This chapter makes two basic points about international coordination of carbon pricing.
The first is that just because climate externalities cause a global free-rider problem (i.e., the reluctance of one country to move ahead unilaterally with carbon pricing because it bears the costs while all countries benefit from a more stable global climate system), this does not mean we have to solve that problem to make a good start on reducing climate externalities. In fact, a significant amount of carbon pricing is in many countries’ own national interests (before even counting the global climate benefits) because the domestic environmental benefits (most important, public health benefits from reduced local air pollution), and perhaps also the domestic fiscal benefits, easily outweigh the climate mitigation costs. So countries can move ahead on their own with carbon pricing and make themselves better off—they do not need to wait for large emitters to act in a coordinated way. This is an important message that policymakers need to communicate to legislators, the general public, and stakeholders as they start to move forward on their mitigation pledges for the 2015 Paris Agreement on climate change.
The second point is that, as countries seek to build on domestic pricing initiatives through international coordination, they should consider the possibility of international carbon price floor arrangements (in preference to linked trading systems). These arrangements provide some protection against free-rider issues and losses in competitiveness, while allowing individual countries the flexibility to set prices higher than the floor, which makes sense if they have relatively high domestic environmental benefits or high fiscal needs, or if higher prices are more politically acceptable in that country than in other countries. Some operationalizing issues need to be worked out (e.g., accounting for broader changes in energy taxes and subsidies and country-level needs for special provisions in carbon pricing programs), but the practicalities should be manageable.
The following two sections elaborate on the above points.
The main focus of this section is on the domestic environmental benefits of carbon pricing1 as the evidence on their magnitude is more solid than for the fiscal benefits (i.e., the economic efficiency benefits from substituting carbon taxes for distortionary taxes on labor, consumption, and capital).
Leaving aside the global climate benefits, carbon pricing can generate substantial domestic environmental benefits, most important the reduction in premature deaths from exposure to local air pollution as carbon pricing reduces the use of coal and other polluting fuels and (less important) the reduction in traffic congestion, accidents, and road damage externalities, to the extent that carbon pricing reduces vehicle use and these externalities are not reflected in road fuel excises.2
Clearly, there are much more efficient instruments than carbon pricing for addressing these domestic environmental externalities. For example, the most efficient way to reduce local air pollution is to impose a Pigouvian tax, either through directly charging emissions or (which may be administratively easier for some countries) an upfront tax on fuel use combined with rebates for entities demonstrating emissions capture during the combustion process. And (for given road infrastructure) traffic congestion is most efficiently addressed through congestion fees on busy roads, rising and falling progressively over the course of the rush hour.
It is also clear, however, that it will take a long time before these ideal charging systems are widely implemented across large carbon emitting countries. With the odd exception (e.g., Chile), countries have yet to introduce a comprehensive set of charges on the major air pollutants with charges aligned to estimates of air pollution costs3 (technology mandates and other regulations are common, but they do not fully internalize the externality). Certainly no country comes anywhere near close to having a set of finely tuned congestion fees, varying with marginal external costs across time of day, on all links in the road network with non-free-flowing traffic.
For the interim, therefore, it is entirely appropriate to account for the unpriced domestic environmental benefits when assessing the welfare effects of near-term carbon pricing schemes. Not doing do violates long-established principles of welfare measurement,4 implying that the welfare impacts of new charges in fuel markets hinge critically on preexisting distortions in those markets, whether they be from prior fuels taxes, externalities, or other sources.
The conceptual framework for assessing the domestic efficiency benefits (or costs) of carbon pricing is basically straightforward. Consider figure 3.1, which shows the demand and supply5 for a fuel product such as coal or gasoline. In the absence of any noncarbon externalities, preexisting fuel taxes or subsidies, or other distortions, the welfare cost of a new carbon charge in this market is the usual darker-shaded triangle—the “Harberger triangle”—with base equal to the reduction in fuel use induced by the carbon charge and height equal to the carbon charge, that is, the fuel’s CO2 emissions factor times the charge on CO2 emissions.
