Economists and policymakers have touted a tax on the carbon content of fossil fuels as an efficient means of encouraging a reduction in carbon dioxide emissions. However, a recent study of the impact of such a tax on U.S. households suggests that the cost burden would be neither trivial nor evenly spread. With even a moderate tax, households would face energy price increases, and the distribution of these increases would vary regionally. Moreover, increased energy prices might not lead to reduced consumption of all energy types, but might increase electricity consumption in some areas.
Economists have suggested energy taxes as a means to encourage reductions in the emission of greenhouse gases. One of these taxes would be aimed at reducing emissions of carbon dioxide, the prime pollutant that contributes to global warming. A tax on the carbon content of fossil fuels (coal, natural gas, and crude oil) seems to be an attractive policy instrument for many reasons. For example, it appears to be the most efficient way to create an incentive for reducing carbon dioxide emissions, which are expected to increase over the next decade. A carbon tax would encourage the substitution of other energy forms for fossil fuels and the increased use of fuels such as natural gas that have a lower carbon content than, say, coal. By slowing the growth of fossil fuel use in general and the use of fossil fuels with the highest carbon contents in particular, a carbon tax would reduce the rate at which carbon dioxide accumulates in the atmosphere.
Carbon taxes also have several appealing fiscal properties. Even moderate carbon taxes (about $50 per ton of carbon) could raise between $50 billion and $75 billion annually. Depending on the policy goal, the revenues generated by a carbon tax could be used to reduce the federal budget deficit, be returned directly to households through a reduction in other taxes, or be targeted at emissions reductions efforts in Third World nations. Transfers to these nations aside, recent studies have shown that the macroeconomic hardship imposed by a carbon tax, while nontrivial, would not be catastrophic. Macroeconomic analyses of carbon taxes conducted by the Energy Information Administration have shown that a tax of approximately $50 per ton of carbon might reduce the real gross national product (GNP) of the United States by only 1 percent annually.
Research on carbon taxes has been directed primarily at macroeconomic analyses of various tax scenarios. Some of the more recent studies include those by the Congressional Budget Office (1990), Dale W. Jorgenson and Peter J. Wilcoxen of Harvard University (1990), and the Energy Information Administration (1991). These studies use models that focus on the implications of various tax schemes for economic growth, production in individual economic sectors, the price of goods and services, and employment. However, analyses of the implications of carbon taxes have been few when compared with the numerous analyses conducted for other federal taxing programs, such as personal and corporate income taxes. Moreover, the analyses that have been undertaken fail to address an important set of economic and political variables—namely, the distribution of the carbon tax burden across U.S. households.
Despite the lack of analysis concerning distribution, carbon taxes have developed a large following among Washington policymakers and politicians. Legislation introduced in the 101st and 102nd Congresses in 1990 and 1991 by Congressman Pete H. Stark of California calls for a carbon tax of approximately $25 per ton of carbon on crude petroleum, natural gas, and coal. One would imagine that the attractiveness of this bill for congressional members would increase if the tax burden were spread thinly and evenly. Perceived inequities in the distribution of the tax could make passage of a carbon tax difficult.
Analyzing distributional effects
Analysis of the distributional effects of a carbon tax necessitates regional data. To capture regional differences in these effects, a recent RFF study divided U.S. households into the nine Census Bureau regions: New England, Middle Atlantic, East North Central, West North Central, South Atlantic, East South Central, West South Central, Mountain, and Pacific. Using 1987–1988 data from the Energy Information Administration, the study focused on the effects of a carbon tax on the price and household consumption of major fuels in these regions. Four categories of household energy consumption were examined: electricity, natural gas, fuel oil (including kerosene and liquefied petroleum gas), and motor gasoline.
The first of two tasks in analyzing the distribution of a carbon tax across U.S. households was to determine how taxes applied to primary energy—namely, crude oil, coal, and natural gas—would affect the prices paid by households for final energy consumption. Because a carbon tax would be applied to coal at the mine mouth, to natural gas at the well-head, to foreign crude oil at the port of entry, and to domestic crude oil at the field, the RFF study modeled how natural gas and electric utilities and petroleum refining companies would pass the tax along to households.
