Article

Quantifying uncertainties influencing the long-term impacts of oil prices on energy markets and carbon emissions

  • Nature Energy 1, Article number: 16077 (2016)
  • doi:10.1038/nenergy.2016.77
  • Download Citation
Received:
Accepted:
Published online:

Abstract

Oil prices have fluctuated remarkably in recent years. Previous studies have analysed the impacts of future oil prices on the energy system and greenhouse gas emissions, but none have quantitatively assessed how the broader, energy-system-wide impacts of diverging oil price futures depend on a suite of critical uncertainties. Here we use the MESSAGE integrated assessment model to study several factors potentially influencing this interaction, thereby shedding light on which future unknowns hold the most importance. We find that sustained low or high oil prices could have a major impact on the global energy system over the next several decades; and depending on how the fuel substitution dynamics play out, the carbon dioxide consequences could be significant (for example, between 5 and 20% of the budget for staying below the internationally agreed 2 C target). Whether or not oil and gas prices decouple going forward is found to be the biggest uncertainty.

Oil prices took a sharp plunge starting in late 2014 and have remained low ever since. Combined with parallel developments in natural gas supply, this plunge has prompted questions regarding what the ‘new normal’ might mean for global markets. How will falling oil and gas prices affect energy decision-making over the long term? Will they damage the business case for renewables? Will they stymie incentives to invest in energy efficiency? How do they change the outlook for coal and nuclear? Does this spell bad news for efforts to mitigate climate change? The International Energy Agency (IEA) recently found that, between now and 2040, lower oil prices will lead to marginally greater oil and gas demand and incrementally smaller renewables and coal demand, which on balance means slightly higher carbon dioxide (CO2) emissions1. However, it is not unlikely that the energy and emissions impacts could be larger than this. What is more, the broader, energy-system-wide impacts of diverging oil price futures are likely to depend on a number of key uncertainties.

Here we present work that unpicks several potentially influential factors, thereby going beyond economic analyses focusing on the very near term impacts of oil prices2,3,4,5,6 and the limited number of scenario analyses for the mid-to-long term1,7,8,9,10. Using the MESSAGE integrated assessment model, we develop and analyse scenarios with wide-ranging oil price levels that are in line with recent market fluctuations. We then employ a suite of sensitivity analyses focusing on numerous climate policy, energy resource, and supply and demand technology uncertainties to explore which future unknowns hold the greatest importance for the global energy system and its resulting emissions. We find that sustained low or high oil prices could have a major impact on the former over the next several decades; and depending on how the fuel substitution dynamics play out, the CO2 consequences could be significant (for example, as little as 5% or as much as 20% of the cumulative emissions allowable for keeping global temperatures under the 2 C threshold). By comparison, our calculated shifts are one to two orders of magnitude larger than those estimated by the IEA. That the net CO2 emissions differences we find are not larger (or smaller) when oil prices either rise or fall is because of parallel responses seen for carbon-intensive coal and low-carbon biomass (that is, their benefits/consequences partially cancel out) and price-induced energy-service demand responses across the end-use sectors (industry, transport, and buildings). Thus, if the goal is to mitigate carbon substantially, high oil prices offer no substitute for climate policies. Whether or not oil and gas prices decouple going forward is found to be the biggest uncertainty influencing the system-wide effects exhibited by our scenarios. The impacts also strongly depend on uncertainties surrounding the future potential of sustainable bioenergy supplies and the costs and availability/scalability of electric vehicles. In short, the energy and CO2 impacts of diverging oil price futures depend not just on prices alone, but rather on a number of uncertain resource-, technology- and policy-related factors.

