This IMF working paper claims that the fossil fuel industry benefits from trillions of dollars of government subsidies every year. It is a working paper intended to "promote discussion" and not the official view of the IMF.

The paper has been widely quoted by climate campaigners who often report it uncritically and describe it as the official IMF view. For example, this BMJ article arguing that medics should be more active in climate lobbying says:

Global fossil fuel subsidies of $5.3tr (£4tr; €4.6tr) annually amount to subsidies on suffering and must stop.

Another BMJ piece argues again quotes the study uncritically:

Governments and taxpayers across the world continue to subsidise fossil fuels to the tune of trillions of dollars every year.

The Guardian reports the study:

Fossil fuel companies are benefiting from global subsidies of $5.3tn (£3.4tn) a year, equivalent to $10m a minute every day, according to a startling new estimate by the International Monetary Fund.

The uncritical reporting of the number is a little strange as it is the same magnitude as the entire industry's global annual revenue according to this estimate. And some commentators have been very critical as this Forbes headline argues:

The IMF's Absurd Calculations Of Fossil Fuel Subsidies: It's Really Not $2,200 For Every American

Is the working paper's conclusion consistent with available evidence?

  • 1
    To the commenters: we don't care about your political opinions. We especially don't care if you agree with environmental economics in this case. Further political comments will be nuked without notice.
    – Sklivvz
    Nov 14, 2017 at 7:15
  • OP, I've clarified that the question is about facts. The original version was being interpreted incorrectly as asking for opinions as evidenced by the dozens of political comments which I've had to nuke and the close vote.
    – Sklivvz
    Nov 14, 2017 at 7:19
  • 2
    @Sklivvz While I'm sure many of the responses were correct closed I think we might also have lost some clarifications that would help explain why the IMF paper was almost universally misinterpreted by the media (especially clarifications about the definition of subsidy in environmental economic which appears to be a little odd).
    – matt_black
    Nov 14, 2017 at 9:13

2 Answers 2


Yes, the fossil-fuel industry benefits from subsidies equivalent to trillions of dollars, each year. Yes, the IMF's figures are broadly consistent with other studies: in particular, the cost of environmental damage from greenhouse gases alone, is of the order of trillions of dollars per year. These are costs which production incurs, but which are not borne directly by producers, and thus not reflected in the price. Environmental economics - for example within the IMF study quoted - treats these indirect subsidies as subsidies.

The industry benefits in several different ways. Merely taking quotes out of context, as the question does, hides the specific nature of the claim. However, the full report from the International Monetary Fund is explicit about the range of subsidies it considers:

the definition of subsidy

Here's what that report says in its abstract (my emphasis):

it focuses on the broad notion of post-tax energy subsidies, which arise when consumer prices are below supply costs plus a tax to reflect environmental damage and an additional tax applied to all consumption goods to raise government revenues.

It goes on further on pages 10-11 to define subsidy in more detail.

In particular:

post-tax consumer subsidies represent the amount by which the cost borne by the consumer falls short of the total economic cost of consumption. This excess cost (or subsidy) is either covered by governments in the form of budgetary support or foregone revenues or passed to the society in the form of environmental damage.


Producer subsidies exist when producers receive either direct or indirect support that increases profitability above what it otherwise would be (that is, the support is not passed forward in the form of lower consumer prices). This support can take many forms, including receiving a price for the output above the supply cost, paying a price for inputs below supply costs, receiving preferential tax treatment, or receiving a direct transfer from the budget

This is consistent with how the term is generally used within the field of Environmental Economics.

See, for example: The Economic Cost of Global Fuel Subsidies, Davis (2014):

Fuel subsidies are different from subsidies in most other markets because of the substantial external costs of driving.

And Energy subsidies: How large are they and how can they be reformed?, Clements et al 2014:

The subsidy estimates capture both those that are explicitly recorded in the budget and those that are implicit and off budget ... Tax subsidies arise when energy taxation, which includes a corrective component for externalities and a component for revenue consideration, is below the efficient level. The efficient taxation of energy requires corrective taxes to capture negative environmental and other externalities owing to energy use.

And Fossil fuel subsidies in the Pacific island context: Analysis of the case of Kiribati, Peltovuori (2017)

Subsidy in this report is defined according to the definition of World Trade Organisation's (WTO) Agreement on Countervailing Measures (ASCM) (Uruguay Round Agreements, 1994). This definition has been recommended to G-20 by the Global Subsidies Initiative (The Global Subsidies Initiative, 2010). The methodology for quantifying fuel subsidies in this study is the price gap approach. In the price gap approach, a cost recovery price for fuels is first estimated based on the world free on board (FOB) market prices. All additional monetary and non-monetary costs such as transport, storage, distribution costs, and corrective taxes to account for externalities are then added on top of the FOB price. This cost recovery price is then compared to the actual price. The gap between the two prices is the amount of subsidy per price unit.

These pollution costs are included within the broader definition of subsidy as used by the IMF, because these negative production externalities are costs incurred by production, that the producer itself does not directly bear. Thus, those costs are not reflected in prices, leading to market failure and an unfair competitive advantage, of the same kind and market impact as if they'd received a cash subsidy on their directly-incurred costs.

The "Economics Help" site explains the issue

enter image description here

The deadweight loss from coal pollution is the red striped triangle: this comes about because coal producers do not have to pay for their pollution, leading to a competitive advantage, and an artificially high quantity sold Q1 at price P1.

consumer subsidies

The International Energy Agency looked at 40 developing countries (rather than the IMF which looked at all countries), and found the consumer-side domestic price subsidies (which are only a fraction of total subsidies) to be USD 325bn in 2015. The year before, that figure had been USD 500 bn.

producer subsidies

The OECD looked at 41 countries, (34 OECD developed countries, plus BRICS + 2) and looked at some of the producer-side subsidies, finding them to be in the range USD 160-240 billion per year, 2010-2014

environmental impact

Global carbon emissions from fossil fuel in 2007 were estimated by Lawrence-Berkeley Labs at 7988 MtC (megatonnes of Carbon) (=8365 minus 377 from cement).

Leading climate impacts economist Chris Hope estimates the social cost of CO2 at around USD106 per tonne CO2 (tCO2), and one tonne of carbon (tC) is equivalent to 3.67 tCO2. Together, these put the environmental impact of CO2 from fossil-fuel emissions at:

7988 million tC/y x 3.67 tCO2/tC x USD106/tCO2 = USD3tn/y.

in summary

The IMF's definition of subsidy was the broadest, including the pollution damage costs, as well as covering all countries, and so is larger than simply the sum of producer- and consumer- subsidies for subset of countries plus some of the environmental impacts.

So, in total:

USD 160-240 bn/y: some of the producer subsidies (OECD)
+ USD 325-500 bn/y: some of the consumer subsidies (IEA)
+ USD 3tn/y: some of the environmental impact (Chris Hope / LBL)

So the IMF estimate of USD 5tn/y (all countries, all producer and consumer subsidies plus all environmental impact) is consistent with this.

The size of the greenhouse-gas impacts alone is in the trillions. This means that the lobbyist's opinion-piece in Forbes is irrelevant to the claim in question: the aspects of the subsidy included, to which the lobbyist objects to, are tiny compared to the environmental impacts - which even they agree are valid.

  • 5
    To the commenters: we don't care about your political opinions. We especially don't care if you agree with environmental economics in this case. Further political comments will be nuked without notice.
    – Sklivvz
    Nov 14, 2017 at 7:13
  • Having not read the opinion-piece on Forbes; what was the objected aspect of the subsidy?
    – Communisty
    Nov 17, 2017 at 9:27
  • 1
    As a point of clarification, the 3T$ figure is considered a subsidy in this analysis because it is a cost the company incurs that the government does not require it to pay for, and removing a cost is equivalent to a payment, right? Nov 19, 2017 at 22:41
  • 1
    @eyeballfrog that's right. As I wrote above: "Thus, those costs are not reflected in prices, leading to market failure and an unfair competitive advantage, of the same kind and market impact as if they'd received a cash subsidy on their directly-incurred costs"
    – 410 gone
    Nov 20, 2017 at 6:43
  • Meta question about this
    – user22865
    Nov 29, 2017 at 8:55

