What is the true social and environmental cost of electricity?

18 July 2018



The Nuclear Energy Agency believes that the market price of electricity does not accurately reflect the true cost of electricity to society and the environment. It has therefore synthesised the most recent research in a new report.


Market prices and production costs account for an important share of the overall economic impacts of electricity, but there has been growing recognition that the market value of electricity as currently constituted does not accurately reflect the cost of its production, or the wider cost to society and the environment. As a result, decisions made regarding supply and future planning based on current market prices fail to capture all of the factors needed to ensure reliable electricity supply in the future.

To try and correct this situation the Nuclear Energy Agency has synthesised the most recent research in this area in a new report The Full Costs of Electricity Provision. Adopting the approach embodied in the report should, it believes, allow policy makers and the public to make better informed decisions along the path towards fully sustainable electricity systems.

Full accounting

Full accounting remains difficult. From researching biophysical dose- response function, calibrating dispersion models and probabilistic assessments to the contentious issue of monetary valuation, different groups of experts need to be co-ordinated in large-scale multi-year efforts to arrive at robust results. Although such a systematic effort is beyond the scope of the report, the issue is too important to be disregarded. The NEA has therefore decided to summarise the most recent research in the present study.

Key concepts

The costs fall into three different categories. The first is plant-level costs, which include the concrete and steel used to build the plant, and the fuel and the manpower to run it. The NEA and the IEA publish a survey of the plant-level costs in OECD countries every five years in the Projected Costs of Generating Electricity series.

The second category concerns the costs at the level of the electricity system, linked through the transmission and distribution grid. It includes the costs that plants impose on the system in terms of extending, reinforcing or connecting to the grid, but also the costs for maintaining spinning reserves or additional dispatchable capacity when the output of some technologies – typically wind and solar photovoltaic (PV) – is uncertain or variable.

The third, broader, category includes items that impact the well- being of individuals and communities outside the electricity sector. These include the impacts of local and regional air pollution, climate change, the costs of major, frequently not fully insurable, accidents, and land use or resource depletion. Social costs also include the impacts of different power technology choices on the security of energy and electricity supply, employment and regional cohesion or on innovation and economic development. If these impacts are negative, they add to the full costs of a technology; if they are positive, in principle, they need to be deducted as a social benefit. The full costs of energy provision now include the totality of the three categories: plant-level costs of generation, grid-level system costs and the external social and environmental costs (Figure 1).

In the case of both grid-level system costs and external costs, the actors who cause them are not those who are primarily affected by them. Grid-level system costs thus have an “external” or “social” component as well. In essence, this means that an outside actor, the government, the regulator or the system operator, needs to step in to ensure that such external costs are not overproduced and are correctly internalised. 

Price concerns

Concerns about higher prices have regularly stunted internalisation efforts. However, it is the responsibility of experts and informed policy makers to insist on internalising social costs, since a reasonable degree of confidence exists that cost internalisation will improve the well-being of society as a whole. Such internalisation will need to take place at the level of the individual technology in order to induce the relevant substitution effects that will lead to an overall system that minimises the full costs of electricity provision. Where necessary, appropriate compensation mechanisms can be devised to overcome unwelcome consequences.

Accounting for full costs based on the measurement of external costs is not an uncontroversial topic. The monetisation of social costs outside a market framework can be misunderstood as an attempt to reduce human well-being to a question of dollars and cents. The large uncertainties involved, which can produce results that change considerably over time or between comparable projects, are also easy targets for detractors. Others have pointed to social factors as one of the impacts that will remain outside the scope of even very comprehensive efforts. Most of these criticisms are based on a misunderstanding of what full cost accounting is trying to achieve. Estimates established for the social cost part of the costs of electricity provision will never be able to mimic the more reliable information about individual and social preferences conveyed by market prices.

Plant-level production costs

Plant-level production costs are the smallest of the three categories indicated in Figure 1. The NEA began reporting plant-level costs in the Projected Costs of Generating Electricity series in 1983, comparing nuclear power plant and coal-fired costs. The IEA joined the NEA in publishing this report in 1989. The two agencies updated the study in later years to evaluate the levelised cost of electricity for a variety of technologies.

The LCOE indicates the discounted lifetime costs for different baseload technologies, averaged over the electricity generated. It can inform the investment choices of electric utilities in regulated systems, but it is less pertinent in deregulated electricity systems where revenues vary periodically over an electricity generator’s lifetime. LCOE is also unable to capture the system costs of certain technologies. Despite all this, it remains an attractive first reference.

Figures 2 and 3 provide estimates of plant-level costs for dispatchable and renewable power generation technologies at capital costs of 3%, 7% and 10%, assuming region-specific fuel prices, an 85% load factor for nuclear, coal and gas, as well as a carbon price of USD 30 per tonne of CO2. The latter assumes that the social costs of climate change due to carbon emissions are at least partially internalised in the policy provisions of OECD countries. With the direct carbon emissions of coal being around one tonne per MWh and those of gas around 400 kg per MWh, their respective median values would be around USD30, and USD12 lower, if strictly no efforts to reduce CO2 emissions were made.

