Effects of U.S. Tax Policy on Greenhouse Gas Emissions (2013)

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 7 Summary of Findings and Recommendations and Use of Tax Policy to Address Climate Change Policy THE COMMITTEE’S CHARGE AND APPROACH TO THE STUDY This report is the response of a committee appointed by the National Acad- emies to a charge by the U.S. Congress to conduct “a comprehensive review of the Internal Revenue Code to identify the types of and specific tax provisions that have the largest effects on carbon and other greenhouse gas emissions and to esti- mate the magnitude of those effects.” The committee is composed of experts in tax policy, energy and environmental modeling, economics, environmental law, and related areas. The charge is extremely broad, encompassing a detailed federal tax code, a vast energy system, an evolving regulatory environment, and a complex net- work of interactions among different parts of the economy. In focusing its work, the committee decided to concentrate its efforts on four groups of tax code pro- visions and closely related spending and regulatory policies that have significant effects on the emissions of CO2 and other greenhouse gases (GHGs). 1. Energy-related tax expenditures. Because the country’s energy sector, dependent as it is on fossil fuels, is the largest source of greenhouse gas emissions, tax provisions specifically targeted toward energy are a log- ical place to start the assessment. Given time and budget constraints, the committee was unable to consider all of the energy-related provi- sions, but it did examine major ones such as the renewable electricity production tax credit and allowance for percentage depletion of petro- leum and natural gas. These provisions are discussed in Chapter 3. 2. Energy-related excise taxes. There are a handful of excise taxes that affect GHG emissions, primarily taxes on gasoline and other motor fuels and taxes on air travel. These are discussed in Chapter 4. 135 

136 Effects of U.S. Tax Policy on Greenhouse Gas Emissions 3. Biofuels provisions. Tax subsidies for biofuels, particularly ethanol, in- teract with a complex set of mandates on renewable fuels, and the com- bined effects of these on energy prices and consumption depend on market interactions that include spillovers to global agriculture markets. These provisions are discussed in Chapter 5. 4. Broad-based tax expenditures. By far the largest tax expenditures are broad-based tax expenditures, including the exclusion of employer- provided health insurance benefits from taxable income, subsidies to owner-occupied housing, and accelerated depreciation of plant and equipment purchased by businesses. Although not targeted to the ener- gy sector, they can affect both overall economic growth and the com- position of output and thereby could potentially have a significant im- pact on U.S. greenhouse gas emissions. These are considered in Chap- ter 6. The committee carried out its work in three stages. We first decided which provisions of the tax code to examine. Next, we reviewed existing research on the selected provisions to gain insights on the channels through which they could affect climate change and determine if prior estimates on impacts exist. Finally, we commissioned new analyses using several existing energy-economic models to estimate the impact of major provisions. This chapter summarizes our principal findings with regard to the green- house gas impacts of the four sets of tax provisions. It then sets out the commit- tee’s recommendations, including both recommendations for improvements in the methods used in our analysis and our guidance for thinking about environ- mental tax policy. CHOICE OF TAX PROVISIONS As described in Chapter 1, we chose for close examination provisions of the tax code that were closely related to energy-intensive activities because the energy sector is the largest source of domestic GHG emissions. The tax code provisions most clearly relevant to the committee’s charge are excise taxes on energy consumption and energy-intensive activities, includ- ing taxes on motor fuels and taxes on air travel. A much larger set of provisions consists of those known as tax expenditures. Tax expenditures are defined by Congress as revenue losses attributable to provisions of the federal tax laws that allow a special exclusion, exemption, or deduction from gross income or that provide a special credit, a preferential rate of tax, or a deferral of tax liability. The U.S. Department of the Treasury (Treasury) and the congressional Joint Committee on Taxation (JCT) each compiles a list of provisions each year. The Treasury list for fiscal year 2014 provides estimates of the revenue impacts for 173 provisions, and the JCT also provides an independent list and set of revenue estimates. 

Summary of Findings and Recommendations 137 Since it was not possible to analyze each of the provisions, the commit- tee’s first task was to select the most important ones in the context of the present study. For this purpose, we selected the provisions with the largest dollar bene- fits to the energy-producing sector. Additionally, we selected some of the large broad-based tax expenditures not targeted at the energy sector that might have substantial impacts on GHG emissions. In the end, the committee analyzed excise provisions that account for 46 percent of all energy-related excise tax revenues as well as those that account for 71 percent of the calculated revenue loss from the 10 largest energy-related tax expenditures in 2011. The broad-based tax expenditures selected by the commit- tee for analysis account for about one-third of the cost of all tax expenditures that year. REVIEW OF EXISTING RESEARCH The next step was to review existing research on the impact of the tax code on greenhouse gas emissions. This was undertaken by the committee, the staff, and a team of consultants hired specifically for this study. One area with a substantial body of research pertains to the impact of gas- oline taxes on fuel consumption and CO2 emissions. Most of these studies are incomplete, however, because they do not incorporate important features of the U.S. tax and regulatory systems, such as biofuels taxes and subsidies, CAFE (Corporate Average Fuel Economy) standards, and regulatory mandates for eth- anol and other biofuels. Moreover, few studies consider the impact of changing gasoline taxes on sectors outside the energy sector (the general-equilibrium ef- fects) or the impacts of the changing revenue streams. Although there is a large literature on tax expenditures, there is virtually no empirical research on the impacts of these provisions on GHG emissions. An exception is a recent study by the Center for Business and Economic Research (CBER) at the University of Nevada, Las Vegas, reviewed in Chapter 2. While the CBER study uses a simplified economic model, its results are useful as a benchmark to compare with the results from more detailed modeling studies. Finally, we observe that there is a substantial recent literature that investi- gates approaches to reducing GHG emissions and achieving national and global climate-change objectives (such as limiting GHG concentrations or temperature increases) in the most efficient way. Both analytical and empirical studies of this type have concluded that the most efficient approach to emissions reductions is through uniform economy-wide taxes or regulations directly targeted on GHG emissions. A uniform carbon price creates incentives for consumers, producers, and innovators to adjust their activities so as to reduce emissions and encourage development of low-emissions technologies. A short discussion of this literature was provided in Chapter 2. 