Among the references cited in that IMF paoer, I found the 2015 ECFIN brief "Measuring Fossil Fuel Subsidies" by Bárány and Grigonytė fairly illuminating in this matter. (To establish authority, ECFIN briefs "are occasional working papers by the European Commission’s Directorate-General for Economic and Financial Affairs which provide background to policy discussions".) Note to moderators: the source explicitly allows reproduction (hence the large quotes) as long as its origin is acknowledged. All bold/emphasis is mine, added for the TLDR people.

Three international organisations (the IEA, the IMF and the OECD) collect data on fossil fuel subsidies in a systematic way, albeit with different methodologies. The International Energy Agency (IEA) provides estimates annually of consumer FFSs for 40 developing countries, including the world's top subsidisers. They are calculated using the pricegap approach, that is, based on the differential between the end user price of a specific fossil fuel and a reference price (the international market price adjusted for transport and distribution costs) of the same fuel. The Organisation for Economic Co-operation and Development (OECD) uses a completely different, inventory based approach to estimate the value of FFSs in its member states. This method identifies all government measures (subsidies and tax breaks) that support fossil fuel production or consumption, and calculates and adds up the value of all these measures based on the government's budget. The International Monetary Fund's (IMF) study provides the most comprehensive pre-tax and post-tax subsidy estimates for 176 countries. Pre-tax subsidies are mostly based on the price-gap approach, and are therefore similar to IEA estimates (although for some OECD countries, producer subsidies are also included). Post-tax subsidies include the negative externalities associated with the use of fossil fuels (that are not internalised through corrective environmental taxes by the government), such as local air pollution, faster climate change and congestion. [...]

Comparison of the different methodologies

The IMF approach is unique in the sense that it considers the inefficient taxation of fossil fuels as subsidisation. Thus the government's failure to deal with a market failure (such as the negative externalities associated with fossil fuel consumption) is itself a form of subsidisation according to the IMF. This approach is the most logical economically, as inefficient taxation (either not taxing fossil fuels enough to control for negative externalities, or taxing energy differently than other consumer products) is just a hidden subsidisation of fossil fuels. As the IMF concept of FFS is broader than that of the IEA or the OECD, the IMF estimates tend to be higher than the estimates of the other two organisations.

As you can see from the other answer, "unique" means unique among these big org estimates; obviously the view that subsidies should be accounted for this way is shared by other researchers (Clemens, etc.) And here are some actual estimates (these are older owing to the ECFIN paper having come out before the newer IMF data, but still helpful for comparison purposes):

The IEA estimates that FFSs in 2013 totalled USD 548 billion, or 5% of the total GDP of the 40 countries included in the analysis (IEA, 2014). The OECD estimates that in the 2005-2011 period an annual average of USD 55-90 billion was spent on fossil fuel (production and consumption) subsidies in its member states (OECD, 2013). This is much lower than the IEA's estimate, but understandably so: governments in developed countries don't set fossil fuel prices (as do some in developing countries), and use sophisticated methods to subsidise fossil fuel production and consumption to a much lesser extent than the countries included in the IEA's analysis. The IMF's estimate for global pre-tax subsidies in 2011 totalled USD 492 billion (or 0.7% of global GDP at the time), relatively close to the IEA's estimate of USD 523 billion for the same year (IEA, 2012). The IMF estimates that global post-tax subsidies amounted to USD 2.0 trillion in 2011, representing 2.9% of global GDP or approximately 8.5% of worldwide government revenue (Clements et al., 2013). Thus the value of the negative externalities associated with the use of fossils fuels is roughly three times as high as actual government support for fossil fuels.

