The adoption of the Paris Agreement in 2015 set a global objective of keeping the global average temperature well below 2°C above pre-industrial times, with efforts to achieve 1.5°C,1 calling for clearer scientific evidence of the impacts of a 1.5°C pathway.2 New energy and climate scenarios have been developed as part of the effort to shed light on this question.2–6 Net-zero emissions targets have since been adopted for 2050, notably in the EU, the UK, Japan and South Korea, and for 2060 in China, which together imply substantial reductions in global fossil fuel use, and large markets for low-carbon technology. Reducing emissions requires increased investment in low-carbon technology, with much debated macroeconomic implications.7–10 Large quantities of fossil fuel reserves and resources are likely to become ‘unburnable’ (stranded) if countries around the world effectively implement climate policies.11–13 The transition may already be underway, and some stranding may even happen irrespective of any new climate policies, in the present trajectory of the energy system, with critical distributional macroeconomic impacts worldwide.10 While concerns over peak oil supply have shaped foreign policy for decades, the main macroeconomic and geopolitical challenges may in fact result from peaking oil (and other fossil-fuel) demand.14,15
Climate policy has traditionally been understood as a ‘Prisoner’s Dilemma’ (PD) game, where the objective of curbing emissions of CO2 is plagued by ‘free-riding’ by those not doing so, but who nevertheless benefit from global mitigation, without the economic burden of environmental regulation.16–19 However, this motive is not supported by the evidence.20,21 But furthermore, the nature of the incentives driving the game may currently be changing: the game may have become about industrial strategy, job creation and success of trade, which is not a simple PD problem. The costs of generating solar and wind energy, depending on location, have already or will soon reach parity with the lowest-cost traditional fossil alternatives,15,22 while investment in low-carbon technologies is generating substantial new employment.23–25
The notion that a country should benefit from free-riding on other countries’ climate policies can also be challenged. Incremental decarbonisation, increasing energy efficiency, and the economic impacts of COVID-19 have led oil and gas prices to decline substantially, affecting the viability of extraction in less competitive regions.15 Fossil fuel exporters can be economically impacted by climate policy decisions of other countries through lower global demand and lower prices, and abandoning climate policies to boost domestic demand or maintain high prices is not sufficient to make up for losses of exports.10
In this article, we ask if the PD game remains an accurate representation of reality, and if not, what the climate policy game has become. Indeed, positive payoffs may arise for fossil energy importers reducing imports while negative payoffs arise for energy exporters losing exports, both being far larger than the actual costs of addressing climate change. A key task for the policy-making and finance communities is to accurately anticipate what the new energy geography will be, and to predict its macroeconomic and geopolitical implications.
Method and scenarios
Understanding the ongoing low-carbon transition and its geopolitical implications requires suitable tools. Most integrated assessment models (IAMs) currently used for assessing climate policy and socio-economic scenarios are based on whole system/utility optimisation algorithms26. IAMs have helped set the global climate agenda by identifying desirable energy system configurations. However, they are unsuitable for studying trends in energy system dynamics, since historical dependences are neglected, while systems optimisation assumes an empirically unsubstantiated degree of system coordination.26,27 By contrast, non-optimisation IAMs, calibrated on time series and driven by system dynamics, can more accurately project energy system transformations.
Here we use the E3ME-FTT-GENIE integrated framework10,28 of highly disaggregated energy, economy and environment models based on observed technology evolution dynamics and calibrated on the most recent time series available (Methods). Our model covers global macroeconomic dynamics (E3ME), S-shaped energy technological change dynamics (FTT),29–31 fossil fuel and renewables energy markets,32,33 and the carbon cycle and climate system (GENIE).6 The framework explores these coupled dynamics in 61 regions covering the globe, 43 sectors of industrial activity with particular focus and detail on the four sectors of highest fossil energy use, representing 88 technologies in power generation, transport, heat and steelmaking (FTT). 29–31 We project changes in output, investment and employment in all sectors and regions of industrial activity, coupled by bilateral trade relationships between regions and input-output relationships between sectors. We simulate endogenous yearly average oil and gas prices using a resource depletion simulation calibrated with a dataset that details 120,000 oil and gas production assets worldwide (Methods). We use a simple game theory framework to identify likely geopolitical motives.
We define four scenarios, from now to 2070, of technology, energy use and economic evolution in 61 regions, on the basis of current policies, developments already under way, and evidence-based expectations regarding future climate policies (Methods).
Technology Diffusion Trajectory (TDT) – We simulate the current trajectory of technology and the economy, based on recently observed trends in technology, energy markets and macroeconomics, exploring the direction of technology evolution with the effects of past and current policies represented implicitly through the data. We interpret the TDT scenario as a realistic representation of business as usual, i.e. not contingent on the adoption of new climate policies such as may be needed to achieve climate targets. It is consistent with a median global average temperature warming of 2.6°C (Methods).
