Chapter 13. Carbon Tax Burdens on Low-Income Households: A Reason for Delaying Climate Policy?
- Benedict Clements, Ruud Mooij, Sanjeev Gupta, and Michael Keen
- Published Date:
- September 2015
- Ian Parry
Without strong measures to mitigate carbon dioxide (CO2) and other greenhouse gases, global temperatures are projected to rise by about 3–4°C over preindustrial levels by the end of this century, with serious risks of catastrophic outcomes.1
Carbon taxes, or similar pricing instruments,2 should be front and center in climate change mitigation. These instruments do the following:
Exploit the full range of emissions mitigation opportunities (shifting to cleaner fuels, reducing energy use, and so on), so long as they are directly targeted at emissions3
Achieve CO2 reductions at lowest cost to the economy, so long as revenues are used productively, most obviously (for advanced countries) to lower the burden of broader taxes on labor and capital
Promote only those emissions reductions for which environmental benefits outweigh the costs, so long as tax rates are aligned with, albeit contentious, environmental damage assessments
Can involve a practical extension of motor fuel excises (long established in most countries and among the easiest of taxes to administer) to other fuel products4
Are in many countries’ own self-interest because of domestic environmental co-benefits—for example, reduced air pollution deaths (for example, Parry, Veung, and Heine, forthcoming)
Require international coordination over one main parameter—a price floor.
Although carbon pricing schemes are emerging in many regions and countries (Figure 1 in World Bank 2014), little over one-tenth of global greenhouse gas emissions in 2013 were formally priced, and most prices are far below what is needed.5 Clearly the politics of carbon taxes are not easy, and at the heart of this is the potential impact on vulnerable (or politically influential) groups. Most debate has surrounded low-income households (the focus of this chapter) as well as trade-exposed, energy-intensive firms (which is beyond the scope).6
Box 13.1Some Prior Literature on Carbon Tax Incidence
There have been several country case studies of carbon and energy tax incidence (mostly before revenue recycling), and some recent cross-country comparisons of energy tax incidence, but not much cross-country assessment of carbon tax incidence. This box provides a quick flavor of some of this literature.
Most work on energy tax incidence focuses on motor fuels, a recent example being Sterner 2012. He finds that these taxes are generally more regressive in the United States than in Europe (primarily because of higher vehicle ownership rates among lower-income households in the United States). And when incidence is measured using consumption expenditure (arguably a better measure—see text) rather than annual income, motor fuel taxes are often progressive, at least across low- to middle-income groups. Poterba (1991) finds similar results for the United States: the burden-to-income ratio for the bottom income quintile is 5.3 times that for the top income quintile with incidence based on annual income, but only 1.5 times that for the top income quintile with incidence based on expenditure.
Casler and Rafiqui (1993) look at a range of taxes on electricity, coal, natural gas, gasoline, and other refined petroleum products for the United States, finding that the tax-burden-to-income ratio overall for the lowest income quintile is only modestly larger than for the top income quintile (in part because indirect price effects are less regressive than the direct price effects). In a similar study, Bull, Hassett, and Metcalf (1994) show that incidence effects of energy taxes are even less regressive with incidence based on consumption. Metcalf (1999) demonstrates that similar energy tax packages can be made distributionally neutral overall for broad income groups (under income and consumption incidence measures) through targeted income and payroll tax reductions.
With regard to country-specific analyses of carbon pricing, U.S. applications include Dinan 2015; Hassett, Mathur, and Metcalf 2009; Morris and Mathur 2015; Parry and Williams 2010; and Rausch, Metcalf, and Reilly 2011, with most focusing on neutralizing adverse distributional effects through revenue recycling. Before recycling, the burden-to-income ratio for low-income households is roughly three to five times that for high-income households, or about 1.5 times as high, based on income and consumption incidence measures, respectively. The two big direct items are motor fuels and electricity, with the budget share for electricity falling most rapidly with higher income. Similar analyses for other countries include Australia (Cornwell and Creedy 1997), Denmark (Wier and others 2005), and Sweden (Brännlund and Nordström 2004).
With regard to cross-country assessments, an early study by Smith (1992), covering a variety of countries and focusing on income-based incidence, finds that carbon and energy taxes are moderately regressive for most countries, but more so for the United Kingdom and Ireland (where, because of colder climates, heating and electricity needs are greater). A study of European countries (measuring incidence against expenditure) by OECD (2014) finds that transportation taxes are generally progressive; heating fuel taxes are regressive, but only slightly so; and electricity taxes are more regressive.
Having a sense of the distributional incidence of carbon taxes across households is critical to informing policy dialogue and aiding in designing accompanying measures—either to protect low-income households or alter the overall distributional equity of the fiscal reform. Policies perceived as broadly fair in this regard are not only desirable for their own sake (more so, given recent trends toward greater inequality in after-tax incomes—see Immervoll and Richardson 2011) but also may have a better chance of being enacted and sustained.
This chapter takes stock of methods for estimating household incidence and applies them, albeit crudely, across advanced countries (where data are more available)—Box 13.1 provides some discussion of the rather limited previous literature on carbon tax incidence.
The main theme of the discussion is that, for several reasons, distributional concerns should not hold up carbon taxes:
Carbon taxes are less regressive (disproportionately burdening low-income groups) than they might first appear. Although the poor generally spend a greater share of their annual income on electricity,
➢ This is less true for transportation and heating fuels and other consumer products whose prices increase indirectly from higher energy costs
➢ Regressive effects are less pronounced when incidence is measured against household consumption (arguably a better measure of household well-being than income)
➢ Cross-country calculations suggest the incidence of carbon taxes can be anything from moderately regressive, to proportional, to moderately progressive.
Undercharging for carbon damages is a highly inefficient way to help low-income households, because the vast majority of benefits (typically about 90 percent) leak away to higher-income groups. Targeted measures are much more effective.