If there are noninternalized, noncarbon externalities associated with use of the fuel, however, like local air pollution damages, then it is possible for the carbon charge to generate a net welfare gain, indicated by the lighter-shaded trapezoid in figure 3.1, equal to the (noninternalized) domestic environmental benefits—damages per unit of fuel use times the reduction in fuel use—less the Harberger triangle. The prospects for an overall welfare gain in the fuel market are clearly greater (1) the greater the size of the noncarbon externality, relative to the carbon charge; and (2) the less any internalization of the externality through preexisting fuel taxes.
In fact, if the fuel is subsidized rather than taxed—as has (at least up until recently) been the case for petroleum products and natural gas in a number of Middle East and North African countries6—and if a carbon charge were superimposed on top of unreformed subsidies, then the prospects for a welfare gain are further enhanced. However, although coal and natural gas are generally not subject to substantial excises, road fuels in most countries are, and in cases (e.g., some European countries) where they may exceed estimates of noncarbon environmental costs, carbon charges will induce welfare losses in road fuel markets despite domestic environmental benefits.
Of course many countries are using domestic (usually regulatory) measures to reduce air emission rates, but this does not eliminate the domestic efficiency gains from carbon pricing. This point is illustrated in figure 3.2, where a regulation, such as requirements for technologies on new coal plants to reduce sulfur dioxide emissions, shifts up the industry average supply curve (inclusive of technology operation costs) for coal generation and shifts down the industry average environmental damages per unit of coal use, but not to zero. For example, plants operating these technologies still emit some sulfur dioxide, the technologies may not always be switched on, older plants may not have these technologies, and plants produce other air emissions, such as nitrogen oxides and direct fine particulates. The carbon charge still has the potential to induce a net welfare gain (indicated by the shaded trapezoid), albeit a smaller one than in the absence of other regulations.
The (second-best) nationally efficient carbon price—that is, the price warranted by domestic environmental benefits (before counting global climate benefits)—is obtained by differentiating the sum of welfare gains (or welfare losses) in the markets for coal, natural gas, and petroleum products with respect to the carbon price. Not surprisingly, the efficient tax (see Parry, Veung, and Heine, 2014) is higher (1) the greater the magnitude of non-CO2 externalities net of any prior fuel taxes, and (2) the greater the share of CO2 reductions that come from fuels with relatively high noninternalized environmental damages.
Parry, Veung, and Heine (2014) estimated nationally efficient carbon prices for large emitters for 2010 using fuel consumption data (from the International Energy Agency), fuel price and tax/subsidy data (from IMF sources), CO2 emissions factors for fuels,7 fuel price elasticities from the empirical literature,8 and estimates of domestic environmental costs by fuel product and by country from an IMF database developed by Parry, Heine, Lis, and Li (2014)9—the appendix to this chapter provides more details on the measurement of these environmental costs.
Ideally, estimates of nationally efficient carbon prices would be projected forward (say to 2030, a typical year for meeting countries’ emissions pledges for Paris), accounting for future changes in the fossil fuel mix, changes in fuel prices (which affect the proportionate change in fuel prices from carbon pricing), tightening domestic environmental regulations, rising valuations of air pollution mortality risks with growth in per capita income, and so on. Nonetheless, the estimates discussed below are still useful in providing some broad sense of the size of domestic environmental benefits and the extent of variation across countries.
Figure 3.3 shows estimates of nationally efficient carbon prices for the top 20 emitting countries in 2010. These estimates indicate the price levels up to which incremental increases in prices are in countries’ own interests because the extra domestic environmental benefits outweigh the extra mitigation costs—only beyond these prices do incremental abatement costs start to rise above domestic environmental benefits (the point at which, in theory, the free-rider problem should start to kick in).
Although the precise numbers in the figure should not be taken too literally, the two key points are first that the nationally efficient CO2 prices can be substantial and second that the efficient prices differ greatly across countries.
Averaging over the top 20, the nationally efficient price is $57 per ton of CO2, which is a large number—about 60% higher than the social cost of carbon in 2010 according to US Inter-Agency Working Group (2013).
For China, the nationally efficient CO2 price is $63 per ton, which essentially reflects the domestic air pollution benefits from reducing coal use—these benefits are relatively high given that China is densely populated with a lot of exposure to coal plant emissions.10 Although air pollution deaths per unit of coal combustion are dramatically lower in the United States (about one-twelfth of that for China), the nationally efficient CO2 price is still $36 per ton, in part because the valuation of health risks is assumed to be about four times higher for the United States than for China, and a more significant portion of the CO2 reductions come from reductions in road fuels, which are substantially undertaxed from a domestic perspective in the United States.
