How to Finance the Energy Transition

The climate policies aimed at achieving the ambitious goal of stabilising global warming at 2°C, stipulated in the 2015 Paris Agreement, are going to cost a lot of money. To finance sustainable solutions, abundant private savings need to be redirected towards long-term investment in infrastructure, encouraged by the public authorities.


For India, which builds most of its infrastructure, a carbon tax of € 50 per tonne would double the price of cement.


The political crisis of the “yellow vests” illustrates the consequences of a climate policy that neglects its inclusion in a coherent economic and social policy. Any systemic transformation of this type (which goes beyond sector technical decisions, such as the debate on the use of nuclear for electricity), requires a new social contract. Ultimately, the cost of climate policies will depend on the quality of this contract.

To understand the challenge, let us point out that the latest Intergovernmental Panel on Climate Change (IPCC) report states that respecting an emission scenario that limits the temperature increase to 1.5°C, requires the use of energy production, transport, infrastructure manufacture techniques, etc. that are three to four times more costly per tonne of avoided carbon than in the 2°C limit scenario. The cost of the last tonne emitted (the “marginal cost” for economists), is expected to be equal to almost €400 per tonne of carbon emitted (for current values) in all countries around the world. A low tax of €50 per tonne would double the price of cement alone (the manufacture of which emits a great deal of CO2) in India, a country which builds most of its infrastructure. For this reason, the economic analysis devoted to this issue since the second IPCC report in 1995, looked at the processes that can be used to minimise the economic and social costs of climate policies, and even transform them into joint benefits.

Rapid decarbonisation of the economy requires low-carbon techniques that are as cheap as possible. This does not seem so difficult if we heed the very optimistic statements on falls in the cost of solar energy, wind turbines and efficient heating systems. However, the notion of cost of a technology is misleading. What needs to be taken into account, is the cost of a technical system, with its complementarities and inertia.

Thus, the cost of a low-carbon source of electricity (nuclear, photovoltaic, wind or hydraulic) depends on how much it is used over the course of one year. The contribution of an intermittent energy is a result of the architecture of the whole system (production, high-voltage transmission, distribution, storage capacities, etc.). This system cannot easily be changed however; once built, it cannot be changed overnight. Any acceleration in this change will thus lead to additional costs, independent of the cost of the low-carbon technologies themselves, as we will be doing without a cheap short-term production capacity, already amortised, and the costs generated by the change have to be included. As such, the closure of the nuclear power station in Fessenheim or changing the high-voltage grid in Germany to transfer production of offshore wind turbines in the north to the south of the country, involve real costs.


These types of constraints are studied by means of national or planetary economic models optimised over time (from today until 2040, for example) that mimic the behaviour of a planner (supposedly all-powerful and benevolent) or of perfect markets (invisible hands able to allocate resources optimally). The situation is different in the transport and mobility sector, where economists more often use local simulations. For a set of varied hypotheses, these models reveal the marginal and the average cost imposed on the technical systems, at any time, by a given CO2 emissions reduction constraint.

The results are guided by technical realities. Let us suppose that country A has an energy system where any drop in emissions costs €10/t because they just need to use a technical solution more, for which that is the additional cost. In this case, the total cost to reduce emissions by 100 tonnes will be 10 x 100 = €1,000 and the average and marginal costs will both be equal to €10. Compare this with another country B which has low-carbon techniques costing €5/t, but of limited potential (hydraulic in France). They can only reduce emissions by 90 tonnes because, beyond that, their expansion capacity is saturated (hydraulic sites). If the last 10 tonnes cost €20 to reduce, the overall bill will be 90 x 5 + 10 x 20 = €650, lower than the previous €1,000, with an average cost of €6.50 instead of €10 per tonne avoided. But (and this is why economists are drawing attention to the marginal cost), it is not certain that the overall cost for country B is more favourable, as everything will depend on how this marginal cost will affect the economy.

