Subsidizing or Taxing Pollution?

Fossil fuels, subsidies, climate change, and the economy

Fossil fuel subsidies should be the low hanging fruit of climate policy.

It’s so neat: eliminate the subsidies and you have a double benefit: combat climate change (by making fossil fuels more expensive) and combat fiscal deficits. What’s not to like?

What looks like a simple thing actually turns out to touch on several of the most important issues of climate change policy – both theoretical and practical.

What are fossil fuel subsidies in the first place, and how are they measured? What’s the right price to put on carbon dioxide emissions? How should we deal with the uncertainty surrounding the future climate impacts when we are designing policy in the here and now?

I’m going to try to briefly touch on all these issues. (Well, as briefly as possible.)

The inspiration – and the ambition to do all this in a single post – comes from some very thoughtful criticisms that Marco Annunziata over at the refreshingly contrarian JustThink substack had on an IMF paper about fossil fuel subsidies (the IMF also has a convenient online summary).

OK, let’s get stuck in.

What’s a fossil fuel subsidy?

This seems an easy question. Surely, it’s money governments give to people to consume fossil fuels. Turns out, that’s only part of the story – those are the explicit subsidies.

The other part of the story is that burning fossil fuels imposes costs on society.

These come in two kinds (oversimplification alert!). Fossil fuel consumption

• creates local pollution which has negative consequences for human health (e.g. respiratory disease and premature death), and

• contributes to climate change as the carbon dioxide (CO2) emitted heats up the atmosphere.

The former are local impacts which occur in the present (not-so-fun fact: outdoor air pollution is estimated to have resulted in 4.5 million premature deaths in 2019 according to a study of the global burden of disease).¹ The latter are global impacts which occur mostly in the future as the earth heats up.

These costs to society are generally not included in the price of fossil fuels, which is why they are called “external costs”.

(Going a bit deeper: The presence of external costs results in market failure. As the free-market price does not “internalize” these costs, the price of fossil fuels is too low and the quantity consumed too high – there’s too much burning of fossil fuels going on. This makes the market outcome “socially suboptimal”. The way to reduce the demand for fossil fuels – and hence reduce pollution – is to raise the price by imposing a tax. Note that not imposing this tax is essentially a subsidy.)

Other than explicit subsidies there are also implicit subsidies. These occur when the retail price of fossil fuels fails to include these “external costs”.

But before getting further into the weeds, let’s start with the IMF numbers to get an idea of the magnitudes.

A trillion here, a trillion there, and soon you’re talking real money

According to the IMF, in 2022 total – explicit and implicit – fossil fuel subsidies were around 7 trillion dollars globally, or a whopping 7% of global GDP.

18% of these, or 1.3 trillion dollars, were explicit subsidies, with the remaining 5.7 trillion dollars being implicit subsidies.

source: IMF

It’s important to repeat that for the implicit subsidies, no money actually changes hands. It’s just that the price of fossil fuels is lower than it should have been compared to a situation where external costs to society are recognized, and included in the price.

Naturally, when barely one out of every five subsidy dollars is explicit, it begs the question of how the IMF arrives at its implicit subsidy number. This is one of Marco’s objections.

Another is that removing an implicit subsidy really means imposing a tax. And given that the implicit subsidy = to-be-levied tax is so large, what would happen to the economy if we did so?

I will start with the last one.

Taxing fossil fuels would plunge the global economy in a recession. Two things on this.

• I don’t think anybody’s talking about hiking fuel taxes instantly – certainly not the IMF. Gradualism would – and should – be the name of the game, precisely to avoid throwing a wrench into the engine of the global economy (indeed the IMF paper assumes a gradual increase in fossil fuel prices).

• The general recommendation of the IMF is that the fossil fuel tax revenue is to be returned to the private sector (save, perhaps, for some R&D subsidies to accelerate technological solutions to climate change) through lowering other taxes and/or raising spending, e.g. via grants to poorer households.

This is, first, to avoid precisely the negative macroeconomic effect Marco is worried about.

Second, to be equitable, low-income households should be shielded from the impact of the tax. That is, they should be compensated for the loss of income resulting from the tax.

Now, if governments do this to plug holes in their budgets - and there are signs that governments are looking at carbon taxes more through the revenue lens - inevitably the result will be to a greater or lesser extent contractionary for the economy. But that’s not the general IMF recommendation. I don’t think it’s consistent to criticize the IMF for its policy prescriptions and then also criticize it for the possibility that governments may not follow them.