In contrast, the nationally efficient CO2 price is more moderate in Australia at $12 per ton, in part reflecting the much more limited exposure to air pollution (due to low population density and the coastal location of power plants, where much of the emissions disperse harmlessly over the oceans). The nationally efficient CO2 price is negative for Brazil, where there is little coal use, and a significant portion of the CO2 reduction comes from reduced gasoline use, which is already overtaxed from a domestic perspective. Also striking are Saudi Arabia and Iran, where domestic subsidies for petroleum and natural gas were large in 2010, implying large efficiency gains in the (unlikely) event that carbon pricing were imposed on top of unreformed subsidies (although both countries have since been scaling back these subsidies).
Besides environmental impacts, carbon pricing also raises substantial amounts of revenue, for example, the carbon prices shown in figure 3.3 would have raised estimated revenues averaging 1.9% of GDP across the top 20 emitters in 2010 (Parry, Veung, and Heine, 2014). This raises the issue of whether—leaving aside environmental benefits—raising these revenues from carbon pricing imposes lower costs on the economy than raising them through broader fiscal instruments. If so, fiscal considerations constitute another reason that carbon pricing is in countries’ own domestic interests.
The combined effect of broader fiscal instruments, including personal income taxes, payroll taxes, corporate income taxes, and general consumption taxes, is to distort the economy in two main ways. First, they reduce the overall level of economic activity by reducing real factor returns, thereby deterring work effort and investments in human and physical capital. Second, they distort the composition of economic activity by shifting activity to the informal sector and promoting excessive spending on tax favored goods (e.g., housing, untaxed fringe benefits). In measuring the efficiency costs of broader taxes, it is standard to capture (in a reduced form way) both types of distortion by estimating the responsiveness of the tax base to higher tax rates (e.g., Saez et al., 2012).
Carbon taxes interact with these broader distortions from the fiscal system in two ways. First, using the revenues from carbon pricing to cut the rates of broader taxes produces efficiency gains—termed the “revenue-recycling effect”—and a key theme of the literature is that these efficiency gains are relatively large, so forgoing them (e.g., through using revenues for low value spending or giving away free allowances in trading systems) greatly increases the overall costs of carbon pricing for the economy (e.g., Parry, Williams, and Goulder, 1999). Second, by increasing energy prices and thereby reducing overall economic activity, carbon taxes tend to compound some of the distortions created by broader taxes—the “tax-interaction effect.”
Up to a point, the revenue-recycling effect tends to dominate the tax-interaction effect, implying a negative cost from carbon tax shifts. This is because cutting broader taxes alleviates distortions both in the overall level of economic activity and its composition, whereas (loosely speaking) higher energy prices compound distortions to the level of economic activity but not its composition. The prospects for negative costs are greater when the composition effect accounts for a substantial portion of the overall efficiency costs of broader taxes, which can apply to countries with large informal sectors (many developing countries) or with large preferences in the tax code (e.g., the United States).11 However, accurately quantifying the efficiency gains from carbon tax shifts, and at what carbon price the incremental efficiency impact changes from positive to negative, is hampered by a lack of systematic empirical work across countries on the absolute efficiency costs of broader taxes and the relative contribution of the level and composition effects. Nevertheless, some rough calculations in Parry, Veung, and Heine (2014) suggest that the efficiency gain could, for some countries, be as large as the efficiency benefits from addressing domestic environmental benefits.
The bottom line from the above discussion is that many countries can start the process of fulfilling their emissions mitigation pledges for the Paris Agreement through unilateral carbon pricing that addresses local (environmental and fiscal) needs while also contributing to relieving a global problem—it is not necessary to wait for other countries to act on carbon pricing.
Viewed another way, once an international agreement to enhance and strengthen domestic initiatives has been implemented, it might be a little less challenging to enforce it than previously thought because countries make themselves worse off by reneging on the agreement. The costs of reneging to the general public might be highly visible, for example, in the form of more severe air pollution in emerging market economies, where urban residents see and deal with the health effects of polluted air on a daily basis and air pollution statistics are routinely reported, or in the form of higher direct and indirect taxes that might be needed to compensate for scaling back a large revenue stream from carbon pricing.
