projected trends across the models support the idea that recovery spending must be strategically aligned with regulatory measures and structural energy system transformations to drive sustained decarbonisation. While a large share of green recovery funding has been allocated to clean power technologies and low-carbon transport, critical hard-to-abate sectors like cement and steel industries or aviation remain underfunded and require additional green investments to ensure alignment with climate targets. The results stress the need for a more balanced approach between investments in low-hanging mitigation measures that can deliver rapid emissions reductions (e.g., solar and wind power) and those requiring bolder structural changes in harder-to-abate sectors that are necessary to meet the Paris temperature goals but may have limited short-term impacts. 3.2. Impacts on electricity generation Assuming continuation of the climate ambition reflected in current policies (CP-EI-GR), a gradual expansion of renewable energy sources for power generation is projected, yet with limited acceleration driven by recovery funding. In line with our findings on emissions (Section 3.1), the two IAMs (GCAM and TIAM) show moderate impacts due to structural constraints and pre-existing market dynamics that limit the speed at which new capacity can be deployed. TIAM, in particular, projects a shift to biomass as policy incentives seemingly favour dispatchable lowcarbon alternatives and bioenergy with carbon capture and storage (BECCS), with the cap on intermittent power entering the grid remaining in place and hence impacting the share of solar negatively. In the additional funding scenario (CP-EI-GR-LO), a more pronounced shift can be observed towards wind and solar. This is accompanied by a reduction in nuclear power in PROMETHEUS, which points to significant acceleration of RES deployment, leading to additional electricity generation Table 2 Key characteristics of the three models comprising the study ensemble (model type, solution horizon, technology choice). Model Name Model Type Solution Horizon Tech choice Key features GCAM 7.0 Partial equilibrium Recursivedynamic (myopic) Logit choice Market equilibrium model, integrated multisector modelling, used for national and international assessments PROMETHEUS V1.1. Energysystem Recursivedynamic (myopic) Logit choice Energy demand and supply simulation, climate change mitigation policies, detailed projections of energy-related variables TIAMGrantham Partial equilibrium Intertemporal optimisation (perfect foresight) Winner takes all Bottom-up, technology-rich, least-cost optimization, multi-regional model, driven by a global CGE model Table 3 Scenario Descriptions. Scenario Description CP-EI Represents the level of short-term ambition that is likely to be materialised through actual policies or concrete policy targets. CP-EI-GR Same climate policy intensity as in CP-EI, but including green recovery spending per technology, sector, and country CP-EI-GR-LO Same as CP-EI-GR, but with increased spending towards low-carbon electricity and transport (50 % of ‘energy affordability’ funds allocated to each option). NDC-LTT It achieves the NDC targets for 2030 and LTT targets (including netzero pledges) for 2050 and beyond NDC-LTT-GR Same climate policy intensity as in NDC_LTT, but including green recovery spending per technology, sector, and country NDC-LTT-GRLO Same as NDC-LTT-GR, but with increased spending towards lowcarbon electricity and transport (50 % of ‘energy affordability’ funds for each option). E. Zisarou et al. Energy Research & Social Science 127 (2025) 104216 5 of approximately 9 and 15 EJ, respectively, by 2050 compared to the baseline. This model also shows stronger alignment with long-term decarbonisation goals, as green recovery investment facilitates faster cost reductions through accelerated learning and grid integration of intermittent renewables, with the additional funds towards transitionoriented technologies yielding reduced nuclear electricity (−2.5 EJ) and coal (−5 EJ) in 2050, relative to the baseline (Fig. 3). The NDC-LTT scenario reflects a concerted push towards decarbonisation driven by the massive deployment of variable RES, such as solar and wind, which dominate power generation capacity. These are complemented by the gradual phaseout of fossil fuels and an expanded role for biomass, nuclear, and hydropower