There are ample opportunities for compensating low-income households in advanced countries through targeted fiscal and spending adjustments and combining carbon taxes with other progressive measures (for example, reducing tax preferences for the wealthy), though the specifics will vary across countries (with the extent of inequality, parameters of the fiscal system, and other factors). The focus should be the distributional impact of the whole policy package, not just the component that raises energy prices.
A second theme of the chapter (though one already receiving much emphasis in the literature) is that diverting carbon tax revenues from the general budget for compensation involves significant costs. Reasonably accurate estimates of burdens on vulnerable groups are therefore important for gauging appropriate compensation levels. And, insofar as possible, compensation instruments that enhance economic efficiency (for example, targeted tax cuts or tax credits that strengthen incentives for work effort) should be used instead of instruments that do not (for example, transfer payments).
The chapter is organized in two main parts. The first deals with conceptual and measurement issues, while the second gives some very broad quantitative sense of the incidence effects of carbon taxes across advanced countries (prior to recycling) and discusses options for simultaneous adjustments to the broader tax and benefit system. The last section provides concluding remarks.
Conceptual and Measurement Issues
The economic burden or incidence of a tax refers to whose economic welfare is reduced by this tax and by how much. Economic incidence is quite different from formal or legal incidence—for example, fuel suppliers may remit carbon tax payments to the government, but bear little incidence if the charge is mainly reflected in higher prices for fuel users. The point in the fossil fuel supply chain at which carbon taxes are levied is largely irrelevant for the ultimate incidence of the tax.
This section presents the most basic analytical model of carbon tax incidence, discusses the measurement of key factors determining incidence, considers some complicating factors, and emphasizes the importance of using carbon tax revenue efficiently.
Basic Model of Carbon Tax Incidence
Consider a static model with the following assumptions (which are relaxed later):
The carbon tax burden is fully passed forward into consumer prices
Co-benefits (for example, better local air quality) from carbon pricing are ignored
Incidence is measured before adjustments in the broader tax and benefit system.
In this setting, carbon tax incidence for a particular household group reflects their loss of consumer surplus, aggregated across all consumer products whose prices rise in response to the higher energy costs caused by the carbon tax.
Consider first a tax on the carbon content of a single fuel, denoted by X, directly consumed by households (for example, gasoline or residential natural gas). Households are classified into i = 1…N income groups, where Ii (defined more carefully below) is average income for group i. The fuel demand curve for group i (taken to be linear over the relevant range) is shown in Figure 13.1, where the height of this curve at any point reflects the benefit to fuel users from an extra unit of consumption.
Figure 13.1Burden of a Fuel Tax for a Household Group
Note: The red trapezoid indicates the burden (loss of consumer surplus) to a household group from the increase in price (caused by a carbon tax) for a particular fuel.
The fuel supply curve is perfectly elastic (that is, fuel is produced competitively under constant returns) with pX0 denoting the per unit production cost or supply price. As drawn in Figure 13.1, there is no preexisting fuel tax, so pX0 is also the initial (precarbon tax) price to fuel users (preexisting taxes are discussed below). Initial fuel consumption for household group i is Xi0, and consumer surplus—benefits to fuel users (the area under the demand curve integrated between the origin and Xi0) less payments to fuel suppliers—is area abc.
Suppose a carbon tax is introduced, increasing the fuel user price to pX1, where pX1 – pX0 equals the tax rate per ton of CO2 times the fuel’s emissions factor (CO2 per unit of X). Fuel consumption falls to Xi1 and consumer surplus falls by trapezoid edbc, consisting of the first order loss edfc, caused by paying a higher price on consumption (the tax payment), plus dbf, or benefits from forgone consumption Xi0 – Xi net of what households would have paid for that consumption.
The consumer surplus loss or tax burden for group i, denoted Bi, can be written as in equation (13.1):
Expressing the burden relative to group i’s income is given by equation (13.2):
This is the proportionate fuel price increase (the same for all households) times group i’s budget share for the fuel. In this highly simplified case, tax incidence is regressive if fuel budget shares are higher for lower-income groups; proportional if all groups have the same budget share; and progressive if higher-income groups have higher budget shares. Alternatively, px0 could be cancelled from equation (13.2), but data at the household level are generally reported on product expenditures (px0Xi0), rather than on physical consumption units (Xi0).
More generally, a carbon tax directly increases consumer prices for several energy products (electricity, heating and transportation fuels) and indirectly for goods in general (by raising the costs of intermediate energy inputs). The formula in (13.2) readily generalizes to the case of multiple consumer price increases, so the burden to income approximation is simply equation (13.3):
in which superscript j indexes a consumer product j, and there are j = 1…M product groupings. The price increase for a particular product such as cars, for example, will reflect the CO2 tax times embodied carbon, that is, CO2 emissions per vehicle from the electricity and fuels used in the manufacture of component parts and assembly.
For advanced countries (see below), carbon taxes appear to be regressive because lower-income households have a relatively high propensity to spend out of current income, and energy products tend to be, albeit weak, necessities,8 though incidence is less regressive when the full range of products whose prices increase are taken into account.
Measurement Issues: A Quick Look
Previous studies (such as those in Box 13.1) have measured incidence effects of carbon taxes (or broader environmental and energy taxes), using formulas like that in equation (13.3), or more sophisticated versions of it (for example, incorporating behavioral responses). Implementing equation (13.3) requires three main pieces of data: price impacts and, by household group, income and product expenditures.
Absolute price increases for energy products directly consumed by households are easily calculated by the tax rate times the CO2 emissions factors for the respective fuels. These factors are essentially fixed for motor fuels and natural gas, and vary very little across countries,9 though CO2 emissions rates per unit of electricity—available by country from the International Energy Agency (IEA)—obviously vary with a country’s power generation fuel mix. Proportionate price increases can be obtained using baseline price data, which are available from IEA for advanced countries and estimated for other countries by Clements and others (2013). Direct effects on energy prices account for about two-thirds of the total estimated burden of a carbon tax for the United States, and likely somewhat more than this for other advanced countries (see below).
