We often talk about climate inequality as a question of finance: who receives climate finance, who can afford clean technologies, and who is protected from the impacts of a changing climate.
But there is another, less visible question that comes earlier: how do we decide what the future energy system should look like?
Governments increasingly use energy-system models to help answer that question. Put simply, these are tools for exploring how future energy demand could be met under different assumptions about technologies, costs, infrastructure and policy. Some identify combinations of power generation, grids, storage and other technologies that minimise overall system costs. Others also represent markets and the behaviour of consumers, firms and investors. However, it is important to remember that a model can only properly compare the technologies, characteristics and choices that it has been designed to represent.
For instance, The European Commission uses the PRIMES energy-system model extensively for its long-term energy projections, reference scenarios and policy impact assessments. PRIMES represents energy demand and supply, investment, technologies, prices and policy measures across individual European countries. Türkiye’s 2022 National Energy Plan uses the Türkiye Energy Model, developed for the Ministry of Energy and Natural Resources. The model makes investment decisions with the objective of minimizing total system cost. The plan projects, among other changes, an increase in solar capacity to 52.9 GW by 2035. However, public documentation does not make clear whether rooftop, prosumer and utility-scale PV are separately represented in the underlying optimization.
When governments adopt sophisticated models to optimize their transitions, the open question is whether those models are detailed enough to compare the full range of energy-system choices relevant to that country.
A recent study examined 60 studies of highly renewable energy systems and found wide variation in the future role assigned to solar PV, identifying both outdated techno-economic assumptions and simplified modelling choices as reasons why solar can be underestimated. Around one third of the studies reviewed represented only a generic solar PV plant. Technologies and configurations such as tracking PV, floating PV, agrivoltaics, prosumer systems and building- or vehicle-integrated PV were much less frequently represented. This diversity matters because solar installed on a rooftop does not necessarily interact with the energy system in the same way as a large solar farm. Distributed generation can supply electricity closer to demand, and depending on the system, this can reduce the need for network expansion and change the overall economics of the energy system.
Another study finds that excluding rooftop PV and prosumers can lead models towards higher-cost system solutions, because some of their system value is not captured. A model that treats all solar as one generic technology may therefore be simpler to operate, but that simplicity can matter for the conclusions it produces.
On its own, incomplete representation of distributed solar does not prove that energy-system modelling causes inequality. But the connection is worth tracing.
Considering decentralized electrification
More than 675 million people globally remain without electricity, while hundreds of millions more experience unreliable supply. A recent review of progress towards SDG 7 highlights decentralized renewable systems, including microgrids and distributed solar, as important options for communities poorly served by conventional electricity networks — estimating that decentralized solar systems could provide electricity to around 41% of the currently unelectrified population.
Rising temperatures are increasing electricity demand just as many countries are trying to expand electricity access and decarbonize their power systems. The SDG 7 review reports that only around 15–25% of low-income households in Asia and Africa have access to air conditioning, compared with around 75% of more affluent households — a gap that leaves many without access to safe indoor temperatures during extreme heat.
A 2023 global compendium by the Council on Energy, Environment and Water reports that 179 million people gained electricity access through decentralized renewable energy solutions in 2021, compared with 35 million in 2012. It documents applications spanning homes, agriculture, healthcare, education and productive uses — including cases where decentralized solar has supported healthcare facilities with cold storage for medicines, vaccines and diagnostic materials.
When we look at the examples around the world it becomes clear that in some places, extending the national grid will be the most effective solution. Elsewhere, a mini-grid, rooftop solar, distributed storage or a combination of centralized and decentralized infrastructure may provide reliable electricity more quickly or at lower total cost. If an energy-system model represents one of those options poorly, the comparison may already be tilted before an investment decision is made.
Representing decentralized technologies more accurately will not, by itself, produce an equitable energy transition. The CEEW review identifies enabling policy, community participation, accessible finance, functioning local markets and innovation as recurring conditions for successful deployment. The SDG 7 literature similarly stresses that financing, governance, technical capacity and social inclusion determine whether decentralized technologies actually improve people’s lives. Many of these factors are difficult to capture in a conventional energy optimisation model.
Energy-system models are enormously useful because they allow governments to compare complex combinations of technologies and infrastructure over decades. But their apparent precision should not obscure the assumptions underneath them. A model can optimise only the choices available inside it.
What should we be asking models to see?
A model designed for the energy transition should certainly ask where electricity can be generated at low cost. But increasingly it may also need to distinguish:
• where generation is located;
• how it interacts with transmission and distribution networks;
• whether generation takes place on land, buildings or other surfaces;
• how consumers themselves invest in and operate energy technologies;
• whether distributed generation changes network investment;
• and which energy services different system configurations can realistically provide.
None of this means modelling every rooftop, farmer or household individually. It means preserving enough of the differences between technologies and system configurations for a model to compare them meaningfully. The people, places and technologies that have historically sat at the margins of the electricity system should not remain at the margins of the models used to design its future.