The study assumed that all carbon taxes would be passed on to consumers, but that the final price of energy to consumers would depend on the amount of intermediate energy processing conducted by electric and gas utilities and petroleum refiners. For example, a primary energy product such as natural gas requires little processing before consumption by households. Thus a carbon tax might be directly added to the price a household would pay for natural gas. However, final energy products—such as fuel oil, gasoline, and electricity—undergo considerable processing before consumption. Thus the RFF study modeled how taxes on the fuels from which these products are derived—coal, crude oil, and natural gas—are passed on as added costs in the course of processing.
Fuel oil and gasoline are refinery products that are manufactured using crude petroleum as a feedstock. These are only two of the many outputs of the process of petroleum refining. Allocating a cost increase for one input—namely, crude oil—to the marginal costs of production of multiple outputs is difficult. Given this difficulty, the RFF study made the simplifying assumption that the carbon tax on crude oil is passed on according to the British thermal unit (Btu) content of fuel oil and gasoline.
The production of electricity uses three carbon-bearing fossil fuels—coal, fuel oil, and natural gas—in addition to nuclear and hydro energy sources. Carbon taxes would not affect nuclear and hydro energy sources but would affect the utility's purchase prices for coal, fuel oil, and natural gas. The use of each type of en-ergy in electricity generation varies by census region. Since each type of energy is taxed at a different rate, changes in electricity prices that include carbon taxes could be expected to vary by region. Regions that produce electricity primarily from fossil energy would be expected to be the hardest hit by a carbon tax. Therefore, the biggest increase in electricity prices should occur in these regions.
The RFF study developed a set of small regional models for the pricing of electricity. These models are driven by generation technologies already existing in a given region, the cost of the various energy inputs, and the ratio of energy cost to non-energy cost (capital and labor) in electricity generation. Using these models, the study simulated the response of regional electricity prices to a carbon tax on the raw energy inputs.
The second task in analyzing the distribution of carbon taxes across U.S. households was to determine how consumers would react to energy price increases. This reaction ultimately depends on the length of time households take to fully respond to the tax and on the response options available. The RFF study examined how households faced with higher energy prices might adjust their energy consumption under three scenarios. Under the most extreme short-term scenario, households would undertake no action to reduce the amount of energy they consume. Under this scenario, the RFF study predicted that consumers would simply absorb the entire tax in higher energy prices. Under the second scenario, households would have a somewhat longer time to contemplate energy price hikes and would have response options available to them. Under this scenario, households would not substitute one type of energy for another, but would reduce their utilization of energy. Households would turn down thermostats in the winter, use their cars less, or reduce their use of labor-saving appliances. Under the third scenario, households would have the most options to alter not only their overall energy use but also the mix of energy they use. In this long-run case, households might substitute one type of energy (fuel oil, electricity, or natural gas) for another.
Changes in Total Residential Energy Expenditures, by Tax Scenario, Assuming Inelastic Demand
Regional effects of two tax rates
Many schemes for carbon tax rates have been proposed. The RFF study analyzed the burden on households of two rates that correspond to the more moderate tax rates proposed. The first tax rate of $25 per ton of carbon is similar to the rate proposed in the 1990 Stark bill. The second tax rate of $41.54 per ton of carbon corresponds to the rate that Jorgenson and Wilcoxen have proposed to slowly reduce emissions of carbon dioxide to 80 percent of the 1990 level by 2005. According to the Jorgenson-Wilcoxen analysis, the tax rate at which this lowered level of emissions is achieved would be maintained from 2005 onward.
The RFF study projected the amount to which these two tax rates would increase expenditures for energy consumption in each of the nine census regions over both the near and the long term. Under the very short-term scenario, the study indicated that a $25 per ton carbon tax would impose the largest cost increase on households in the West North Central region. These households would suffer an increase of $153.00 per year, while those in the Pacific region would suffer the smallest increase—$97.76 per year. With a tax of $41.54 per ton of carbon, households in the West North Central and Pacific regions would again face the largest and smallest overall annual increases—$265.94 and $163.46, respectively (see table, p. 7). Although the higher tax rate imposes the higher tax burden, these figures suggest that the regional distribution of that burden would vary by as much as 60 percent with either the higher or the lower tax rate.