Energy and emission impacts of alternative oil price futures

Our scenario exercise leads to several insights with implications for energy–climate policy, technology and markets. First, sustained low or high oil prices could have a major impact on the global energy mix (that is, ‘substitution effects’) between now and 2050 (see Fig. 2, focusing on the results with our central case assumptions, that is, ignoring the uncertainty ranges for the moment). As expected, low oil prices lead to considerably greater (cumulative) use of oil in both the baseline and climate policy scenarios. Similarly, if natural gas prices remain coupled to oil prices across all regional markets, then the future might also see a similar expansion of gas use in a low-oil-price world. For both resources, utilization of unconventionals16,27 is greater in the low-oil-price scenario. This dynamic is consistent with a future of strong technological change (over the long term) in unconventional oil extraction, leading to lower costs of production, whereas the high-oil-price scenario represents a future with less innovation in oil extraction. Meanwhile, coal and low-carbon energy (namely biomass) see greater deployment under high oil (and gas) prices. This is due to biomass- and coal-based energy carriers (liquid fuels, gases, electricity) both reaching cost-competitiveness with oil- and gas-derived energy forms. Synfuels production (without carbon capture and storage in the baseline; with it in the climate policy scenario) is, in fact, responsible for a considerable amount of the increased coal consumption seen in the high-oil/gas-price case. Electric and biofuel vehicles also make much greater inroads by mid-century in the high-oil-price case; natural gas vehicles, in contrast, fail to do the same. Perhaps surprisingly, according to our analysis, the other low-carbon primary resource options (solar, wind, hydro, geothermal, and nuclear) experience only a slight uptick in the high-oil/gas-price case, principally because the substitution possibilities for nuclear and non-biomass renewables are largely restricted to electricity generation, and in this sector inexpensive coal represents a more cost-effective option (so long as carbon pricing remains relatively moderate up to 2050). Finally, we note that the fuel substitution dynamics discussed here are also found under various other levels of climate policy stringency (that is, comparing the low- and high-oil-price cases for the same carbon price; see Supplementary Tables 1 and 2 and Supplementary Fig. 1 for details).

Figure 2: Cumulative energy demand from 2010 to 2050 by primary resource under low or high oil prices.
Figure 2

ad, Energy demand in the case of no climate policy (‘Baseline’) and the reference climate policy (‘Mitigation’) scenarios under low or high oil prices for crude oil (a), natural gas (b), low-carbon resources (biomass, non-biomass renewables and nuclear) (c) and coal (d). In a,b, crude oil and natural gas are sub-divided into conventional and unconventional resources (for example, oil sands, shale oil and gas, and tight gas), following the definitions of refs 16,27. Uncertainty ranges are given by the small grey bars of varying shades overlaid along the tops of the main bars. These reflect minimum/maximum values obtained for individual scenarios within the relevant set of sensitivity cases, numbered according to the definitions in Table 1. Error bars positioned at the top of each main bar reflect the full range of uncertainty across the entire suite of sensitivity cases. Note the different y-axis scales of the charts. e,f, Percentage shares of cumulative energy demand by resource type for no climate policy (‘Baseline’) (e) and the reference climate policy (‘Mitigation’) (f) scenarios. Conventional and unconventional oil/gas are combined here. The colours are common to all panels, denoted by the key. In the mitigation cases, a minority share of the coal- and gas-based energy is equipped with carbon capture and storage (coal: 7–12%, gas: <1%). Supplementary Tables 1 and 3 contain the data for the charts. One zettajoule (ZJ) is equal to one sextillion (1021) joules; for reference, annual global primary energy production in 2010 was approximately 0.5 ZJ.

These findings regarding fuel substitution dynamics under either low or high oil prices vary across our different supply- and demand-side sensitivity cases. One sees this by noting the uncertainty ranges overlaid along the tops of the bars in Fig. 2 (see Supplementary Tables 1–4 for numerical details). Eight separate sensitivities are presented (see Table 1, excluding the climate policy sensitivities); of these, the one leading to the largest variation in our results (that is, for the different fuels in the various scenarios) is the uncertainty surrounding the future correlation between oil and gas prices (#5). If prices do manage to decouple globally, then natural gas deployment stands to gain considerably from sustained high oil prices (because gas would remain moderately priced, midway between the low- and high-case levels), whereas the opposite would be true in a low-oil-price world. Although these dynamics may be expected in the directional sense, the magnitude of the swing is arguably pronounced. We note, for instance, that in both baseline and mitigation scenarios the total natural gas consumption under high oil prices becomes significantly greater (1 ZJ cumulative to 2050) than under low prices—a complete reversal from the scenarios with our central case assumptions. This is largely explained by gas replacing coal, non-biomass renewables, and nuclear for power generation, along with some substitution of gas for oil and coal in industry and for oil in buildings applications (either for heating or as a chemical feedstock). Another important sensitivity relates to the availability and potential of sustainable biomass for energy purposes (#1). If global supplies are constrained to just 100 EJ yr−1 at all points in time (that is, at the lower end of the potential assessed by the IPCC (ref. 26); in the climate policy scenarios this sees all regions running up against their bioenergy limits from 2020 onwards), then cumulative biofuels demand could be as much as 1 ZJ lower, with oil- and coal-based liquid fuels filling the gap. In contrast, uncertainties surrounding the future costs and availability/scalability of biofuels and fossil synfuels production (#2, 3, 4) are found to affect the energy mix in a relatively minor way across the various scenarios. Similar observations are made when assessing the demand-side sensitivity cases focusing on the future costs and availability/scalability of advanced transport technologies (#6, 7, 8). The one exception relates to electric vehicles: cumulative oil demand swings through a range approaching 1 ZJ, depending on the future competitiveness of this nascent class of technologies.

