Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Chapter 4 Energy-Related Excise Taxes INTRODUCTION All of the other analyses in this report consider the impact of subsidies from tax expenditures. The present chapter considers the impact of two energy-related excise taxes: highway motor fuels excise taxes and taxes on commercial aviation fuel. These taxes are important because the sectors to which they apply are highly GHG-intensive. Emissions from passenger vehicles, including cars and light trucks, account for more than 20 percent of U.S. GHG emissions, and airlines add approximately 2.5 percent of U.S. GHG emissions (U.S. Environmental Protection Agency, 2012). The federal government taxes gasoline for on-road use at $0.184 per gal- lon and on-road diesel at $0.244 per gallon.1 It is important that motor fuels are taxed on a volumetric basis rather than on the basis of energy content. The committee asked all four economic modeling contractors to estimate impacts on greenhouse gases of removing these federal highway motor fuels excise taxes. Comparing results across models provides a basis for understanding the mecha- nisms at work in the models as well as an estimate of the uncertainty of the es- timates due to model differences. Summary results are presented in Table 4-1 below. Several types of taxes are currently levied on U.S. domestic air travel, in- cluding the federal ticket tax, the flight segment fee, the passenger facility charge, and the federal security fee (Borenstein, 2011). In addition, commercial aviation pays an excise tax of $0.044 per gallon of jet fuel. Jet fuel for non- commercial aviation is taxed at a higher rate of $0.219 per gallon. According to Airlines for America, the total taxes levied on a typical $300 round-trip ticket is $61 (20 percent), but fuel taxes account for a very small share of the total tax 1 Internal Revenue Code sections 4041 and 4083 impose a tax on fuels to pay into the Highway Trust Fund (authorized in IRC section 9503) and a tax to fund the Leaking Un- derground Storage Tank Trust Fund. The rates quoted here are the combination of both taxes. 81
82 Effects of U.S. Tax Policy on Greenhouse Gas Emissions burden (Airlines for America, 2012). Unfortunately, none of the models used for this study was able to analyze the federal ticket tax, the flight segment fee, the passenger facility charge, or the federal security fee. Moreover, only one model analyzed the excise tax on aviation fuel. FINDINGS FROM PRIOR LITERATURE Unlike the other provisions analyzed in this study, there is a large litera- ture on the tax on highway motor fuels, particularly on gasoline. There are liter- ally dozens of studies that have estimated the price-elasticity of demandâshort run, long run, or bothâfor gasoline in the U.S. (Gillingham K. , 2011) Price elasticity of demand measures the sensitivity of the quantity demanded of a good when the price of the good changes, other things held constant. Though demand elasticities do not directly determine greenhouse gas impacts, they are an inter- mediate step to computing emissions. The committee uncovered 11 studies that quantify the impacts of price on vehicle miles traveled. These studies all find that, as the cost of driving increases, vehicles are driven less, as one would ex- pect. When vehicles are driven less, consumption of motor fuels and GHG emis- sions are expected to decrease. Another vein in the literature finds that increasing fuel costs can accelerate turnover in vehicle stock, influencing consumers to purchase more fuel efficient vehicles when they purchase new vehicles (Busse et al. 2013). More fuel effi- cient vehicles are expected to lead to lower emissions per quantity of fuel con- sumed, though not necessarily less fuel consumed if the rebound effect is large enough. (The rebound effect in this context refers to the tendency of drivers to increase the number of miles driven when the cost per mile driven declines as cars become more fuel efficient. (Greening et al., 2000 and Small and Dender, 2007). In contrast to the rich literature that we reviewed on the excise tax on mo- tor vehicle fuel, our literature review did not uncover any studies of the taxes on aviation fuel or passengers airline tickets. ANALYSIS OF EXCISE TAXES ON HIGHWAY FUELS The Internal Revenue Code levies a tax of $0.184 per gallon of gasoline or alcohol fuel for on-road use. This tax applies whether the fuels are pure or blended, such as the common gasoline plus ethanol mixtures.2 The Code also levies on a tax on diesel fuels at $0.244 per gallon for diesel and kerosene and $0.197 per gallon for diesel-water fuel emulsion.3 2 See Internal Revenue Code section IRC 4081. 3 See Internal Revenue Code section IRC 4081.
Energy-Related Excise Taxes 83 Unlike the other provisions we considered, highway fuels excise taxes are a revenue source rather than a tax expenditure. The Internal Revenue Serviceâs Statistics of Income division collects and reports information on excise tax re- ceipts. In 2010, federal excise taxes on gasoline totaled approximately $25 bil- lion, or 53 percent of all federal excise tax receipts. On-road diesel accounted for an additional $8.6 billion. Combined highway fuels taxes account for $33.7 billion, or over 71 percent of all federal excise taxes (U.S. Internal Revenue Service, 2012). We also note that the impacts of the highway motor fuels taxes interact with other taxes and regulations in this sector. In particular, the tax credits for biofuels and the renewable fuels standards (RFS) will be major constraints on the effects of the highway taxes. A fuller discussion of these is contained in Chapter 5. Modeling Results The committee asked the four modeling teams that were engaged to esti- mate the greenhouse gas impacts of removing the highway motor fuels excise taxes. Summary results are presented in Table 4-1. All four models find that GHG emissions will increase if the taxes are removed. However, the estimates differ greatly across different models. Variations in the results across the models highlight differences in both the modeling approaches and assumptions used. TABLE 4-1 Summary Impacts of Removing Federal Highway Fuels Taxes Across Four Models IGEM NEMS CBER FAPRI Assumption Modeling Period 2010-2035 2010-2035 2010-2035 2014-2021 Real GDP Growth 2.7% 2.6% 2.6% 2.6% Energy Cons. Growth 0.48% 0.4% 0.4% 0.4% GHG Included CO2,CH4, CO2 CO2 CO2,CH4, N2O N2O,HGWP Regional coverage US US World (ltd) World a Result: Increase in Cumulative CO2 emissions (MMT) 2,158 88 1,400 78.6 Avg. annual CO2-e emissions (MMT/year) 83 3.5 54 9.8 Cumulative CO2-e emissions (MMT) 2,173 NA NA NA a Note that NEMS and CBER include only CO2 emissions, while the other models include other non-CO2 greenhouse-gas emissions.
84 Effects of U.S. Tax Policy on Greenhouse Gas Emissions Discussion of Modeling Results on Highway Fuels Excise Taxes It will be useful to begin with some general discussion about the models. NEMS, CBER and FAPRI are all partial equilibrium models. This means they describe aspects of the modeled sectorâin this case use of liquid fuelsâwith considerable detail. They do not, however, reflect the interactions between the energy sector and other parts of the economy. For instance, in the highway mo- tor fuels excise taxes scenarios, NEMS, CBER and FAPRI do not model the impact of the increased federal revenues used to replace lost excise tax receipts on the composition of output and do not model emissions from increased con- sumer spending on non-energy goods and services. The results therefore primar- ily reflect the first-order response of consumers and producers to lower transpor- tation fuel expenses, which in turn reflect embedded price elasticities of demand and supply for transportation fuels. There are also differences across the three partial equilibrium models. NEMS has more fine-grained details on many aspects of the energy sector but focuses on the U.S. alone, while FAPRI and CBER model aspects of world en- ergy markets. While FAPRI and CBER allow the world oil price to change in response to U.S. consumption, NEMS holds the world oil price fixed. FAPRI was designed with particular attention to biofuels, so it has more detail on the agricultural sector than NEMS or CBER. CBER is a more stylized model than either FAPRI or NEMS. This makes it less realistic but more transparent. Additionally, the NEMS-NAS and CBER models include only CO2 emis- sions, while the other models include other non-CO2 greenhouse gases. This omission is very small in the models that include non-CO2 emissions. One advantage of the partial equilibrium models is that they represent spe- cific components of the energy sector. For instance, both NEMS and FAPRI distinguish between taxes on gasoline and taxes on ethanol, and both sets of results suggest this is important in modeling the removal of the highway fuels taxes. The taxes on highway fuels are assessed per gallon, and ethanol has lower energy content per gallon than gasoline. This means that when the highway tax- es are removed, the price per unit energy of ethanol declines by more in percent- age terms than the price per unit energy of gasoline. By contrast, CBER does not distinguish between gasoline and ethanol, which explains in part why the CBER estimates are larger than the other two partial equilibrium models. In addition, CBER assumes more price-elastic demand for highway fuels than the other two models. Given that the results from the partial equilibrium models mainly reflect highway fuelsâ demand and supply elasticities, it is useful to consider what be- havior and decisions the elasticities capture (see Gillingham K. T., 2011 for more detail). Economists usually distinguish between long- and short-run price elasticities. For instance, in the case of gasoline, the short-run price elasticity reflects adjustments that consumers and businesses make that do not involve adjusting the type of vehicles they own. Consumers may drive fewer miles, per- haps because they opt not to take discretionary trips, they carpool to work or
Energy-Related Excise Taxes 85 they take public transportation. Consumers may also adjust their driving habits, for instance, driving more slowly on highways, which would reduce fuel use per mile traveled. Or, households with multiple cars may shift their usage to their more fuel efficient car. Long-run price elasticities include adjustments to the stock of vehicles. If fuel prices are lower, as in the scenarios where federal highway fuels excise taxes are removed, consumers may purchase less fuel- efficient cars and may retire their less fuel-efficient cars more slowly. Conceptu- ally, IGEM and CBER apply long-run elasticities to do their calculations, while NEMS-NAS and FAPRI have time-varying elasticities. Appraisal of Individual Model Results We now discuss and assess the individual model results. We begin with the IGEM results. IGEM has estimates of the impacts of removing the highway fuels excise taxes that are larger than those of the other studies. Having reviewed the calculations, the committee concludes that the IGEM model cannot accurate- ly capture the structure of the motor fuels tax provisions and the associated regu- lations and therefore cannot provide reliable results for these provisions. IGEM does not contain a detailed sectoral description of the transportation sector. It does not have a detailed treatment of gasoline or highway fuels, of the properties of vehicles, of vehicle-miles travelled, or of the substitute fuels. It does not re- flect the different energy content per gallon of different fuels. The IGEM exper- iment changed taxes on refined petroleum products, not on highway fuels. Final- ly, IGEM does not include the renewable fuel standards (RFS), so it cannot capture the RFSâs constraints on inter-fuel substitution and on the mix between gasoline and ethanol. The strengths of IGEM â the capabilities to capture the impacts of the rest of the economy and the recycling of revenues â cannot offset its shortcomings in analyzing the effects of highway fuels taxes. The second approach â the CBER model â has the advantage of transpar- ency and reliance on estimates in the literature for its price-elasticities. Howev- er, in this context, it has three important shortcomings. First, it is a static model, and its elasticities are long-run rather than short-run. While the extent of the overestimate will depend upon the dynamics, it is likely that this would lead to an overestimate of the response by a substantial margin. Second, as will be dis- cussed for the next two models, the CBER model does not contain a realistic representation of the renewable fuel standards, which are likely to constrain pro- duction and reduce the impact of taxes on emissions. Third, the price-elasticity of demand for petroleum in the transportation sector is assumed to be -0.52. This elasticity is applied to the price of crude oil rather than the price of gasoline, which would imply that it is too large and the response of quantity of gasoline demanded is therefore also too large. Taking these three factors together sug- gests that the CBER model is likely to overestimate the response of GHG emis- sions to the removal of highway fuel taxes by a large margin.