to balance the grid which is dominated by variable renewable energy sources. GCAM once again shows a small impact of recovery funding in terms of accelerating decarbonisation, as capital investments alone are found insufficient to markedly alter clean technology uptake without complementary regulatory and structural reforms. In contrast, PROMETHEUS projects an uptick of wind energy (+7 EJ) by 2030 in the NDC-LTT-GR scenario, compared to the baseline, while the reshuffled funds (NDC-LTT-GR-LO) would slightly boost this growth (+11 EJ) in the same timeframe—suggesting that additional green investment could accelerate the upscale of wind by overcoming investment barriers and enhancing infrastructure readiness. In addition, solar power is favoured more in NDC-LTT-GR-LO as additional funding would yield cost reductions, improved grid integration, and enhanced storage capacity (+10 EJ), hinting at the potential of investment interventions to create a more supportive environment for higher solar PV deployment. Similarly, coal and nuclear would see larger reductions in NDC-LTT-GR-LO than in NDC-LTT-GR (Fig. 4), as increased financial support to the transition could expedite the retirement of high-emission and inflexible baseload technologies. In TIAM, early-stage ramping up of RES technologies would lead to an increase in the share of wind (+8.5EJ by 2030), followed by a trend reversal by 2050 (−8.3 EJ), compared to the baseline (NDC-LTT), due to a shift towards biomass (+10.3 EJ). 3.3. Trends in transport energy use Accounting for 23 % of global GHG emissions [41], transport is a critical sector that needs to be decarbonised on the way to ‘well-below 1.5 ◦C’. However, sector-specific decarbonisation plans remain underdeveloped in many NDCs, with only one-third of countries including explicit CO 2 mitigation targets for transport [42]. Our modelling results show that, while green recovery measures could yield emissions reductions in the transport sector, these would be limited under current policies. GCAM and TIAM show modest gains in electrification and a slow shift to cleaner fuels, suggesting that the Fig. 2. Differences in global CO 2 emissions compared to the baseline (CP-EI-GR and CP-EI-GR-LO vs. CP-EI as well as NDC-LTT-GR and NDC-LTT-GR-LO vs. NDCLTT) in 2030–2050. Fig. 3. Differences in the electricity generation mix compared to the baseline (CP-EI-GR and CP-EI-GR-LO vs. CP-EI). E. Zisarou et al. Energy Research & Social Science 127 (2025) 104216 6 COVID recovery funding is not very effective to achieve transport decarbonisation, and additional measures should be considered to avoid/prohibit investments in carbon-intensive transport. Despite its benefits, the shift towards renewable electricity would not immediately displace fossil fuels in transportation, which would expectedly remain heavily dependent on petroleum products. In contrast, PROMETHEUS shows a more optimistic impact (Fig. 5), suggesting increases of electricity by 3–4 EJ in the next decade (CP-EI-GR), indicating a more substantial role in promoting the adoption of electric vehicles and charging infrastructure, relative to the baseline; nonetheless, this effect is focused only on road passenger segment (private cars) and does not lead to a fundamental transformation of the entire transport system, as infrastructure and technological adoption barriers remain. The increased green recovery funding (CP-EI-GR-LO) can further accelerate the transport electrification trend, with electricity use rising further (+6.5 EJ in 2030 and +8.5 EJ in 2050 in PROMETHEUS results) replacing the use of fossil fuels. Meanwhile, liquid fuel reductions are far more pronounced in this scenario (−34 EJ in 2040), as opposed to the CP-EI-GR case (−11 EJ), highlighting the pivotal role of enhanced investment in accelerating the decline of fossil-based transport fuels. Many recovery-funded infrastructure projects (e.g. refuelling networks, EV supply chains) initiated during the 2020–2030 decade), become operational in 2030s, but their full effect materialises in the 2040s, driving deeper fuel substitution. By 2050, much of the achievable fuel switch has already occurred and the rate of change slows down especially in TIAM. GCAM results show minimal changes in transport emissions on top of the NDC-LTT pathway, suggesting that short-term recovery funding alone is insufficient to drive deep decarbonisation in the sector. PROMETHEUS