Our second key insight is that, depending on how the fuel substitution dynamics play out, the potential impacts of sustained low or high oil prices on CO2 emissions could be significant. The lower the oil price, the stronger the carbon price signal that is needed to motivate a given level of CO2 abatement; or put another way, for the same carbon price schedule, less abatement is achieved under lower oil prices. In our reference scenarios (either baseline or climate policy), future differences in global annual CO2 emissions (from fossil fuels and industrial processes) between the two oil price cases are 3.5 to 4 GtCO2 yr−1 in 2030 and 6 to 7 GtCO2 yr−1 in 2050 (see Fig. 3). For the latter year, this represents about 10% of emissions in the baseline scenarios and 20% in the climate policy scenarios. From a cumulative CO2 perspective (2010–2050), emissions in the low-oil-price cases are nearly 140 GtCO2 higher—approximately three to four years’ worth of global emissions at current rates28, or roughly 15% of the 2010–2050 budget for staying below the 2 C target, according to the IPCC (ref. 29). To put these numbers further into context, we note that recent analyses have estimated the global greenhouse gas (GHG) emission reductions resulting from countries’ submitted (as of early-October 2015) Intended Nationally Determined Contributions (INDCs) to be 3.6 GtCO2eq yr−1 in 2030 (range: 0 to 7.5 GtCO2eq yr−1), relative to the levels expected under pre-INDC policies30. Viewed from these different perspectives, the emission differences brought about by vastly diverging oil price futures are certainly non-trivial; on the other hand, they are considerably smaller than the CO2 reductions needed to safely achieve the 2 C target30,31,32. What all of this suggests is that global mitigation efforts would be moderately hampered by sustained low oil prices and moderately boosted by sustained high prices.

Figure 3: Fossil fuel and industrial process CO2 emissions under low or high oil prices.
Figure 3

No climate policy baseline: solid lines. Reference climate policy scenario: dashed lines. These are the same scenarios as in Fig. 2. Uncertainty ranges reflect minimum/maximum annual emissions levels obtained for individual scenarios within the relevant set of sensitivity cases. Thin solid and dashed grey lines present emissions for an intermediate oil price case (see Supplementary Table 5 for more information about this case). One gigatonne (Gt) is equal to one billion (109) metric tonnes. Historical data from ref. 28.

That the net CO2 emissions differences between our reference low- and high-oil-price cases are not larger (or smaller) is primarily due to the parallel responses of coal and biomass that we find (see Fig. 2). Coal is relatively more carbon-intensive than biomass; hence, a simultaneous increase or decrease in the use of both fuels leads to a partial cancelling out of the emissions benefits/consequences of one or the other. In addition to these countervailing fuel–emission dynamics, some of the difference in CO2 between the low- and high-oil-price cases is simply due to greater or lesser energy-service demands, respectively, (that is, price-induced demand responses) in countries’ end-use sectors: industry, transport and buildings (further details can be found in Supplementary Tables 1–4 and Supplementary Fig. 2; see refs 33,34 for similar discussions). More specifically, our scenario analysis indicates that energy efficiency and conservation efforts are likely to suffer if oil prices remain low for an extended period of time, thereby putting further upward pressure on emissions (see also refs 8,9).

