Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Business as unusual

Energy models provide powerful insights for decision-makers, but more care is needed around the choice of reference scenarios and the transparency of assumptions.

The future is unwritten. While this creates optimism for positive change, it isn't very helpful for making decisions about future energy systems. Governments need to understand how to meet policy goals, the possible effects of policy decisions and the impacts of economic trends. Businesses need to make investment decisions over different time horizons that ensure they still exist into the future. Academics need to inform both groups using the best evidence they can obtain. If we set ourselves specific goals then we had better have some ideas about how to achieve them, and how they will impact on the different facets of our lives.

Energy models thus play a vital role as crystal balls through which we can make predictions about energy systems. To do this, they have become increasingly complex and sophisticated. Yet their mounting centrality to decision-making makes it important to stop and think about what energy models can — and cannot — tell us.

A key component of any energy modelling exercise is the central reference case. This typically takes the form of the ‘business as usual’ scenario, in which no low-carbon policies are adopted and we carry on as we did in the past. However, low-carbon policies are growing in number and ambition and events in recent years have unfolded rapidly and unexpectedly, so that the past no longer feels like a sound template to work from. Arguably, therefore, it's hard to justify the use of a business-as-usual framing any more.

Yet, reference scenarios are useful to benchmark model outputs (and different models) against one another, and to understand the impacts of policies relative to counterfactual cases without them. It's useful to think about the framing of the questions being asked of energy models. Having decided, based on scientific evidence, to make changes to limit global warming to well below 2 °C, should we continue to countenance scenarios that do not work towards that goal? Shouldn't our reference points be only those other scenarios trying to meet the same goal but by different means? Choosing a reference point that takes account of where we find ourselves today offers a more realistic way to interpret the alternatives and make sound choices. This is the approach adopted in a recent report from Carbon Tracker and the Grantham Institute for Climate Change and the Environment, which takes the Nationally Determined Contributions of the Paris Agreement as the starting point for its analysis of future emissions and energy technology pathways1. It also formed the basis of a workshop they organised2 in July 2017.

Model inputs must also be up to date. Energy models that don't take account of rapidly changing capacity additions or price changes in the last few years, or that have inaccurate learning rates for different technologies, will become quickly outdated. They will also underestimate or over-estimate the potential of different energy technologies, which risks presenting too pessimistic or too optimistic an outlook, leading to a lack of preparedness that could ultimately prove more costly. Indeed, as discussed by Felix Creutzig and colleagues in this issue (article no. 17140), the high deployment rate of photovoltaics in the last decade or so exceeded that predicted by many models, sometimes within a few years of the model's publication.

If energy models quickly become outdated even with recent data, then perhaps we should examine the range of possible futures being considered. As the future is inherently unknowable, it's easy to be conservative when thinking ahead. BP projected3 that, despite significant growth, electric vehicles would still only account for 100 million cars (6%) in the global fleet by 2035. Using their updated baseline and technology data, the Carbon Tracker and Grantham Institute study projected a share of 35% instead. Both predictions could be forgiven for not foreseeing that France4 and the UK5 would ban sales of new petrol and diesel cars by 2040, or that Volvo would announce6 that all their new cars would be at least partially battery powered from 2019. But decisions like these will already be changing the future vehicle fleet. If disruption and change are part of the fabric of the energy system now, then we should consider a broader scope of possibilities for what may come to pass and think carefully about what we predict is most likely to happen.

Assumptions also play a critical role in energy models. Models are built by different groups to serve different needs. Like any crystal ball, the material they’re made from and its imperfections will refract, skew, and obscure the future differently. Given the energy system's inherent complexity, energy models have to make assumptions and simplifications at different levels of temporal or geographical scale, aggregation, and understanding about specific elements. In the integrated assessment modelling community, establishment of the shared socioeconomic pathways has attempted to harmonize the socioeconomic assumptions going into models. Yet, recent debates in the energy systems modelling literature79 have discussed the transparency and feasibility of assumptions used there. Defining what is feasible warrants significant further discussion and establishment of common criteria. Nonetheless, it's clear that transparency of assumptions is an essential part of presenting any set of energy scenarios. Users and readers need to understand what assumptions and simplifications have been made so that they can properly grasp the limitations of the outputs. Towards that end, we encourage authors to include clear descriptions of assumptions in the main text of their paper rather than in the Supplementary Information.

Energy models are a valuable way to try to understand the future energy system and make informed decisions. Their growing sophistication suggests the insights to be gleaned from them will only increase in time. However, through greater consideration and discussion of reference points, data updates, and assumptions, among other factors, we can add yet more power to their predictions.


  1. 1

    Expect the Unexpected: The Disruptive Power of Low-Carbon Technology (Carbon Tracker, Grantham Institute, 2017).

  2. 2

    Business as Usual Energy Modelling Workshop: Note of Discussion (Grantham Institute, 2017);

  3. 3

    BP Energy Outlook: 2017 Edition (BP, 2017);

  4. 4

    Chrisafis, A. & Vaughn, A. France to ban sales of petrol and diesel cars by 2040. The Guardian (6 July 2017);

  5. 5

    Castle, S. Britain to ban new diesel and gas cars by 2040. The New York Times (26 July 2017);

  6. 6

    Vaughan, A. All Volvo cars to be electric or hybrid from 2019. The Guardian (5 July 2017);

  7. 7

    Clack, C. T. M. et al. Proc. Natl Acad. Sci. USA 114, 6722–6727 (2017).

    Article  Google Scholar 

  8. 8

    Jacobson, M. Z., Delucchi, M. A., Cameron, M. A. & Frew, B. A. Proc. Natl Acad. Sci. USA 114, E5021–E5023 (2017).

    Article  Google Scholar 

  9. 9

    Heard, B. P., Brook, B. W., Wigley, T. M. L. & Bradshaw, C. J. A. Renew. Sustain. Energy Rev. 76, 1122–1133 (2017).

    Article  Google Scholar 

Download references

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Business as unusual. Nat Energy 2, 17150 (2017).

Download citation


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing