Evaluating fuel tax equity: direct and indirect distributional effects.

A recent proposal by the Clinton administration calls for a general tax on all energy products. Along with raising revenue, such a tax has the desirable effects of conserving finite resources, reducing pollution emissions, and reducing dependency on foreign supplies of energy. In addition, the

Most studies of the distributional effects of energy price increases are partial equilibrium in nature. Attention focuses on a particular fuel or fuel aggregate, and effects on other commodities are ignored. Examples are seen in the work of Zupnick (1975), Palmer et al. (1976), Stucker (1977), Henderson (1988), and Poterba (1991), who find that gasoline and other energy taxes place disproportionate burdens on low income and nonwhite consumers.(2)

Partial equilibrium studies are incapable of capturing the entire range of relationships among sectors of the economy. This fact is recognized in Solow's (1985) energy-based general equilibrium study.(3) Solow states that energy taxes are likely to distort input choices and cause changes in the composition of output.(4) However, the ability of Solow's model to measure such effects is restricted. The economy is composed of three sectors, including a single fuel aggregate, and it is assumed that the capital stock and labor force are fixed. Solow's study illustrates how data constraints impose a trade-off between the number of sectors that can be included and the degree of flexibility attainable in estimating parameter values. Among general equilibrium approaches, this fact explains why the input-output (I-0) model is used to measure highly disaggregated effects. Under the fixed-proportions assumption, direct and indirect sectoral relationships are captured by the Leontief inverse. Bahl and Shellhammer (1969) discuss the range of uses for the I-0 model to evaluate tax incidence and other tax effects.(5)

Applications that examine the relationship between energy consumption and income are seen in work by Herendeen (1974) and Hannon (1975). They each find that accounting for both direct and indirect energy consumption results in a more uniform proportion of expenditure on energy across income levels.(6) When all effects are included, a tax on energy is more likely to be proportional than regressive.

This study also uses the I-0 approach. To inform the current debate surrounding energy taxation, distributional effects are measured using the most recent and disaggregated energy and dollar flows among producing sectors. We thereby incorporate the effects on industrial energy requirements and consumer expenditure of the two oil price shocks of the 1970s and the major decline in oil prices in the 1980s. By measuring the potential regressivity of taxes placed on gasoline, coal, refined petroleum products besides gasoline, electricity, and natural gas, the list of fuels found in past studies is expanded. To evaluate effects across political constituencies, a broad range of consumer classifications is considered: relative burdens are examined for income quintiles, age categories, and rural and urban areas.

THE FUEL TAX AND REGRESSIVITY: DIRECT AND INDIRECT EFFECTS

Suppose that a gasoline or other fuel tax is imposed on the fuel producer.(7) The tax is assumed not to affect the relative amounts of physical outputs purchased by consumers; real consumption shares are fixed. Consistent with I-0 assumptions, ratios of input use per dollar of output are constant for all industries. With fixed I-0 coefficients and consumption shares, demand elasticities are zero for all producers and consumers in the economy. A fuel tax is fully shifted forward to consumers.(8) Shifting occurs directly, as fuel producers raise their prices to account for the tax, and indirectly, as all producers raise their prices to cover the increased cost of fuels and other inputs. In the end, each product's price rises in proportion to its direct and indirect use of the taxed fuel. For commodity j, if the taxed fuel is gasoline, the price change is the product of the tax rate, t, and the coefficient [gamma.sub.gj] which measures commodity j's direct and indirect gasoline requirement per dollar of output:(9)

1 [[delta]P.sub.j] = [t[gamma].sub.gj] Assuming consumption of the original market basket, with commodity prices initially normalized to one, the increase in expenditure for commodity j equals [[delta]P.sub.j][C.sub.j]. The total increase in expenditure caused by a gasoline tax equals the value of direct and indirect gasoline required to satisfy consumer expenditure on all products, [[chi.sub.g], weighted by the gasoline tax rate:

2 [Mathematical Expression Omitted] To measure regressivity, total consumption is decomposed into expenditures by higher income groups (rich), [C.sup.R], and lower income groups (poor), [C.sup.P]. The term [Mathematical Expression Omitted] measures total direct and indirect gasoline consumption by the rich, while [Mathematical Expression Omitted] measures this consumption by the poor. Using the current market basket, a tax is regressive if the ratio of direct and indirect gasoline consumption to total expenditure for the poor exceeds this ratio for the rich.(10) The tax is regressive when

3 [Mathematical Expression Omitted] Alternatively, the tax is regressive if the ratio of poor to rich direct and indirect gasoline shares is greater than one. Because it cancels, the tax rate is unimportant in determining regressivity.(11)

THE DATA BASE

The fuels we consider are gasoline, coal, refined petroleum products besides gasoline, electricity, and natural gas. Dollar-based I-0 data are from the Bureau of Economic Analysis (BEA) (1990). To account for specific energy products and energy intensive commodities, the 79 order BEA aggregation is expanded to 89 commodities.(12) The latest detailed energy flows are for 1985, from the National Energy Accounts (NEA). The energy-based model described by Miller and Blair (1985) and Casler and Hannon (1989) is used.(13)

Consumption expenditures for 1985 are from the Consumer Expenditure Survey (CES) (1989) and show spending on commodities by income and other classifications.(14) In forming expenditure vectors, a computer printout of detailed survey data ensured the closest possible match between BEA and CES product categories.(15)

DISTRIBUTIONAL EFFECTS ACROSS INCOME QUINTILES

The top portion of Table 1 shows direct expenditure shares for fuels by income category. The first column shows average shares for all consumers. Shares by income quintiles appear in the other columns. Spending on gasoline constitutes the largest share across income groups, followed by electricity, natural gas, other refined petroleum products, and coal. Spending on all fuels comprises approximately 10 percent of average household expenditures. The number in parentheses below each 5hare is the ratio of the share to the corresponding share from the highest income quintile. A ratio greater than one implies that the lower quintile allocates a higher proportion of its expenditure directly to the fuel considered. Because all lower quintile ratios exceed one, an ad valorem or per unit tax on any fuel is regressive.

[TABULAR DATA 1 OMITTED]

Expenditure shares for direct and indirect energy consumption appear in the bottom portion of Table 1. Shares divided by the highest income quintile's direct and indirect shares are in single parentheses. In double parentheses are the ratios of direct energy consumption to direct plus indirect energy consumption. Except for refined petroleum products used by the lowest three quintiles, all direct and indirect share ratios are greater than one; fuel taxes are generally regressive even after accounting for indirect effects. However, there is far less variation compared with the direct expenditure data; all of the share ratios have moved significantly toward one. When both direct and indirect effects are considered, fuel taxes are much more neutral.

Relative to the highest quintile, direct and indirect expenditure shares decline steadily for coal, electricity, and natural gas; the first quintile bears the highest direct and indirect burden for these fuels. For gasoline and refined petroleum products, shares rise until the last quintile. The amount by which relative shares exceed one for the lowest two quintiles shows that a tax on natural gas is most regressive. Among fuels, the ratio of direct to direct plus indirect gasoline consumption is the largest across all quintiles. Accounting for indirect effects does the least to reduce the regressivity implied by the direct burden of a gasoline tax.

The final row in Table 1, "Total Btu per dollar of expenditure," shows total direct and indirect energy consumption in British thermal units (Btus) per dollar of expenditure.(16) It is used to assess the burden of a uniform tax on all primary energy products.(17) This burden rises between the first two quintiles and declines for the last three. Relative to the highest quintile, the lowest would pay 7.7 percent more tax per dollar of expenditure. This compares with 35.3 percent more when only direct energy purchases are taxed. For the lowest income quintiles, share ratios for total direct and indirect energy are smaller than ratios for most individual fuels. In terms of distributional effects, total direct and indirect energy consumption is a good candidate for taxation.(18)

DISTRIBUTIONAL EFFECTS BY LOCATION

Direct energy expenditure shares for urban and rural consumers appear in the top portion of Table 2. Numbers in parentheses are expenditure shares relative to shares for rural consumers. Direct expenditure shares for rural consumers exceed those for urban consumers for gasoline, coal, other refined petroleum products, and electricity; taxing these fuels places a larger relative burden on rural residents. However, examining direct and indirect effects in the bottom of the table, ratios of urban to rural direct and indirect expenditure shares are much closer to one; when all effects are considered, a tax applied to any fuel becomes more neutral. Gasoline is least affected when indirect consumption is included, and a gasoline tax has the greatest negative distributional impact in rural areas. From the "Total Btu per dollar of expenditure" row, urban households consume 87.2 percent of the energy consumed by rural households per dollar of expenditure. This compares with 76.9 percent for direct energy alone.

                 TABLE 2
        1985 ENERGY EXPENDITURE SHARES
                BY AREA(a)

                   Urban       Rural
           Direct Expenditure Shares
Gasoline                 4.348      5.950
                        (0.731)    (1.000)
 Coal                    0.009      0.032
                        (0.277)     1.000)
Refined petroleum        0.427      0.662
                        (0.645)     1.000)
Electricity              2.828      4.092
                        (0.691)     1.000)
Natural gas              1.565      1.198
                        (1.306)     1.000)
Total                    9.176     11.934
                        (0.769)    (1.000)
Direct and Indirect Expenditure Shares
Gasoline                 5.233      6.809
                        (0.769)    (1.000)
                        (0.831))   (0.874))
Coal                     0.866      1.088
                        (0.796)    (1.000)
                        (0.010))   (0.029))
Refined petroleum        9.120     10.045
                        (0.908)     1.000)
                        (0.047))   (0.066))
Electricity              6.165      7.674
                        (0.803)    (1.000)
                        (0.459))   (0.533))
Natural gas              3.578      3.458
                        (1.035)    (1.000)
                       ((0.437))   (0.347))
Total Btu per            31218      35815
dollar of expenditure   (0.872)    (1.000)
                        (0.289))   (0.310))
(a) Numbers in parentheses represent expenditure shares
divided by rural expenditure shares. Numbers in double
parentheses represent the ratio of direct energy to direct
plus indirect energy for the fuel shown.

DISTRIBUTIONAL EFFECTS ACROSS AGE CATEGORIES

The top portion of Table 3 shows direct energy expenditure shares by age group. in parentheses are the shares for each age category divided by shares for the oldest group of consumers. Direct burdens tend to rise with age for refined products, electricity, and natural gas. Gasoline shares fall dramatically for those over 65. For individual fuels, greater neutrality again results when direct and indirect effects are accounted for. Those older than 65 face the lowest direct and indirect burden from a gasoline tax; the burden is heaviest on the youngest age category. For other age groups, the burden of a gasoline tax is fairly uniform. With a tax on each Btu of direct and indirect energy, the youngest group of Americans would pay 2.7 percent more than the amount paid by the oldest group for each dollar of expenditure.

[TABULAR DATA 3 OMITTED]

A TAX ON INDIRECT FUEL CONSUMPTION

Results in Tables 1-3 show that accounting for both direct and indirect energy use is not sufficient to achieve complete neutrality or progressivity. However, the consistent manner in which indirect effects lead to more proportional energy consumption per dollar of expenditure suggests an alternative strategy-taxation of indirect energy use alone. By exempting direct energy purchases from taxation, the most regressive component is removed. Producers using fuels as inputs will pay the tax, which is fully shifted to consumers. With direct consumer purchases of energy exempted, the prices of nonenergy products are mostly affected.(19)

Measures of tax regressivity based on indirect energy consumption are shown in Table 4. Ratios of indirect fuel consumption shares to the indirect shares for the highest income category are shown on top; indirect ratios by region appear in the middle; the bottom of the table shows ratios by age groups. To see changes in regressivity, these ratios can be compared with those in single parentheses in Tables 1-3.

[TABULAR DATA 4 OMITTED]

Across quintiles, when only indirect energy consumption is considered, a significant movement toward neutrality occurs for natural gas. Taxes on gasoline and electricity become progressive, and a tax on refined petroleum products becomes more progressive than when both direct and indirect effects are considered. Because most coal use is indirect, there is little change in regressivity. A uniform tax on all energy products is slightly progressive for the lowest income quintile and virtually neutral for all other quintiles. For regional categories, an improved pattern of neutrality occurs for all fuels except natural gas. For age categories, the effects of a tax on indirect energy consumption result in greater neutrality for gasoline and natural gas. Declines in indirect consumption of electricity and refined petroleum products over the life cycle lead to relatively lower burdens for the oldest group of Americans.

Conclusions

Our findings show the likely degree of vertical inequity after imposition of an energy tax. Our study also provides information for analyzing the distributional effects of energy price changes from sources such as oil price shocks, changes in natural gas regulation, or discoveries of alternative energy supplies. Finally, two distinct outcomes emerge regarding fuel tax regressivity, depending on whether exemptions are allowed for direct energy purchases.