But what form should such an international agreement take?
Underpricing from an international perspective is familiar from situations where countries compete for mobile tax bases, in which context some progress has been made through tax agreements, for example, in the European Union for value added taxes and excises on alcohol, tobacco, and energy products. A key lesson here is that it seems easier for countries to agree on tax floors than tax rates, not surprisingly given, for example, heterogeneity among countries in the public health benefits from these taxes (depending on the prevalence of smoking, alcohol abuse, etc.), their fiscal needs, and the political acceptability of higher taxes.
The carbon pricing analog would be a coordinated CO2 price floor among a coalition of willing countries, which could be pursued alongside the Paris process. This arrangement would provide some degree of protection for industries competing with imports from other countries that are party to the agreement and some protection against free-riding and cross-border fuel smuggling. Yet individual countries could set prices exceeding the floor, which is efficient if they have relatively high domestic environmental benefits or fiscal needs. More generally, the political acceptability of carbon pricing differs considerably across countries, and those countries willing and able to set higher prices should not be held back. In contrast, linked emissions trading systems (in their pure form) impose a uniform price across participating countries, and policies (e.g., the UK carbon tax floor) to raise the overall domestic carbon price in one country will only reduce allowance prices without any effect on system-wide emissions, which are fixed by the regional cap.
A carbon price floor agreement could initially be negotiated among a limited number of willing (preferably large emitting) countries and progressively expanded over time with additional participants.12 The arrangement could accommodate both countries with carbon taxes and with trading systems, although in the latter case the systems would need to have explicit mechanisms that permanently withdraw allowances from the system to prevent the price falling below a target level.13
One challenge is how to account for idiosyncratic, special provisions among countries with carbon pricing programs, such as exemptions or reduced rates for influential or vulnerable sectors. Another is how to account for changes in existing energy taxes or subsidies that can enhance or offset the emissions impact of a direct price on carbon.
In principle, both challenges could be addressed through monitoring “effective carbon prices,” which convert direct carbon pricing programs and existing taxes/subsidies on fossil fuel products, electricity, and possibly other products (e.g., vehicles) into an aggregate carbon price. This involves converting all carbon pricing schemes and broader energy taxes/subsidies into an economy-wide average carbon price equivalent. For a carbon price scheme with incomplete coverage, the price needs to be weighted by the fraction of emissions reductions that would come from the covered sector under economy-wide pricing.14 For broader taxes/subsidies on energy products, these should be converted to carbon price equivalents by dividing the tax/subsidy rate by CO2 emissions per unit of the product and then weighting by the share of reductions that would come from that product under economy-wide carbon pricing.15
Figure 3.4 shows some calculations of effective carbon prices across selected countries for 2010. The calculations are highly simplified—just accounting for taxes/subsidies on fossil fuel products16—and somewhat outdated (e.g., Mexico and Indonesia have since scaled back fuel subsidies), but the main point here is that effective carbon prices differ substantially across countries. For example, effective CO2 prices are relatively high (more than $40 per ton) in France, where road fuels account for a relatively high share of nationwide emissions and there are already high road fuel excises, but effective carbon prices are below $10 per ton in most other cases. Given such wide dispersion, achieving convergence in effective carbon prices is likely impractical. But nor is it desirable, as fuel taxes may in part be addressing domestic environmental and fiscal needs rather than CO2 emissions.
Instead, an agreement might focus on raising effective carbon prices in each country by a target amount (e.g., $20 per ton in 2020 and $70 per ton by 2030) relative to the effective price in that country in a (previous) baseline year. Effective carbon prices would need to be independently assessed and consistently measured across countries, although the practicalities should be manageable once countries have agreed to analytical conventions (e.g., over fuel price responsiveness assumptions).
This chapter pushes back on two notions about global carbon policy implied by economic models that do not account for noninternalized, noncarbon externalities in fossil fuel markets and broader distortions to economic activity from the fiscal system.
The first is the rather pessimistic notion that acting unilaterally on carbon pricing inevitably makes an individual country worse off, and, as a consequence, meaningful progress on carbon pricing will not occur until a pricing agreement among large emitters with credible enforcement mechanisms is in place. In contrast, this chapter argues that carbon pricing is in many countries’ own national interests when account is taken of noninternalized local externality benefits, which (up to a point) exceed domestic mitigation costs. Fiscal considerations reinforce this argument to the extent that the efficiency costs of raising revenues from carbon pricing are initially lower than those for broader fiscal instruments (when the full range of distortions created by the latter are properly considered).
It is not inconceivable that policymakers can get these important but arcane-sounding points across to stakeholders with more accessible messages about estimated lives saved from breathing healthier air, improvements in their international urban air quality rankings, specific commitments to cut burdensome taxes or address underinvestment in hospitals, schools, and infrastructure, with the proceeds from carbon pricing, and so on. In this way, significant headway on carbon pricing at the local level might be made, whereas at the international level policymakers and organizations continue dialogue on the practicalities of coordinated regimes for eventually building on local pricing initiatives.
The second notion is that the long-term goal should be a uniform global carbon price. That has always looked impractical, given the principle of common but differentiated responsibilities, and the reluctance (especially in current times of historically high fiscal pressures) of advanced countries to transfer large side payments to developing countries that might otherwise be unwilling to impose the same carbon price. A key point from the current chapter is that a globally uniform carbon price is not economically efficient either, given the wide dispersion across countries in the magnitude of noncarbon externalities from fuel use and the dim prospects as far as the eye can see for perfectly internalizing them through other pricing policies, as well as differences in the efficient amount of carbon pricing from a fiscal perspective.
In short, there are both pragmatic and economic arguments for flexible international regimes built around carbon price floor arrangements. Discussions about these regimes should be welcomed by delegates to the annual UN Framework Convention on Climate Change meetings because they would complement and strengthen the mitigation commitment process already initiated by the landmark 2015 Paris Agreement.
Appendix: Procedures for Measuring Domestic Environmental Externalities
Parry, Heine, Lis, and Li (2014) used four main steps to quantify the air pollution damage from coal plants.
First, data on the geographical location of coal plants in different countries was mapped to granular data on the number of people living at different distance classifications from each plant (up to 2,000 km away, given the potential long-range transport of emissions from tall smokestacks). These data are used to extrapolate “intake fractions”—the average portion of a particular pollutant that ends up being inhaled (as fine particulates) by exposed populations—for different pollutants, from a widely cited study for China, adjusting for population exposure in other countries relative to that in China. Although this approach does not account for differences in meteorological conditions (e.g., wind speeds) between other countries and China, which affects regional pollution formation, some cross-checks with air quality models in a limited number of cases suggest the bias from omitting these factors is not necessarily large and does not follow a systematic pattern.
The second step is to obtain elevated mortality risks by country from additional pollution emissions by linking intake fractions to two pieces of information from the World Health Organization’s Global Burden of Disease project. One is baseline mortality rates in different regions for illnesses (heart, pulmonary and lung diseases, and strokes) whose prevalence is increased by exposure to pollution.17 The other is evidence on the relationship between pollution exposure and additional mortality risk, or “concentration-response” functions.18 One noteworthy issue here is that, although at lower pollution concentrations the concentration-response function is approximately linear, some (although not all) evidence suggests it may flatten out at especially severe concentration levels as people’s channels for absorbing pollution become saturated. Paradoxically, this would imply (given other factors) lower marginal environmental benefits for small pollution reductions in severely polluted regions—this possibility is not taken into account in the estimates presented here.
The third step is to monetize mortality risks. For this purpose, Parry, Heine, Lis, and Li (2014) use estimates of the value per premature mortality for the average OECD country ($3.7 million, updated to 2010) and the elasticity of mortality valuation with respect to income (0.8)—both based on the literature—to extrapolate mortality values for all countries.
The final step is to convert damages per ton of emissions into damages per unit of coal use using a country-level database of coal plant air emissions factors compiled by the International Institute for Applied Systems Analysis. The results presented earlier are based on those for a representative sample of plants with emissions control technologies and are therefore lower than the industry average emission rates (the latter including plants with no control technologies), a possible justification being that industry average emission rates will gradually converge to the former over time as older, dirtier plants are retired from the fleet.
