Planes, trains and automobiles

Tim Worstall has challenged, in a recent post, the logic of the DfT's suggestion that Air Passenger Duty (APD) needs to be increased further to take full account of the contribution of aviation to carbon-emissions. By Tim's calculations, taking Stern's figure of $85 as the social cost of a tonne of carbon-dioxide emissions, the cost of carbon emitted by air travel is "around and about the amount currently charged" under the APD.

One could raise a number of reservations about this estimation, but let's say (for the sake of argument) that he is right. It raises interesting questions, either about Stern's calculations, or about this approach to welfare economics.

Tim cites approvingly the DfT's explanation of the principles underlying Pigovian taxation. The DfT does indeed reduce the idea to its most basic essentials, but it seems to me that the essence has been corrupted or lost to some extent in the act of reduction. Let me see if I can set out the idea clearly, so we can judge mechanisms like the APD for how well they embody the economic theory that underlies them.

Economics, if it treats only of monetary values, is subject to the Wildean criticism that it knows the cost of everything and the value of nothing. Economists have therefore felt it important to consider the impact of actions on utility (or, simplistically, happiness). Welfare is the notional sum of utility or happiness amongst a group (ignoring for the moment that it is ridiculous to aggregate something incommensurable like happiness). Welfare economics is the study of economics from the perspective of maximizing welfare.

In theory, if all participants in a transaction agree voluntarily to its terms, then they must all feel themselves at least to be no worse off from the transaction, and some of them presumably feel themselves to be better off, or why would they have agreed to it voluntarily? This satisfies the condition needed to demonstrate that an action has increased general welfare. By definition therefore, free markets increase welfare. This (and the inability to make the same case for coerced actions, such as those obliged by the state) is the basis of the libertarian philosophy.

In practice, however, it is often the case that people not party to a transaction may be affected by it. The classic example is a factory whose effluent is discharged into a river. No matter whether all parties engaged in commercial transactions with the factory-owner do so voluntarily, the contribution of the factory to the general welfare may be negative, if the benefit to the direct participants is outweighed by the impact on those who use the river downstream from the factory. The pollution of the river is, in the language of economists, an externality - a value (negative in this case, but they can be positive too) that is attributable to someone's activities, which they are not able to capture (if positive) or obliged to incur (if negative).

Externalities skew decisions, such that the options that provide the greatest benefit to society (maximize welfare) may not be the ones that offer the greatest benefits to the participants in transactions. The way to ensure that markets provide the optimal outcome for society is to internalize the externalities. In effect, this means ensuring that the party responsible for the externality is attributed the cost or value of that externality.

A Pigovian tax (named after Arthur Pigou, the Cambridge economist who first set out a detailed exposition of welfare economics) is a mechanism whereby the cost of a negative externality can be internalized to the producer of that externality. If the impact of the effluent from the factory results in losses to those downstream at a rate of £100/litre, a government may impose a Pigovian tax of £100/litre on the factory, so that the factory-owners' decisions take into account the full costs of any choice, and not just those incurred directly. The Pigovian tax provides an incentive for the factory-owner to minimize the discharge of effluent or even to locate elsewhere where the external costs of disposing of the effluent are lower.

However, the application of the Pigovian tax is not sufficient in its own right to ensure that welfare is maximized. It may be that the factory-owner has few feasible alternatives, and decides to continue to discharge and pay the tax. In that case, the damage (or disutility) to those downstream is no less than before, and the utility of the factory-owner has also been reduced by application of the tax. The utility of the government (or taxpayers generally) may have been increased by the revenue from the Pigovian tax, but it is not the government or taxpayers generally who were being harmed by the externality. The Pigovian tax can only be said to have achieved its objective if it represents an accurate valuation of the harm and if it is distributed as compensation to those on whom the harm is inflicted, proportionately to their share of the harm. In other words, a tax (or charge) cannot be said to be an effective mechanism to internalize externalities unless it either deters the externality or fully compensates those who experience the effect of the externality (or a bit of both).

Greenhouse-gas (GHG, or, in the vernacular, carbon) emissions are an externality. They are a particularly difficult example, because the impact of the harm is uncertain and because most of those who may experience the harm are remote, both geographically (e.g. inhabitants of vulnerable countries) and temporally (i.e. future generations). It is important that mechanisms to internalize the carbon externality reflect the remoteness and uncertainty.

If we applied that tax of £100/litre to the factory-owner in the above example, but recycled the revenues to him by some subsidy or tax-break, there would have been no point applying the Pigovian tax in the first place. Similar could be said if we recycled the tax-revenues to his customers - the amount that he had to increase his prices would be balanced by the extra amount that his customers were able to pay for them. The tax would be neither an effective deterrent, nor an effective means of compensating those who experience the harm. It would simply be an inefficiency in the market, perhaps effective at salving consciences, but of no real value. Indeed, if it is effective at salving consciences, it may exacerbate the harm, because customers may consume more of the good, freed (superficially) from the responsibility of considering the impact of their choices on those affected by the externality.