Sensitivity analyses focusing on the previously described uncertainties indicate that the CO2 emissions impacts of oil prices may vary, with the spreads across the different cases being marginally greater in the reference climate policy scenarios than in the baseline scenarios (that is, 13.5 and 0 US$ per tCO2eq carbon price in 2030, respectively; see Fig. 3 uncertainty ranges, as well as Supplementary Tables 1 and 2, for cumulative values). Emissions differences are larger, for example, in both the optimistic biofuels production and pessimistic fossil synfuels production sensitivity cases, relative to the corresponding scenarios with central case assumptions: cumulative CO2 under low oil prices is >140 GtCO2 larger than under high prices. This is intuitive, considering that the former makes less carbon-intensive biomass more competitive whereas the latter makes more carbon-intensive coal less competitive. Similar observations and reasoning apply to the demand-side sensitivity cases making more optimistic assumptions for natural gas and hydrogen vehicles, as these back out oil- and coal-based liquid fuels in transport. In the opposite direction, the two sensitivity cases leading to smaller emissions differences (as low as 112 GtCO2) are those related to limited biomass availability and oil-to-gas price coupling. Across the full suite of uncertainties, the difference in cumulative CO2 between the low-oil-price sensitivity with minimum (maximum) emissions and the high-oil-price sensitivity with maximum (minimum) emissions is 97 (158) GtCO2 in the baseline and 55 (194) GtCO2 under climate policy (see Supplementary Tables 1–4). In other words, sustained low oil prices could lead to greater cumulative emissions that are as little as 5% or as much as 20% of the 2010–2050 budget for staying below the 2 C target. Moreover, we note that, depending on the carbon price schedule, the emissions difference between low and high oil prices ranges from 110 to 139 GtCO2. (Our most stringent climate policy scenario—that is, with the highest carbon prices we tested, 61 US$ per tCO2eq in 2030—sees global temperatures peaking at slightly above 2 C towards the end of the century; see Supplementary Figs 3 and 4 for details.)

The final insight stemming from our analysis is that if the stringency of global climate policy remains moderate over the next several decades to 2050, fluctuations in future oil prices could be as least as big a swing factor for crude oil, natural gas, coal and low-carbon energy demand—if not bigger—than climate policy itself (Fig. 2). For oil and gas in particular, we find that the quantities of these resources left in the ground (that is, unburned35) by mid-century could be driven more by their own base prices (excluding any carbon price add-on) than by mitigation efforts. In contrast, climate policy would become the more dominant driver of energy system change if that policy would be fairly stringent (that is, of the type necessary for holding global temperature rise to around 2 C). By our estimates with MESSAGE, this would be consistent with carbon prices in 2030 of 40 US$ per tCO2eq or greater, rising over time. (Supplementary Tables 1 and 2 and Supplementary Fig. 1 present energy mix results across a range of carbon price cases.) To be clear, we do not mean to suggest that carbon pricing at less substantial levels is unimportant for mitigating CO2; indeed, our analysis shows that it will be critical, given that such policy instruments specifically target carbon-intensive fossil resources and can thus drive declines in emissions. What our analysis instead highlights is that an extended period of either low or high oil prices would impact both fossil and non-fossil resources at the same time and in different ways; and this could have mixed effects on CO2.

Conclusions

In summary, by employing the MESSAGE integrated assessment model, this study finds that sustained low or high oil prices could have a major impact on the global energy system over the next several decades; and depending on how the fuel substitution dynamics play out, the carbon dioxide consequences could be significant (for example, between 5% and 20% of the budget for staying below the internationally agreed 2 C target). The variance in the impacts depends on a suite of critical uncertainties, chief among them the coupling between oil and gas prices going forward.

Although not entirely comparable, a recent analysis by the IEA1 looked at diverging oil price futures where prices slowly return to either high (128 US$ per bbl) or mid (85 US$ per bbl) levels by 2040. That analysis arrives at findings similar, in the directional sense, to those we obtain: lower oil prices lead to greater cumulative oil and gas demand and lesser renewables and coal demand. In terms of magnitudes, however, the energy demand shifts we estimate (moving from higher to lower prices) are substantially larger (by one to two orders of magnitude) for each of the various energy sources: from +1.2 to +2.0 ZJ for oil, −0.6 to +2.3 ZJ for gas, −0.4 to −1.9 ZJ for coal, and −0.1 to −1.2 ZJ for renewables and nuclear (to 2040; values spanning all sensitivity cases), compared to +0.13 ZJ for oil, +0.01 ZJ for gas, −0.10 ZJ for coal, and −0.03 ZJ for biomass and non-hydro renewables according to the IEA’s assessment (approximate calculations based on numbers shown in Table 4.1 of ref. 1). Such trivial shifts in the energy mix in the latter probably explain why cumulative CO2 emissions are estimated to be a mere 3 GtCO2 greater in the IEA’s lower-oil-price case, whereas we calculate the increase to be in the range of 50 to 98 GtCO2 (to 2040). Inter-study discrepancies so immense point to deep uncertainties in how critical factors will drive energy system development, and by extension climate change mitigation, over the twenty-first century.