and TIAM display a more dynamic response (Fig. 6), projecting increased electrification and gradual decline in liquid fuel use—more moderate in the former but considerably bolder in the latter, compared to the CP-EIcases. TIAM notably projects a drop of about 13.4 EJ in fossil fuel use in the transport sector by 2030 and PROMETHEUS a drop of about 8,6 EJ). PROMETHEUS projects smaller fossil fuel reductions by 2050 compared to 2030s (~5EJ), as much of the nearterm mitigation stems from early, cost-effective actions, while in the NDC-LTT context the sectoral transformation towards electrification (and other clean fuels) is almost complete by 2050 and the marginal impact of green recovery funding is limited. These results again suggest that the transport sector decarbonisation requires a combination of longterm ambitious climate targets providing clear and predictable signals to investors for the transformation with strong, well-defined near-term low-carbon investment to accelerate transport electrification combined with recharging infrastructure. 3.4. Transformation in the buildings sector In the current policy framework, recovery funding could influence the energy mix in the built environment towards energy efficiency and electrification, while reducing the use of fossil fuels. According to GCAM, recovery funding would result in reduced electricity demand, particularly in the long run, driven by energy efficiency improvements Fig. 4. Differences in the electricity mix compared to the baseline (NDC-LTT-GR and NDC-LTT-GR-LO vs. NDC-LTT). Fig. 5. Differences in the transport energy consumption by fuel compared to the baseline (CP-EI-GR and CP-EI-GR-LO vs. CP-EI). E. Zisarou et al. Energy Research & Social Science 127 (2025) 104216 7 (building retrofits, insulation improvements, high-efficiency appliances, etc.) and despite increased electrification. In parallel with gas phaseout, solid fuels (in the form of biomass, not coal) would still play a role in heating due to legacy effects of persistent residential biomass use and inefficient systems remaining in operation in certain regions. Additional green recovery funding (CP-EI-GR-LO) would yield some further electrification gains by 2030, boosting the adoption of electric heating systems, heat pumps, and efficient appliances; post-2040, however, electricity consumption would slightly decline compared to the baseline (CP-EI) as energy efficiency impacts outweigh the electrification gains. PROMETHEUS and TIAM, on the other hand, suggest that recovery funding can lead to an increase in electricity use in the building sector, especially post-2040, with a parallel drop in gas consumption (−5 EJ in 2030, −7.6 EJ in 2040, and −10 EJ in 2050; PROMETHEUS). Reshuffling funds from energy affordability initiatives towards decarbonisation efforts (CP-EI-GR-LO) could further strengthen the electrification trend, with electricity demand increasing by 3.3 EJ in 2040 (PROMETHEUS) compared to the baseline (Fig. 7); despite a boosted shift to cleaner alternative sources, small amounts of liquid fuels remain in the mix in this scenario due to higher penetration of biofuels and/or continued reliance on oil-based heating in certain developing regions. In the NDC-LTT context, the impact of recovery funding depends on how effectively investments are directed towards electrification, energy efficiency, and alternative clean heating technologies. GCAM shows negligible effects of recovery funding in this case: the energy efficiency measures already implied in the NDC-LTT trajectory are sufficient to Fig. 6. Differences in the transport energy consumption by fuel compared to the baseline (NTC-LTT-GR and NTC-LTT-GR-LO vs. NTC-LTT). Fig. 7. Differences in the building sector's energy mix compared to the baseline (CP-EI-GR and CP-EI-GR-LO vs. CP-EI). E. Zisarou et al. Energy Research & Social Science 127 (2025) 104216 8 limit energy demand growth in buildings. Both PROMETHEUS and TIAM paint a more vibrant picture, showing stronger impacts of recovery funding on building electrification when applied to a trajectory aligned with longer-term ambitions (+1–2 EJ range by 2050), meaning that financial incentives drive a more aggressive adoption of electric heating systems, heat pumps, and energy-efficient appliances. They also show a decrease in gaseous fuels (up to −4 EJ in 2050; PROMETHEUS), reinforcing the shift away from fossil gas for cooking