138 Effects of U.S. Tax Policy on Greenhouse Gas Emissions COMMISSIONED MODELING STUDIES Committees of the National Research Council are generally asked to re- view and synthesize the existing scientific literature in a specified area. Howev- er, because of the limited existing research in the area of the committee’s focus and the need to use a unified set of baseline assumptions to compare the effects of different tax provisions, the committee concluded that it would be necessary to commission new studies. The committee determined that the most useful approach to analyzing tax provisions is the use of computable energy-economy models. It is virtually im- possible to calculate the impacts of existing tax provisions on GHG emissions without computational simulations. Examining the effects of a particular tax provision over time entails projecting its impact on investment and consumption decisions, the impact of those decisions on energy production and consumption, and the resulting effects on GHG emissions. Analysis of each of these compo- nents requires using a formal model to simulate economic, revenue, and GHG outcomes under different policies. Limited time and resources as well as best practice pointed to the use of existing integrated economic models. Moreover, models developed over many years represent best judgments of modeling teams about the historical paths, current state, and likely future paths of important variables as well as their re- sponse to external shocks or changes such as tax policies. There is simply no substitute for careful modeling for analyzing the problems the committee was asked to address. The committee applied the following criteria in choosing among alterna- tive models to be used for this report. First, the structure of the models needed to be sufficiently flexible so that the pertinent tax provisions could be introduced. Second, the models chosen should be widely used in other studies and subjected to peer review, and it will be helpful if they are familiar to researchers and poli- cy analysts. Third, the analysis should be reproducible and the data on which it relies should be available to other investigators. Fourth, the committee attempted when possible to obtain results from mul- tiple models to provide a robustness check on the results. Fifth, none of the models should be under the primary operation or control of members of the committee. Sixth, the modelers were commissioned to perform the analyses only if they could run the models and deliver results to meet time and budget con- straints. As a final precaution, in advance of making firm contractual commitments with the modelers, we submitted the proposed protocol for the analysis along with the specific models and modeling assumptions to outside experts not affili- ated with the committee. The experts were asked to analyze the protocol. They provided written comments, and the committee had a meeting to discuss their comments with the experts, and then responded to their critiques in writing. 

Summary of Findings and Recommendations 139 Using these criteria, the committee selected four different modeling teams to undertake its analyses. Chapter 2 and Appendix A describe the models and the rationale for their selection. Like all economic models, the selected models have strengths and limitations. Two of the models employed (National Energy Modeling System for the National Academy of Sciences [NEMS-NAS] and Food and Agricultural Policy Research Institute at the University of Missouri [FAPRI-MU]) were detailed analyses of the energy and related sectors using a partial equilibrium (PE) framework. PE models describe the focal sectors in considerable detail, but do not fully capture its interactions with all other parts of the economy, a special concern because the energy sector is so large and plays such a key role in so much economic activity. A third model (the Intertemporal General Equilibrium Model [IGEM]) en- compassed the entire U.S. economy using a general equilibrium (GE) frame- work. GE modeling in principle solves some of the limitations of PE models, but it operates at such a high level of aggregation that it cannot capture many of the important features of the energy sector. IGEM in particular does not have highly resolved detail on choices among methods of electrical generation, nor does it capture the intricate regulations of the biofuels sector. The fourth model we em- ployed (CBER) was a simplified PE energy supply and demand analysis that was specifically designed to estimate the GHG impacts of tax expenditures, but did so with a highly stylized modeling structure. An important reservation about most analysis of the impact of tax provi- sions is that it pertains only to U.S. GHG emissions. Of the models used for this report, only the FAPRI-MU model estimated the global impacts of U.S. tax pol- icy as they related to international agriculture, crop, and biofuel markets. Cli- mate change is a global phenomenon that depends upon global emissions. While this shortcoming is primarily important for tradable goods like grains and oil, the committee notes this shortcoming with respect to the current modeling struc- tures. A second important concern is that current empirical models cannot relia- bly capture the impact of the tax code on innovation and technological change, or what is sometimes called induced innovation. A change in prices and output of a specific technology, such as solar photovoltaic or offshore wind, will gener- ally lead to increased research and development (R&D), commercialization, and experience or “learning by doing.” These will in turn lower the costs of produc- tion, the prices of products that use that technology, and spur further production. These considerations suggest that a subsidy may lead to increased R&D and learning by doing, and thereby promote specific technologies. Two of the mod- els employed (IGEM and NEMS-NAS) included induced innovation in their structure. There is little consensus, however, on the mechanisms behind induced innovation or the magnitude of the response in the energy sector. For this rea- son, the committee did not attempt a separate analysis of this issue. For each of the models, the committee specified a set of baseline assump- tions on gross domestic product (GDP) growth, oil prices, the regulatory envi- ronment, as well as the tax system. The assumptions about U.S. GDP growth 

140 Effects of U.S. Tax Policy on Greenhouse Gas Emissions and path of world oil prices are derived from the Energy Information Admin- istration’s 2011 Energy Outlook (AEO11), which is widely used in the modeling community. The tax code and regulatory environment of 2011 chosen by the committee as the basis of its analysis were also included in the Energy Infor- mation Administration’s 2011 baseline. In addition to the baseline, some of the provisions were analyzed using alternative assumptions for GDP growth and oil prices to provide sensitivity analyses. These alternatives were higher macroeco- nomic growth, low natural gas prices, and high petroleum prices. In analyzing the tax provisions, the committee established a common set of tax and regulatory benchmarks for the analysis. The time frame, set to allow for the effect of investment decisions on GHG emissions, covered in most in- stances 2010-2035. We assumed a stable tax code in which the 2011 tax code provisions remain in place indefinitely. Note, therefore, that tax provisions that expired at the end of 2011 or 2012 are assumed to be extended indefinitely in our base case. This implies that the fiscal changes included in the American Taxpayer Relief Act of 2012 (enacted in January 2013) are not reflected in our analysis. Similarly, the regulatory framework was assumed to take the regulations in place in 2011, and no regulatory measures beyond those adopted as of 2011 are imposed. The Clean Air Interstate Rule, for example, is in force for the pur- pose of our analysis, but other proposed regulations affecting power plants are not implemented. Our baseline assumption excluded important pending or re- cently finalized regulations such as the Cross-State Air Pollution Rule, the Mer- cury and Air Toxics Standards, and the proposed New Source Performance Standard for CO2 on new power plants. Environmental Protection Agency vehi- cle emission standards in place as of 2011 are assumed to continue, meaning that the analysis does not include the impact of the vehicle standards adopted in Au- gust 2012. The Renewable Fuel Standards under the Energy Independence and Security Act are included, although these are modified to allow for a plausible potential waiver scenario. To estimate the impacts of each of the specific tax provisions of interest, the selected models were run with a baseline assumption that all of the 2011 tax code provisions remained in place indefinitely, and then the models were rerun by removing each tax provision one at a time. RESULTS OF THE COMMITTEE’S MODELING STUDIES At the outset, it is useful to put the results described in this section in the context of projected GHG emissions under business-as-usual conditions and the magnitude of GHG reductions in certain proposed targets. For this purpose, we consider the difference between the emissions trajectory in our baseline and an “emissions reduction target” necessary to meet the climate-change targets pro- posed by the Obama administration in 2009 or analyzed in a recent comprehen- sive report by the National Academies in 2010. For the baseline emissions, we 

Summary of Findings and Recommendations 141 rely on the AEO 2011 estimates. The AEO 2011 baseline used for this report projects that CO2 emissions will grow by 8 percent from 2010 to 6,105 MMT in 2035. This baseline reflects trends that are projected to occur without specific carbon-pricing policies such as cap-and-trade restrictions or taxes on GHG emissions. (We note as well that the AEO 2011 baseline has a lower emissions trajectory than many other integrated assessment models.) To meet the emis- sions reduction target would require that emissions be reduced 42 percent below 2005 levels by 2030. This is estimated to be a 60 percent reduction below the AEO11 baseline. Energy-related Tax Expenditures The committee identified 10 major energy tax expenditure provisions as candidates for analysis. Of these, the committee evaluated 5, which account for 71 percent of the revenue loss from the 10 provisions. This section provides a summary of four of the provisions, while the fifth (biofuels subsidies) is covered in subsequent sections. Production Tax Credits for Renewable Electricity The production and investment tax credits for renewable electricity pro- vide a tax credit of 2.3 cents per kWh of power for the first 10 years of electrici- ty production generated from qualifying renewable sources (primarily solar, wind, and biomass) or a credit equal to 30 percent of investment in qualifying equipment. These credits lower the cost of electricity generated from renewable resources, encouraging their substitution for fossil fuels and thereby tend to re- duce GHG emissions. The committee’s analysis using the NEMS-NAS model indicates that the- se provisions lower CO2 emissions. This finding holds for both the Reference and High-Macroeconomic-Growth scenarios. This finding confirms that de- creasing the costs of low-carbon renewable fuels will lead to substitution away from high-carbon fossil fuels in the electricity sector. However, the impact is small, amounting to a reduction of 0.3 percent of U.S. CO2 emissions compared to the reference scenario. Moreover, these tax expenditures are among the most costly that the committee examined in terms of revenue forgone per ton of CO2 reduced. Oil and Gas Depletion Allowances The depletion allowance permits owners of oil and gas wells to deduct a value equal to the decline in the value of their reserves as oil or gas is extracted and sold—a method known as cost depletion. Under current law, some taxpay- ers may use percentage depletion as an alternative to cost depletion. Under per- 

142 Effects of U.S. Tax Policy on Greenhouse Gas Emissions centage depletion, taxpayers deduct a percentage of gross income associated with the sale of the produced commodity. This deduction can exceed the cost of the original investment over the life of the property. Percentage depletion for oil and gas is currently available only for domestic production by independent (i.e., nonintegrated) companies up to a maximum of 1,000 barrels of oil per day (6 MMcf per day of natural gas) and cannot exceed half the net income from the property. The depletion rate is set at 15 percent of gross revenues associated with production. In our modeling, removing the percentage depletion allowances (and sub- stituting the lower valued with cost depletion) has virtually no effect on GHG emissions. The primary impact of percentage depletion on emissions comes from an increase in the production of natural gas and the associated ripple ef- fects of higher natural gas production in other markets. Surprisingly, even though the depletion allowance is generally associated with oil, the NEMS-NAS model projects that it has virtually no impact on oil production. Although natural gas production goes down as the tax preference is re- moved, the complex substitution patterns lead to largely offsetting forces and to a minimal impact on overall emissions. The average effect on GHG emissions over the time horizon of the models is too small to accurately estimate, or even determine, if the sign of the change is positive or negative. Home Energy-efficient Improvement Credits The committee examined qualitatively two other provisions. The first is Credits for Energy Efficiency Improvements to Existing Homes. Analysis of this provision proved difficult. The committee did not find, and was unable within its time and budget constraints to produce, detailed and reliable estimates of its impacts on GHG emissions. We expect that as the provision is currently struc- tured, it is unlikely to produce major savings in GHG emissions. However, giv- en the magnitude of the tax expenditure and the evidence of unexploited savings in this sector, we believe that understanding the impacts of tax incentives on household energy consumption should be a high priority for future research. Nuclear Decommissioning Tax Preference Another provision that was analyzed qualitatively was the special tax rate on reserves set up to decommission nuclear power plants at the end of their life- time. The committee could find no detailed published studies of the impact of this provision on GHG emissions. Based on the available evidence, including the projections of nuclear power under different scenarios using the NEMS-NAS model, we find that the decommissioning provision is likely to have little impact on greenhouse gas emissions. The underlying reasoning is straightforward. The capital costs for a new power reactor are high, about $7 billion in the most recent estimates (AEO 

Summary of Findings and Recommendations 143 2012). Moreover, there have been only two reactors commissioned since the Three Mile Island accident in 1979. Projections used for this study found that there would be few or no new nuclear power plants built in the study period even with this provision in place. If the provision were removed, lowering the profitability of new nuclear plants, there would still be few or no plants built. Other factors might enter to change this conclusion, such as operating costs, lifetimes, and the expected price of natural gas for competing base-load power. But on the whole, the nuclear decommissioning tax preference provision is not considered significant enough to influence the decision by a utility company to make such a large capital expenditure. Excise Taxes In revenue terms, federal excise taxes are small compared to total tax ex- penditures. For fiscal year 2011, for example, total federal excise tax collections were $72 billion, while total tax expenditures amounted to $1,226 billion. How- ever, most of the federal excise taxes are energy related, while only a small frac- tion of tax expenditures are energy related. Chapter 4 considers two such excise taxes—highway motor fuel taxes and taxes on air travel. Highway Motor Fuel Taxes The Internal Revenue Code levies highway fuels taxes of $0.184 per gal- lon of gasoline or alcohol fuel for on-road use, whether pure or blended. The Code also levies on a tax on diesel fuels at $0.244 per gallon for diesel and kero- sene and $0.197 per gallon for diesel-water fuel emulsion. This chapter reviewed four commissioned studies of the effect of remov- ing the excise taxes on highway fuels. All four models find that removing the excise taxes on highway fuels would result in increasing greenhouse gas emis- sions. But the magnitude of the estimated effects varies dramatically for the dif- ferent models. 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 pro- duction and consumption to meet the RFS mandates, from the volumetric bias of highway fuels taxes, and from application of the tax within each model’s struc- ture. A close examination of the results leads the committee to conclude that the NEMS-NAS and the FAPRI models 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 remov- ing highway fuels taxes on GHG emissions is estimated to be very small be- cause of special features of the taxes and the market. The volumetric bias of the taxes means that removing them favors ethanol, which will reduce the GHG 

144 Effects of U.S. Tax Policy on Greenhouse Gas Emissions 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 percent and 0.17 percent of annual U.S. CO2 emissions, respectively. 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). If the RFS were to disappear tomorrow, or if the regulations on E85 were to change drastically, or the volumetric bias of highway taxes were to be removed, the projected impacts of removing the gaso- line tax might be substantially different and would probably be significantly larger. The magnitude of the differences across models leads the committee to caution against using the precise numerical results from a single model and rec- ommends drawing only broad conclusions about the nature and direction of im- pacts. Policy makers and analysts should rely on multiple models, methodolo- gies, and estimates in calculating impact of the tax code and other policies on greenhouse-gas emissions and climate change. Aviation Fuel Taxes The jet fuel excise tax is $0.043 per gallon for commercial aviation and $0.193 per gallon for noncommercial aviation. Even though airline excise tax rates have been raised in recent years, little research has been undertaken into their impacts, and particularly their impacts on GHG emissions. Only one of the modelers we commissioned, CBER, estimated the impact of removing the tax on jet fuel. We believe that more work is needed in analyzing the economic impact and structure of aviation taxation. This is a sector producing rapidly growing GHG emissions, and as yet there are no ready substitutes for fossil-fuel-based jet fuel. Although the GHG emissions from aviation fuels are smaller than for mo- tor fuels today, our results suggest that the GHG reductions per dollar of tax revenue foregone might be higher in this sector than for motor fuels. Biofuels Provisions One particularly important set of tax provisions involves the use of ethanol and other biofuels, particularly as substitutes for petroleum products. These pro- visions involve a complex combination of taxes, tax expenditures, import tariffs, and regulatory mandates that interact to change the composition of fuels. Prior to 2012, the Internal Revenue Code provided an array of tax credits for biofuels. The most important was that alcohol fuels blended with gasoline or used in pure form as a fuel both qualified for a $0.45 per gallon credit under the Volumetric Ethanol Excise Tax Credit (VEETC). There were also tax credits for 

Summary of Findings and Recommendations 145 ethanol from small producers, for producers of cellulosic biofuels, for biodiesel, and for small agri-biodiesel producers. In addition, there was a $0.54 per gallon tariff on imported ethanol. These subsidies were the largest among all the ener- gy-related ones. The VEETC and biodiesel provisions expired at the end of 2011, but under the committee’s methodology, each of these provisions is in- cluded in our base case. The excise tax exemption and credits lowered the cost of biofuels and therefore should have encouraged their substitution for petroleum motor fuels, reducing GHG emissions. Because biofuels are almost always sold as a blend with petroleum fuels, however, the subsidies also lowered the delivered price of the petroleum-biofuel blend, thereby encouraging additional consumption of motor fuels. Given the two factors operating in opposite directions and the fact that ethanol production has positive GHG emissions, the overall impact of the subsidy on GHG emissions is ambiguous. The committee analyzed the biofuels provisions with two different mod- els, although it concentrated its analysis on the FAPRI-MU model, which had the most detailed treatment of the biofuels sector. The findings indicate that re- moving all tax code provisions and the import tariff would result in a decrease of emissions of 5 MMT per year of CO2 equivalent globally. This is less than 0.02 percent of global emissions. On a global basis, removal of the provisions results in a decrease in global ethanol use and an increase in global gasoline use. The results are complicated by the mandates for renewable fuels. If the mandates are removed along with the subsidies, the estimated emissions are smaller than the estimates with the mandates. Therefore, as is intuitive, the mandates reduce (in absolute value) the size of the impact of the subsidy on GHGs. The results of the other modeling studies are consistent with the central FAPRI estimates. These results show the often-counterintuitive nature of the effects of tax subsidies. Although it might seem obvious that subsidizing biofuels should re- duce CO2 emissions because they rely on renewable resources rather than fossil fuels, many studies we reviewed found the opposite. As structured, the biofuels tax credits encouraged the consumption of motor fuels overall because they lower prices, and this effect appears to offset any reduction in the GHG intensity of motor fuels when switching from gasoline to biofuels. (The GHG intensity is calculated as the ratio of the emissions of GHG per unit output.) Broad-based Tax Expenditures The committee examined a limited set of broad-based tax expenditures in addition to energy-sector provisions. The broad-based provisions are important primarily because of their impact on the size and composition of the overall economy. First, they may shift output from industries with lower GHG intensi- ties to industries with higher GHG intensities or vice versa. Second, they may affect the rate of growth of national output and therefore change emissions simp- ly because the economy is larger or smaller. 