A caveat to the analysis described here is that a single model was employed to answer the questions posed and, thus, the results are conditional on the chosen framework. Although previous work36 has analysed the long-term fossil resource dynamics of baseline and climate policy scenarios within a multi-model comparison context, no such comparisons have yet been carried out on the topic of oil prices, at least not as framed in this paper. Such an exercise could certainly be fruitful for the global modelling community, as it would eliminate yet another source of uncertainty on top of the numerous parametric sensitivity analyses that we conduct (that is, the structural assumptions of models). Another important caveat to our analysis is that it considers only sustained low or high oil prices, whereas the combined dynamics of oil demand and oil field exploration and development will probably ensure that future oil prices are more volatile than the intentionally stylized paths crafted here. Future work might therefore consider studying, for example, the energy and carbon ‘lock-in’ effects of oil prices that remain low for a time but then rise to much higher levels afterwards (that is, resource/technology/policy decisions made myopically; see ref. 37). Finally, the topic of fossil fuel subsidies represents an area the global modelling community could continue to explore going forward: how do subsidies distort markets, and what might be the impacts of reforming them in various countries over the coming years?

Methods

Overview of the integrated modelling framework.

The MESSAGE (Model for Energy Supply Strategy Alternatives and their General Environmental Impact) integrated assessment framework is comprised of several inter-linked models. (Version ‘V.5a’ of MESSAGE was used for this paper.) At its core is a global energy–economic model based on a linear programming optimization (cost-minimization) approach which is used for medium-to-long-term energy system planning and policy analysis12,13. For each of its eleven regions, the model provides information on the utilization of domestic resources, energy imports and exports and trade-related monetary flows, investment requirements, the types of production or conversion technologies selected (technology substitution), pollutant emissions, and fuel substitution processes, as well as temporal trajectories for primary, secondary, final, and useful energy and their respective prices. At the primary level in particular, regionally specific resource supply curves and extraction technologies (for crude oil, natural gas, coal, uranium, biomass, and other renewables) are specified as input assumptions (the oil supply curves vary by oil price case; see Supplementary Methods). This leads to an endogenous calculation of resource prices, which then contributes to the endogenous calculation of other commodity prices (at the secondary, final, and useful levels), considering costs for energy conversion and transport/distribution, as well as energy subsidies, taxes, and other price mark-ups.

For the estimation of price-induced changes of energy demand, iterations between MESSAGE and the macro-economic model MACRO are relied on38. In MACRO, capital stock, available labour, and energy inputs determine the total output of the economy according to a nested constant elasticity of substitution (CES) production function. Through the linkage to MESSAGE, internally consistent projections of gross domestic product and energy demand are calculated in an iterative fashion, taking into account price-induced changes of both. Six different end-use demand categories are represented in MACRO: electric and thermal heat demands in the industrial and residential/commercial sectors (1–4), non-energy feedstock demands for industrial applications (5), and mobility demands in the transportation sector (6). Prices are calculated uniquely for each of these six demands, and therefore the macro-economic responses seen in the different sectors capture both technological and behavioural measures in each sector uniquely (at a high level of aggregation). MACRO is run for all eleven MESSAGE regions simultaneously.

MESSAGE is used in conjunction with MAGICC (Model for Greenhouse gas Induced Climate Change) version 6.8 (ref. 39) for calculating climatic indicators such as atmospheric gas concentrations, radiative forcing, and annual-mean global surface air temperature.

Much more detailed information about the MESSAGE modelling framework can be found in the Supplementary Methods and in refs 12,13. The following paragraphs focus on the innovative features that were implemented in MESSAGE to undertake the analysis described in the current paper.

Constructing the low- and high-oil-price cases.