The same steps as discussed previously were used to assess environmental damages for natural gas used in power generation.
Local air emissions from ground-level sources—principally vehicles and residential heating—tend to stay locally concentrated (rather than being transported long distances), which simplifies assessment of their intake fractions. Parry, Heine, Lis, and Li (2014) obtain (from other studies) ground-level intake fractions for air emissions from more than 3,000 cities, extrapolate these to the country level, and then follow the three steps mentioned previously.
In the absence of better data, Parry, Heine, Lis, and Li (2014) regress travel delays from a city-level database on various transportation indicators and then extrapolate delays to the national level using the regression coefficients and country-wide measures of those same indictors. Marginal delay (the delay one extra kilometer driven in one vehicle imposes on other road users) is assumed to be four times the average delay, loosely based on specifications commonly used by transportation engineers. The result is scaled by vehicle occupancy and monetized based on literature, suggesting the value of congested travel time is around 60% of the wage. These estimates likely understate marginal congestion costs; for example, cars impose greater delay to other road users when buses (which have high vehicle occupancies) are a significant share of vehicles on the road.
Parry, Heine, Lis, and Li (2014) estimate accident externalities by country using data on people killed in traffic accidents and assumptions about the decomposition between internal and external risks, for example, injury risks to occupants in single-vehicle collisions are assumed to be internal, whereas risks to pedestrians/cyclists and a portion of injuries in multivehicle collisions are taken to be external. Other components of external costs (e.g., medical and property damages borne by third-parties, a portion of nonfatal injuries) are extrapolated from several country case studies. Road damage costs, which apply to heavy (diesel) vehicles, are obtained from data on road maintenance expenditures (where it is available and extrapolated from other countries where it is not) and assumptions about the portion of wear and tear due to vehicle traffic as opposed to other factors such as climate.
Mileage-related externalities (congestion, accidents, road damage, and local pollution where emissions regulations are expressed on a per mile basis) are scaled back by around 50% in computing Pigouvian taxes, given that only about half of the long run reduction in fuel use from higher road fuel taxes comes from reductions in vehicle mileage (the other half coming from long-run improvements in average fleet fuel economy, which essentially have no effect on mileage-related externalities).
- Bento, A., M. Jacobsen, and A. A. Liu. 2012. Environmental Policy in the Presence of an Informal Sector. Discussion paper. Ithaca, NY: Cornell University Press.
- Clements, B., D. Coady, S. Fabrizio, S. Gupta, T. Alleyene, and C. Sdralevich, eds. 2013. Energy Subsidy Reform: Lessons and Implications. Washington, DC: International Monetary Fund.
- Coady, D., I. Parry, L. Sears, and B. Shang. 2015. How Large Are Global Energy Subsidies? Working Paper 15/105. Washington, DC: International Monetary Fund.
- Harberger, A. C. 1964. The measurement of waste. American Economic Review 54:58–76.
- Organization for Economic Cooperation and Development. 2015. Taxing Energy Use 2015: OECD and Selected Partner Economies. Paris, France: OECD Publishing.
- Parry, I. W. H., and A. M. Bento. 2000. Tax deductions, Environmental policy, and the “double dividend” hypothesis. Journal of Environmental Economics and Management 39:67–96.
- Parry, I. W. H., R. C. Williams, and L. H. Goulder. 1999. When can carbon abatement policies increase welfare? The fundamental role of distorted factor markets. Journal of Environmental Economics and Management 37:52–84.
- Parry, I. W. H., C. Veung, and D. Heine. 2014. How Much Carbon Pricing is in Countries’ Own Interests? The Critical Role of Co-Benefits. Working Paper 14174. Washington, DC: International Monetary Fund.
- Parry, I. W. H., D. Heine, S. Li, and E. Lis. 2014. Getting Energy Prices Right: From Principle to Practice. Washington, DC: International Monetary Fund.
- Saez, E., J. Slemrod, and S. H. Giertz. 2012. The elasticity of taxable income with respect to marginal tax rates: A critical review. Journal of Economic Literature 50:3–50.
- US Inter-Agency Working Group. 2013. Technical Update of the Social Cost of Carbon for Regulatory Impact Analysis Under Executive Order 12866. Washington, DC: US Inter-Agency Working Group.