The marginal cost of avoided carbon is a decisive indicator; this is the additional cost producers will bill their customers, in the same way as oil producers sell theirs at the market price, whatever the cost of production, real or inferior. This cost will have an impact in all sectors of the economy, from energy to cement, steel to fertilizer, from the production of buildings to cars and agriculture… The propagation effect will raise the price of consumer goods, lower purchasing power and worsen unemployment, because made in France goods will be more expensive than imported goods. It is better that the effect is on €10 than on €20. These mechanisms are described well in “general equilibrium” models (see insert) that represent all economic flows respecting their accounting coherence, used to study the systemic consequences of a price system change and, for some, technical changes. This is the type of model that can be used to study the consequences of an oil shock on the whole economy (public and trade deficits, effect on salaries).

Added to the cost propagation is the issue of income redistribution. In our example, country A will make no additional profit. The producer in country B will bill an additional €20 on all his products; for a constant level of sales, he will receive additional income of €2,000 although his additional costs are only €650. Some of the additional €1,350 profit may be considered necessary for the technical reconversion of the industry, but another part is pure income taken to the detriment of downstream sectors, hence the risk of non-optimal allocation of collective resources. The money that was to have been used for energy transition is deviated from its goal.

Let’s start with a simple principle: it is better not to remove a tonne of carbon that costs €150 when we can reduce the same tonne elsewhere for €20. It is important to equalise the marginal costs of carbon abatement. In the absence of an omniscient central planner, there are two means to achieve this goal using a “price signal” by a tax on tonne of carbon emitted, or by negotiable emissions permits sold by auction (this more complicated process can fail if the freely allocated part is too large, as is the case of the European emission quota exchange system). By doing so, the tax has a dual advantage: preventing the formation of income in the hands of polluting activities (those arising from the difference between the marginal cost and the average cost) and raising income to reduce the most penalising taxes for the economy. It is indeed to minimise the economic and social costs of energy price propagation that for a quarter of a century, economic analysis has proposed a carbon tax that is equal to the marginal cost of avoided carbon emissions.

The priority consists of offsetting any carbon tax by a similar fall in another tax that contributes to production costs. In French conditions, this means reducing the gap between gross salary and net salary by lowering social contributions where there is obvious room for manoeuvre. Another part of the tax could be allocated to reducing VAT on essential products to offset its negative effects on the lowest incomes. This is how ecological tax reform can include loss of population well-being for the lowest earners and the budgets most affected by inescapable expenditures (housing, transport, food). The “yellow vest” conflict confirms the political risk of separating ecological tax and social justice. In the 1990s, economic literature showed how, by substituting economy-dampening taxes with a carbon tax, a double dividend could be created, i.e. an increase in income and in employment. This literature shows that such a result, which is based heavily on reducing the “oil bill” of private individuals and companies in France, is not automatic, as it arises from very specific implementation conditions. They must include tax reduction for a part of emissions to enable the most energy-intensive industries to adapt (steel, cement) and additional compensation for the most vulnerable social classes (low-income and fringe populations). There is a consensus however, in all of the IPCC group 3 reports, on the need to outweigh, compensate rising energy costs by transforming tax systems.

The effectiveness and acceptability of this transformation depends on the consultation mechanisms implemented. Recent political debate in France has got it the wrong way around; raise a tax and, then, in a bid for justice, redistribute the income by means of compensatory transfers. This approach is based on the idea that it is possible to separate the effectiveness of a pollution-abatement policy and the fair distribution that goes along with it. It only applies in very specific conditions however; at the margin of a given economy, where the way income is redistributed does not interfere with its structure. Offsetting a fuel tax with a ‘green cheque’ for populations living outside the city centres would only aggravate urban expansion if the rise in house prices is not changed. Reducing payroll taxes can only be effective after negotiations between social partners. Otherwise, if it is only companies who benefit from the drop, employees’ purchasing power will fall and lead to a net loss, rather than a double dividend.