86 Effects of U.S. Tax Policy on Greenhouse Gas Emissions The third approach, NEMS-NAS, has the most detailed structure of any of the models (see the description in Appendix A). Additionally, and importantly, it has a detailed treatment of inter-fuel substitution and the different regulations and mandates, particularly the renewable fuel standards. It has the shortcomings of the other partial-equilibrium models of excluding spillover spending effects on outside the energy sector. Additionally, the price-elasticities are low com- pared to many studies. The major and surprising result of the NEMS-NAS estimates is that the impact of removing the highway taxes is very small. Here is the reason for the surprising result. The key factor at work is the âvolumetric biasâ of highway fuels taxes. This signifies that the highway fuels taxes are imposed on a volu- metric basis. Because ethanol, and particularly E85 (which contains 85% ethanol by volume), has a lower energy-to-volume ratio than gasoline, removing the highway-fuels excise taxes has the effect of favoring ethanol-based fuels. Addi- tionally, because ethanol use is constrained by the renewable fuel standards (RFS), as described in detail in Chapter 5, the removal of highway taxes favors E85 over gasoline. According to the NEMS-NAS simulations, total energy use in the trans- portation sector would rise by 0.32% over the period when highway fuels excise taxes are removed. However, because of the volumetric bias and RFS, gasoline use is slightly lower over the entire period, while E85 rises substantially. An increase in liquid fuel consumption would lead to an increase in CO2 emissions except for the shift toward E85, which has a lower GHG emission rate than blended gasoline. The committee notes an important reservation at this point concerning the increased use of E85 in these calculations as well as those in the FAPRI model below. E85 is used in Brazil, but it has not been in widespread use in the U.S. The NEMS-NAS calculations project a hundred-fold increase in the use of E85 over the next two decades. This increase is highly contingent on the RFS man- dates continuing in force and on the development of an E85 car fleet and the associated fueling infrastructure. The key result for NEMS-NAS is that removing the highway excise taxes results in a very small increase in CO2 emissions of 3.5 MMT per year, or about 0.07% of average annual U.S. CO2 emissions over the 2010-2035 period. The FAPRI model has a detailed analysis of the structure of the biofuels industry and mandates. The petroleum sector is relatively aggregated. There are four separate markets in the petroleum sector: petroleum, gasoline, diesel, and residual oils. Petroleum is refined into the three petroleum products. Final de- mands consist of transportation, agriculture, and other. There is no detail of the transportation capital stock or fuel-efficiency standards. FAPRI has two regions, the U.S. and the rest of the world, and thus can calculate the impacts on global GHG emissions. Overall, FAPRI is well-designed to test for policies that work primarily through the biofuels subsidies and mandates, as well as the complex interactions of the different grades of ethanol. FAPRI assumes that the RFS mandates will apply (subject to waiving some of the advanced mandates). To
Energy-Related Excise Taxes 87 meet the mandates requires that increasing amounts of E10 and eventually E85 will be produced and used in motor vehicles. As we note below, the growth rate of E85 use is extremely ambitious. Readers who wish to understand the full de- tails should see Appendix A and the FAPRI model documentation referenced in Appendix A. For highway motor fuels tax, there was no sensitivity analysis per- formed on the effects of relaxing or removing the RFS mandates. Chapter 5 con- siders that sensitivity analysis for biofuels tax subsidies, and shows that the FAPRI and NEMS-NAS models have similar behavior when subsidies are re- moved. FAPRI has a similar effect to NEMS