In the case of carbon emissions, most of us, metaphorically, own shares in the factory, and most of us live downstream. But some own more shares in the factory and suffer less of the impact of the pollution than others (e.g. rich world vs poor world, this generation vs future generations). If we impose a Pigovian tax on the factory, but recycle the revenues to those who own most of the shares, we have not effectively internalized the externality. We may have exacerbated the harm by letting the factory's customers think that they now need not worry about the impact of the factory's externalities. This is exactly what most of the rich-world mechanisms to internalize carbon externalities, including the APD, are doing.

Carbon is carbon is carbon. The environment doesn't care where or how the greenhouse gases are emitted. It has an equal cost to the environment or society. Pigovian taxes (or equivalent mechanisms) should value all emissions from all sources equally.

Let's say that we apply Stern's $85/tonne of CO2 to all sources of emissions equally, through the various mechanisms like the APD.

  • If Tim is right, £10 per short-haul trip or £40 per long-haul trip (the cost of the APD) does the trick for aviation. (Strange how all short-haul trips produce the same emissions, regardless of distance and occupancy levels, and all long-haul trips produce four times as much, and that a preferential rate of VAT on the fuel doesn't count, but anyway...)
  • The current price of carbon imposed on road users is, by any reasonable assessment, very much higher than $85/tonne. To accurately internalize Stern's cost of carbon, we'd do away with vehicle excise duty, and more than halve fuel excise duty. The costs of owning and running a car would be significantly lower than today, with petrol costing perhaps 50-60p/litre, rather than today's 95p/litre.
  • Costs of rail transport would be significantly higher, as equivalent duty (VAT and carbon-tax) was placed on the fuel used by cars and trains, and subsidies were removed so that people incurred the full cost of their transport choices.
  • Costs of heating homes with gas would be around 40% higher, as the low-rate of VAT was removed and a carbon-tax of $85/tonne were applied.
  • Costs of heating homes with other fossil fuels would be higher still
  • Costs of gas, oil and coal for industrial processes, and of electricity generated from fossil fuels, would be marginally higher, as the combined cost of the Climate-Change Levy (CCL), the EU Emissions Trading Scheme (EU-ETS) and of the Renewables Obligation (RO, only relevant to electricity) does not come to $85/tonne of CO2. Oil and coal would increase in expense relative to gas, which would be barely more expensive than today. The impact on costs to consumers, and therefore on demand and levels of consumption of the products, would be less, as energy-costs make up only a portion of overall costs.

In this world, then, where the carbon externality is fully internalized at the price calculated by Stern, aviation would be no more expensive than today, road-use would be much cheaper, rail-use and heating homes would be more expensive, and electricity and manufactured products would be marginally, but not significantly, more expensive.

Would this represent an effective deterrent to producers of the externality? Of course not. The only significant reduction of emissions would be likely to be in home heating. In other significant sectors, emissions would be little more constrained, and in the case of driving less constrained, than they are today.

Would the revenues be an effective means of compensating those who suffer the effects of the externalities (i.e. those whose utility is reduced by them)? Revenues go into the Government's budget for that year. They do not go proportionately to compensating those in countries that contribute less to the problem but that are more exposed to the consequences. And they are not put away in provision for the future costs imposed on future generations. The revenues from these Pigovian taxes are, in effect, recycled to the majority shareholders in the factory.

There are two ways of looking at this from the neo-classical perspective. Either we go with Stern's figures for the cost of carbon, in which case we have to change the mechanisms through which the cost is applied, so that the revenue is apportioned proportionately to those on whom the consequences fall. Or we live with the mechanisms, in which case we have to accept that, if (say) 80% of the revenue from the Pigovian taxes is recycled to the polluters, we need to multiply Stern's figure by five to get the cost of carbon that should be used when pricing the mechanisms.

There is another perspective. Quantifying, comparing and aggregating interpersonal utility is rubbish. Conventional welfare economics is rubbish. Calculating social costs is rubbish. Forget about Stern and all the other attempts to calculate the social cost of carbon. His figure is not too high or too low, it is too specific. Values should be discovered, not calculated, in markets that are based on free choice within the constraints of strict respect for property rights. Externalities represent indirect infringements on property rights (or inability to capture those rights, in the case of positive externalities). We need institutions that prevent indirect infringements on property rights, in the same way that we require law & order to prevent direct infringements on property rights. The purpose of the institutions will be to enable people to choose either not to have external costs inflicted on them, or contract the price at which they accept them. Given the ubiquity of some externalities (such as greenhouse gases), this may require approximate rather than specific mechanisms. But the test of legitimacy will be whether those whose property rights are at risk of being infringed by the externality are protected from, or compensated to their satisfaction, for the infringement.

By this test, all our current mechanisms to internalize the externalities of carbon emissions are badly-designed failures. But while we are stuck with them, let's at least stop pretending that they provide adequate internalization of the externalities.