Input/output tables can be used to estimate the indirect impacts of carbon taxes on the prices of other consumer products. These tables provide, for different industries, the value of outputs for final and intermediate products, and inputs—both energy (fuels, electricity) and non-energy (labor, capital, raw materials, and so on). Dividing fuel and electricity purchases by fuel prices, and applying emissions factors, the embodied CO2 per dollar for each intermediate product, and ultimately each final product, can be inferred, and multiplied by the CO2 tax to give the proportionate price increases.10
One complication is that the CO2 emissions factor for power generation, and embodied carbon in various non-energy consumer products, will fall somewhat in response to carbon pricing as firms adopt energy-saving technologies, power generators switch to cleaner fuels, and so on, though (for similar reasoning as noted above) it may be reasonable to ignore this dampening effect on consumer prices for the scale of carbon taxes considered here.
Incidence Relative to Income or to Expenditure?
The appropriate definition of income against which tax burdens, including carbon tax burdens, should be measured for different household groups is somewhat unsettled. Annual income is problematic given that many people with low annual income (for instance, students, retirees with high accumulated savings, people temporarily laid off or on maternity leave) are not poor in a life-cycle context, yet they contribute greatly to disparities in annual income across households.11 This problem is partly (though, because of constraints on consumption smoothing across the lifecycle, not fully) alleviated by measuring incidence against annual consumption expenditure rather than income. Incidence studies based on expenditure suggest that the potentially regressive impacts of energy or carbon taxes are much less pronounced (Poterba 1991; Hassett, Marthur, and Metcalf 2009).12
Spending by household income groups on energy and non-energy products is often available from household expenditure surveys. Many advanced countries routinely conduct these surveys;13 the World Bank’s Living Standards Measurement Study compiles them for approximately 40 developing countries; various other developing countries (for example, Bangladesh, Cambodia, India) administer surveys themselves. There are some concerns about the accuracy of the surveys (collecting data from thousands of households is difficult and costly), including the ability of households to remember or accurately record their expenditures, the design of the survey instrument, and data collection or entry (see Xu and others 2007).
Some Further Considerations
This subsection discusses the possible passback of carbon taxes in lower supply prices and the incidence implications of co-benefits from carbon pricing.
Passback of Carbon Taxes in Lower Supply Prices
Some minor fraction of the burden of carbon taxes may be passed backward in lower producer prices, to the extent that fuel supply curves are upward sloping in the medium to longer term (for example, due to scarce inputs or long-lived, sector-specific capital). To the extent this reduces the net-of-tax return to capital, some of the carbon tax burden is borne by owners of capital (in lower equity values or dividends) with progressive effects because the better off earn a relatively high share of their income from capital (Metcalf and others 2008). But if, as seems plausible even for large economies like the United States, net-of-tax returns to capital are largely determined in world markets, the burden of lower supply prices will tend to be borne by labor in the form of lower wages. The incidence implications here, which depend on whether pollution-intensive industries disproportionately employ low- or high-wage workers; and on substitution elasticities between labor, capital, and polluting inputs and so on; become difficult to estimate (Fullerton and Heutel 2011). On balance, studies for the United States suggest that the passback of carbon taxes reduces the regressivity of carbon taxes (Rausch, Metcalf, and Reilly 2011), though the empirical effects are model specific.
Carbon pricing can produce significant environmental co-benefits, for example, reduced air pollution from coal combustion and externalities like congestion from motor vehicles, at least until these other externalities are fully priced through other policies. Co-benefit estimates can be quite large, averaging $57.5 per ton of CO2 across the top 20 emitters (Parry, Veung, and Heine, forthcoming), though with substantial cross-country variation (due to sharp differences in population exposure to pollution, for instance).
Suppose, not unreasonably, that peoples’ valuation of air pollution and other externality benefits is roughly proportional to income (OECD 2012). Then the distribution of local air pollution co-benefits across income groups within a country may be progressive if exposure risks decline with income (because the wealthy reside in areas with cleaner air). Vehicle externality benefits (largely borne by motorists as a group) are also progressive (though to a lesser extent) given that miles driven generally rise by less than proportional to income (Pickrell and Schimek 1997). These effects are not definitive, however—for example, reduced air pollution might positively affect local property values, with adverse effects for low-income renters.
Implications and Importance of Revenue Recycling
Accounting for the use of carbon tax revenues is critical because it is the overall incidence of the policy change, including accompanying adjustments to the broader tax and social safety net system, that matters, not just the component reflecting higher energy prices. This subsection discusses how revenue recycling might affect distributional incidence and underscores the potentially high opportunity costs to diverting substantial revenues to compensation schemes.
Recycling and Incidence
in which αi is household group i’s gain (approximated by the first order transfer) from broader tax and benefit adjustments as a proportion of their income. Clearly, the regressivity of the tax could be reduced if this gain were larger as a fraction of income for low-income households than for high-income households.
Unfortunately, accurate estimates of the distributional benefits from alternative adjustments of tax and benefit schedules through carbon tax recycling are not presently available for most countries. These schedules are typically complex (with numerous tax brackets subject to different marginal rates, various means-tested tax credits and benefits, and so on) so microsimulation data sets (readily available, for example, for the United States and the United Kingdom, but less so for other countries) are really needed to run individual household incomes (which can be heterogeneous) through these schedules to compute net tax liabilities, before aggregating over household samples within a broad income category.
Even these data-based fiscal incidence studies involve inaccuracies in that they do not typically account for behavioral responses to tax and benefit changes, nor the complicated ways in which relative prices (for example, for labor and capital) might be affected, with second-round implications for household burdens (Boadway and Keen 2000). Computable general equilibrium models can incorporate these effects, though the results are typically sensitive to structural and parameter assumptions, and these models cannot take advantage of the highly disaggregated data reflecting intricate details of the tax and benefit systems used in data-based studies.