because of the volumetric bias of highway fuels taxes and the RFS. The FAPRI model estimates that removing the highway fuels taxes would lead to an increase in GHG emissions of 9.8 MMT per year of CO2-equivalent, or about 0.17% of U.S. CO2 emissions over the 2014-2021 period. GHG emissions for the U.S. are slightly higher than the glob- al total, while rest-of-the-world GHG emissions decline slightly when the high- way taxes are removed. The basic factor leading to low GHG emissions is simi- lar to that in the NEMS-NAS model: because ethanol has lower energy per gallon, reducing the highway excise taxes increases the use of ethanol relative to gasoline. While the basic forces at work in FAPRI and NEMS-NAS are similar, FAPRI has a slight rise in gasoline consumption rather than the small decline in gasoline consumption in NEMS-NAS. The difference depends upon the time peri- od and details of the specification and is probably not reliably resolved. Table 4-2 summarizes these points on the four models in a succinct fash- ion. AVIATION FUEL EXCISE TAX As with highway fuels, the Internal Revenue Code imposes an excise tax on aviation fuels. Fuel for use in commercial aviation is taxed at $0.043 per gal- lon, while fuel for non-commercial use (that is, private use) is taxed at $0.193 per gallon for gasoline and $0.218 per gallon for jet fuel. In 2010 the IRS re- ported receipts of nearly $390 million from the tax on commercial use fuel and approximately $22 million from the tax on non-commercial fuel. For compari- son, the IRC also imposes several taxes on passenger air transport with 2010 receipts totaling $7.6 billion. Modeling Results Only one of the modelers, CBER, estimated the impact of removing the tax on jet fuel. That modelâs projections suggest that cumulative CO2 emissions would increase by over 70 MMT over the time period from 2010 to 2035, con- siderably less than their estimate of the impact of removing the highway fuels tax. On the other hand, the implied estimate of the change in government reve-
88 Effects of U.S. Tax Policy on Greenhouse Gas Emissions nue per ton change in emissions is smaller for the jet fuel calculation than it was for the highway fuels excise tax, suggesting that for the same change in govern- ment revenue, a policy adjustment to air travel would have a bigger impact on emissions. The CBER model assumes that the demand elasticity for jet fuel is the same as for all other transportation fuels: just above -0.5. The existing empirical literature on airlines suggests that the demand for air travel is more price elastic than the demand for gasoline and other oil products. Benchmark price-elasticity estimates for air travel are around -1 (Borenstein, 2011). TABLE 4-2 Summary Appraisal of Studies of Impact of Removing Highway Fuels Taxes Average CO2 Modeling emmisions Model Period (MMT/year) Advantages Disadvantages Net Appraisal IGEM 2010-2035 83 General No gasoline Not applicable equilibrium or vehicle because of lack approach; sector; no of sectoral details, econometric representation highway fuels estimates of of regulations taxes, and many parameters or biofuels mandates. mandates; no highway fuels taxes NEMS 2010-2035 3.5 Highly detailed Partial Most appropriate structural model of equilibrium; modeling approach. vehicles and fuel U.S. emissions Result depends sector; contains only. critically on the detail of RFS and presence of ethanol products; volumetric bias vintage model of and renewable investment fuel standards. CBER 2010-2035 54 Transparent; Long-run Likely to some interfuel elasticities; overestimates substitution; world no reprentation he impact on petroleum market of biofuels GHG emissions mandates; by a large factor. elasticities high; partial equilibrium FAPRI 2014-2021 9.8 Highly detailed Highly stylized Appropriate model of biofuels treatment of modeling approach. sector; global petroleum Result depends impacts. demand; partial critically on the equilibrium. presence of volumetric bias and renewable fuel standards.