146 Effects of U.S. Tax Policy on Greenhouse Gas Emissions Although there are many large tax expenditures, here as elsewhere the committee had to narrow the scope of its inquiry. After examining the largest tax expenditures (see Table 1-3), the committee decided to examine three broad types of provisions: (1) tax incentives that affect investment generally, (2) tax incentives for owner-occupied housing, and (3) tax incentives for provisions of health insurance. These revenue losses from these provisions totaled $370 bil- lion in fiscal year 2012, or about one-third of all tax expenditures. We undertook our analysis using IGEM, which is a full-employment, multisector, general-equilibrium model. The large revenue increases entailed in removing these provisions were offset or recycled in two different ways—by lump-sum transfers and by changes in individual and corporate tax rates. Because the model has large incentive effects of changes in the tax structure, IGEM sug- gested that some of the largest estimated impacts of changes in the tax provisions on GHG emissions come through changes in overall economic activity. Accelerated Depreciation Accelerated depreciation is one of the largest business tax expenditures in the federal income tax code. This set of provisions allows businesses to write off the value of their capital assets at a rate that is faster than the estimated econom- ic depreciation. This provision is the only one of the broad-based provisions for which the committee has confidence that the model is able to capture the princi- pal channels through which GHG emissions are affected. The model runs show that eliminating accelerated depreciation would reduce the GHG intensity of national output by shifting production away from GHG-intensive activities such as coal mining and electric power generation to low-GHG activities such as communications. However, the net effect depends upon how the resulting revenues are re- cycled. Eliminating accelerated depreciation would reduce the capital stock and national output. The effect on the capital stock would be partially offset if labor supply increases. Furthermore, if the additional revenue were used to finance cuts in individual and corporate marginal tax rates, this might eliminate any net impact on GHGs. If, on the other hand, the higher revenues are returned to tax- payers in lump-sum rebates, the analyses show that overall emissions are about 2 percent lower, reflecting the combination of lower emissions per dollar of na- tional output and lower national output. Owner-occupied Housing Provisions The federal income tax provides significant tax incentives for investments in owner-occupied housing. These include deductibility of mortgage interest and property taxes and exclusion from taxation of the first $250,000 ($500,000 for couples) of capital gains on home sales. Defenders of these preferences often justify them as encouraging more people to own homes, although some econom- 

Summary of Findings and Recommendations 147 ic studies indicate that the major effect is on home size rather than housing ten- ure choice. These are large tax preferences, totaling $147 billion in 2011, ac- cording to the Treasury. According to the estimates prepared for the committee, eliminating the tax subsidies for owner-occupied housing and using the revenue to lower marginal tax rates would improve the efficiency of allocation of the capital stock and in- crease national output. GHG emissions would increase at about the same rate as GDP increases. These results are consistent with earlier analyses. However, the simulation does not fully capture the effects of the housing subsidies on emissions intensity. In particular, it does not capture the effects on housing size and in turn household consumption of electricity, natural gas, and home heating oil. It also is unable to capture potential effects of the housing tax preferences on the spatial distribution of housing and patterns of automobile use and gasoline consumption. Even the relationship between housing subsidies and changes in the composition of output are somewhat obscure in the IGEM model. We therefore find the analysis of the housing provisions inconclusive and be- lieve that the effects on GHG emissions of eliminating or reducing the housing tax preferences are a high-priority topic for further research. Employer-provided Health Care Provisions The largest single tax expenditure in the Internal Revenue Code is the ex- clusion of employer-provided health insurance (EPHI) benefits from taxable income of employees. The Treasury Department estimates that exclusion of EPHI benefits will reduce income tax receipts by $180 billion in 2012. The only model considered by the committee that had the potential to es- timate the impact of health care provisions was IGEM. However, this model does not capture the pathways by which the exclusion of EPHI from the tax on employee compensation might affect demand for health care services. There- fore, simulations for the committee using IGEM treated the removal of the health tax preference as equivalent to imposing a new excise tax on personal and business services (of which health is a sector). The model found that eliminating the EPHI exemption would raise national output if offset by cuts in marginal tax rates, but would reduce national output slightly if revenues are rebated in a lump-sum fashion. The model’s findings of the effect of changes in the composition of output on GHG intensity were anomalous. We expected that eliminating health care subsidies would raise GHG emissions per unit of output because the health care sector is less GHG intensive than the rest of the economy. The IGEM model results show the opposite effect, however, with a small decrease in GHG intensi- ty. Our inability to understand the structural features of the model that explain these results lead us to conclude that the impact of the health provisions on GHG emissions remains an open question and an important subject for future research. 