Net-zero CO2 globally in 2050 (Net-zero) – We add new climate policies (e.g. technology subsidies, feed-in tariffs, fuel taxes, public procurement, and an increasing exogenous carbon price, see Methods) by either increasing the stringency of what already exists in reach region, or by implementing policies that may be reasonably expected in each regional context, based on two indications, the regional technological compositions and an extrapolation of policies adopted in other regions with similar contexts. The UK, EU, China, Japan and South Korea reach net-zero emissions independently in 2050. Moderate amounts of negative emissions from biomass power generation linked to carbon capture and storage (BECCS) are used to offset residual emissions in other industrial sectors. This scenario achieves a median warming of 1.5°C.
Net-zero in Europe and East-Asia (EU-EA Net-zero) – We use the same policies to achieve net-zero emissions for Europe and East Asia (China, Japan, South Korea) but assume TDT policies for all other nations. This represents a second baseline in which TDT is augmented to include new net-zero targets adopted by major economies. This scenario achieves a median warming of 2.0°C.
Investment Expectations (InvE) – We switch off our energy technology evolution model (FTT) and replace its variables by prescribing exogenously all final energy demand from data, in which energy markets grow over the simulation period, to reflect expectations of relatively slow or delayed decarbonisation by a major subset of investors in energy systems. We use data from the IEA’s World Energy Outlook 2019 current policies scenario,34 run the macroeconomic model (E3ME) alone but determine fossil fuel prices using our fossil fuel resource depletion model consistent with the demand. This scenario is consistent with a median warming of 3.5°C, and is similar to standard baselines (e.g. RCP 8.535) used widely.
Changes in energy systems
Figure 1 shows the evolution of technology globally for electricity generation, passenger road transport, household heating and steelmaking, as modelled using the FTT components, covering 58% of global final energy carrier use, and 66% of global CO2 emissions. In the InvE scenario, the technology composition is derived from the IEA scenario data. Global fuel combustion and industrial emissions in all sectors are also shown.
We observe that the InvE baseline sees coal and natural gas use dominate power generation, petrol and diesel use in road transport translate into a steady growth of oil demand, while technology remains relatively unchanged for heating and steelmaking and other parts of the economy. Note that the InvE scenario projection is not likely to be realised as it features substantially lower than already-observed growth rates in solar, wind, electric vehicles and heat pumps (Suppl Note 1).
In stark contrast, TDT scenario projects a relatively rapid continued growth, at the same rates as observed in the data, of some low-carbon technologies (solar, wind, hybrids and electric vehicles, heat pumps, solar heaters) while others continue their existing moderate growth (biomass, geothermal, hydroelectricity, CNG vehicles). Some technologies have already been in decline for some time, such as coal-based electricity and diesel cars (UK, EU, US), coal fireplaces and oil boilers in houses, and some inefficient coal-based steelmaking technologies (most countries).
Through a positive feedback of learning-by-doing and diffusion dynamics (Suppl. Fig. 1), solar photovoltaics (PV) becomes the lowest cost technology soon after 2025-2030 in all but the InvE scenario, depending on regions and solar irradiation. Electric vehicles display a similar type of winner-take-all phenomenon, although at a later period. Lastly, heating technologies evolve as the carbon intensity of households gradually declines. The trajectory of technology in the TDT scenario, as observed in recent data, suggests that the absolute value of energy consumed in the next three decades is substantially lower than what InvE suggests, as the relatively wasteful and costly thermal conversion of primary fossil fuels into electricity, heat or usable work stops growing even though the whole energy system continues to grow. In the Paris-compliant Net-zero scenario, technology transforms at a comparatively faster pace to reach global carbon neutrality, while in the EU-EA Net-zero scenario, low-carbon technology deployment in regions with net-zero targets accelerates cost reductions for all regions, inducing faster adoption even in regions without climate policies.
We comprehensively model the global demand for all energy carriers in all sectors in 61 regions, shown in Figure 2 (sectoral details are given in Suppl. Fig. 2, regional details in Suppl. Fig. 3-4; see Suppl. Dataset 1). We observe a peaking in the use of fossil fuels and nuclear by 2030 and concurrent rise of renewables in all but the InvE scenario (Fig. 2a,b). PV takes most of the market, followed by biomass, which serves as a negative emissions conduit, and wind, which in our scenarios is gradually outcompeted by PV. The growth of hydro is limited by the number of undammed rivers that can be dammed, while other renewables have lower potentials or lack competitiveness (geothermal and ocean-related systems). Cost trajectories are dictated by the interaction between diffusion and learning-by-doing.