Reproducing real-world price behaviour in global IAMs has historically presented a challenge for modellers, and earlier studies have shown that oil and gas prices can diverge widely between frameworks, even in the base year36. IAMs have rather focused their attention on relative price differences between fuels. However, in undertaking the analysis with MESSAGE described in this paper, it was important for the model to represent oil and gas prices as precisely as possible. Hence, a novel methodological development enabling our analysis is that the endogenously calculated energy prices (both primary and final) in MESSAGE were adjusted so that they reproduce observed prices (for all regions, energy sectors and fuels). Towards this goal, an extensive data set of historical prices, subsidies and taxes was compiled from a number of sources (see refs 40,41 for details). In fitting this data into MESSAGE, we assumed for both crude oil and coal that there is a single global price, because these fuels are globally traded. For natural gas, three separate regional market prices were used as benchmarks for the different MESSAGE regions, thereby reflecting regionally fragmented markets. Wherever possible, prices in the data set represent average prices from 2006 to 2010, in all cases converted to US$2005 GJ−1. The strategy for harmonizing the MESSAGE prices and the historical prices relied on the use of ‘price adjustment factors’, which were applied to fuels at both the resource extraction and end-use levels. A hypothetical example illustrates how this works. If the model previously calculated the price of a given fuel in a particular sector and region to be 8 US$ GJ−1 (based on the costs of resource extraction, conversion and distribution activities), but we know the price to be 12 US$ GJ−1 from the historical data set, then a price adjustment factor of +4 US$ GJ−1 (=12–8 US$ GJ−1) was applied to increase the endogenous MESSAGE price to the observed price. At the resource extraction (primary energy) level, this translates into a shifting of the crude oil and natural gas supply curves in the vertical (price) dimension. The fuel/region/sector-specific price adjustment factors estimated for the base year were then held constant—in a given oil price case—throughout the time horizon of the model. The adjustment factors can be interpreted in the following way: they capture all components embedded in the price of fuels beyond their technology-related costs (that is, things that models like MESSAGE are well-suited to represent, such as resource extraction, conversion and transportation costs). The factors explicitly represent both energy subsidies and taxes; residual terms then cover additional components such as producer rents/profits and speculation, among other things.

Alternative oil price cases were created by lowering or raising the price adjustment factors on oil until the desired price level was reached in 2020 (40 US$ per bbl in the low case, 110 US$ per bbl in the high case). Endogenously determined price dynamics then take over in the years after 2020, so that prices rise gradually in line with more costly grades of oil being consumed. The same is true of natural gas, depending on whether or not its prices were assumed to be coupled to oil; if not coupled, then gas prices remain at a moderate level in between the low/high extremes. In addition, we assumed that year-2020 subsidy rates for oil and gas (on both the supply and demand sides) scale proportionately with their respective 2020 prices; in other words, a given %-change in the price corresponds to a given %-change in the subsidy rates. This scaling algorithm is intended to reflect the dynamics of real-world subsidy schemes, which tend to fluctuate up and down with energy prices (see ref. 1). Once the subsidy rate is set, it is held constant throughout the time horizon of the model.

Modelling the link between crude oil and natural gas prices.

In the past, oil and gas prices tended to be correlated because crude oil (and refined oil products) and natural gas were competitive substitutes in several energy and industrial sectors, the two resources were often produced using similar technologies by firms possessing similar expertise, and/or many gas supply contracts (particularly for liquefied natural gas, LNG) were indexed on oil prices42. So when oil prices rose (or declined), gas prices tended to as well, even if their absolute price levels differed considerably. However, the past several years have shown that these relationships could be undergoing a transition. In the United States, for example, oil and gas prices have recently decoupled, owing, at least partly, to hydraulic fracturing techniques for gas production22. Price correlation meanwhile remains strong in most of Europe and elsewhere21,23; although, this too could change over time if the fragmented gas markets of today become more globalized (with LNG being shipped over long distances, as is the case at present with crude oil). The emergence of these new market dynamics is a key uncertainty for the future—hence the sensitivity cases we run for the future coupling between oil and gas prices. In all instances, consistent with past observations41, we assumed that subsidy rates for oil and gas (on both the energy supply and demand sides) scale proportionately with their respective prices.

Developing climate policy scenarios for the analysis.

Climate policy, or ‘mitigation’, scenarios were run by imposing a globally harmonized carbon price that begins in 2020 and grows with an interest rate of 5% yr−1 until the end of the century. (‘Mitigation’ meaning that CO2 emissions are reduced below those of the no climate policy baseline.) For instance, the ‘reference climate policy’ scenario focused on in this paper assumes a carbon price that begins in 2020 at 8.3 US$ per tCO2eq and grows with an interest rate of 5% yr−1, reaching 13.5 US$ per tCO2eq in 2030 and 36 US$ per tCO2eq in 2050, before continuing into the hundreds of dollars later in the century. Such moderate carbon pricing in the MESSAGE framework leads to roughly 2.6–2.7 C warming (median likelihood) above pre-industrial levels by 2100 (with temperatures peaking soon afterwards) and atmospheric GHG concentrations of approximately 615–635 ppm CO2eq in the same year. Results for more stringent climate policy scenarios (that is, with elevated carbon price schedules; some that come closer to achieving the 2 C target) are presented in Supplementary Tables 1 and 2 and Supplementary Figs 1, 3 and 4.