Energy-Related Excise Taxes 89 How a change in air travel will translate into a change in the amount of jet fuel consumed is complicated, and depends on the length of flights, number of takeoffs and landings, aircraft used and other factors. Also, changes to the price of jet fuel could eventually cause airlines to make adjustments to the way they do business, including changing the speed of flights, the number of flights, capi- tal investments in existing aircraft (such as the installation of winglets) and the fuel efficiency of the aircraft used. Understanding the relationship between fuel prices and the fuel efficiency of air travel remains an important area for future research. If one were to account for more elastic demand for air travel as well as adjustments made by airlines in response to changes in their input costs, the im- plied impact of changes to jet fuel taxes would likely be larger than reflected in the CBER modeling runs. Summary on Aviation Taxes While the total GHG impact of removing the tax on jet fuel is small, jet fuel taxes are a small component of the total taxes on air travel. For example, the federal ticket tax is 7.5 percent of federal revenue from excises, while fuel taxes are less than 0.5 percent of revenue. Adjustments that reflect taxes in addition to the tax on jet fuel would have a commensurately larger impact on GHG emis- sions, although they may not have the same impact on airlinesâ decisions about fuel efficiency. Given the potential impacts in this sector, it remains an im- portant area for future research. OVERALL EVALUATION ENERGY-SECTOR EXCISE TAXES This chapter has reviewed research on the impacts of excise taxes in the energy sector on greenhouse-gas emissions. There are two important sets of ex- cise taxesâthose on highway fuels and those on air travel. Most of the research in this and earlier studies has focused on the taxation of highway fuels, and this summary pertains primarily to that sector. This chapter reviewed four commissioned studies of the effect of remov- ing the excise taxes on highway fuels. (For a discussion of the different models and their treatment of fuel demands, see Appendix A.) All four models find that removing the excise taxes on highway fuels would result in increasing green- house gas emissions. This result occurs because a lower post-tax price for high- way fuels generates higher demand for highway fuels, which are largely derived from petroleum products. But the magnitude of the estimated effects varies dramatically for the dif- ferent models. The committee notes that large differences in projections of dif- ferent energy-economic models have been seen in other model-comparison stud-
90 Effects of U.S. Tax Policy on Greenhouse Gas Emissions ies, so there is ample precedent for divergent results.4 Having studied the model results and the broader literature, the committee concludes that the differences among the models are large and incompletely understood. The differences arise from the types and values of price elasticities used by the different models, from assumptions about increasing biofuels production and consumption to meet the RFS mandates, from the volumetric bias of highway fuels taxes, and from appli- cation of the tax within each modelâs structure. A close examination of the re- sults leads the committee to conclude that the NEMS-NAS and the FAPRI mod- els capture the forces at work in this sector most reliably and therefore form the basis of our estimates. Taking these two modeling results together produces a striking conclusion: The impact of removing highway fuels taxes on GHG emis- sions is estimated to be very small because of special features of the taxes and the market. The volumetric bias of the taxes means that removing the taxes fa- vors ethanol, which will reduce the GHG impacts of increasing highway fuel consumption. Additionally, the renewable fuel standards constrain the use of ethanol. According to the two models, the effect of removing the highway fuels taxes is 4 MMT per year (NEMS) and 10 MMT per year (FAPRI). These are 0.07% and 0.17% of annual U.S. CO2 emissions, respectively. The third model (CBER) is similar to these two models when adjustments are made to account for the upward bias in its methods. The committee emphasizes the contingent nature of the model projections. They are contingent because the results depend upon the structure, timing, and implementation of the renewable fuels standards (RFS) as well as a quirk in the tax structure (its volumetric bias). Moreover, the impact works through E85, which has not yet entered significantly in U.S. fuel consumption. If the RFS were to disappear tomorrow, or if the regulations on E85 were to change drastically, or the highway motor fuels taxes were levied by energy content instead of by vol- ume, the projected impacts of removing the gasoline tax might be substantially different and would probably be significantly larger. Additionally, as discussed in the next chapter, there are many uncertainties about how the most recent version of the RFS (RFS2) will be implemented. Finally, it should also be noted that the complex structure of the RFS may imply that large tax increases will have differ- ent effects from the small tax decreases that the current study examined. The magnitude of the differences between models leads the committee to caution against relying on specific numerical results from a single model and recommends drawing only broad conclusions about the nature and direction of impacts. Policy makers and analysts should rely on multiple models, methodol- ogies, and estimates in calculating impact of the tax code and other policies on greenhouse-gas emissions and climate change. 4 The Energy Modeling Forum (EMF) at Stanford University has undertaken several model comparison studies in energy, oil, and climate change. Projections of energy con- sumption in the recent EMF-22 comparison were found to differ by almost a factor of two among models between 2010 and 2050 (see Clarke et al. 2010).