One further point relates to the potential compensation for low-income households through automatic indexing of transfer and other benefit programs to higher general price levels, which is considered in some incidence studies. For example, in the United States, indexing of social security payments to seniors and federal income taxes to the consumer price level, and food purchase vouchers for low-income households to food prices, would offset about 10 percent of the burden of a carbon tax on the bottom income quintile according to Dinan 2015.14
Costs from Revenue Diversion
The key point from the literature on interactions between environmental taxes and the broader fiscal system is the critical importance of efficient revenue use—most obviously cutting other distortionary taxes on labor, capital, and consumption—for containing policy costs and improving social welfare (for example, Goulder and others 1999). The point is even more valid in light of public finance literature (Feldstein 1999; Saez, Slemrod, and Giertz 2012), underscoring that broader taxes significantly distort not only the level of economic activity (by reducing the returns to work effort and capital accumulation) but also its composition (by promoting informality and excessive spending on tax-preferred goods like housing and fringe benefits).
Figure 13.2 underscores the point with illustrative calculations of the efficiency costs of a carbon tax (excluding environmental benefits), expressed as a percentage of GDP, for a representative, large-emitting economy, under plausible, though very crude, assumptions about the price-responsiveness of emissions and the efficiency costs of broader taxes. The main point is that with revenue recycling, the efficiency costs of carbon taxes that reduce emissions by about 20 percent or less (that is, CO2 taxes up to about $60 per ton) are very small, and perhaps even slightly negative,15 while deadweight costs are substantial—0.75 percent of GDP for a 20 percent emissions reduction—when carbon tax revenues are returned lump sum or in other ways that do not increase economic efficiency.
Figure 13.2Illustrative Cost of a Carbon Tax, with and without Revenue Recycling
Source: Based on formulas in Parry and Williams 2010.
Note: The figure assumes a CO2 intensity of 0.5 tons per $1,000 of GDP (about average for the top 20 CO2-emitting countries—see Parry, Veung, and Heine, forthcoming); each successive $3 per ton increase in the CO2 tax reduces emissions by 1 percent; and a marginal efficiency cost of $0.3 per extra dollar of revenue raised from broader taxes, with 60 percent and 40 percent of this due to distortions in the level of, and composition of, economic activity, respectively. Costs with recycling are moderately negative over some range, which differs from the general tax results derived in Diamond and Mirlees 1971, due to the inclusion here of additional distortions (affecting the composition of economic activity) from the broader fiscal system, beyond those in factor markets (affecting only the level of economic activity).
A key implication is the potentially high opportunity costs from diverting revenues from the general budget to compensation schemes that do not enhance economic efficiency. Avoiding unnecessary compensation is therefore important and, where possible, compensation measures should be used that promote economic efficiency (for example, targeted reductions in personal income and payroll taxes, and earned income tax credits, all of which increase returns from formal work effort).
A first-pass assessment of carbon tax incidence in many different countries, before revenue recycling, can be obtained based on CO2 emissions factors, energy prices, input/output tables (to measure indirect consumer price effects), and expenditure patterns for household income groups. Although there are complicating factors (for instance, behavioral responses, changes in producer prices, co-benefits from carbon pricing), it is difficult to make general statements about how they affect overall incidence, even directionally. Broader adjustments to the tax and benefit system that accompany carbon taxes need to be integrated into incidence analyses, while making transparent that revenue diversion from the general budget for compensation involves significant costs, for example, by reducing opportunities to cut other taxes that distort economic activity.
Box 13.2Carbon Tax Incidence in Developing Countries
This box makes three broad points about carbon tax incidence in developing countries.
First, carbon taxes might be less regressive in developing countries than in advanced countries if access to the power grid and vehicle ownership is skewed toward higher-income groups (though low-income households are affected by diesel fuel taxes indirectly through higher bus fares). Some suggestive evidence to back up this possibility comes from incidence studies of petroleum product subsidies, which are largely concentrated in the Middle East and North Africa. According to Figure 3.12 in Clements and others (2013), on average, the bottom income quintile receives only 3 percent of the benefits from gasoline subsidies, 7 percent from diesel fuel subsidies, and 4 percent from liquefied petroleum gas subsidies.
Second, the case for including carbon taxes as part of the fiscal system could be stronger in developing countries than in advanced countries if large informal sectors constrain the revenue bases of broader fiscal instruments such as personal income taxes and value-added tax. Carbon and more general energy taxes can help broaden the tax base into the informal sector (Bento, Jacobsen, and Liu 2012).
Third, however, targeted compensation for low-income households can be more challenging, at least if many people are not formally registered as taxpayers or benefit recipients. Besides strengthening social safety nets, use of carbon tax revenues for broader spending on health, education, housing, job programs, clean fuel alternatives, and so on may be needed to maintain equity objectives in light of higher energy prices, though this likely involves greater leakage of program benefits to the non-poor.
Incidence Across Advanced Countries: A Broad Picture
The subsections here pull together various data sources to paint a very broad picture of overall carbon tax incidence for advanced countries, before and after revenue recycling. Box 13.2 provides some brief remarks for developing countries.
Incidence before Recycling
The discussion starts with the first-order, economy-wide burden of a CO2 tax using an illustrative tax rate of $35 per ton16 and the impacts of the tax on various prices for both energy and other products. These results are then matched to data on household budget shares.
Figure 13.3 shows, for selected advanced countries in 2012 (and ignoring behavioral responses), the first-order burden of the $35 CO2 tax—the tax revenue as a percentage of GDP, decomposing the tax that would be paid by primary fuels.17
Figure 13.3First-Order Burden from a $35 per Ton CO2 Tax, Selected Countries, 2012
The tax burdens vary from less than 0.5 percent of GDP in Denmark, France, Norway, Sweden, and Switzerland—countries where there is relatively little use of coal in particular, but also natural gas—to more than 1.5 percent of GDP in the Czech Republic, Korea, and Poland—which are all relatively heavy coal users.
Figure 13.4 provides some flavor of the direct impacts of a $35 per ton CO2 tax on household energy prices in advanced countries. Specifically, it shows (assuming full pass through and calculated as discussed above) the percentage increase in energy prices for 2012 (or thereabouts) for residential electricity, natural gas, gasoline, and diesel.