Selecting the sensitivity cases to run.

The sensitivity cases run for this study are summarized succinctly in Table 1. We of course recognize that, if not for computational constraints, an essentially limitless number of parametric assumptions could have been tested using our modelling framework. The subset of factors focused on here was selected after careful consideration of the fuel substitution possibilities (for oil and gas) present in the transport, industry, buildings and power sectors.

References

  1. 1.

    World Energy Outlook 2015 (International Energy Agency, 2015).

  2. 2.

    ,  & Macroeconomic impacts of oil prices and underlying financial shocks. J. Int. Financ. Mark. Inst. Money 29, 1–12 (2014).

  3. 3.

    ,  & Macroeconomic impacts of oil price shocks in Asian economies. Energy Policy 86, 867–879 (2015).

  4. 4.

     & The impacts of global oil price shocks on China’s fundamental industries. Energy Policy 68, 394–402 (2014).

  5. 5.

    , ,  & The Great Plunge in Oil Prices: Causes, Consequences, and Policy Responses (World Bank Group, 2015).

  6. 6.

     & Cheap Oil? Making Sense of a Competitive Oil Market (Inter-American Development Bank, 2015).

  7. 7.

    International Energy Outlook 2014 Report Number DOE/EIA-0484(2014) (U.S. Energy Information Administration, 2014).

  8. 8.

     & Oil and natural gas prices and greenhouse gas emission mitigation. Energy Policy 37, 4797–4808 (2009).

  9. 9.

    et al. Impact assessment of the increase in fossil fuel prices on the global energy system, with and without CO2 concentration stabilization. Energy Policy 36, 3477–3484 (2008).

  10. 10.

    Uncertainties in the Outlook for Oil and Gas. PhD thesis, Univ. London (2013).

  11. 11.

    Global energy–climate scenarios and models: a review. WIREs 3, 363–383 (2014).

  12. 12.

    et al. Global Energy Assessment—Toward a Sustainable Future 1203–1306 (Cambridge Univ. Press and the International Institute for Applied Systems Analysis, 2012).

  13. 13.

    ADVANCE ADVANCE Integrated Assessment Model Wiki Documentation (University College London and the International Institute for Applied Systems Analysis, 2015);

  14. 14.

    European Brent Spot Price FOB (U.S. Energy Information Administration, 2015);

  15. 15.

    The Prize: the Epic Quest for Oil, Money, and Power (Simon Schuster, 1991).

  16. 16.

    et al. Global Energy Assessment—Toward a Sustainable Future 423–512 (Cambridge Univ. Press and the International Institute for Applied Systems Analysis, 2012).

  17. 17.

    Production costs of global conventional and unconventional petroleum. Energy Policy 64, 134–140 (2014).

  18. 18.

    ,  & The relationship of natural gas to oil prices. Energy J. 29, 47–65 (2008).

  19. 19.

     & What drives natural gas prices? Energy J. 29, 45–60 (2008).

  20. 20.

    Have oil and gas prices got separated? Energy Policy 49, 707–718 (2012).

  21. 21.

    ,  & Gas versus oil prices the impact of shale gas. Energy Policy 47, 117–124 (2012).

  22. 22.

    Correlations Between Daily Futures Price Changes of Crude Oil and Other Commodities Generally Rose in Recent Years (U.S. Energy Information Administration, 2015);

  23. 23.

    European Union The Impact of the Oil Price on EU Energy Prices (Directorate General for Internal Policies, 2014).

  24. 24.

    et al. The role of technology for achieving climate policy objectives: overview of the EMF 27 study on global technology and climate policy strategies. Climatic Change 123, 353–367 (2014).

  25. 25.

    , , ,  & Transport electrification: a key element for energy system transformation and climate stabilization. Climatic Change 123, 651–664 (2014).

  26. 26.

    et al. in Climate Change 2014: Mitigation of Climate Change (eds Edenhofer, O. et al.) 811–922 (IPCC, Cambridge Univ. Press, 2014).