The newer IMF paper being discussed in the question here also stresses these points:

A key factor in estimating the magnitude of current subsidies is which definition of “subsidies” is used. [...] Post-tax energy subsidies are dramatically higher than previously estimated—$4.9 trillion (6.5 percent of global GDP) in 2013, and projected to reach $5.3 trillion (6.5 percent of global GDP) in 2015. Post-tax subsidies are large and pervasive in both advanced and developing economies and among oil-producing and non-oil-producing countries alike. But these subsidies are especially large (about 13–18 percent) relative to GDP in Emerging and Developing Asia, the Middle East, North Africa, and Pakistan (MENAP), and the Commonwealth of Independent States (CIS). Among different energy products, coal accounts for the biggest subsidies, given its high environmental damage and because (unlike for road fuels) no country imposes meaningful excises on its consumption. [...] These findings must be viewed with caution. Most important, there are many uncertainties and controversies involved in measuring environmental damages in different countries—our estimates are based on plausible—but debatable—assumptions.

As you can probably guess from the above, the picture is particularly skewed in China; Bárány and Grigonytė have this visual comparison:

enter image description here

And here is the expanded methodology summary by agency from Bárány and Grigonytė:


The International Energy Agency defines an energy subsidy as "any government action directed primarily at the energy sector that lowers the cost of energy production, raises the price received by energy producers or lowers the price paid by energy consumers" (IEA, 2014, p. 315). The price-gap approach is used by the IEA to estimate fossil fuel consumption subsidies for developing countries. This approach looks at the difference between a reference price and the price paid by end users. If this difference is positive, the particular fossil fuel is subsidised. The reference price is equal to the import parity price (the price of the fossil fuel at the nearest international hub, with transport and distribution costs and the VAT added) for fossil fuel importing countries, while it equals the export parity price (the price of the fossil fuel at the nearest international hub, minus transport and distribution costs and the VAT) for exporters. Many energy exporting countries and the Organisation of the Petroleum Exporting Countries (OPEC) disagree with this methodology. According to their view, the reference price for an energy exporting country should be the cost of production of the fossil fuel, and not the international market price (IEA et al., 2010). According to economic theory however, the legitimate reference price is equal to the export parity price for OPEC members, as these countries are making implicit losses (or not realizing implicit profits) on each barrel of oil that is not exported.


The OECD defines a subsidy as "any measure that keeps prices for consumers below market levels, or for producers above market levels or that reduces costs for consumers or producers" (OECD, 2005, p.114). Even though this is in line with the IEA's definition, the OECD takes a different approach to estimate the extent of consumption and production subsidies together in its member states, using a "broad concept of support that encompasses direct budgetary transfers and tax expenditures that provide a benefit or preference for fossil-fuel production or consumption, either in absolute terms or relative to other activities or products." In other words, the OECD aims to measure all FFSs that are explicitly included in the general government budget. In its Inventory, the OECD analysed over 550 government measures that can be considered fossil fuel subsidisation. The 2010 Joint Report of the IEA, the OECD, and the World Bank distinguishes seven basic types of FFSs, based on the official type of government intervention. These seven types are as follows: (1) trade instruments such as tariffs; (2) regulations such as price controls that result in consumer prices being below market level; (3) tax breaks either for consumers or producers of fossil fuels; (4) credit to fossil fuel producers; (5) direct financial transfer either to reduce end user prices or to lower the costs of producers; (6) risk transfer such as loan guarantees; (7) energy-related services provided by the government at less than full cost.