148 Effects of U.S. Tax Policy on Greenhouse Gas Emissions Further Observations on the Broad-based Tax Expenditures The committee’s major finding is that the broad-based provisions influ- ence GHG emissions primarily through their effects on overall national output. In most cases, the change in GHG emissions was close to or equal to the change in national output induced by removing the tax provision (see particularly Table 6-2). This is a plausible conclusion following from the relatively modest effects of all the tax provisions on the ratio of emissions to national output. If tax law changes do not affect the emissions-output ratio markedly, then their effect on output will be the principal driver of their impact on emissions. A second finding is that the way revenues generated by eliminating tax preferences are recycled significantly affects output and emissions. Recycling new revenues through reductions in individual and company tax rates is likely to raise national output and therefore will also increase GHG emissions. Estimates from the IGEM model indicate that removing the broad-based tax expenditures might increase national output and related GHG emissions in the order of 1 to 2 percent of baseline emissions. In contrast, lump-sum tax rebates do not increase incentives to save, invest, work more, or use resources more efficiently and will have little effect on national output. This result also emphasizes the importance of examining the sources of the revenue when analyzing the impact of different tax expenditures and subsidies. A third finding is that the broad-based provisions generally have little ef- fect on emissions intensities, ranging from -0.3 percent to +0.2 percent over the period 2010-2035. Thus, whatever the uncertainty of our results, the one com- mon finding is that the impacts of these provisions on emissions are not large. Finally, we need to reiterate our reservations about the simulations of the broad-based tax expenditures. The results are highly sensitive to assumptions about how tax revenues from eliminating the provisions are returned to the economy. Moreover, we were able to use only a single model to calculate the impacts of most of these provisions, and that model is not transparent about some of the pathways through which tax provisions affect GHG emissions changes. Additionally, in the case of accelerated depreciation, where we were able to compare IGEM’s results with other modeling results, they were diver- gent. We conclude that changes in broad-based tax provisions are likely to have a small impact on overall GHG emissions, mainly by changing national output. Even the largest impact on emissions, however, is tiny relative to the projected growth in GHG emissions over the period 2010–2035. In its baseline simulation, the IGEM model projects a 43 percent growth in emissions over that period. The largest change in emissions from any of the broad-based provisions was in the order of 2 percent. Therefore, the tax provision with the largest emissions impact changes the growth of emissions over the period by only a tiny fraction. 

Summary of Findings and Recommendations 149 Comparison with CBER Modeling Results We compared the results of our detailed modeling studies with those of a comprehensive study of energy tax expenditures by the CBER modeling group at the University of Nevada, Las Vegas. Their initial report was released at about the time our study was launched. The committee reviewed their work and commissioned additional simulations using their model. As we observed in Chapter 2, the advantage of the CBER approach is its comprehensive nature. It is also a transparent modeling structure, in which the assumptions are laid out clearly in a way that allows users to understand key driving forces. It has the disadvantages of any partial equilibrium model; in ad- dition, it lacks many of the important regulatory and fine-grained structures of the more detailed energy-sector models. The study also includes some non-tax direct expenditures, such as spending on low-cost residential weatherization. It studies the energy tax expenditures in effect for the period 2005–2009, which is different from the other models. On the other hand, the average annual cost of those expenditures was $19 billion, approximately the same as those in our study. The committee used the CBER model to obtain an order-of-magnitude es- timate of the impact of all energy-related subsidies. Under the methods and as- sumptions of that study, if all tax subsidies would have been removed, then net CO2 emissions would have decreased by 30 MMT per year over the 2005-2009 period. This total represented about ½ percent of total U.S. CO2 emissions over this period. The CBER study found that several of the tax policies reduced CO2 emis- sions, but that direct expenditures reduced emissions more than tax subsidies. Consistent with our modeling results, CBER found the biofuels credits increased CO2 emissions. The committee concludes that the basic finding of the CBER study that removing all energy-sector subsidies would increase GHG emissions is plausi- ble, but it is not robust and is subject to a large margin of error. The CBER re- sults are consistent with the basic findings of the detailed modeling studies we conducted—that the overall effect of current energy tax subsidies is close to zero. We summarize the results of the commissioned economic modeling simu- lations in Table 7-1. SUMMARY OF FINDINGS The committee’s report on the impacts of the U.S. tax code on greenhouse gas emissions has relied on analyses that ranged from simplified modeling ap- proaches through detailed energy-sector modeling to general equilibrium econ- omy-wide modeling. The estimates have major uncertainties for reasons that we have described at various points in this report. Nevertheless, despite these uncer- tainties, we draw the following general conclusions. 

150 TABLE 7-1 Summary of Modeling Results Model Parameters NEMS CBER FAPRI IGEM Baseline Parameters AEO 20111 AEO 2012 2 AEO 2012 AEO 2011 for GDP and and Assumptions emissions baseline Other variables as specified in Chapter 6 Gases Included in Model Only CO2 Only CO2 CO2, CH4, N2O, LUC/ILUC CO2, CH4, N2O, HGWPs Policies Modeled and Indicative Results Model Policy NEMS CBER FAPRI IGEM Tax Credits for Production of Decrease Decrease Not Modeled Not Modeled Renewable Electricity Excess of Percentage Over Increase Increase Not Modeled Not Modeled Cost Depletion Credits for Energy Efficiency Not Modeled Decrease Not Modeled Not Modeled Improvements to Existing Homes Special Tax Rate on Nuclear Not Modeled Decrease Not Modeled Not Modeled Decommissioning Reserve Funds 1 AEO 2011 Assumptions: GDP 2.7%; GDP components of final demand – Consumption 2.4%; Investment 4.6%; Government 0.7%; Ex- ports 6.3%; Imports 4.6%. 2 AEO 2012 Assumptions: GDP 2.6%; GDP components of final demand – Consumption 2.3%; Investment 4.2%; Government 0.4%; Ex- ports 6.0%; Imports 4.1%. 