Grid-level system costs

While system costs have always existed in unbundled electricity systems, the topic has moved into focus over the last few years with the deployment of significant amounts of variable renewable energy (VRE) sources in many OECD countries. Such system effects are often divided into the following three broad categories:

Profile costs are related to the variability of VRE output, and they are able to demonstrate that in the presence of VRE generation it is generally more expensive to provide the residual load. The overall system thus becomes more expensive even if the plant-level costs of VRE are comparable to those of dispatchable technologies.

Balancing costs are related to the uncertainty of power production due to unforeseen outages or to production forecasting errors.

Uncertainties in VRE production may also lead to an increase in ramping and cycling of conventional power plants, to inefficiencies in plant scheduling and, overall, to higher system costs.

Grid and connection costs reflect the effects on the T&D grid infrastructure due to the locational constraint of generation plants. The impacts of siting restrictions are more significant for VRE. Because of their geographic location constraint, it could be necessary to build new transmission lines or to increase the capacity of existing infrastructure (grid reinforcement) in order to transport the electricity from centres of production to load. Also, high shares of distributed PV resources may require sizeable investment in the distribution network, in particular to allow the inflow of electricity from the producer to the grid when the electricity generated exceeds demand. Connection costs (ie the costs of connecting the power plant to the nearest connecting point of the transmission grid) can also be significant, especially if distant resources have to be connected, as is sometimes the case for offshore wind. Any estimate of system costs is therefore bound by significant uncertainty and cannot be easily extrapolated to a different system or to a different context.

Figure 4 provides an example of the reconstruction of grid-level system costs for dispatchable and renewable technologies, based on a survey of the literature and the NEA study Nuclear Energy and Renewables: System Effects in Low-carbon Electricity Systems (NEA, 2012). The results continue to hold up well despite the evidence provided by the growth of variable renewables since then.

Climate change impacts

The desire to reduce greenhouse gas (GHG) emissions in order to prevent or mitigate the impacts of anthropogenic climate change has been a high priority for policy makers in many countries for the past two decades. However, this priority has not translated into an ability to quantify and monetise the impacts of fossil fuel combustion. There are three major issues in this context: i) different dimensions of uncertainty ii) discounting future impacts and iii) equity issues among different stakeholders. The multilateral process has thus chosen a different approach because of the factors mentioned above. Rather than estimating the marginal social costs, the amount of emissions considered socially optimal has been the target. Such quantitative targets can be formulated in terms of annual GHG emissions, their resulting concentration in the earth’s atmosphere or in terms of the global temperature increase that would be caused.

Air pollution

Air pollution constitutes the biggest uninternalised cost of generation. According to the World Health Organisation, it is the world’s largest single environmental health risk. WHO studies from 2014 and 2016 find that in 2012 more than 7 million deaths were caused by air pollution. About 3 million deaths are due to outdoor air pollution, to which electricity is a significant contributor, and 4.3 million deaths to household air pollution. Even if air pollution is mainly an issue in developing countries, OECD countries are also affected.

A recent OECD study estimated the social welfare loss in OECD countries owing to air pollution is well above one trillion USD, corresponding to about 3% of the gross domestic product. The most carefully studied sources of air pollution are particulate matter of different sizes, ground-level ozone (O3), sulphur oxides (SOx), nitrogen oxides (NOx) and lead. These emissions arise during the combustion of fossil fuels, coal, oil, gas or biomass, and impact primarily the respiratory system leading to bad health (morbidity) or premature death (mortality). In both cases, large uncertainties remain.

The costs of major accidents

The reported damage – not necessarily the number of fatalities – caused by both natural catastrophes and human-made accidents has continuously increased in the last three decades. Many factors have contributed to this trend and have increased the vulnerability of societies to accidents and catastrophe hazards: growth of the population and the global economy, industrialisation, urbanisation and development of coastal and other risk-prone areas, as well as the growth of more complex and interrelated infrastructures.

Better reporting may also have contributed to such vulnerability. Natural catastrophes impose the largest toll in terms of human fatalities and economic consequences. If only human-made accidents are considered, the energy sector is the second-largest contributor, with transportation causing about 60% of all mortalities. For all energy technologies, however, the external costs associated with severe accidents are several orders of magnitude lower than those caused during normal operation from pollution and carbon emissions.

Risks of severe accidents in all energy chains should not be neglected, however, as they have the potential to cause large-scale and long-term impacts to human health and to the environment.

Land-use change and natural resource depletion

Although electricity generation can have large and lasting impacts on the land an recources it uses, the exact nature of land-use change is largely site and technology-specific. The most significant external cost of land-use changes are the effects on the ecosystems of natural areas. Most electricity sources have significant land requirements when the whole fuel cycle is considered, including fuel extraction, generation and waste disposal. The fuel that has the highest by far land-use requirements is biomass. Land use is part of the larger category of natural resource use, which includes water pollution and natural resource depletion. Despite these concerns, the depletion of non-renewable resources, such as fossil fuels and uranium, should not be a major consideration in policy making.

The security of energy and electricity supply

Unsurprisingly, governments of many countries are concerned with understanding the factors influencing the security of energy and electricity supplies and are seeking to develop policy frameworks and strategies to enhance them. Discussions about energy supply security have for a long time lacked meaningful quantification.