Figure 13.4Increase in Energy Prices for a $35 per Ton CO2 Tax, Selected Countries, 2012
The absolute price increases for natural gas, gasoline, and diesel are essentially uniform for each fuel across countries (given uniform emissions factors), though proportionate price increases are larger in countries with relatively low energy prices. For example, household natural gas prices increase by about 20 percent in Canada, Mexico, and the United States (where supply costs and taxes are low), but less than 6 percent in Chile, Denmark, Greece, Japan, and Sweden (where supply costs are higher and, for Denmark and Sweden, specific taxes are levied in addition to value-added tax). Similarly, gasoline and diesel prices increase by more than 8 percent in Mexico and the United States (where fuel taxes are relatively low), but these prices rise by less than 5 percent in most other cases (the percentage increase in diesel prices is a bit larger than for gasoline because of the higher emissions factor and, in many cases, lower tax rate for diesel).
The absolute price increase differs substantially for electricity, from more than US$0.025 per kilowatt hour (kWh) in Estonia, Greece, and Israel (where power generation is fossil-fuel intensive) to about US$0.0025 per kWh or less in France, Norway, Sweden, and Switzerland (where there is little reliance on fossil-fuel generation). Percent price increases exceed about 15 percent in Estonia, Israel, Mexico, Korea, and the United States, while they are less than 2 percent in France, Norway, Sweden, and Switzerland.
With regard to the indirect impacts of carbon taxes on the prices of other consumer products, for the United States, Morris and Mathur (2015) estimate these price effects for a CO2 tax of $15 per ton in 2010. The prices of other products rise, but in most cases by a fairly modest percentage compared with energy price increases. For example, some consumer prices (clothing and health, for instance) rise by less than 0.5 percent, and others (for example, mass transit, household supplies) by between 0.5 and 1 percent, though in some cases (such as auto purchases and parts, air and transit travel) price increases are somewhat more significant at between 1 and 2 percent.
Figure 13.5 shows the contribution of direct and indirect price effects to the burden of a carbon tax, expressed relative to income and consumption, as borne by different income quintiles in the United States. Burdens from indirect price effects are smaller than those from the direct price effects and smaller in relative terms for lower-income groups (the size of the indirect burden relative to the direct burden falls from 54 percent for the top income quintile to 42 percent for the bottom income quintile for the consumption-based measure). Therefore, accounting for indirect effects moderately increases progressivity.
Figure 13.5Direct and Indirect Burden of a Carbon Tax by Income Quintile in the United States, 2010
Source: Morris and Mathur (2015).
Note: Calculations are a simple average over data for income deciles.
Speaking very loosely, the increases in non-energy consumer products in other advanced economies might be expected to either follow a pattern broadly similar to those in Morris and Mathur 2015 for countries like Estonia, Korea, and Mexico (where proportionate increases in electricity prices are broadly similar to those for the United States) or to rise more moderately in the majority of cases (where proportionate increases in electricity prices are less pronounced than in the United States). In the latter cases, this moderate rise implies an even smaller burden from indirect price effects relative to the burden from direct price effects.
A consistent, cross-country database of carbon tax incidence is not available from previous studies, because these studies either have an individual-country focus or look at the incidence of energy taxes (see Box 13.1). Here the chapter crudely extrapolates carbon tax incidence for advanced countries as follows:
The analysis starts with data, by income quintile, on the share of consumption paid in electricity, heating, and motor fuel taxes, reported in OECD 2014. These figures are divided by tax rates (from IEA 2014) and multiplied by the absolute energy price increases underlying Figure 13.4 (as a proxy for heating by natural gas) to give the burden by quintile of the direct impacts on energy prices of a $35 per ton CO2 tax. Finally, burdens from indirect price increases are inferred using, by quintile, the ratio of burdens from indirect and direct price increases from Figure 13.5 (for consumption), after scaling these ratios by the proportionate increase in electricity prices in other countries relative to those in the United States.18
The result, for selected countries, is shown in Figure 13.6. Cross-country differences reflect differences in carbon-tax-induced energy price impacts (relatively high for Estonia and the United Kingdom and low for Belgium and France—see Figure 13.4) and initial budget shares for energy. The main point is that, within countries, effects vary from moderately regressive (for example, in Austria, the Czech Republic, Estonia, Poland, the United Kingdom), to roughly proportional (Belgium, Finland, Germany), to moderately progressive (Slovenia, Turkey).
Incidence after Recycling
This section considers distributional impacts from the use of carbon tax revenues, first if used for reductions in other energy excises (on electricity and vehicles) and then for adjustments to the broader tax and benefit system.
Reducing Other Energy Taxes
One possibility for offsetting, in a transparent way, some of the carbon-tax-induced burden on households from higher energy prices might be to lower preexisting taxes on energy that the carbon tax makes redundant on climate grounds. The main targets here are excises on residential electricity consumption, and on vehicle sales, that are applied across many advanced countries.19
Figure 13.7 shows, for a selection of advanced countries, what portion of the increase in residential electricity prices, and motor fuel prices (weighting across gasoline and diesel fuel prices) induced by a $35 per ton CO2 tax could be offset by lowering preexisting excises, where they apply. For 11 out of 23 countries, the burden on households from higher electricity prices could be completely neutralized by offsetting reductions in electricity excises, or put another way, current excises exceed (often by well over 100 percent) the impact of carbon taxes on residential electricity prices.20 And in 18 cases, the burden of higher motor fuel prices could be neutralized, on an annualized basis, by lowering excises on motor vehicle sales.
Figure 13.7Burden of Higher Household Energy Prices that Could Be Offset by Lowering Other Energy Taxes, Advanced Countries, 2010
Note: For electricity, the bars show the proportion of the increase in residential electricity prices from a $35 per ton CO2 tax that could be offset by lowering a preexisting excise tax on residential electricity consumption. For motor fuels, the bars show the proportion of the increase in motor fuel prices from the same carbon tax that could be offset (on an annualized basis) by lowering excises on vehicle sales.
However, many countries have little scope for offsetting the burden of higher natural gas prices on households, or the indirect impacts on the prices of other consumer products, by cutting preexisting excises. And of course, scaling back electricity and vehicle excises will moderately offset some of the emissions reductions from carbon taxes and reduce, quite substantially, the net revenues from the carbon tax—potentially in proportion to the share of residential electricity and motor fuels in total CO2 emissions.21
Adjusting the Broader Tax and Benefit System
The distributional impacts of potential adjustments to the broader tax and benefit system accompanying a carbon tax will depend on country specifics, such as the underlying distribution of pretax income and rates and brackets in personal income tax schedules; therefore, the discussion here offers only some general remarks, along with some statistics for the United States.
Figure 13.8 shows the share of income from various federal taxes, and a prospective carbon tax, in the United States by income quintile. Payroll taxes are approximately proportional up to the top income quintile, where they become regressive (because of the cut-off beyond which the marginal tax rate drops to zero). In contrast, the personal income tax is quite progressive, because of tax credits at the bottom end (making the average tax burden negative), tax allowances, and rising marginal rates. Less important with regard to tax burdens are excises that are regressive and corporate income taxes that are progressive at the upper end (though accurately measuring the incidence of this tax is difficult; see below).
Figure 13.8Share of Federal Taxes by Income Quintile, United States, 2009
With regard to personal income taxes, a proportional reduction in marginal tax rates will perform well on efficiency grounds (improving incentives for work effort and savings, and reducing incentives for tax sheltering) but will have regressive effects. In contrast, raising personal income tax thresholds benefits low-income households disproportionately relative to their income, though the tax reduction (the increase in the threshold times the household’s marginal tax rate) is larger in absolute terms for higher-income households. Extending tax credits (for child care, for instance) is a bit better in this regard because the value of the credit is the same for all taxpayers. Better still (for targeting low-income households) is expanding earned income tax credits, which phase out as income rises. None of these measures reach all low-income households, however, because some of them do not pay income tax. Moreover, the economic efficiency gains from these tax reductions are smaller than from cuts in marginal tax rates—they promote labor force participation and shifting from informal to formal work effort, both of which depend on average tax rates, but not hours worked on the job or shifting from tax-preferred to ordinary spending, both of which depend on marginal tax rates. Efficiency gains from cutting payroll rates are likely to be somewhat smaller than those for cutting income tax rates because there are fewer exemptions and deductions for payroll taxes (for example, fringe benefits are exempt but mortgage interest is not).
The burden of corporate income taxes largely falls on labor in countries with globally integrated capital markets (where the net-of-tax return to capital cannot be pushed down below that available in other countries). Although the burden across workers at different earnings levels is unclear, capital and skills may be complementary so that cutting corporate income taxes benefits high-wage workers disproportionately. Cutting capital taxes likely produces the biggest source of efficiency gains, given the international mobility of the tax base, though in large part at the expense of welfare losses in other countries suffering the capital flight; indeed, tax theory (Kanbur and Keen 1993) suggests that other countries will respond by cutting their own capital taxes, thereby dampening efficiency gains in the domestic economy.
Compared with tax systems, benefit programs potentially offer a far more efficient means for redistributing to lower-income households. The drawback of most benefit programs (for example, social security, unemployment, housing, family-dependent benefits) is that they do not increase economic efficiency. Nonetheless, at least for the United States, only a minor fraction of the 10 percent or so of carbon tax revenues needed to keep the bottom income quintile whole (40 percent, or 4 percent of carbon tax revenue, according to Dinan 2015) need come from these types of instruments.
Finally, in addition to recycling carbon tax revenues in tax cuts and benefit increases, numerous other accompanying fiscal adjustments could alter the incidence of a fiscal package containing a carbon tax, most obviously scaling back tax expenditures that disproportionately benefit the better off. For example, cutting relief for pension contributions, estates transferred at death, state and local income taxes, mortgage interest, charitable giving, capital gains and dividends, could all have progressive effects (CBO 2013).
The general theme of the above discussion is that distributional concerns are potentially manageable. For one thing, the energy price impacts of carbon taxes may not be that regressive, at least if incidence is measured against household consumption, and may be proportional, or even progressive in some countries. For another, there are numerous opportunities in advanced countries to compensate low-income households by recycling carbon tax revenues in targeted tax cuts and benefit increases and to combine carbon taxes with other progressive measures, though the specifics need to be carefully examined on a country-by-country basis. Failing to charge for carbon damages is, in fact, a very inefficient distributional policy because the vast majority of benefits leak away to higher-income groups, or put another way, only a small fraction of the carbon tax revenues are needed to compensate low-income households. Even for this group, a large portion of the compensation could take the form of tax cuts and earned income tax credits (that have some efficiency benefits) rather than transfer payments (that do not). With regard to the recycling of other revenues, policymakers typically need to trade off efficiency and distributional objectives (tax reductions with higher efficiency gains tend to be more regressive) though there is always scope for other adjustments (for example, reforming tax preferences) to promote distribution neutrality in a broad fiscal package.
The general message of this chapter is that, although distributional concerns are potentially important for both fairness and the politics of reform, they should not hold up the establishment of a robust price on CO2 emissions.
This discussion concludes by noting a couple of carbon pricing schemes that seem needlessly costly. One is emissions trading systems in which allowances are given away for free to existing emitters. Not only does this policy forgo large economic efficiency gains from recycling revenues in other tax reductions, it does nothing to improve distributional outcomes across households—in fact, it may greatly worsen them by transferring windfall profits to wealthy households via higher equity values for firms receiving free allowances (Dinan and Rogers 2002; Parry 2004). The other scheme, “tax and dividend,” also forgoes large efficiency benefits, by returning 100 percent of carbon tax revenues to households in equal lump-sum transfers. As emphasized above (for the United States), only about 10 percent of revenues are needed to compensate bottom income quintiles, and only a minor portion of this need take the form of instruments (like transfer payments) that do not increase economic efficiency.
2012. “Environmental Policy in the Presence of an Informal Sector.” Discussion paper, Cornell University, Ithaca, New York..
2000. “Redistribution.” In Handbook of Income Distribution, edited by A.Atkinson and F.Bourguignon (ed). Amsterdam: North Holland..
2012. “Unilateral Climate Policy Design: Efficiency and Equity Implications of Alternative Instruments to Reduce Carbon Leakage.” Energy Economics34(Supplement 2): S208–S217..
2004. “Carbon Tax Simulations Using a Household Demand Model.” European Economic Review48(1): 211–33..
1994. “Who Pays Broad-Based Energy Taxes? Computing Lifetime and Regional Incidence.” Energy Journal15(3): 145–64..
2015. “Administration of a US Carbon Tax.” In Implementing a US Carbon Tax: Challenges and Debates, edited by I.Parry, A.Morris, and R.Williams, (ed)38–61. New York: Routledge..
1993. “Evaluating Fuel Tax Equity: Direct and Indirect Distributional Effects.” National Tax Journal46(2): 197–205..
Clements, Benedict, DavidCoady, StefaniaFabrizio, SanjeevGupta, TrevorAlleyene, and CarloSdralevich, (ed) eds. 2013. Energy Subsidy Reform: Lessons and Implications. Washington: International Monetary Fund.
Congressional Budget Office (CBO). 2012. The Distribution of Household Income and Federal Taxes, 2008 and 2009. Washington: Congressional Budget Office.
Congressional Budget Office (CBO). 2013. The Distribution of Major Tax Expenditures in the Individual Income Tax System. Washington: Congressional Budget Office.
1997. “Measuring the Welfare Effects of Tax Changes Using the LES: An Application to a Carbon Tax.” Empirical Economics22(4): 589–613..
1971. “Optimal Taxation and Public Production I: Production Efficiency, and II: Tax Rules.” American Economic Review61: 8–27, 261–78..
2015. “Offsetting a Carbon Tax’s Burden on Low-Income Households.” In Implementing a US Carbon Tax: Challenges and Debates, edited by I.Parry, A.Morris, and R.Williams, (ed)38–61. New York: Routledge..
2002. “Distributional Effects of Carbon Allowance Trading: How Government Decisions Determine Winners and Losers.” National Tax Journal55(2): 199–222..
1998. “Gasoline Demand Revisited: An International Meta Analysis of Elasticities.” Energy Economics20(3): 273–95.
2004. “Turning on the Lights: A Meta Analysis of Residential Electricity Demand Elasticities.” Journal of Agricultural and Applied Economics36(1): 65–81..
1999. “Tax Avoidance and the Deadweight Loss of the Income Tax.” Review of Economics and Statistics81(4): 674–80..
2015. “Carbon Taxes and Energy Intensive Trade Exposed Industries: Impacts and Options.” In Implementing a US Carbon Tax: Challenges and Debates, edited by I.Parry, A.Morris, and R.Williams, (ed)159–77. New York: Routledge..
2011. “Analytical General Equilibrium Effects of Energy Policy on Output and Factor Prices.” The B.E. Journal of Economic Analysis & Policy10(2): 1–26..
1999. “The Cost-Effectiveness of Alternative Instruments for Environmental Protection in a Second-Best Setting.” Journal of Public Economics72(3): 329–60..
2009. “The Incidence of a US Carbon Tax: A Lifetime and Regional Analysis.” Energy Journal30(2): 155–78..
International Energy Agency (IEA). 2014. World Energy Statistics and Balances. Paris: International Energy Agency.
International Monetary Fund (IMF). 2011. “Promising Domestic Fiscal Instruments for Climate Finance.” Background Paper for the Report to the G20 on “Mobilizing Sources of Climate Finance,” International Monetary Fund, Washington. www.imf.org/external/np/g20/pdf/110411b.pdf.
International Monetary Fund (IMF). 2015. World Economic Outlook Database. International Monetary Fund, Washington. www.imf.org/external/pubs/ft/weo/2014/02/weodata/index.aspx.
2011. “Redistribution Policy and Inequality Reduction in OECD Countries: What Has Changed in Two Decades?” IZA Discussion Paper No. 6030, Institute for the Study of Labor, Bonn, Germany..
1993. “Jeux Sans Frontiers: Tax Competition and Tax Coordination When Countries Differ in Size.” American Economic Review83: 877–92..
1977. “Inequality: Earnings versus Human Wealth.” American Economic Review67(2): 42–53.
1999. “A Distributional Analysis of Green Tax Reforms.” National Tax Journal52: 665–81.
2008. “Analysis of US Greenhouse Gas Tax Proposals.” NBER Working Paper No. 13980, National Bureau of Economic Research, Cambridge, Massachusetts..
2015. “The Distributional Burden of a Carbon Tax: Evidence and Implications for Policy.” In Implementing a US Carbon Tax: Challenges and Debates, edited by I.Parry, A.Morris, and R.Williams, (ed)38–61. New York: Routledge..
2013. The Climate Casino: Risk, Uncertainty, and Economics for a Warming World. New Haven, Connecticut: Yale University Press.
Organisation for Economic Co-operation and Development (OECD). 2012. Mortality Risk Valuation in Environment, Health and Transport Policies. Paris: Organization for Economic Cooperation and Development.
Organisation for Economic Co-operation and Development (OECD). 2014. The Distributional Incidence of Energy Taxes. Paris: Organization for Economic Cooperation and Development.
2004. “Are Emissions Permits Regressive?” Journal of Environmental Economics and Management47(2): 364–87.
2000. “Tax Deductions, Environmental Policy, and the ‘Double Dividend’ Hypothesis.” Journal of Environmental Economics and Management39(1): 67–96..