  27. 27.

    An assessment of world hydrocarbon resources. Annu. Rev. Energy Environ. 22, 217–262 (1997).

  28. 28.

    Trends in Global CO2 Emissions: 2014 Report (PBL Netherlands Environmental Assessment Agency and European Commission, Joint Research Centre, 2014).

  29. 29.

    IPCC: Summary for Policymakers. In Climate Change 2014: Mitigation of Climate Change (eds Edenhofer, O. et al.) (Cambridge Univ. Press, 2014).

  30. 30.

    Synthesis Report on the Aggregate Effect of the Intended Nationally Determined Contributions (United Nations Framework Convention on Climate Change, 2015).

  31. 31.

    et al. Locked into Copenhagen pledges—implications of short-term emission targets for the cost and feasibility of long-term climate goals. Technol. Forecasting Soc. Change 90, 8–23 (2015).

  32. 32.

    The Emissions Gap Report 2014. A UNEP Synthesis Report (United Nations Environment Programme, 2014).

  33. 33.

    et al. Limited impact on decadal-scale climate change from increased use of natural gas. Nature 514, 482–485 (2014).

  34. 34.

    et al. The impact of hydrocarbon resources and GDP growth assumptions for the evolution of the EU energy system for the medium and long term. Energy Strategy Rev. 6, 64–79 (2015).

  35. 35.

     & The geographical distribution of fossil fuels unused when limiting global warming to 2 C. Nature 517, 187–190 (2015).

  36. 36.

    , , ,  & Fossil resource and energy security dynamics in conventional and carbon-constrained worlds. Climatic Change 123, 413–426 (2014).

  37. 37.

     & Short term decisions for long term problems—the effect of foresight on model based energy systems analysis. Energy 35, 2033–2042 (2010).

  38. 38.

     & MESSAGE-MACRO: linking an energy supply model with a macroeconomic module and solving it iteratively. Energy 25, 267–282 (2000).

  39. 39.

    ,  & Emulating coupled atmosphere-ocean and carbon cycle models with a simpler model, MAGICC6—Part 1: Model description and calibration. Atmos. Chem. Phys. 11, 1417–1456 (2011).

  40. 40.

    , ,  & Report on Improving the Representation of Existing Energy Policies (Taxes and Subsidies) in IAMs Deliverable 3.1. (ADVANCE, 2014);

  41. 41.

    World Energy Outlook 2014 (International Energy Agency, 2014).

  42. 42.

     & The Relationship Between Crude Oil and Natural Gas Prices (Energy Information Administration, Office of Oil and Gas, 2006).

Download references

Acknowledgements

The authors acknowledge funding provided by the ADVANCE project (FP7/2007–2013, grant agreement No. 308329) of the European Commission. The International Energy Agency (in particular A. Bromhead, L. Cozzi, N. Selmet and G. Zazias) provided critical data support, which made the price calibration possible in the model. P. Kolp and M. Strubegger of IIASA are also recognized for their assistance with model code development.

Author information

Affiliations

  1. International Institute for Applied Systems Analysis (IIASA), Schlossplatz 1, 2361 Laxenburg, Austria

    • David L. McCollum
    • , Jessica Jewell
    • , Volker Krey
    •  & Keywan Riahi
  2. University of Tennessee, 1640 Cumberland Avenue, Knoxville, Tennessee 37996, USA

    • David L. McCollum
  3. The World Bank, 1818 H Street, NW, Washington DC 20433, USA

    • Morgan Bazilian
    •  & Marianne Fay
  4. Graz University of Technology, Inffeldgasse, 8010 Graz, Austria

    • Keywan Riahi

Authors

  1. Search for David L. McCollum in:

  2. Search for Jessica Jewell in:

  3. Search for Volker Krey in:

  4. Search for Morgan Bazilian in:

  5. Search for Marianne Fay in:

  6. Search for Keywan Riahi in:

Contributions

D.L.M, J.J., V.K. and K.R. designed the research. J.J. contributed data for the modelling. D.L.M. and V.K. implemented the modelling. M.F. and M.B. provided feedback on the scenarios, in particular assisting with the framing. All authors contributed to writing the manuscript.

Competing interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to David L. McCollum.

Supplementary information

PDF files

  1. 1.

    Supplementary Information

    Supplementary Tables 1–7, Supplementary Figures 1–7, Supplementary Discussion, Supplementary Methods and Supplementary References.