An indicator of the security of supply for OECD countries over 40 years was thus developed by the NEA – the simplified supply and demand index or SSDI. The SSDI shows a remarkable improvement of the security of energy supplies for the great majority of OECD countries over the 40-year time frame of the study. The value of the SSDI significantly increased between 1970 and 2007 in most economies in the study: Australia, Canada, Finland, France, Japan, the Netherlands, Sweden, the United Kingdom and the United States. This improvement resulted from the introduction of nuclear power for electricity generation, decreasing energy intensity and increased diversification of imported fuels such as coal, oil and gas. In general, all low carbon technologies such as nuclear energy, hydro, wind and solar possess a number of attractive characteristics in terms of external energy supply security. They differ, however, with respect to the contribution to the internal or technical security of supply, in particular in electricity systems. Governments should thus create frameworks that allow all low-carbon technologies to make their contribution to the security of energy supplies.

Employment generated in the electricity sector

Since the employment required for different technologies in competitive labour markets is the result of competitive, firm cost minimisation, one might ask why employment should be considered as a positive externality. In addition to constituting an economic cost, it is because high employment rates can contribute to social cohesion and general well-being at the societal level. From this perspective, not only the quantity but also the quality of the labour that is required by different technologies should be taken into consideration. If operations and manufacturing are included, indications are that nuclear power is more labour-intensive than other forms of generation, has higher education requirements and generally pays higher salaries.

The impact of energy innovation

Technological change in the energy sector contributes to the macroeconomy in terms of i) value added, income and employment, ii) the functioning of the economy, firms and households that are dependent on cheap and reliable energy supply, and iii) the waves of innovation and the spillovers that are generated on both the supply and demand sides, which constitute the principal reason why governments fund basic research and development (R&D) in energy.

Trends in R&D funding have changed remarkably. Since 2000, the public budget for R&D on renewables has been multiplied by five, and for energy efficiency by two. For nuclear energy, there has been a sharp decrease from about USD 8 billion per year in 1980, largely for fission, to less than 3 billion today, with fusion now taking the bigger part (EC, 2016a). R&D funding is often most successful if combined with other instruments. In climate change policy, for instance, pollution pricing should be complemented with specific support for clean innovation (e.g. through additional R&D subsidies).

Promising new clean technologies deserve the highest possible attention in terms of policy support, even if this would mean reducing R&D support targeted on improving existing dirty technologies. Policies should thus support a wide range of low-carbon technologies, as no one, single silver bullet exists.

Large uninternalised costs

Air pollution, climate change and system costs constitute the largest uninternalised costs. The different chapters in this report converge on one single insight: the external costs of the normal operations of electricity generation exceed the costs of other phases of the life cycle of electricity generation – upstream or downstream of operations – as well as the costs of major accidents by at least one order of magnitude. In terms of the back end of the life cycle, decommissioning and the storage of waste constitute significant costs for nuclear power indeed. However, these are economic costs, for which provisions exist to be internalised through the funds that are constituted by electricity producers and that are passed on in customer prices and tariffs.

Major accidents of energy structures are fortunately rare during the life cycle of all power generation technologies and thus do not figure heavily in the accounting of full costs. The problem for policy making is, of course, that such accidents receive an extraordinary amount of attention from the media and the general public. The greatest number of fatalities is recorded in coal mining and hydroelectricity, two technologies which do not generate widespread public concerns.

Air pollution constitutes the biggest uninternalised cost of electricity generation. Worldwide, the deaths of 3 million people per year are attributed to ambient air pollution, of which power generation contributes a significant share. The full costs of climate change come with high uncertainties but are routinely characterised by analysts to be in the trillions of US dollars or euros. Climate change action has a unique role in this context. Public awareness, media focus and political attention are intense, but have failed thus far to translate into effective GHG emission reductions.

Policy makers

The public, the media and policy makers are prone to attention bias. An accident with 50 fatalities once every ten years will get infinitely more media and policy attention than 1 000 premature deaths coupled with increased morbidity in a large population because of a constant level of pollution over the same time span. While individual human suffering cannot be calculated and compared, dispassionate reflection with an aim to improve general welfare would suggest that the far larger number of casualties due to air pollution would demand at least as much attention as rare accidents. Policy instruments that should be considered fall into three broad categories:

1. Price- and market-based measures such as taxes, prices, subsidies, the allocation of property rights and market creation.

2. Norms, standards and regulations, which are the default measure of policy making.

3. Information-based measures, including R&D support, are not minor add-ons but are at the heart of internalisation. Whatever the chosen instrument, governments must be the primary driver behind implementation. 

NEA Figure 1. Different cost categories that make up the full costs of electricity provision (Source: NEA 2012)
NEA Figure 2. Plant-level costs for different power generation technologies (USD per MWh)
NEA Figure 3. Plant-level costs for despatchable power generation technologies (USD per MWh)
NEA Figure 4. Grid-level system costs of selected generation technologies for shares of 10% and 30% of VRE generation


Linkedin Linkedin   
Privacy Policy
We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.