The cost of transition to a low-carbon world will depend on the transformation of technical systems and their global cost. Current economic models, used to simulated future decarbonisation pathways, describe this transformation using parameters that are mostly fixed on the costs of low-carbon technologies. These models are made either by linking these parameters to investment in research and development, or by simulating learning by doing mechanisms (the more a technique is used, the more its cost decreases). Today, we have more than 1,000 economic scenarios covering a universe of possibilities. They all require significant investment. In France, this investment is insufficient. Until the middle of the century, investment in energy systems alone should go from today’s 2% to 2.8% of world gross domestic product. If we take all investment in infrastructure (including transport, housing and investment upstream of industry, especially for material transformation), we arrive at incremental investment needs that amount to 2.5% of gross fixed capital formation by 2035, or on average 650 billion per year.

Added to this need for investment in production equipment and capacity, are the so-called transaction costs (organisation, negotiation, lifting of legal constraints). In many African countries, the latter represent an amount equal to equipment purchase and lowering them depends on solving institutional and legislative problems. This is the case of plans for the photovoltaic sector in Western Africa, with the negotiation of contracts, the search for funding and even the remuneration of permits by the political authorities.


The availability of sufficient financial resources is decisive in reducing costs. These resources cannot be provided by carbon tax income, as it must be used to minimise the effects of energy costs and offset the loss of purchasing power for vulnerable populations.
But this need for funding could be a false problem. Funding the transition runs up against no quantitative constraint; it is established that one of the problems in the global economy is over-abundant savings that flit from market to market looking for the fastest maximum profit. The current financial system (banks, pension funds, insurance, bonds markets) does not know how to redirect these savings into long-term infrastructure investment likely to enable the transition to a low-carbon economy. The IPCC has nevertheless evaluated the required redirection at between 5.6% and 8.3% of current annual capital income (rise in capital value and interest received), in other words quite a low proportion of financial flows.

How can this redirection come about? It requires public policies to lower the level of risk of investment in low-carbon infrastructure. One possibility is in the form of public guarantees that weigh less on public budgets as they are only exercised in the event of project failure. If savings could be redirected in this way, climate transition would not be a cost but a lever to reduce the current deficit in productive investment in favour of short-term property and asset investments and to sort the global financial system out of the short-sighted behaviours  that led to the 2008 subprime crisis.

Albert Camus once said that to misname things was to add to the woes of the world. That is the case here. By juggling lightly with the notions of the cost of a technique, a system, the cost of well-being and the cost of investment, we are forbiding ourselves from understanding how to transform a real and unavoidable technical cost into a collective benefit an effort that, like when you stick to a regular run results in a health benefit. Instead we fail to detect obstacles to overcome for a successful low-carbon transition.




The most recent integrated economic evaluation models rely on the meeting of two traditions. On the one hand, energy prospective models developed in the 1970s (from optimisation models). Quentin Perrier, an economist at the CIRED, proposes an energy application to test for the optimal electricity mix according to parameters chosen by the user (another class of model is based on simulations that do not postulate optimal behaviours). On the other hand, is the tradition of general equilibrium models that hark back to the Norwegian economist Leif Johansen in 1960. They require the construction of social accounting matrices used to coherently enter the value streams in an economy, by imposing an accounting balance between lines and columns without supposing that these accounting balances are optimal. In the 1990s, the explosion of computing capabilities enabled the use of “Computable General Equilibrium” models which, for reasons of technical simplicity mostly supposed equilibrium in markets and remuneration of production factors (labour, capital, raw materials) at their marginal productivity. This simplification supposes economies that are close to their “optimal equilibrium”, making it difficult to take account of economies that are out in disequilibrium. Since the 2000s, a new class of “hybrid” models has emerged. These hybrid models include engineering data (energy, transport, land occupation) in general equilibrium models, and do not hypothesize on perfect markets or optimal growth.




Jean-Charles Hourcade
Jean-Charles Hourcade is emeritus director of research at the CNRS and emeritus director of study at the EHESS. He has been Conviening Lead Author for group 3 of the IPCC.

Share on Facebook
Share on Twitter
Please reload


November 11, 2018

November 11, 2018

Please reload