and heating, as well as in liquids, albeit at a slower rate. Fossil fuel use would remain mostly in developing regions and in harder-to-electrify building applications. Interestingly, TIAM projects that additional funds for transitionrelated interventions among the recovery packages announced and/or implemented worldwide (NDC-LTT-GR-LO) would bear increased fruit (e.g., in terms of electrification), reaching 5,6 EJ in 2030 and in 2040 (Fig. 8), while mid-century minimize this trend (+1,5 EJ in 2050), while PROMETHEUS shows similar effects but smaller on magnitude. This also aligns with findings from previous studies [41–43] that emphasise the importance of regulatory measures, technology innovation, and consumer behaviour shifts combined with financial incentives in driving building-sector transformation. 3.5. The impact of COVID-19 recovery spending on industry decarbonisation The industrial sector faces large and complex challenges towards decarbonisation, as several low-emission technologies that can potentially reduce industrial emissions are technically immature and experience high costs, while competitiveness issues pose additional challenges for industry decarbonisation. Although the identified industry-specific recovery measures are much smaller compared to other sectors, observable shifts in industrial energy demand emerge due to systemic interactions, through spillovers from investments in clean electricity, reflecting the integrated nature of the energy system in transition as captured by IAMs. GCAM suggests that recovery funding would have a minimal impact on industrial energy consumption patterns with only marginal deviations from the baseline scenario; over time, industrial energy use would remain largely unchanged, indicating that both the announced/official and the hypothetical/reallocation recovery scenarios fail to induce meaningful shifts in the sector's trajectory, suggesting that these funding scales are inadequate to overcome existing structural barriers. Instead, PROMETHEUS depicts a more responsive industrial system (Fig. 9), where recovery funding could influence fuel substitution patterns mostly away from coal, particularly in the near term. TIAM suggests that industries would pivot towards gas as a transitional fuel, leveraging its cost-competitiveness and availability, but only in this decade (+2.3 EJ by 2030) since deeper decarbonisation pressures and technology shifts would gradually moderate gas reliance. Solid fuel use follows downward trajectories in TIAM and PROMETHEUS (−4–8 EJ range by 2050) in CP-EI-GR, due to a combination of factors such as carbon pricing impacts and industrial process electrification, all of which reduce the sector's dependence on coal and other solid fuels—and slightly more (−7–9 EJ range) in CP-EI-GR-LO, pointing to somewhat reinforced coal phaseout trends. Liquid fuels could increase in the reallocation funding scenario (CP-EI-GR-LO) compared to the baseline (+2 EJ; PROMETHEUS by 2050) which showed almost no effect. Funding redistribution in the CP-EI-GR-LO scenario would not substantially alter industrial energy choices beyond the shifts already observed in CP-EI-GR, as the redistribution away from energy affordability benefits only the electricity and transport sector investment. Under the NDC-LTT assumptions, industry—which faces structural challenges in reducing emissions while maintaining competitiveness—is expected to undergo a fundamental shift, with increasing electrification, a gradual decline in fossil fuels, and the integration of cleaner technologies and fuels like hydrogen and Carbon Capture and Storage options. The effectiveness of this transition depends on climate policy ambition and on the scale and direction of financial support, particularly in terms of how recovery funding is allocated. In this context, GCAM results indicate only a slight increase in electricity demand, particularly in the funding reallocation scenario (NDC-LTT-GR-LO); additional funding could promote electrification to some extent but could not fundamentally alter the sectoral energy mix. PROMETHEUS shows larger implications but still not as significant as in the current policy scenario; a common trend in the NDC-LTT-GR Fig. 8. Differences in the building sector's energy mix compared to the baseline (NDC-LTT-GR and NDC-LTT-GR-LO vs. NDC-LTT). E. Zisarou et al. Energy Research & Social Science 127 (2025) 104216 9