The International Monetary Fund (IMF) distinguishes producer and consumer subsidies to energy. "Consumer subsidies arise when the prices paid by consumers, including both firms (intermediate consumption) and households (final consumption), are below supply costs, including transport and distribution costs. Producer subsidies arise when prices are above this level"(Clements et al., 2013, p.5). The benchmark price is the international market price - adjusted for transportation and distribution costs - for internationally traded products - while it equals the cost-recovery price for energy products that are not internationally traded. The IMF further distinguishes pre-tax subsidies and tax subsidies for fossil fuel consumption. Pre-tax subsidies are defined in a similar way to the IEA's approach to subsidies, i.e. the difference between the oppor-tunity cost of supplying a consumer with fossil fuel (the international market price) and the price paid by the end user. The tax subsidy is the difference between the efficient level and the actual level of taxation for a given fossil fuel. The efficient level of taxation means first that the tax controls for the externalities associated with the use of the fossil fuel such as pollution and its effects on health, environmental costs, congestion, all of which reduce over-all welfare but are not taken into account by the user of the fossil fuel. This approach has a direct consequence for assessing post-tax subsidies for coal which is the most polluting fossil fuel, hence the negative externalities associated with the use of coal are by far the largest. Second, efficient taxation implies that fossil fuels are taxed the same way as other consumer products. Intuitively, the sum of pre-tax and tax subsidies is equal to the overall subsidy to the par-ticular fossil fuel, called the post-tax subsidy. While pre-tax subsi-dies have mostly been phased out in the developed world, they are still common in developing countries. Tax subsidies are prevalent in both developed countries and emerging economies. While the IMF provides the most comprehensive source of data for FFSs; some estimates are still missing from the study, as subsidies are not estimated for each country or for each fossil fuel. Petroleum subsidies are estimated most thoroughly for all 176 countries included in the analysis, using the price-gap approach, with production subsidies included for some OECD countries. Natural gas and coal subsidies are estimated for only 56 countries (with production subsidies to coal included for some OECD countries). Subsidies to electricity are estimated for 77 countries using multiple approaches and multiple data sources.

So, who is right?

In general, taxing negative externalities (as the IMF paper implicitly proposes) is called a Pigovian tax (a term found in the IMF paper itself). And the Forbes piece is a good hint that a certain part of the political-economic spectrum (like Hayek and so forth) don't dig that. In contrast others like Stiglitz have asked for a tax of $50-$100 per cabon tonne, approximately 10 to 20 times higher than in the current EU system.

If you want decision by majority of (US) economists, here is a (somewhat dated) snippet from a proponent:

In a 2006 survey of Ph.D. members of the American Economic Association, 65.0 percent agreed that ‘‘the US should increase energy taxes’’ [Whaples 2006]. Similarly, the Wall Street Journal asked business economists in 2007, ‘‘What is the most economically sound way for the government to encourage development of alternatives to fossil fuels?’’ The poll found 54 percent advocating ‘‘taxes that raise the cost of purchasing fossil fuels’’ [Izzo 2007].

There are also much more recent surverys of the US public (rather than just economists) showing support for a carbon tax. Also for quite a number of purposes (in federal regulatios) a Pigovian carbon tax already exists in the US as the "Social Cost of Carbon", and it has withstood a challenge in a federal court as non-arbitrary. However the battle between the Obama-era regulations/legislation and the Trump administration's reinterpretation thereof continues, this time in quantitative terms, from $45/ton (by 2020) of the Obama-era to $1-$6/ton in Trump administration's view. The large difference comes from whether the SCC should include world-wide effects of US emission or just effects in the US, as well as from a time-discounting issue.

And we're probably getting a bit too far afield here, but there's some recent evidence from Britain that such Pigovian taxes do achive their goal. The UK has a £23 tax per tonne of carbon that "tops up" the EU one. And as a result:

The fresh research shows that Britain has climbed from a 2012 ranking of 20th out of 33 industrialised countries to 7th on the low-carbon electricity league table.

“Britain is reducing its carbon emissions from electricity faster than any other major country, and this has happened because the carbon price and lower gas prices have forced coal off the system – the amount of coal-fired power generation in Britain has fallen 80pc between 2012 and 2016,” said Dr Iain Staffell, from Imperial College London.

While coal generation has fallen in the UK, Dutch coal-[f]ired power plants have ramped up due to sluggish European carbon price, causing emissions in the Netherlands to rise by 40pc between 2012 to 2016.

And no, you won't get my own opinion on who is right.

  • 3
    I have no particular objection to the length of your quotes, but more explanatory summary would help me and other readers follow the outline of the argument.
    – matt_black
    Nov 16, 2017 at 18:40

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .