Gas taxes for thee, but not for me

Why the popular opposition to Pigouvian taxes?

Paris and other French cities have recently been the site of large and heated (there were fires) protests. The most prominent symbol of the protests is the high-visibility gilets jaunes which all French motorists are required to own. This symbol is fitting for a series of protests that coalesced in opposition to an increase in the gas tax. While this protest (and any other protest) isn’t about a single grievance, it certainly seems to signify something about the popular opinion of gas taxes.

These protests pair somewhat uncomfortably¹ with the popular march against climate change in Paris just three months ago.

This pattern seems common: Acknowledgment that global warming and excessive fossil fuel use are problems which must be addressed followed by reluctance or opposition to concrete responses (Leiserowitz 2006).

Other examples include:

Washington Initiative 732

A 2016 revenue neutral carbon tax that would have used revenues from the tax to reduce the (highly regressive) state sales tax and increase a low-income tax credit. It lost despite Washington’s reputation for environmental concern.

Washington Initiative 1631

A 2018 carbon tax that would have used revenue from the tax to invest in green initiatives and other infrastructure projects. Again, the measure did not pass.²

Australian carbon tax

Introduced in 2012 by the Labor government, it was repealed in 2014 by the Liberal government.

(Carbon taxes aren’t always doomed to failure. British Columbia has had a carbon tax of $35 per tonne in place since 2008. Sweden has the world’s highest carbon tax at about $138 per tonne. California recently voted down a repeal of recent gas tax increases.)

Pigouvian taxes

What are Pigouvian taxes?

So gas and carbon taxes are controversial with the public. Not so for economists and technocrats. They are both examples of Pigouvian taxes. In brief, Pigouvian taxes seek to correct market failures arising from negative externalities. A negative externality is any cost generated by a market activity that is not priced into the market exchange.

For example, local air pollution caused by combustion of gasoline harms other nearby people—think especially of those who don’t even own a car—who receive nothing in exchange for bearing this harm. Because these costs aren’t included in the price of gasoline, gas is cheaper than it ought to be, all things considered. This leads to overconsumption of gasoline.

A Pigouvian tax tries to remedy this market failure by making the ‘invisible’ costs of externalities manifest via a tax. If the total external costs of driving—due to externalities like greenhouse warming, congestion and local pollution—amount to $1.76 per gallon (an estimate from (Parry, Walls, and Harrington 2007)), a tax of $1.76 on each gallon forces the gasoline purchaser to ‘internalize’ those costs—account for the social harm they are creating with their actions.

A tax only an economist could love

“Economists are in almost universal agreement that, in concept, pollution taxes are the most cost-effective means of reducing pollution” (Hsu 2009). In a direct survey, 65% of economists attested that there should be an overall increase in energy taxes (Whaples 2006). Gregory Mankiw keeps a partial list of prominent economists that support Pigouvian taxes in the form of the membership rolls of the Pigou Club.

Contrariwise, only 5% of the general public supported a tax on driving in a 2007 survey.

For a (very accessible) fuller discussion in support of Pigouvian taxes from an economist, see (Mankiw 2009).

Reasons to oppose

I’ll confess myself to be on the side of economists when it comes to Pigouvian taxes. I’m then left with the question, why are they so unpopular with the general public? There are many possible reasons which we’ll now rehearse (Though we will stick to arguments where there’s something interesting to say about Pigouvian taxes specifically; we won’t rehash arguments about whether the government should ever intervene in the market, can be trusted, etc.) . Since the case for Pigouvian taxes rests on both positive (these are the effects of Pigouvian taxes) and normative (these effects are good) premises, we’ll categorize³ the objections as:

1. Misapprehensions of the positive
2. Disagreements with the positive
3. Misapprehensions of the normative
4. Disagreements with the normative

Positive misapprehensions

Visible costs, hidden benefits

When a new Pigouvian tax is instituted, the costs it imposes are certain and immediate. The promised benefits only dangle from the end of a causal chain (i.e. tax → increased cost → reduced consumption → benefit). This compares unfavorably to direct regulations like CAFE where the benefits are obvious and the costs are hidden. It takes more deliberate and careful analysis to reverse first impressions and come to the conclusion that, all things considered, Pigouvian tax A is preferable to regulation B (a conclusion economists reach much more often than the public). (Barthold 1994)

Metric effect

Gas taxes are usually proposed in terms of currency per volume—e.g. cents per gallon. Apparently⁴, people tend to overweight these kinds of taxes in comparison to taxes presented in percentage terms. So this works against gas taxes. (Hsu 2009)

Isolation effect

Gas taxes are relatively complicated. The isolation effect says that in complicated decision scenarios, people will focus on a single dominant aspect and neglect other considerations. With gas taxes, the dominant consideration may be the cost of buying gas since this a common, salient experience for many. The other benefits then will be relatively neglected and overall evaluations of the gas tax will be imbalanced. (Hsu 2009)

Distributional impact

The biggest concern by far (according to my sense of the public discussion) is about the distributional impacts. It’s often believed that gas taxes (and other Pigouvian taxes) are unfair and regressive. The story is that poor people spend a larger fraction of their income on gas, so increasing the gas tax flatly (i.e. without regard to income) ends up taxing poor people at a higher percentage of their income. There are several possible responses to this:

Flat denial

The regressive argument looks at incomes. But income is only part of the story. Arguably, actual individual consumption is a better measure (Fitoussi, Sen, and Stiglitz 2009). The total value of people’s consumption is determined not only by their private incomes but also by the assistance they receive from the government and non-profits. If we account for this additional income analogue, “low-expenditure households devote a smaller share of their budget to gasoline than do their counterparts in the middle of the expenditure distribution” (Poterba 1991). (Cronin, Fullerton, and Sexton 2019) is a more recent article making a similar point about a general carbon tax.

Less regressive than alternatives

If we treat the amount of revenue to be raised as fixed (a mildly reasonable assumption), then the question changes from “Is a gasoline tax regressive?” to “Is a gasoline tax more regressive than alternative sources of funding?”. Is it more regressive than a progressive income tax? Probably. Is it more regressive than a flat sales tax? Usually not. So a shift to a gas tax from a sales tax—a common revenue-neutral proposal—is actually progressive.

Depends on how revenue is spent

If the revenue raised by a gas tax is spent on, for example, low-income tax credits or increasing funding for public transit, it can easily be made progressive. “Lump-sum rebates are […] much more progressive, benefiting the three lower income quintiles even when ignoring environmental benefits” (Williams et al. 2014).

Distribution of benefits

The externalities that Pigouvian taxes seek to reduce don’t have equal incidence across the income spectrum. For example, low-income people likely work jobs which are less flexible making them more sensitive to unpredictable congestion during commutes (Arnott, De Palma, and Lindsey 1994). They also have reduced ability to choose housing in areas with good air quality. Consideration of factors like these mean that the benefits may accrue disproportionately to the poor which could make the overall scheme progressive even if the tax itself is regressive.

Concluding remarks on distributional impacts

If you’re still concerned, (Schweitzer 2009) and (Metcalf 1999) address the issues fairly comprehensively.

I’ll also note here that the gas tax the gilets jaunes opposed was not revenue neutral—the revenue was to be used to reduce the deficit. So it was actually regressive at the most straightforward level.

Even though concerns about the distributional impact are primarily normative, I categorize this issue under positive misapprehensions because that’s where the error lies. People aren’t wrong to be concerned about the distributional impacts of policy; they just happen to be wrong as to how (at least some) gas taxes affect different income groups.

Positive disagreements

Non-uniform marginal cost

Carbon taxes are essentially an ideal case for Pigouvian tax because the harm being done is the degradation of a global common resource. This means that the marginal social cost of any additional carbon is nearly constant so a flat, uniform tax closely aligns with costs. On the other hand, a Pigouvian tax on soda would be difficult to implement well because the social cost of consumption varies widely with the consumer—nothing at all for those who are healthy and much higher for those who are likely to rely on socialized health care to remediate unhealthy eating. (If we really want to be obnoxious, we can call these Pigouvian taxes which apply a uniform tax despite varying marginal harms Procrustean Pigouvian taxes.) This consideration only limits the scope of Pigouvian taxes though; it does nothing to argue against the efficacy of Pigouvian taxes that are applied in fitting circumstances. (Fleischer 2015)

Normative misapprehensions

Status quo bias

Many people already drive and have established expectations about the benefits and costs of doing so. An increased gas tax increases the costs of present activities while providing new and uncertain benefits. Because people overvalue the present state of affairs, this factor tends to oppose the gas tax. ((Hsu 2009) points to the endowment effect but it strikes me as a bit weird to call the habit of driving a possession.)

(I call this a misapprehension due to my belief that no popular moral theory endorses stasis as an intrinsic good. Presumably, people could be cajoled into overcoming this bias by coming to a fuller understanding of themselves, their preferences, and alternative states of the world.)

Normative disagreements

Doing vs. allowing harm

Imposing a gas tax seems to proactively harm intensive consumers and current producers while the less-taxed status quo seems to harm only by accident. People generally seem to find allowing harm preferable to doing harm. See also the doctrine of double effect and people’s reluctance to redirect the trolley in the trolley problem. ((Hsu 2009) talks about this as a psychological bias called the do-no-harm effect.)

Moral circle

The costs and benefits of a Pigouvian tax are not always evenly spread. We see some of this dynamic⁵ in the foot-dragging of the United States in regards to global warming. A flat global carbon tax would be more costly for Americans as the average American produces 16.5 tonnes of CO₂ annually compared to the global average of 5.0 tonnes. Additionally, the US is likely to suffer less from global warming than other countries because of its climate and wealth. If Americans look only to the costs and benefits for themselves and their fellow citizens, they may be perfectly rational (in a particular narrow sense) in rejecting carbon taxes and other global warming mitigations. A similar dynamic plays out any time some egocentric subgroup is harmed more by a Pigouvian tax than helped by reduced externalities.⁶

Discount rate and population ethics

Some of the harms which a Pigouvian tax would avert peak decades in the future. A carbon tax, for example, would be paid now with the biggest benefits coming decades in the future. In those cases, an individual would pay a cost now for benefits they’d reap far in the future, if it all. If their discount rate is high or if they value present persons more than potential future persons (Parfit 1984), it would be fully rational (again, in a particular sense) for them to disfavor the tax.