Messaging issues

Conceptually, not imposing the tax is equivalent to a subsidy – that’s why, in my view, it is legitimate to talk about implicit subsidies.

But I agree with Marco that this is not an innocuous messaging choice. To the casual reader, the statement “we will cut x bn of subsidies” sounds very different from “we will hike taxes by x bn”.

I’m also sympathetic that the headline number, without mention of the distinction between explicit and implicit subsidies, may create the wrong impression.

That said, I don’t think the solution could have been not mentioning implicit subsidies at all. Accounting for the negative externalities from fossil fuel use – pricing them correctly – is a key issue for the functioning of market economies, for fiscal policy, as well as for climate change. Quite simply, the IMF (or any policy institution) would not be doing its job if it swept this under the rug.

So perhaps a better messaging choice could have been: “There are explicit fossil fuel subsidies of $x; there are implicit subsidies of $y resulting from local damages of fossil fuel use not being in the price; and there are additional implicit subsidies of $z due to the emissions contributing to climate change.”

(Now picture the IMF press officer rolling her eyes trying to turn this into a press release!)

Who benefits from the subsidy? Marco also objects that the IMF is not clear that the (bulk of) the explicit fossil fuel subsidies are given to consumers (hence creating an impression that governments subsidize “big oil” on a large scale).

Here, I disagree. The subsidy generally does end up with the producers: they get to sell more because prices are lower, and (as long as the selling price stays above cost), their profit increases.

What about the distributional argument for fossil fuel subsidies? For large parts of the developing world it’s probably reasonable if the government helps households with their energy bills to buttress living standards - although there are serious targeting issues: these subsidies also help a lot of people who don’t need help.

But precisely due to these targeting issues, I don’t buy the argument for developed countries. My observation is that the bulk of fossil fuel subsidies in advanced economies go to the middle class and certain well-connected interest groups (which is not to deny that there is fuel poverty in advanced economies as well).

Marco also has this objection:

The implicit subsidy number is an estimate, and as such uncertain because the impact of climate change is highly uncertain. That is certainly true, but the devil is in the detail.

Let’s start by looking under the implicit subsidy bonnet and see what’s included.

It turns out that globally, the estimated damage from fossil fuel consumption via the local pollution channel is on par with the damage from the estimated contribution to climate change (see chart below on global fossil fuel subsidies by component). Why does that matter?

Because a big chunk of the IMFs implicit subsidy (nearly two thirds, if you also include foregone consumption tax revenue and road congestion externalities) has nothing to do with climate change.

source: IMF

What price is right?

Now that we’ve gotten some important issues out of the way we can close in on the prize: how to value carbon emissions; and – more broadly – how to deal with the uncertainty surrounding estimates of global warming and associated economic damages (spoiler: there is no definite answer, but there is a reasonable way to go about it).

Just to pick up the thread again: fossil fuel subsidies can be explicit – the government picking up the bill for part of its citizens’ consumption – or implicit. The latter come from the market price not including the social cost of burning fossil fuels. This cost comes in two forms: monetized damages from local air pollution; and monetized damages from (the contribution of the associated emissions to) climate change.

So let’s now look at the second big chunk of social costs, the damages from climate change. Unlike with local air pollution, the bulk of these damages occur in the future – and are much more uncertain, a point I’ll return to.

Here, there’s no way to get around a difficult concept: the social cost of carbon (SCC). The SCC is the cost of emitting an additional ton of CO2 today (expressed in today’s dollars). To calculate it, we need

• an estimate of how much additional warming this ton of CO2 will cause

• convert the physical consequences of this warming (increased frequency of extreme weather events, droughts etc.) into monetary damage – think lost GDP

• as these damages occur in the future, we need to turn future dollars into today’s dollars by way of discounting.

Once we have this number, we have our carbon tax, i.e. by how much we need to tax fossil fuels – remember that the price of fossil fuels is too low because it doesn’t incorporate costs to society from climate change.

In turn, if we know the right level of the carbon tax, we know what the price of fossil fuels should be, which means that we know the level of the implicit subsidy that we are after.

Turns out that when using this – theoretically correct – way to estimate the implicit subsidy on fossil fuels, i.e. by using the global discounted damages from emitting an additional ton of CO2, the IMF obtains a much larger value. The total subsidy becomes 50 percent larger and amounts to 11 percent of global GDP!

By contrast, the 7 percent of global GDP number the IMF paper headlines with comes from using that carbon price which would reduce emissions to a path consistent with the 2°C warming Paris goal (or: that level of carbon tax which, if levied, would achieve Paris) by 2030.

Marco suggests that the IMF shied away from emphasizing the (theoretically correct), higher estimate because that may appear implausibly high.

In my view, the price isn’t quite as arbitrary.

First, they are upfront about it being a conservative estimate. Second, the reason may be technical: they may have wanted to avoid their results being vulnerable to criticism about the discount rate they used in the theoretically correct calculation (damage estimates are highly sensitive to the choice of discount rate).

Finally, and this relates to the second point, thinking in terms of practical policy recommendations, I think it is legitimate to go for something that is more politically feasible even if not theoretically ideal, so long as it is conceptually plausible and they are transparent about it.

How to deal with uncertainty in climate change

But I still haven’t addressed the elephant in the room, which is of course how to deal with the uncertainty in climate change.

First off, let’s be clear about what’s not uncertain.

Contrary to what deniers of climate change like to believe, there is no uncertainty about the qualitative predictions of the science. Science has established (since the nineteenth century) that an increased concentration of carbon dioxide in the atmosphere will increase the earth’s temperature.

So the mechanism of global warming is well understood. What there is uncertainty about is the exact numbers:

• The link between atmospheric concentration of CO2 and the degrees of warming. Estimates of the degree of warming resulting from this concentration are subject to very large uncertainty. Why? The complexity of the climate system is such that the precise values for the important feedback parameters cannot be known with any degree of accuracy.

• The link between warming and economic damages. Suppose we actually knew the exact temperature increase resulting from the emission of CO2. Estimates of the resulting economic damages are, again, subject to very large uncertainty. Why? In short, because we have no past experience to rely on.

So what should we do?

One way could be to say that we wait until uncertainty is reduced. Scientists may come up with better estimates of warming and economists with better estimates of damages. In the meantime, it’s business as usual.

I think it’s fairly obvious that this is not a prudent strategy. First, every passing minute means higher emissions, which gets us closer to a – potentially catastrophic – climate outcome. Second, it is far from clear that the uncertainty will narrow down meaningfully over any relevant time horizon: scientists may never know the “true” value of crucial climate parameters.

What’s the way forward?

Let me start by proposing two principles.

Principle #1: prudence - let’s take out insurance

It’s true that the science is uncertain.

We could be lucky and get benign outcomes because it turns out, say, that the climate is not very sensitive to the atmospheric concentration of carbon dioxide. At the same time, catastrophic outcomes – warming of 4, or 6, or 8 or more degrees - are well within the realm of the possible. In other words, humanity is faced with a – hopefully small – but certainly non-negligible probability of climate disaster.²

In other words, when we look at the climate, we shouldn’t just look at the most probable outcome, or the mean outcome, but also at the dispersion of outcomes, and, in particular, at the right tail of the distribution of outcomes. (Investors in particular are familiar with this way of thinking.)

And this is a good place to start from. Because we know how to deal with the prospect of a low-probability, but highly damaging event: we take out insurance. Which means we pay a small amount, the insurance premium, to avoid a large loss. (Investors call this “hedging”.)

(Politicians: this to me looks a much better way of framing decarbonization. Every voter is familiar with the concept, as the majority of people take out some insurance. The other thing of course that needs to be clear is that the cost of insurance – the premium – must be spread fairly in society.)

Principle #2: cost minimization

Another principle is to minimize the cost of climate policies, because they are substantial - I completely agree with Marco on this one. (I also agree that politicians haven’t always been clear to people about the costs involved. Although for every politician of this kind, there is also one of another kind who is not clear about the costs of not acting!)

From principles to policy

How do we turn all this into practical policy prescriptions?

Turns out this is something where economics (the discipline) can actually help.

Faced with large uncertainty about key parameters of the climate-economy system, a “good policy” is one that produces acceptable outcomes for a wide range of plausible values for these parameters.

Broadly, we can make two kinds of mistakes:

• enact an ambitious policy that spends too much on mitigation but the climate outcome is benign because the crucial climate sensitivies are low; and

• spend too little on mitigation and end up facing a very adverse climate outcome because the climate sensitivities are high.