Without exemptions, patterns of regressivity for direct and indirect energy expenditure resemble those for direct energy. Thus, if interest is focused upon the hierarchy of tax burdens and not the exact levels of burdens, direct effects provide this information. However, taxes on fuels are far less regressive than direct expenditure data indicate. Accounting for both direct and indirect effects reduces the energy tax burden from 20.2 to 5.8 percent, on average.(20)

Although greater neutrality results for individual fuels when direct and indirect effects are considered, a tax on any single fuel can reduce horizontal equity.(21) This is the advantage of taxing all fuels, as in the Clinton proposal.(22) In addition to a general Btu tax, policymakers are likely to continue targeting gasoline for additional taxation; it is the fuel of choice for private transport across all income categories. Even though indirect use of this fuel is the smallest and rural and lower income consumers are adversely affected, the overall equity concerns and the potential public reaction to a tax on gasoline are far less than those likely to arise from a tax on any other specific energy product considered here.

From an equity standpoint, a much better alternative is a tax that exempts all direct energy purchases by consumers. Such a tax is neutral or progressive for all fuels except coal and natural gas. Compared with the average regressivity of a Btu tax on total direct and indirect energy consumption, the relative burden falls from 5.8 to 0.4 percent. Such a tax is progressive for the lowest income quintile and virtually neutral for the others. Exempting direct consumer expenditure still subjects the preponderance of economy-wide energy use to taxation: fuel use by the industrial sector accounts for approximately 32 percent of all gasoline, 99 percent of all coal, 89 percent of other refined petroleum products, 64 percent of all electricity, and 69 percent of all natural gas.(23) By exempting direct energy expenditures for consumers, taxes on specific fuels are possible without obvious implications for horizontal equity. Finally, consumers might be much less inclined to oppose an energy tax that does not apply directly to them.

ENDNOTES

(1) See National Tax Association (1978) for discussion of the wide range of issues involved in the regulation and taxation of energy production. (2) Bowman and Mikesell (1983) and Shmanske (1990) investigate the factors that underlie determination of state gasoline tax rates. (3) See Harberger's (1962) seminal article on tax incidence in a general equilibrium setting Goulder and Summers' (1989) general equilibrium approach uses a five sector model that includes aggregate energy (4) These distortions and changes are discussed by Sweeney 1984), (5) Bhatia (1988) uses a highly aggregated I-0 model with flexible coefficients. (6) Indirect energy consumption refers to the energy it takes to produce energy and nonenergy products. (7) To treat imports and exports in an equivalent manner and to avoid reducing the international competitiveness of U.S. firms, the tax can be applied to the fuel content of imports and can be rebated for exports. (8) It is also assumed that the tax has no effect on wages and other factor prices, because the distribution of in come could be altered (9) Equation 1 is from the value-based Leontief price model Product prices initially equal one. See Miller and Blair (1985, pp. 354 56) (10) Problems in estimating economic income to determine distributional effects are considered by Wallace et al. (1991). Because of transitory income, Poterba (1991, p 156) observes that current expenditure is likely to provide a more reliable measure of long-run well-being than current annual income. Therefore, it is a more consistent basis for determining relative tax burdens. (11) Determining long-run regressivity involves estimating de tailed future consumption paths, a difficult task given data constraints and the absence of models capable of capturing the dynamics of change. (12) Gasoline poses a special problem, because it is part of BEA's refined petroleum aggregate. However, gasoline sales to I-0 sectors in the NEA enable us to modify BEA data and to treat gasoline as a separate input. (13) Adjustments to the BEA and NEA data required for a consistent energy-based model are discussed by Casler 1989). (14) In forming direct and indirect effects, expenditure data are converted to producer values using trade and transportation margins from the BEA. Following BEA conventions, the imputed rental value of owner occupied homes is used to measure consumption expenditure on housing capital. (15) For example, the published form of the expenditure survey aggregates all forms of public transportion. Detailed

(21) For example, a tax on refined petroleum products would severely penalize households heating with oil. Residents of the northeast would face a higher burden than residents of the midwest (22) Clinton's proposal exempts some fuel types, such as solar and geothermal energy. (23) Because so little gasoline is consumed by the industrial sector, its potential for raising revenue, conserving resouces, and decreasing pollution emissions is reduced. However, there is plenty of room for achieving these objectives by taxing other fuels.