2014. Getting Energy Prices Right: From Principle to Practice. Washington: International Monetary Fund..
Forthcoming. “How Much Carbon Pricing Is in Countries’ Own Interests? The Critical Role of Co-Benefits.” Climate Change Economics..
2010. “What Are the Costs of Meeting Distributional Objectives for Climate Policy?” B.E. Journal of Economic Analysis and Policy10(2, Symposium): Article 9..
1997. “Trends in Personal Motor Vehicle Ownership and Use: Evidence from the Nationwide Personal Transportation Survey.” In Proceedings from the Nationwide Personal Transportation Survey Symposium, October 29–31. No. 17 of Searching for Solutions: A Policy Discussion Series. Washington, DC: U.S. Federal Highway Administration, 85–127..
1991. “Is the Gasoline Tax Regressive?” In Tax Policy and the Economy, edited by DavidBradford, (ed)5. Cambridge, Massachusetts: National Bureau of Economic Research.
2011. “Distributional Impacts of Carbon Pricing: A General Equilibrium Approach with Micro-Data for Households.” Energy Economics33(S1): S20–S33..
2012. “The Elasticity of Taxable Income with Respect to Marginal Tax Rates: A Critical Review.” Journal of Economic Literature50(1): 3–50..
1992. “The Distributional Consequences of Taxes on Energy and the Carbon Content of Fuels.” European Economy, Special Edition No. 1: The Economics of Limiting CO2 Emissions, 241–68..
2012. “Distributional Effects of Taxing Transport Fuel.” Energy Policy41: 75–83.
2015. “An Illustrated User Guide to the World Input–Output Database: The Case of Global Automotive Production.” Review of International Economics. doi: 10.1111/roie.12178..
United States Interagency Working Group on Social Cost of Carbon (US IAWG). 2013. Technical Support Document: Technical Update of the Social Cost of Carbon for Regulatory Impact Analysis under Executive Order 12866. Washington: Interagency Working Group on Social Cost of Carbon, United States Government.
1999. “Distributional Aspects of an Environmental Tax Shift: The Case of Motor Vehicles Emissions Taxes.” National Tax Journal52(1): 53–65..
2005. “Are CO2 Taxes Regressive? Evidence from the Danish Experience.” Ecological Economics52(2): 239–51..
World Bank. 2014. State and Trends of Carbon Pricing. Washington: World Bank.
The author is grateful to Steve Clark, Terry Dinan, Tarun Narasimhan, and Ruud de Mooij for help with this chapter.
Including much higher warming from feedbacks in the climate system, dramatic sea level rises from collapsing ice sheets, dramatic local climate change from changing ocean currents, and so on.
The discussion in this chapter focuses on carbon taxes, though the distributional incidence of their emissions trading analogs (with allowance auctions) is equivalent for the same emissions coverage, prices, and revenue use.
For the United States a carbon tax is about five times as effective at reducing CO2 as a comparable tax on electricity or incentives for renewable power generation (Figure 3.1 in Parry and others 2014).
In the United States, a tax on the carbon content of fossil fuels need only involve monitoring about 1,500–2,000 taxpayers (Calder 2015).
Prices in the largest program, the European Union (EU) Emissions Trading Scheme, have been below the equivalent of US$10 per ton in recent years, substantially less than estimated environmental damages, or levels (if globally applied) consistent with aggressive long-term climate stabilization (see below).
For example, the approximation overstates the loss of consumer surplus by only 5 percent when fuel use is reduced by 10 percent.
Meta analyses by Espey (1998) and Espey and Espey (2004) put median income elasticities for transport fuels and electricity at 0.83 and 0.92, respectively, in advanced countries, though there is considerable variation across studies. Unfortunately, not much is known about how income elasticities vary by income class, which would provide a more refined sense from this information of how budget shares for energy products vary across the income distribution.
See, for example, the spreadsheet tool at www.imf.org/external/np/fad/environ/data/data.xlsx (based on data from the International Institute for Applied Systems Analysis).
Up to one-half of the inequalities in annual income across households might be attributed to variations in income over their life cycle rather than differences between life-cycle income (Lillard 1977).
Regressive impacts would be milder still if a measure of lifetime income were used instead of expenditure (Walls and Hanson 1999).
For example, the United States carries out the Consumer Expenditure Survey annually, the United Kingdom the Living Costs and Food Survey annually, and Eurostat the Household Budget Surveys every five years for every EU member country.
Only about 40 percent of U.S. households in the bottom income quintile receive social security benefits, and 20 percent receive food purchase vouchers. Moreover, indexing of benefits would only provide full compensation for a particular household if all (rather than a fraction) of its income was from benefits, and if its budget shares for energy were representative of (rather than higher than) those of the average household (because the average household determines the formula for calculating cost-of-living increases in response to higher energy prices).
The negative cost result differs from the standard implication of Diamond and Mirrlees (1971), that swapping product taxes for broader income taxes will, under neutral assumptions, increase efficiency costs. The reason is that Diamond and Mirrlees do not incorporate distortions from the broader fiscal system, beyond those in factor markets. See Parry and Bento 2000 for more explanation.
This figure is based on the central case of US IAWG (2013). If applied to leading CO2-emitting countries, this starting price would be broadly consistent with keeping long-term, mean projected warming to about 2.5°C (Nordhaus 2013, 228).
The CO2 tax is superimposed on top of any existing CO2 pricing (for example, about $10 per ton for covered emissions in the EU trading system).
This latter adjustment does not substantially alter overall incidence patterns, given that the indirect burdens are minor relative to direct burdens.
These taxes are imposed in addition to taxes (value-added or similar general sales taxes) on consumer goods in general.
For many EU countries, cutting electricity excises would not conflict with the EU Energy Tax Directive, given that it imposes a relatively modest floor tax on electricity equivalent to about US 0.07 cents per kWh.
The burden of higher energy prices might also be offset by introducing “subsistence” thresholds for energy consumption below which no tax is paid. Although this measure is progressive, it is similar to using revenues for lump-sum transfers (see the discussion below), which yields no efficiency gains.