What’s the “least regret” policy? When you ask the models, it turns out that the good policy is an ambitious policy, because it offers cheap insurance against a catastrophic climate outcome. It is worth repeating this: compared to the costs of a climate disaster, the insurance is cheap.

Where the rubber meets the road: carbon prices

OK, fine, you might say. But that doesn’t help me much in setting the right level of the carbon tax!

Let’s now be relentlessly practical.

Where is the carbon price now? An estimate for 2019 has the global (implicit) price of emissions at twenty dollars per ton of CO2. A few years on, with at least some prices higher, like in the EU (see chart), back of the envelope I would reckon the current value for the global price of carbon is between 25 and 50 dollars.³ (EDIT: a recent estimate covering 22% of global greenhouse gas emissions is just short of $25.)

source: ICE EUA, Macrobond

Where does the price need to be by the end of the decade for us to achieve 2 degrees? Indicatively, a few estimates are: 92 dollars,⁴ 115 dollars,⁵ 195 dollars⁶ (there’s even an estimate that the current SCC is 1000 dollars!).

So it seems to me that the uncertainty in the estimates of where we should be is completely dwarfed by the distance from even the lowest value! Current carbon prices are simply an order of magnitude too low. And given the importance of local pollution, they would still be even if we stopped caring about climate change altogether.

Which means that the general direction is clear: up! How quickly? As quickly as feasible, given a) the economy b) political majorities.

I personally think that both innate human myopia (more tactfully: Mark Carney’s “tragedy of the horizon”) as well as the normal calculus of politicians – future generations don’t vote! – virtually ensure that we won’t do too much, too fast. Too little, too late seems a great deal more likely to me.

The point is: given where we are, a little should go a long way.

An analogy might help.

We know that smoking is bad for our health. We know the why, and the how – medical science understands the causal mechanisms. But medicine can’t predict the consequences of smoking for each individual, nor forecast their life expectancy. We’ve all heard about people who lived until ripe old age even though they smoked a packet of cigarettes per day.

Does that mean I can smoke a packet of cigarettes per day? Maybe I’ll get away with it. Or it may drastically shorten my life expectancy. Does it make sense to stop smoking instantly? Absolutely. But perhaps that’s not immediately feasible. If I smoke a packet a day, do I really need to know whether it’s smoking five cigarettes, or three, or none that would likely save my life?

Renewables

The above has implications for the renewables discussion. This is really worth a separate post, but I’ll try to briefly address Marco’s points on this one.

His objections are that statements about the competitiveness of renewables based on the IMF fossil fuel subsidy estimates are questionable, and that one should not conclude that removing the subsidies would immediately provide us with cheap, clean energy.

Now, solar PV and onshore (although not offshore) wind are, in general, already cheaper on a (levelized) cost basis than fossil fuels, according to the International Energy Agency.⁷ And, as I’ve tried to argue, you don’t need to stick to the IMF numbers, nor place much weight on climate-related estimates altogether to see that carbon prices must rise meaningfully. In turn, that would further boost the competitiveness of renewables.

On the second objection, again I haven’t seen anyone suggesting that we should just jack up carbon prices massively and switch to renewables instantaneously, and indeed a key reason that’s not feasible is that we lack renewables capacity (relative to energy needs).

But that’s precisely where carbon pricing comes in. It provides the price signal to steer energy demand away from fossil fuels as well as to encourage an increase in supply of clean energy by incentivizing further development of renewables capacity. We can get there, we just need to get on with pricing carbon!

Thank you for persevering.

(Marco, I hope I have presented your views fairly, and accurately.)

1. Cited in the IMF paper.

2. Robert Pindyck’s excellent book Climate Future updates a study summarising 131 scientific papers on the value of the climate sensitivity parameter (degrees warming resulting from a doubling of the atmospheric CO2 concentration). Ignoring outliers gives a range of 0.5 to 7.0 degrees.

3. The price of emissions in the European Union’s Emissions Trading System has increased by about 150% between 2019 and the first half of 2024. Assuming all other carbon prices in the world have remained constant implies a lower bound estimate of 25 dollars. Assuming all other carbon prices have increased by the same as in the EU – surely an overestimate – gives the upper bound estimate of 50 dollars.

4. Stiglitz-Stern High Level Commission on Carbon Prices (average).

5. IPCC 6^th Assessment Report.

6. Network for the Greening of the Financial System. These values are all inflation adjusted (expressed in 2023 dollars) from the World Bank.

7. World Energy Outlook, p. 301.