REFERENCES

Bahl, Roy W. and Kenneth L. Shellhammer. "Evaluating the State Business Tax Structure: an Application of Input-Output Analysis." National Tax Journal 22 (June, 1969): 203-16. Bhatia, K.B. "Tax Incidence in a Hierachial Model." Journal of Public Economics 37 (1988): 221-42. Bowman, J.H. and J.L. Mikesell. "Recent Changes in State Gasoline Taxation: A, Analysis of Structure and Rates" National Tax Journal 36 2 (June, 1983): 163-82. Casler, S. "Energy Flows through the U. S. Economy; 1980, 1982, and 1985" Report prepared for the U.S. Congress, Office of Technology Assessment, May, 1989. Casler, S. and B. Hannon. "Readjustment Potentials in industrial Energy Efficiency and Structure." Journal of Environmental Economics and Management 17 (3) (1989) 93-108. Goulder, L.H. and L.H. Summers. "Tax Policy, Asset Prices, and Growth, A General Equilibrium Analysis." Journal of Public Economics 38 (1989); 265-96. Hannon, Bruce. "Energy Conservation and the Consumer." Science 189 (4197) (July, 1975): 95-101. Harberger, A. "The Incidence of the Corporation Income Tax." Journal of Political Economy 70 (1962): 215-40. Herendeen, Robert A. "Affluence and Energy Demand." Mechanical Engineering (October, 1974): 18-22. Henderson, Lenneal J. "Energy Policy and Socioeconomic Growth in Low-Income Communities." The Review of Black Political Economy (Fall, 1988): 11-27. Miller, R.E. and P.D. Blair. Input-Output Analysis: Foundations and Extensions. Englewood Cliffs, NJ; Prentice-Hall, 1985. Jack Faucett Associates. "National Energy Accounts" Prepared for the Department of Commerce, JACKFAU-84-316, December, 1984. National Tax Association. "Energy Taxation--Conflicting Federal, State, and Local Interest." National Tax Journal 31 3 (September, 1978). Palmer, J.L., J.F. Todd and H.P. Tuckman. "The Distributional Impact of Higher Energy Prices; Ho, Should the Federal Government Respond?" Public Policy 24 (4) (Fall, 1976): 545-68. Poterba, James M. "Is the Gasoline Tax Regressive?" In Tax Policy and the Economy, edited by David Bradford. Cambridge: The MIT Press, 1991, vol. 5, pp. 145-64. Shmanske, S. "The Determinants of State Gasoline Taxation in the 1970's." Resouces and Energy 12 (4) (1990); 339-52. Stucker, James P. "The Distributional Implications of a Tax on Gasoline." Policy Analysis 3 (2) (Spring, 1977): 171-86. Solow, John L. "General Equilibrium Incidence of Energy Taxation" Southern Economic Journal 51 (April 1, 1985): 1018-30. Sweeny, James L. "The Response of Energy Demand to Higher Prices; What Have We Learned)" The American Economic Review, 74, 2 (May, 1984): 31-37. U.S. Department of Commerce, Bureau of Economic Analysis. "Annual Input Output Accounts of the U.S Economy, 1985." Survey of Current Business, January, 1990. U.S. Department of Labor, Bureau of Labor Statistics. "Consumer Expenditure Survey: Integrated Survey Data, 1984-1986" Bulletin 2333, August, 1989. Wallace, S., M. Wasylenko and D. Weiner. "The Distributional Implications of the 1986 Tax Reform." National Tax Journal 44 2 (June, 1991) 181-98. Zupnick, Jan William. "The Short Run Incidence of a Tax Induced Rise n the Price of Gasoline." (Journal of Economic Issues 9 2 (June, 1975): 409-14. data are to capture distinctions: fuel inverse coefficients vary greatly for transport by bus,rail and air; expenditure shares these forms of public tansport differ accross income group. (16) Because of double counting, the sums of direct and indirect fuel shares are not meaning and are not presented. For example,the direct and indirect cost of share for refined petroleum includes the refined petroleum used to produce electricity. The measure "Total Btu per dollar of expenditure" is calculated to avoid double counting. See Casler and Hannon (1989). (17) Primary energy products include coal,crude petroleum, natural gas, and hydro and nuclear power. These fuels are often transformed, used as inputs to produced other fuels before final use. For example, crude petroleum is transformed into refined petroleum Products ducts and coal is mostly used to produce electricity. (18) A wellhead tax ,crude petroleum and natural gas and a mine-mouth tax on coal would accomplish this goal, because most energy products come from these sources Electricity requires special treatment because of nuclear, hydro, and other nonfossil fuel-based sources of generation. (19) For consumers, fuel prices, will somewhat because the indirect energy used to produce fuels will be subject to taxation. (20) In measuring energy content of imports and exports, accounting for indirect energy use is essential. To keep U.S. products competitive abroad requires a tax rebate of all taxes--payed directly or indirectly Taxing imports for their direct fuel requirements alone undervalues their en ergy content and leaves U.S. goods less competitive.