Figure 2c,d,e shows the evolving geography of the global supply and demand of primary fossil energy and renewables. Since fossil energy is widely traded internationally but renewable energy is primarily consumed in local electricity grids (Suppl. Note 2), the geographies of demand and supply differ substantially for fossil fuels while they are essentially identical for renewables. The observed rapid diffusion of renewables substantially decreases the value of regional energy trade balances, without replacement by new equivalent sources of trade. While renewable technical potentials are mostly dependent on the area of nations, fossil fuel production and decline are concentrated in a subset of geologically suited regions.32
Distributional impacts and geopolitics
International fossil fuel trade forms a key source of economic power in the current geopolitical order. The demise of fossil fuel markets is therefore unlikely to proceed without important changes in economic and political power, and it is critical to explore the various ways in which this could play out. For that, it is necessary to first understand what comparative market power each producer region wields, and second, what macroeconomic and fiscal implications market strategies can have.
We show in Figure 3 the cost distribution of global oil and gas resources according to the Rystad36 database, which comprehensively documents over 120,000 oil and gas assets covering most existing resources worldwide (Methods and Suppl. Dataset 2), aggregated here in eight key regions. In the TDT scenario, our model projects cumulative global oil and gas use up to 2050 of 890 and 630 Gbbl respectively (480 and 370 Gbbl in the Net-zero scenario). Saudi Arabia and other OPEC countries together possess over 650 and 202 Gbbl of resources of oil and gas, characterised predominantly by substantially lower costs of production (below $20 per barrel in many cases), compared to the resources left in the US, Canada and Russia, occurring at substantially higher production costs (between $20 and $80 per barrel). This suggests that should they wish to do so, under the expectation of limited future oil and gas demand, OPEC countries could together or independently decide to flood markets to price out other participants from fossil fuel markets by maintaining or increasing their levels of production.
We define two scenario variants that represent two opposite OPEC courses of action. At one end of the spectrum, in a scenario of oil and gas asset fire-sale (denoted SO for ‘sell-off’), OPEC ramps its production to reserve ratio up to a sufficiently high level to gradually acquire a large fraction of global demand as it peaks and declines, effectively offshoring what would otherwise be production losses. At the other extreme, in a scenario of strict quotas (denoted QU for ‘quotas’), OPEC limits production to maintain a constant share of the peaking and declining global demand, keeping its traditional role of stabilising markets.14 Figure 4a shows changes in in prices for all scenarios, and Figure 4b,c changes in quantities for the EU-EA Net-zero scenario originating from current technological trajectories and the existing net-zero pledges, relative to the expectations benchmark in InvE. We observe that, whereas in the QU EU-EA Net-zero scenario the production losses are more evenly distributed between nations, in the SO EU-EA Net-zero scenario, the US, Canada, South America, and to a lesser extent Russia, are gradually excluded from oil and gas production as it concentrates towards OPEC countries (Methods).
The prices of fossil fuels are estimated in E3ME-FTT by identifying the marginal cost of the resource production that matches demand at every time point, which in the case of oil and gas, uses a depletion algorithm based on the Rystad data. Depending on decisions made, long-term oil prices could remain at values as low as $35/bbl for long periods as the expected economic viability of higher cost resources (such as tar sands, oil shales, arctic and deep offshore) deteriorates.
Changes in oil and gas prices, combined with slumps in production, may therefore have disruptive structural effects on high-cost fossil fuel exporters such as the US, Canada, Russia and South America. Meanwhile shedding expensive imports benefits GDP and employment in large importer regions such as the EU, China and India, as money not spent on expensive energy imports is spent domestically, while output is boosted by major low-carbon investment programmes. Figure 4d,e,f shows this using percent changes in government royalties, GDP and total employment between the Net-zero and the InvE scenarios. These transformations arise from changes in fossil and energy production sectors, their dependent supply chains and other recipients of spending income in unrelated sectors, including government royalties. Losses of jobs and output in exporter countries are in general not overcompensated by the job and output creation effect of renewables deployment, while in importer countries, net gains are observed. Supply chain effects amplify output changes that originate from the energy sector (manufacturing, construction, services). For clarity of analysis, we assume no compensatory effect from any deficit spending (Suppl. Note 3).
Economic changes implied by the new net-zero pledges (the EU-EA Net-zero scenario against InvE) are given in Figure 5, showing output, exports, investment and lost fossil fuel production discounted by 6% and cumulated over the next 15 years (see Suppl. Fig. 5, Suppl. Tables 1-2 and Suppl. Dataset 3 for comparison variants).
Lastly, we find, using in a simple two-by-two game theory framework (Table 1, Suppl. Note 4, Suppl. Fig. 6) that if one assumed that strategic climate and energy policy decisions were taken solely on the basis of the GDP or employment outcomes, and that these were known in advance by policy-makers, the EU-EA Net-Zero SO would be a stable Nash equilibrium. Here, the decision by importers to decarbonise is a dominant strategy, as is that of OPEC to flood markets.