In the short term, neither tax rate would lead to changes in the level of carbon dioxide emissions because households would not adjust their energy mix or consumption in response to proportional increases in energy prices. Over the long term, a carbon tax would be expected to reduce these emissions, but how this reduction would come about is unclear. Consider the impacts of a tax of $41.54 per ton of carbon on energy prices and consumption in the Mountain, South Atlantic, and New England census regions when households have both the time and the options for changing the mix and amount of energy they use (see figure, p. 8). Due to regional variations in the prices of fuels currently consumed by households, the options for energy substitution, and the composition of existing electricity generation technologies, energy prices and substitution patterns in these regions could differ significantly.
The carbon tax would raise all fuel prices, but according to the RFF study the changes in these prices would not be the same in each of the above-mentioned census regions. Although the amount of increase in the prices of gasoline, fuel oil, and natural gas as a result of the tax would not vary among these regions, regional differences in the prices that households pay for these fuels would be observed because of regional variations in initial fuel prices. For example, in 1987 the price that households in the New England region paid for natural gas was higher than that paid by households in the Mountain region. If a tax of $41.54 per ton of carbon had been imposed in that year, households in the New England region would have faced an 8 percent increase in the price of natural gas as compared with a 14 percent increase in the Mountain region.
The impact of this carbon tax on electricity prices is complicated by the variation in carbon content of the various raw energy types used to produce electricity across the country. The pre-tax prices of electricity in the South Atlantic and the Mountain regions are similar, but the electricity generation mix in the South Atlantic region is much less carbon-intensive. Thus the percentage increase in the price of electricity due to the carbon tax would be lower in the South Atlantic region than in the Mountain region. The tax would raise the price 11 percent in the South Atlantic region and 14 percent in the Mountain region.
The impact of price changes on the pattern of energy consumption is difficult to predict. Three of the four fuel types in the RFF analysis could be substituted for one another over the long term. Household heating needs could be met by electricity, natural gas, or fuel oil. However, gasoline could not be used for home heating; and electricity, natural gas, and fuel oil have not been substituted for gasoline in transportation. Therefore, the response of households in all three census regions to a carbon tax on gasoline would be conservation.
Percentage changes in price and consumption of fuels, residential sector, 5 to 7 years after imposing a tax of $41.54 per ton of carbon
The RFF study suggests that over the long term energy price increases would not necessarily lead to reduced consumption of all three household energy forms (fuel oil, natural gas, and electricity). In the Mountain region, the options for home heating are restricted to electricity and natural gas. Higher prices for these two fuel types would lead to conservation. In the New England region, household consumption of electricity would be reduced, but the higher price for fuel oil would lead some households to switch from fuel oil to electricity. Thus, despite electricity price increases, there would be little change in the total quantity of electricity consumed. In the South Atlantic region, consumers would abandon use of fuel oil and natural gas. While the price of electricity would rise, so would the level of electricity consumed. The effect this increased electricity consumption would have on carbon dioxide emissions would depend, in a complex way, on the type of plant used to generate electricity in response to rising demand. If added demand were met with coal-fired electric plants, carbon emissions would be expected to rise.
Several conclusions can be drawn from the RFF analysis of the relative size and distribution of taxes on the carbon content of fossil fuels. First, the burden on households of moderate carbon taxes would not be trivial; with less moderate taxes, it would rise dramatically. Second, the distribution of the burden would be decidedly uneven and could vary regionally by as much as 60 percent. Third, households would both conserve energy and alter their energy consumption to a significant extent over time. However, changes in energy use would vary from region to region. In the case of electricity, consumption would be reduced in some regions and increased in others.
Raymond J. Kopp is director of and a senior fellow in the Quality of the Environment Division at RFF. Diane E. DeWitt is a former research assistant in the division.
A version of this article appeared in print in the October 1991 issue of Resources magazine.
A version of this article appeared in print in the October 1991 issue of Resources magazine.