Highway Motor Fuels Decrease Decrease Decrease Decrease Excise Tax Volumetric Ethanol Very small impact Increase Increase Not Modeled Excise Tax Credit Ethanol-specific Tariff Very small impact Not Modeled Increase Not Modeled Biodiesel Excise Tax Credit Very small impact Decrease Increase Not Modeled Tax Incentives for Home Not Modeled Not Modeled Not Modeled Uncertain Ownership Tax Incentives for Health Not Modeled Not Modeled Not Modeled Uncertain Insurance and Health Care Accelerated Depreciation Decrease Not Modeled Not Modeled Increase 151 

152 Effects of U.S. Tax Policy on Greenhouse Gas Emissions First, the combined effect of current energy-sector tax expenditures on GHG emissions is very small and could be negative or positive. The most comprehensive study available suggests that their combined impact is less than 1 percent of total U.S. emissions. If we consider the estimates of the effects of the provisions we analyzed using more robust models, they are in the same range. We cannot say with confidence whether the overall effect of energy-sector tax expenditures is to reduce or increase GHG emissions. (Chapters 3-5) Second, individual energy-sector tax expenditures in some cases con- tribute to, and in other cases subtract from, U.S. or global GHG emissions. The subsidies on ethanol that expired at the end of 2011 are estimated to in- crease global GHG emissions. By contrast, the production and investment tax credits for renewable electricity appear to reduce U.S. GHG emissions. The de- pletion allowance has virtually zero impact on emissions. (Chapters 3-5) Third, the best existing analytical tools are unable to determine in a reliable fashion the impact of some important subsidies. Important tax ex- penditures that have resisted analysis include ones subsidizing residential energy efficiency. The difficulties in this case involve such factors as the discount rate consumers apply to future fuel savings, the strength of any rebound effect, and the extent to which consumers understand and respond to tax law changes. (Chapters 3-5) Fourth, the revenues foregone by energy-sector tax subsidies are sub- stantial in relation to the effects on GHG emissions. The Treasury estimates that the revenue loss from energy-sector tax expenditures in fiscal years 2011 and 2012 totaled $48 billion. Few of these were enacted to reduce GHG emis- sions. To the extent that they include as a primary objective reducing GHG emissions, however, they are inefficient. Very little if any GHG reductions are achieved at substantial cost with these provisions. (Chapters 1 and 3-5) Fifth, the impacts of the broad-based tax expenditures on GHG emis- sions come primarily through their impact on the level of national output. Broad-based tax expenditures entail roughly 50 times more revenues foregone than the energy-sector subsidies. We investigated a subset of provisions repre- senting about one-third of the revenue losses from tax expenditures—subsidies to equipment investment through accelerated depreciation, to health care, and to owner-occupied housing. Except for accelerated depreciation, we were unable to reach a definite conclusion on whether they increase or decrease GHG emissions per unit of output. Rather, the principal effect of these provisions is on the size of national output and the allocation of resources among sectors. If the revenue impacts of removing broad-based subsidies were offset by reducing distortion- ary taxes, the resulting increase in national output would be accompanied by increased GHG emissions. If the subsidies were replaced with lump-sum tax cuts that do not reduce distortions, there would likely be little effect on national output or emissions. (Chapter 6) Sixth, it is difficult to estimate the impact of the broad-based tax ex- penditures on GHG emissions intensity. The committee examined the existing literature and commissioned modeling studies to estimate the effects of changes 

Summary of Findings and Recommendations 153 in the broad-based provisions on the overall GHG intensity of the economy. The results were not judged to be sufficiently reliable to draw firm conclusions. (Chapter 6) Seventh, the effects of many tax provisions are complicated by their interaction with regulations. Very few tax provisions take place in a regulatory vacuum. Particularly in the energy sector, energy and environmental regulations overlay and interact with tax provisions. Prime examples are the interaction of tax provisions with the CAFE standards for light-duty vehicles, the air pollution standards for the mix of electricity generation, the Renewable Portfolio Stand- ards for electricity generation, and the Renewable Fuel Standards for motor fuels blended from petroleum and ethanol. An important finding of our studies is that regulatory and environmental constraints generally reduce the size of the im- pacts of tax provisions on GHG emissions. (Chapters 3-5) Eighth, energy excise taxes reduce GHG emissions, but the impact is limited because of special features of the tax and because of regulatory con- straints. Analysts have studied the effects of gasoline taxes for many years. The committee’s estimates show unambiguously that highway fuel excise taxes re- duce fuel consumption and GHG emissions. The analysis for this report finds that the current highway fuels taxes have a relatively small impact on GHG emissions because of the volumetric bias of the taxes as well as the constraints imposed by the renewable fuels standards. (Chapter 4) RESEARCH RECOMMENDATIONS After reviewing various approaches, the committee concludes that the only reliable methodology for estimating the impacts of tax policy on GHG emissions is through the use of energy-economic modeling. However, existing models are primarily useful for understanding the economic and energy-sector linkages that produce emissions changes and for suggesting the likely sign of those effects. In very few cases are existing models able to determine with precision the quantita- tive impacts of tax provisions on GHGs. Moreover, there are numerous short- comings to existing models. The following recommendations to the Congress, the modeling community, the research support agencies, as well as the broader community provide guidance on the areas where the committee finds that more attention is needed. The committee recommends continued support of energy-economic modeling to better understand the impacts of taxes and other public policies on greenhouse gas emissions and the broader economy. Particular attention should be given to improving current models in the following ways: First, models need to be made more transparent by clarifying both their assumptions and their structure. Second, models should include measures of economic welfare that can be used to measure the efficiency and distributional impacts of policies. 

154 Effects of U.S. Tax Policy on Greenhouse Gas Emissions Third, there should be more work to integrate partial equilibrium models with general equilibrium models so that the impact of revenue recy- cling and overall economic impacts can be more reliably estimated. At present, the impacts of taxes and tax expenditures on GHG emissions are difficult to measure reliably. This proved to be the case across the entire spectrum of provisions evaluated in this study. The detailed analyses in Chapters 3 through 6 show the difficulties the committee encountered in arriving at relia- ble estimates of GHG impacts, and in some cases the committee was unable to determine whether the sign of the impact was negative or positive. The major way to determine the reliability of large energy-economic models today is through comparing different approaches. This leads to a final research recom- mendation: Fourth, the committee recommends increased attention to studies that compare energy-economic models as a tool for improving understanding of models, narrowing the range of estimates, and improving model reliability. GUIDANCE FOR SCORING GHG EMISSIONS The Congressional Budget Office is required to estimate (“score”) the budget impact of proposed legislation. Some have suggested that there be a par- allel procedure for estimating the effects of legislation on GHG emissions and climate change. The committee finds that attempting to institute scoring is premature. Estimating the impacts of tax provisions on greenhouse gas emis- sions is difficult because the mechanisms are so complex; there are so many interacting forces; the regulatory and market environments change quickly and unpredictably; and regulatory and tax arbitrage across national boundaries tends to reduce or increase the impact measured from a global vantage point. Today’s modeling capabilities are not yet up to this task, particularly on the rapid sched- ule necessary for legislative action. Simply put, we do not believe that GHG accounting of taxes and expenditures is ready for prime time; indeed, it is not even ready for a short cameo spot on “The Accounting Channel” in the wee hours of the night. Because of the difficulties and resources required to provide reliable estimates, the committee discourages requiring the formal scoring of tax proposals for their impacts on GHG emissions. Much further work needs to be done before it can be accomplished routinely and reliably. USE OF TAX POLICY TO ADDRESS CLIMATE CHANGE In addition to estimating the impacts of the tax code on GHG emissions, the committee was asked to examine broader implications of taxes and climate- change policy. The congressional Joint Committee on Taxation, in its report on the legislation authorizing this study, suggested that the National Academies’ report “discuss the importance of controlling carbon dioxide and greenhouse gas 

Summary of Findings and Recommendations 155 emissions as part of a comprehensive national strategy for reducing U.S. contri- butions to global climate change, and evaluate the potential for changes in the Code to reduce carbon dioxide emissions” (Joint Committee on Taxation, 2009). Although the committee does not make any recommendations about specific changes, the analysis undertaken for this report leads to several important in- sights and cautions about tax policy in the context of climate change. First, current tax expenditures and subsidies are a poor tool for re- ducing greenhouse gases and achieving climate-change objectives. The committee has found that several existing provisions have perverse effects, while others yield little reduction in GHG emissions per dollar of revenue loss. This is not surprising, because most of the provisions act indirectly rather than directly on emissions and most provisions were not structured with reduction of GHG emissions as a primary policy objective. The feedback effects within the energy sector (e.g., the fuel substitution effects when tax policy favors one source over others) or the international spillover effects (e.g., shifts in trade flows due to differential tax treatment) can offset or even reverse the expected direct effects of these policies. Such leakages and regulatory and tax arbitrage are common features of indirectly targeted provisions. For example, it may seem obvious that a subsidy for renewable fuels will reduce GHG emissions, but the response of different sectors can undermine the goal, as our analysis of biofuels shows. Thus, if tax expenditures are to be made an effective tool for reducing GHG emissions, much more care will need to be applied to designing the provi- sions to avoid inefficiencies and perverse offsetting effects. (Chapters 2-6) Second, some tax expenditures are more efficient than others. Some tax policies have fewer leakages and offsetting forces than others and therefore may be more effective. In contrast with the ethanol subsidies, where the emis- sions from biofuels more than offset the reduction in petroleum emissions, the renewable production tax credits for electricity appear to have the desired effects of increasing the production of renewables in the mix of energy sources and reducing GHG emissions. At their current scale, however, they achieve small emission reductions and are costly per unit of emissions reduction. (Chapters 3-5) Third, the committee’s reservations about tax expenditures and sub- sidies do not necessarily apply to tax incentives directly targeted on activi- ties such as research and development on technological advances. Govern- mental support for innovation will help the nation and the world transition to a low-carbon energy system. We did not review these expenditures nor ex- plore the effectiveness of the current incentives to promote low-carbon technol- ogies, or the best way to deliver them, but there is a substantial literature justify- ing R&D subsidies on the basis of its positive spillovers and private underinvestment in research in the energy sector. (Chapter 2) Fourth, tax reforms that increase the economic efficiency of our econ- omy may increase GHG emissions, but the increased output is likely much more than sufficient to pay for reducing the higher emissions if efficient climate-change policies are employed to reduce emissions. Our general equi- 

156 Effects of U.S. Tax Policy on Greenhouse Gas Emissions librium modeling suggests that substituting a less distortionary tax regime for current tax expenditures may well raise GHG emissions as the economy grows faster. This is not an argument against removing inefficient tax expenditures, however. Effective tax reform will increase the nation’s productivity and living standards, thereby providing more than sufficient resources to pay for reducing the additional GHG emissions. (Chapter 2, EIA, 2009; National Research Coun- cil, 2010; Clarke et al., 2010) Finally, many studies have found that the most reliable and efficient way to achieve given climate-change objectives is to use direct tax or regula- tory policies that create a market price for CO2 and other greenhouse gas emissions. A central finding of many studies in this area is that the most effi- cient way to reduce GHG emissions is through policies that create a market price for CO2 and other GHGs (National Research Council, 2010; Resources for the Future, 2010). This can be accomplished either by tradable GHG-emissions allowances or by taxes on GHG emissions. The national and global emissions reductions necessary to meet internationally agreed-upon climate objectives are many times larger than those resulting from current tax subsidies. The commit- tee finds that tax policy can make a substantial contribution to meeting the na- tion’s climate-change objectives, but that the current approaches will not ac- complish that. In order to meet ambitious climate-change objectives, a different approach that targets GHG emissions directly through taxes or tradable allow- ances will be both necessary and more efficient. (Chapter 2, EIA, 2009; National Research Council, 2010; Clarke et al., 2010)